SPOP replication tests.

The old version of SPOP with "count" argument used an API call of dict.c
which was actually designed for a different goal, and was not capable of
good distribution. We follow a different three-cases approach optimized
for different ratiion between sets and requested number of elements.

The implementation is simpler and allowed the removal of a large amount
of code.

.gitignore

@@ -4,6 +4,7 @@
 dump.rdb
 redis-benchmark
 redis-check-aof
+redis-check-rdb
 redis-check-dump
 redis-cli
 redis-sentinel

CONTRIBUTING

@@ -20,7 +20,7 @@ each source file that you contribute.
 
 # How to provide a patch for a new feature
 
-1. Drop a message to the Redis Google Group with a proposal of semantics/API.
+1. If it is a major feature or a semantical change, write an RCP (Redis Change Proposal). Check the documentation here: https://github.com/redis/redis-rcp
 
 2. If in step 1 you get an acknowledge from the project leaders, use the
    following procedure to submit a patch:
@@ -31,4 +31,6 @@ each source file that you contribute.
     d. Initiate a pull request on github ( http://help.github.com/send-pull-requests/ )
     e. Done :)
 
+For minor fixes just open a pull request on Github.
+
 Thanks!

README

@@ -26,6 +26,25 @@ After building Redis is a good idea to test it, using:
 
     % make test
 
+Fixing build problems with dependencies or cached build options
+—--------
+Redis has some dependencies which are included into the "deps" directory.
+"make" does not rebuild dependencies automatically, even if something in the
+source code of dependencies is changes.
+
+When you update the source code with `git pull` or when code inside the
+dependencies tree is modified in any other way, make sure to use the following
+command in order to really clean everything and rebuild from scratch:
+
+    make distclean
+
+This will clean: jemalloc, lua, hiredis, linenoise.
+
+Also if you force certain build options like 32bit target, no C compiler
+optimizations (for debugging purposes), and other similar build time options,
+those options are cached indefinitely until you issue a "make distclean"
+command.
+
 Fixing problems building 32 bit binaries
 ---------
 

deps/Makefile

@@ -58,7 +58,7 @@ ifeq ($(uname_S),SunOS)
 	LUA_CFLAGS= -D__C99FEATURES__=1
 endif
 
-LUA_CFLAGS+= -O2 -Wall -DLUA_ANSI $(CFLAGS)
+LUA_CFLAGS+= -O2 -Wall -DLUA_ANSI -DENABLE_CJSON_GLOBAL -DREDIS_STATIC='' $(CFLAGS)
 LUA_LDFLAGS+= $(LDFLAGS)
 # lua's Makefile defines AR="ar rcu", which is unusual, and makes it more
 # challenging to cross-compile lua (and redis).  These defines make it easier

deps/lua/src/Makefile

@@ -25,9 +25,10 @@ PLATS= aix ansi bsd freebsd generic linux macosx mingw posix solaris
 LUA_A=	liblua.a
 CORE_O=	lapi.o lcode.o ldebug.o ldo.o ldump.o lfunc.o lgc.o llex.o lmem.o \
 	lobject.o lopcodes.o lparser.o lstate.o lstring.o ltable.o ltm.o  \
-	lundump.o lvm.o lzio.o strbuf.o
+	lundump.o lvm.o lzio.o strbuf.o fpconv.o
 LIB_O=	lauxlib.o lbaselib.o ldblib.o liolib.o lmathlib.o loslib.o ltablib.o \
-	lstrlib.o loadlib.o linit.o lua_cjson.o lua_struct.o lua_cmsgpack.o
+	lstrlib.o loadlib.o linit.o lua_cjson.o lua_struct.o lua_cmsgpack.o \
+	lua_bit.o
 
 LUA_T=	lua
 LUA_O=	lua.o

deps/lua/src/fpconv.c

@@ -0,0 +1,205 @@
+/* fpconv - Floating point conversion routines
+ *
+ * Copyright (c) 2011-2012  Mark Pulford <mark@kyne.com.au>
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
+ * a copy of this software and associated documentation files (the
+ * "Software"), to deal in the Software without restriction, including
+ * without limitation the rights to use, copy, modify, merge, publish,
+ * distribute, sublicense, and/or sell copies of the Software, and to
+ * permit persons to whom the Software is furnished to do so, subject to
+ * the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be
+ * included in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+ * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+ * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+ * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
+ * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+ * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+ * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/* JSON uses a '.' decimal separator. strtod() / sprintf() under C libraries
+ * with locale support will break when the decimal separator is a comma.
+ *
+ * fpconv_* will around these issues with a translation buffer if required.
+ */
+
+#include <stdio.h>
+#include <stdlib.h>
+#include <assert.h>
+#include <string.h>
+
+#include "fpconv.h"
+
+/* Lua CJSON assumes the locale is the same for all threads within a
+ * process and doesn't change after initialisation.
+ *
+ * This avoids the need for per thread storage or expensive checks
+ * for call. */
+static char locale_decimal_point = '.';
+
+/* In theory multibyte decimal_points are possible, but
+ * Lua CJSON only supports UTF-8 and known locales only have
+ * single byte decimal points ([.,]).
+ *
+ * localconv() may not be thread safe (=>crash), and nl_langinfo() is
+ * not supported on some platforms. Use sprintf() instead - if the
+ * locale does change, at least Lua CJSON won't crash. */
+static void fpconv_update_locale()
+{
+    char buf[8];
+
+    snprintf(buf, sizeof(buf), "%g", 0.5);
+
+    /* Failing this test might imply the platform has a buggy dtoa
+     * implementation or wide characters */
+    if (buf[0] != '0' || buf[2] != '5' || buf[3] != 0) {
+        fprintf(stderr, "Error: wide characters found or printf() bug.");
+        abort();
+    }
+
+    locale_decimal_point = buf[1];
+}
+
+/* Check for a valid number character: [-+0-9a-yA-Y.]
+ * Eg: -0.6e+5, infinity, 0xF0.F0pF0
+ *
+ * Used to find the probable end of a number. It doesn't matter if
+ * invalid characters are counted - strtod() will find the valid
+ * number if it exists.  The risk is that slightly more memory might
+ * be allocated before a parse error occurs. */
+static inline int valid_number_character(char ch)
+{
+    char lower_ch;
+
+    if ('0' <= ch && ch <= '9')
+        return 1;
+    if (ch == '-' || ch == '+' || ch == '.')
+        return 1;
+
+    /* Hex digits, exponent (e), base (p), "infinity",.. */
+    lower_ch = ch | 0x20;
+    if ('a' <= lower_ch && lower_ch <= 'y')
+        return 1;
+
+    return 0;
+}
+
+/* Calculate the size of the buffer required for a strtod locale
+ * conversion. */
+static int strtod_buffer_size(const char *s)
+{
+    const char *p = s;
+
+    while (valid_number_character(*p))
+        p++;
+
+    return p - s;
+}
+
+/* Similar to strtod(), but must be passed the current locale's decimal point
+ * character. Guaranteed to be called at the start of any valid number in a string */
+double fpconv_strtod(const char *nptr, char **endptr)
+{
+    char localbuf[FPCONV_G_FMT_BUFSIZE];
+    char *buf, *endbuf, *dp;
+    int buflen;
+    double value;
+
+    /* System strtod() is fine when decimal point is '.' */
+    if (locale_decimal_point == '.')
+        return strtod(nptr, endptr);
+
+    buflen = strtod_buffer_size(nptr);
+    if (!buflen) {
+        /* No valid characters found, standard strtod() return */
+        *endptr = (char *)nptr;
+        return 0;
+    }
+
+    /* Duplicate number into buffer */
+    if (buflen >= FPCONV_G_FMT_BUFSIZE) {
+        /* Handle unusually large numbers */
+        buf = malloc(buflen + 1);
+        if (!buf) {
+            fprintf(stderr, "Out of memory");
+            abort();
+        }
+    } else {
+        /* This is the common case.. */
+        buf = localbuf;
+    }
+    memcpy(buf, nptr, buflen);
+    buf[buflen] = 0;
+
+    /* Update decimal point character if found */
+    dp = strchr(buf, '.');
+    if (dp)
+        *dp = locale_decimal_point;
+
+    value = strtod(buf, &endbuf);
+    *endptr = (char *)&nptr[endbuf - buf];
+    if (buflen >= FPCONV_G_FMT_BUFSIZE)
+        free(buf);
+
+    return value;
+}
+
+/* "fmt" must point to a buffer of at least 6 characters */
+static void set_number_format(char *fmt, int precision)
+{
+    int d1, d2, i;
+
+    assert(1 <= precision && precision <= 14);
+
+    /* Create printf format (%.14g) from precision */
+    d1 = precision / 10;
+    d2 = precision % 10;
+    fmt[0] = '%';
+    fmt[1] = '.';
+    i = 2;
+    if (d1) {
+        fmt[i++] = '0' + d1;
+    }
+    fmt[i++] = '0' + d2;
+    fmt[i++] = 'g';
+    fmt[i] = 0;
+}
+
+/* Assumes there is always at least 32 characters available in the target buffer */
+int fpconv_g_fmt(char *str, double num, int precision)
+{
+    char buf[FPCONV_G_FMT_BUFSIZE];
+    char fmt[6];
+    int len;
+    char *b;
+
+    set_number_format(fmt, precision);
+
+    /* Pass through when decimal point character is dot. */
+    if (locale_decimal_point == '.')
+        return snprintf(str, FPCONV_G_FMT_BUFSIZE, fmt, num);
+
+    /* snprintf() to a buffer then translate for other decimal point characters */
+    len = snprintf(buf, FPCONV_G_FMT_BUFSIZE, fmt, num);
+
+    /* Copy into target location. Translate decimal point if required */
+    b = buf;
+    do {
+        *str++ = (*b == locale_decimal_point ? '.' : *b);
+    } while(*b++);
+
+    return len;
+}
+
+void fpconv_init()
+{
+    fpconv_update_locale();
+}
+
+/* vi:ai et sw=4 ts=4:
+ */

deps/lua/src/fpconv.h

@@ -0,0 +1,22 @@
+/* Lua CJSON floating point conversion routines */
+
+/* Buffer required to store the largest string representation of a double.
+ *
+ * Longest double printed with %.14g is 21 characters long:
+ * -1.7976931348623e+308 */
+# define FPCONV_G_FMT_BUFSIZE   32
+
+#ifdef USE_INTERNAL_FPCONV
+static inline void fpconv_init()
+{
+    /* Do nothing - not required */
+}
+#else
+extern void fpconv_init();
+#endif
+
+extern int fpconv_g_fmt(char*, double, int);
+extern double fpconv_strtod(const char*, char**);
+
+/* vi:ai et sw=4 ts=4:
+ */

deps/lua/src/lua_bit.c

@@ -0,0 +1,189 @@
+/*
+** Lua BitOp -- a bit operations library for Lua 5.1/5.2.
+** http://bitop.luajit.org/
+**
+** Copyright (C) 2008-2012 Mike Pall. All rights reserved.
+**
+** Permission is hereby granted, free of charge, to any person obtaining
+** a copy of this software and associated documentation files (the
+** "Software"), to deal in the Software without restriction, including
+** without limitation the rights to use, copy, modify, merge, publish,
+** distribute, sublicense, and/or sell copies of the Software, and to
+** permit persons to whom the Software is furnished to do so, subject to
+** the following conditions:
+**
+** The above copyright notice and this permission notice shall be
+** included in all copies or substantial portions of the Software.
+**
+** THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
+** EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
+** MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
+** IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
+** CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
+** TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
+** SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+**
+** [ MIT license: http://www.opensource.org/licenses/mit-license.php ]
+*/
+
+#define LUA_BITOP_VERSION	"1.0.2"
+
+#define LUA_LIB
+#include "lua.h"
+#include "lauxlib.h"
+
+#ifdef _MSC_VER
+/* MSVC is stuck in the last century and doesn't have C99's stdint.h. */
+typedef __int32 int32_t;
+typedef unsigned __int32 uint32_t;
+typedef unsigned __int64 uint64_t;
+#else
+#include <stdint.h>
+#endif
+
+typedef int32_t SBits;
+typedef uint32_t UBits;
+
+typedef union {
+  lua_Number n;
+#ifdef LUA_NUMBER_DOUBLE
+  uint64_t b;
+#else
+  UBits b;
+#endif
+} BitNum;
+
+/* Convert argument to bit type. */
+static UBits barg(lua_State *L, int idx)
+{
+  BitNum bn;
+  UBits b;
+#if LUA_VERSION_NUM < 502
+  bn.n = lua_tonumber(L, idx);
+#else
+  bn.n = luaL_checknumber(L, idx);
+#endif
+#if defined(LUA_NUMBER_DOUBLE)
+  bn.n += 6755399441055744.0;  /* 2^52+2^51 */
+#ifdef SWAPPED_DOUBLE
+  b = (UBits)(bn.b >> 32);
+#else
+  b = (UBits)bn.b;
+#endif
+#elif defined(LUA_NUMBER_INT) || defined(LUA_NUMBER_LONG) || \
+      defined(LUA_NUMBER_LONGLONG) || defined(LUA_NUMBER_LONG_LONG) || \
+      defined(LUA_NUMBER_LLONG)
+  if (sizeof(UBits) == sizeof(lua_Number))
+    b = bn.b;
+  else
+    b = (UBits)(SBits)bn.n;
+#elif defined(LUA_NUMBER_FLOAT)
+#error "A 'float' lua_Number type is incompatible with this library"
+#else
+#error "Unknown number type, check LUA_NUMBER_* in luaconf.h"
+#endif
+#if LUA_VERSION_NUM < 502
+  if (b == 0 && !lua_isnumber(L, idx)) {
+    luaL_typerror(L, idx, "number");
+  }
+#endif
+  return b;
+}
+
+/* Return bit type. */
+#define BRET(b)  lua_pushnumber(L, (lua_Number)(SBits)(b)); return 1;
+
+static int bit_tobit(lua_State *L) { BRET(barg(L, 1)) }
+static int bit_bnot(lua_State *L) { BRET(~barg(L, 1)) }
+
+#define BIT_OP(func, opr) \
+  static int func(lua_State *L) { int i; UBits b = barg(L, 1); \
+    for (i = lua_gettop(L); i > 1; i--) b opr barg(L, i); BRET(b) }
+BIT_OP(bit_band, &=)
+BIT_OP(bit_bor, |=)
+BIT_OP(bit_bxor, ^=)
+
+#define bshl(b, n)  (b << n)
+#define bshr(b, n)  (b >> n)
+#define bsar(b, n)  ((SBits)b >> n)
+#define brol(b, n)  ((b << n) | (b >> (32-n)))
+#define bror(b, n)  ((b << (32-n)) | (b >> n))
+#define BIT_SH(func, fn) \
+  static int func(lua_State *L) { \
+    UBits b = barg(L, 1); UBits n = barg(L, 2) & 31; BRET(fn(b, n)) }
+BIT_SH(bit_lshift, bshl)
+BIT_SH(bit_rshift, bshr)
+BIT_SH(bit_arshift, bsar)
+BIT_SH(bit_rol, brol)
+BIT_SH(bit_ror, bror)
+
+static int bit_bswap(lua_State *L)
+{
+  UBits b = barg(L, 1);
+  b = (b >> 24) | ((b >> 8) & 0xff00) | ((b & 0xff00) << 8) | (b << 24);
+  BRET(b)
+}
+
+static int bit_tohex(lua_State *L)
+{
+  UBits b = barg(L, 1);
+  SBits n = lua_isnone(L, 2) ? 8 : (SBits)barg(L, 2);
+  const char *hexdigits = "0123456789abcdef";
+  char buf[8];
+  int i;
+  if (n < 0) { n = -n; hexdigits = "0123456789ABCDEF"; }
+  if (n > 8) n = 8;
+  for (i = (int)n; --i >= 0; ) { buf[i] = hexdigits[b & 15]; b >>= 4; }
+  lua_pushlstring(L, buf, (size_t)n);
+  return 1;
+}
+
+static const struct luaL_Reg bit_funcs[] = {
+  { "tobit",	bit_tobit },
+  { "bnot",	bit_bnot },
+  { "band",	bit_band },
+  { "bor",	bit_bor },
+  { "bxor",	bit_bxor },
+  { "lshift",	bit_lshift },
+  { "rshift",	bit_rshift },
+  { "arshift",	bit_arshift },
+  { "rol",	bit_rol },
+  { "ror",	bit_ror },
+  { "bswap",	bit_bswap },
+  { "tohex",	bit_tohex },
+  { NULL, NULL }
+};
+
+/* Signed right-shifts are implementation-defined per C89/C99.
+** But the de facto standard are arithmetic right-shifts on two's
+** complement CPUs. This behaviour is required here, so test for it.
+*/
+#define BAD_SAR		(bsar(-8, 2) != (SBits)-2)
+
+LUALIB_API int luaopen_bit(lua_State *L)
+{
+  UBits b;
+  lua_pushnumber(L, (lua_Number)1437217655L);
+  b = barg(L, -1);
+  if (b != (UBits)1437217655L || BAD_SAR) {  /* Perform a simple self-test. */
+    const char *msg = "compiled with incompatible luaconf.h";
+#ifdef LUA_NUMBER_DOUBLE
+#ifdef _WIN32
+    if (b == (UBits)1610612736L)
+      msg = "use D3DCREATE_FPU_PRESERVE with DirectX";
+#endif
+    if (b == (UBits)1127743488L)
+      msg = "not compiled with SWAPPED_DOUBLE";
+#endif
+    if (BAD_SAR)
+      msg = "arithmetic right-shift broken";
+    luaL_error(L, "bit library self-test failed (%s)", msg);
+  }
+#if LUA_VERSION_NUM < 502
+  luaL_register(L, "bit", bit_funcs);
+#else
+  luaL_newlib(L, bit_funcs);
+#endif
+  return 1;
+}
+

deps/lua/src/lua_cmsgpack.c

@@ -7,14 +7,38 @@
 #include "lua.h"
 #include "lauxlib.h"
 
-#define LUACMSGPACK_VERSION     "lua-cmsgpack 0.3.0"
+#define LUACMSGPACK_NAME        "cmsgpack"
+#define LUACMSGPACK_SAFE_NAME   "cmsgpack_safe"
+#define LUACMSGPACK_VERSION     "lua-cmsgpack 0.4.0"
 #define LUACMSGPACK_COPYRIGHT   "Copyright (C) 2012, Salvatore Sanfilippo"
 #define LUACMSGPACK_DESCRIPTION "MessagePack C implementation for Lua"
 
-#define LUACMSGPACK_MAX_NESTING  16 /* Max tables nesting. */
+/* Allows a preprocessor directive to override MAX_NESTING */
+#ifndef LUACMSGPACK_MAX_NESTING
+    #define LUACMSGPACK_MAX_NESTING  16 /* Max tables nesting. */
+#endif
 
-/* ==============================================================================
- * MessagePack implementation and bindings for Lua 5.1.
+/* Check if float or double can be an integer without loss of precision */
+#define IS_INT_TYPE_EQUIVALENT(x, T) (!isinf(x) && (T)(x) == (x))
+
+#define IS_INT64_EQUIVALENT(x) IS_INT_TYPE_EQUIVALENT(x, int64_t)
+#define IS_INT_EQUIVALENT(x) IS_INT_TYPE_EQUIVALENT(x, int)
+
+/* If size of pointer is equal to a 4 byte integer, we're on 32 bits. */
+#if UINTPTR_MAX == UINT_MAX
+    #define BITS_32 1
+#else
+    #define BITS_32 0
+#endif
+
+#if BITS_32
+    #define lua_pushunsigned(L, n) lua_pushnumber(L, n)
+#else
+    #define lua_pushunsigned(L, n) lua_pushinteger(L, n)
+#endif
+
+/* =============================================================================
+ * MessagePack implementation and bindings for Lua 5.1/5.2.
  * Copyright(C) 2012 Salvatore Sanfilippo <antirez@gmail.com>
  *
  * http://github.com/antirez/lua-cmsgpack
@@ -29,23 +53,27 @@
  * 20-Feb-2012 (ver 0.2.0): Tables encoding improved.
  * 20-Feb-2012 (ver 0.2.1): Minor bug fixing.
  * 20-Feb-2012 (ver 0.3.0): Module renamed lua-cmsgpack (was lua-msgpack).
- * ============================================================================ */
+ * 04-Apr-2014 (ver 0.3.1): Lua 5.2 support and minor bug fix.
+ * 07-Apr-2014 (ver 0.4.0): Multiple pack/unpack, lua allocator, efficiency.
+ * ========================================================================== */
 
-/* --------------------------- Endian conversion --------------------------------
- * We use it only for floats and doubles, all the other conversions are performed
+/* -------------------------- Endian conversion --------------------------------
+ * We use it only for floats and doubles, all the other conversions performed
  * in an endian independent fashion. So the only thing we need is a function
- * that swaps a binary string if the arch is little endian (and left it untouched
+ * that swaps a binary string if arch is little endian (and left it untouched
  * otherwise). */
 
 /* Reverse memory bytes if arch is little endian. Given the conceptual
- * simplicity of the Lua build system we prefer to check for endianess at runtime.
+ * simplicity of the Lua build system we prefer check for endianess at runtime.
  * The performance difference should be acceptable. */
 static void memrevifle(void *ptr, size_t len) {
-    unsigned char *p = ptr, *e = p+len-1, aux;
+    unsigned char   *p = (unsigned char *)ptr,
+                    *e = (unsigned char *)p+len-1,
+                    aux;
     int test = 1;
     unsigned char *testp = (unsigned char*) &test;
 
-    if (testp[0] == 0) return; /* Big endian, nothign to do. */
+    if (testp[0] == 0) return; /* Big endian, nothing to do. */
     len /= 2;
     while(len--) {
         aux = *p;
@@ -56,30 +84,44 @@ static void memrevifle(void *ptr, size_t len) {
     }
 }
 
-/* ----------------------------- String buffer ----------------------------------
- * This is a simple implementation of string buffers. The only opereation
+/* ---------------------------- String buffer ----------------------------------
+ * This is a simple implementation of string buffers. The only operation
  * supported is creating empty buffers and appending bytes to it.
  * The string buffer uses 2x preallocation on every realloc for O(N) append
  * behavior.  */
 
 typedef struct mp_buf {
+    lua_State *L;
     unsigned char *b;
     size_t len, free;
 } mp_buf;
 
-static mp_buf *mp_buf_new(void) {
-    mp_buf *buf = malloc(sizeof(*buf));
-    
+static void *mp_realloc(lua_State *L, void *target, size_t osize,size_t nsize) {
+    void *(*local_realloc) (void *, void *, size_t osize, size_t nsize) = NULL;
+    void *ud;
+
+    local_realloc = lua_getallocf(L, &ud);
+
+    return local_realloc(ud, target, osize, nsize);
+}
+
+static mp_buf *mp_buf_new(lua_State *L) {
+    mp_buf *buf = NULL;
+
+    /* Old size = 0; new size = sizeof(*buf) */
+    buf = (mp_buf*)mp_realloc(L, NULL, 0, sizeof(*buf));
+
+    buf->L = L;
     buf->b = NULL;
     buf->len = buf->free = 0;
     return buf;
 }
 
-void mp_buf_append(mp_buf *buf, const unsigned char *s, size_t len) {
+static void mp_buf_append(mp_buf *buf, const unsigned char *s, size_t len) {
     if (buf->free < len) {
         size_t newlen = buf->len+len;
 
-        buf->b = realloc(buf->b,newlen*2);
+        buf->b = (unsigned char*)mp_realloc(buf->L, buf->b, buf->len, newlen*2);
         buf->free = newlen;
     }
     memcpy(buf->b+buf->len,s,len);
@@ -88,11 +130,11 @@ void mp_buf_append(mp_buf *buf, const unsigned char *s, size_t len) {
 }
 
 void mp_buf_free(mp_buf *buf) {
-    free(buf->b);
-    free(buf);
+    mp_realloc(buf->L, buf->b, buf->len, 0); /* realloc to 0 = free */
+    mp_realloc(buf->L, buf, sizeof(*buf), 0);
 }
 
-/* ------------------------------ String cursor ----------------------------------
+/* ---------------------------- String cursor ----------------------------------
  * This simple data structure is used for parsing. Basically you create a cursor
  * using a string pointer and a length, then it is possible to access the
  * current string position with cursor->p, check the remaining length
@@ -102,7 +144,7 @@ void mp_buf_free(mp_buf *buf) {
  * be used to report errors. */
 
 #define MP_CUR_ERROR_NONE   0
-#define MP_CUR_ERROR_EOF    1   /* Not enough data to complete the opereation. */
+#define MP_CUR_ERROR_EOF    1   /* Not enough data to complete operation. */
 #define MP_CUR_ERROR_BADFMT 2   /* Bad data format */
 
 typedef struct mp_cur {
@@ -111,22 +153,15 @@ typedef struct mp_cur {
     int err;
 } mp_cur;
 
-static mp_cur *mp_cur_new(const unsigned char *s, size_t len) {
-    mp_cur *cursor = malloc(sizeof(*cursor));
-
+static void mp_cur_init(mp_cur *cursor, const unsigned char *s, size_t len) {
     cursor->p = s;
     cursor->left = len;
     cursor->err = MP_CUR_ERROR_NONE;
-    return cursor;
-}
-
-static void mp_cur_free(mp_cur *cursor) {
-    free(cursor);
 }
 
 #define mp_cur_consume(_c,_len) do { _c->p += _len; _c->left -= _len; } while(0)
 
-/* When there is not enough room we set an error in the cursor and return, this
+/* When there is not enough room we set an error in the cursor and return. This
  * is very common across the code so we have a macro to make the code look
  * a bit simpler. */
 #define mp_cur_need(_c,_len) do { \
@@ -136,7 +171,7 @@ static void mp_cur_free(mp_cur *cursor) {
     } \
 } while(0)
 
-/* --------------------------- Low level MP encoding -------------------------- */
+/* ------------------------- Low level MP encoding -------------------------- */
 
 static void mp_encode_bytes(mp_buf *buf, const unsigned char *s, size_t len) {
     unsigned char hdr[5];
@@ -219,7 +254,7 @@ static void mp_encode_int(mp_buf *buf, int64_t n) {
         }
     } else {
         if (n >= -32) {
-            b[0] = ((char)n);   /* negative fixnum */
+            b[0] = ((signed char)n);   /* negative fixnum */
             enclen = 1;
         } else if (n >= -128) {
             b[0] = 0xd0;        /* int 8 */
@@ -299,7 +334,7 @@ static void mp_encode_map(mp_buf *buf, int64_t n) {
     mp_buf_append(buf,b,enclen);
 }
 
-/* ----------------------------- Lua types encoding --------------------------- */
+/* --------------------------- Lua types encoding --------------------------- */
 
 static void mp_encode_lua_string(lua_State *L, mp_buf *buf) {
     size_t len;
@@ -314,13 +349,26 @@ static void mp_encode_lua_bool(lua_State *L, mp_buf *buf) {
     mp_buf_append(buf,&b,1);
 }
 
+/* Lua 5.3 has a built in 64-bit integer type */
+static void mp_encode_lua_integer(lua_State *L, mp_buf *buf) {
+#if (LUA_VERSION_NUM < 503) && BITS_32
+    lua_Number i = lua_tonumber(L,-1);
+#else
+    lua_Integer i = lua_tointeger(L,-1);
+#endif
+    mp_encode_int(buf, (int64_t)i);
+}
+
+/* Lua 5.2 and lower only has 64-bit doubles, so we need to
+ * detect if the double may be representable as an int
+ * for Lua < 5.3 */
 static void mp_encode_lua_number(lua_State *L, mp_buf *buf) {
     lua_Number n = lua_tonumber(L,-1);
 
-    if (floor(n) != n) {
-        mp_encode_double(buf,(double)n);
+    if (IS_INT64_EQUIVALENT(n)) {
+        mp_encode_lua_integer(L, buf);
     } else {
-        mp_encode_int(buf,(int64_t)n);
+        mp_encode_double(buf,(double)n);
     }
 }
 
@@ -328,7 +376,11 @@ static void mp_encode_lua_type(lua_State *L, mp_buf *buf, int level);
 
 /* Convert a lua table into a message pack list. */
 static void mp_encode_lua_table_as_array(lua_State *L, mp_buf *buf, int level) {
+#if LUA_VERSION_NUM < 502
     size_t len = lua_objlen(L,-1), j;
+#else
+    size_t len = lua_rawlen(L,-1), j;
+#endif
 
     mp_encode_array(buf,len);
     for (j = 1; j <= len; j++) {
@@ -345,7 +397,7 @@ static void mp_encode_lua_table_as_map(lua_State *L, mp_buf *buf, int level) {
     /* First step: count keys into table. No other way to do it with the
      * Lua API, we need to iterate a first time. Note that an alternative
      * would be to do a single run, and then hack the buffer to insert the
-     * map opcodes for message pack. Too hachish for this lib. */
+     * map opcodes for message pack. Too hackish for this lib. */
     lua_pushnil(L);
     while(lua_next(L,-2)) {
         lua_pop(L,1); /* remove value, keep key for next iteration. */
@@ -367,30 +419,43 @@ static void mp_encode_lua_table_as_map(lua_State *L, mp_buf *buf, int level) {
  * of keys from numerical keys from 1 up to N, with N being the total number
  * of elements, without any hole in the middle. */
 static int table_is_an_array(lua_State *L) {
-    long count = 0, max = 0, idx = 0;
+    int count = 0, max = 0;
+#if LUA_VERSION_NUM < 503
     lua_Number n;
+#else
+    lua_Integer n;
+#endif
+
+    /* Stack top on function entry */
+    int stacktop;
+
+    stacktop = lua_gettop(L);
 
     lua_pushnil(L);
     while(lua_next(L,-2)) {
         /* Stack: ... key value */
         lua_pop(L,1); /* Stack: ... key */
-        if (lua_type(L,-1) != LUA_TNUMBER) goto not_array;
-        n = lua_tonumber(L,-1);
-        idx = n;
-        if (idx != n || idx < 1) goto not_array;
+        /* The <= 0 check is valid here because we're comparing indexes. */
+#if LUA_VERSION_NUM < 503
+        if ((LUA_TNUMBER != lua_type(L,-1)) || (n = lua_tonumber(L, -1)) <= 0 ||
+            !IS_INT_EQUIVALENT(n))
+#else
+        if (!lua_isinteger(L,-1) || (n = lua_tointeger(L, -1)) <= 0)
+#endif
+        {
+            lua_settop(L, stacktop);
+            return 0;
+        }
+        max = (n > max ? n : max);
         count++;
-        max = idx;
     }
     /* We have the total number of elements in "count". Also we have
-     * the max index encountered in "idx". We can't reach this code
+     * the max index encountered in "max". We can't reach this code
      * if there are indexes <= 0. If you also note that there can not be
-     * repeated keys into a table, you have that if idx==count you are sure
+     * repeated keys into a table, you have that if max==count you are sure
      * that there are all the keys form 1 to count (both included). */
-    return idx == count;
-
-not_array:
-    lua_pop(L,1);
-    return 0;
+    lua_settop(L, stacktop);
+    return max == count;
 }
 
 /* If the length operator returns non-zero, that is, there is at least
@@ -405,6 +470,7 @@ static void mp_encode_lua_table(lua_State *L, mp_buf *buf, int level) {
 
 static void mp_encode_lua_null(lua_State *L, mp_buf *buf) {
     unsigned char b[1];
+    (void)L;
 
     b[0] = 0xc0;
     mp_buf_append(buf,b,1);
@@ -413,33 +479,70 @@ static void mp_encode_lua_null(lua_State *L, mp_buf *buf) {
 static void mp_encode_lua_type(lua_State *L, mp_buf *buf, int level) {
     int t = lua_type(L,-1);
 
-    /* Limit the encoding of nested tables to a specfiied maximum depth, so that
+    /* Limit the encoding of nested tables to a specified maximum depth, so that
      * we survive when called against circular references in tables. */
     if (t == LUA_TTABLE && level == LUACMSGPACK_MAX_NESTING) t = LUA_TNIL;
     switch(t) {
     case LUA_TSTRING: mp_encode_lua_string(L,buf); break;
     case LUA_TBOOLEAN: mp_encode_lua_bool(L,buf); break;
-    case LUA_TNUMBER: mp_encode_lua_number(L,buf); break;
+    case LUA_TNUMBER:
+    #if LUA_VERSION_NUM < 503
+        mp_encode_lua_number(L,buf); break;
+    #else
+        if (lua_isinteger(L, -1)) {
+            mp_encode_lua_integer(L, buf);
+        } else {
+            mp_encode_lua_number(L, buf);
+        }
+        break;
+    #endif
     case LUA_TTABLE: mp_encode_lua_table(L,buf,level); break;
     default: mp_encode_lua_null(L,buf); break;
     }
     lua_pop(L,1);
 }
 
+/*
+ * Packs all arguments as a stream for multiple upacking later.
+ * Returns error if no arguments provided.
+ */
 static int mp_pack(lua_State *L) {
-    mp_buf *buf = mp_buf_new();
+    int nargs = lua_gettop(L);
+    int i;
+    mp_buf *buf;
 
-    mp_encode_lua_type(L,buf,0);
-    lua_pushlstring(L,(char*)buf->b,buf->len);
+    if (nargs == 0)
+        return luaL_argerror(L, 0, "MessagePack pack needs input.");
+
+    buf = mp_buf_new(L);
+    for(i = 1; i <= nargs; i++) {
+        /* Copy argument i to top of stack for _encode processing;
+         * the encode function pops it from the stack when complete. */
+        lua_pushvalue(L, i);
+
+        mp_encode_lua_type(L,buf,0);
+
+        lua_pushlstring(L,(char*)buf->b,buf->len);
+
+        /* Reuse the buffer for the next operation by
+         * setting its free count to the total buffer size
+         * and the current position to zero. */
+        buf->free += buf->len;
+        buf->len = 0;
+    }
     mp_buf_free(buf);
+
+    /* Concatenate all nargs buffers together */
+    lua_concat(L, nargs);
     return 1;
 }
 
-/* --------------------------------- Decoding --------------------------------- */
+/* ------------------------------- Decoding --------------------------------- */
 
 void mp_decode_to_lua_type(lua_State *L, mp_cur *c);
 
 void mp_decode_to_lua_array(lua_State *L, mp_cur *c, size_t len) {
+    assert(len <= UINT_MAX);
     int index = 1;
 
     lua_newtable(L);
@@ -452,6 +555,7 @@ void mp_decode_to_lua_array(lua_State *L, mp_cur *c, size_t len) {
 }
 
 void mp_decode_to_lua_hash(lua_State *L, mp_cur *c, size_t len) {
+    assert(len <= UINT_MAX);
     lua_newtable(L);
     while(len--) {
         mp_decode_to_lua_type(L,c); /* key */
@@ -466,34 +570,44 @@ void mp_decode_to_lua_hash(lua_State *L, mp_cur *c, size_t len) {
  * a Lua type, that is left as the only result on the stack. */
 void mp_decode_to_lua_type(lua_State *L, mp_cur *c) {
     mp_cur_need(c,1);
+
+    /* If we return more than 18 elements, we must resize the stack to
+     * fit all our return values.  But, there is no way to
+     * determine how many objects a msgpack will unpack to up front, so
+     * we request a +1 larger stack on each iteration (noop if stack is
+     * big enough, and when stack does require resize it doubles in size) */
+    luaL_checkstack(L, 1,
+        "too many return values at once; "
+        "use unpack_one or unpack_limit instead.");
+
     switch(c->p[0]) {
     case 0xcc:  /* uint 8 */
         mp_cur_need(c,2);
-        lua_pushnumber(L,c->p[1]);
+        lua_pushunsigned(L,c->p[1]);
         mp_cur_consume(c,2);
         break;
     case 0xd0:  /* int 8 */
         mp_cur_need(c,2);
-        lua_pushnumber(L,(char)c->p[1]);
+        lua_pushinteger(L,(signed char)c->p[1]);
         mp_cur_consume(c,2);
         break;
     case 0xcd:  /* uint 16 */
         mp_cur_need(c,3);
-        lua_pushnumber(L,
+        lua_pushunsigned(L,
             (c->p[1] << 8) |
              c->p[2]);
         mp_cur_consume(c,3);
         break;
     case 0xd1:  /* int 16 */
         mp_cur_need(c,3);
-        lua_pushnumber(L,(int16_t)
+        lua_pushinteger(L,(int16_t)
             (c->p[1] << 8) |
              c->p[2]);
         mp_cur_consume(c,3);
         break;
     case 0xce:  /* uint 32 */
         mp_cur_need(c,5);
-        lua_pushnumber(L,
+        lua_pushunsigned(L,
             ((uint32_t)c->p[1] << 24) |
             ((uint32_t)c->p[2] << 16) |
             ((uint32_t)c->p[3] << 8) |
@@ -502,7 +616,7 @@ void mp_decode_to_lua_type(lua_State *L, mp_cur *c) {
         break;
     case 0xd2:  /* int 32 */
         mp_cur_need(c,5);
-        lua_pushnumber(L,
+        lua_pushinteger(L,
             ((int32_t)c->p[1] << 24) |
             ((int32_t)c->p[2] << 16) |
             ((int32_t)c->p[3] << 8) |
@@ -511,7 +625,7 @@ void mp_decode_to_lua_type(lua_State *L, mp_cur *c) {
         break;
     case 0xcf:  /* uint 64 */
         mp_cur_need(c,9);
-        lua_pushnumber(L,
+        lua_pushunsigned(L,
             ((uint64_t)c->p[1] << 56) |
             ((uint64_t)c->p[2] << 48) |
             ((uint64_t)c->p[3] << 40) |
@@ -524,7 +638,11 @@ void mp_decode_to_lua_type(lua_State *L, mp_cur *c) {
         break;
     case 0xd3:  /* int 64 */
         mp_cur_need(c,9);
+#if LUA_VERSION_NUM < 503
         lua_pushnumber(L,
+#else
+        lua_pushinteger(L,
+#endif
             ((int64_t)c->p[1] << 56) |
             ((int64_t)c->p[2] << 48) |
             ((int64_t)c->p[3] << 40) |
@@ -581,13 +699,14 @@ void mp_decode_to_lua_type(lua_State *L, mp_cur *c) {
     case 0xdb:  /* raw 32 */
         mp_cur_need(c,5);
         {
-            size_t l = (c->p[1] << 24) |
-                       (c->p[2] << 16) |
-                       (c->p[3] << 8) |
-                       c->p[4];
-            mp_cur_need(c,5+l);
-            lua_pushlstring(L,(char*)c->p+5,l);
-            mp_cur_consume(c,5+l);
+            size_t l = ((size_t)c->p[1] << 24) |
+                       ((size_t)c->p[2] << 16) |
+                       ((size_t)c->p[3] << 8) |
+                       (size_t)c->p[4];
+            mp_cur_consume(c,5);
+            mp_cur_need(c,l);
+            lua_pushlstring(L,(char*)c->p,l);
+            mp_cur_consume(c,l);
         }
         break;
     case 0xdc:  /* array 16 */
@@ -601,10 +720,10 @@ void mp_decode_to_lua_type(lua_State *L, mp_cur *c) {
     case 0xdd:  /* array 32 */
         mp_cur_need(c,5);
         {
-            size_t l = (c->p[1] << 24) |
-                       (c->p[2] << 16) |
-                       (c->p[3] << 8) |
-                       c->p[4];
+            size_t l = ((size_t)c->p[1] << 24) |
+                       ((size_t)c->p[2] << 16) |
+                       ((size_t)c->p[3] << 8) |
+                       (size_t)c->p[4];
             mp_cur_consume(c,5);
             mp_decode_to_lua_array(L,c,l);
         }
@@ -620,20 +739,20 @@ void mp_decode_to_lua_type(lua_State *L, mp_cur *c) {
     case 0xdf:  /* map 32 */
         mp_cur_need(c,5);
         {
-            size_t l = (c->p[1] << 24) |
-                       (c->p[2] << 16) |
-                       (c->p[3] << 8) |
-                       c->p[4];
+            size_t l = ((size_t)c->p[1] << 24) |
+                       ((size_t)c->p[2] << 16) |
+                       ((size_t)c->p[3] << 8) |
+                       (size_t)c->p[4];
             mp_cur_consume(c,5);
             mp_decode_to_lua_hash(L,c,l);
         }
         break;
     default:    /* types that can't be idenitified by first byte value. */
         if ((c->p[0] & 0x80) == 0) {   /* positive fixnum */
-            lua_pushnumber(L,c->p[0]);
+            lua_pushunsigned(L,c->p[0]);
             mp_cur_consume(c,1);
         } else if ((c->p[0] & 0xe0) == 0xe0) {  /* negative fixnum */
-            lua_pushnumber(L,(signed char)c->p[0]);
+            lua_pushinteger(L,(signed char)c->p[0]);
             mp_cur_consume(c,1);
         } else if ((c->p[0] & 0xe0) == 0xa0) {  /* fix raw */
             size_t l = c->p[0] & 0x1f;
@@ -654,54 +773,163 @@ void mp_decode_to_lua_type(lua_State *L, mp_cur *c) {
     }
 }
 
-static int mp_unpack(lua_State *L) {
+static int mp_unpack_full(lua_State *L, int limit, int offset) {
     size_t len;
-    const unsigned char *s;
-    mp_cur *c;
+    const char *s;
+    mp_cur c;
+    int cnt; /* Number of objects unpacked */
+    int decode_all = (!limit && !offset);
 
-    if (!lua_isstring(L,-1)) {
-        lua_pushstring(L,"MessagePack decoding needs a string as input.");
-        lua_error(L);
+    s = luaL_checklstring(L,1,&len); /* if no match, exits */
+
+    if (offset < 0 || limit < 0) /* requesting negative off or lim is invalid */
+        return luaL_error(L,
+            "Invalid request to unpack with offset of %d and limit of %d.",
+            offset, len);
+    else if (offset > len)
+        return luaL_error(L,
+            "Start offset %d greater than input length %d.", offset, len);
+
+    if (decode_all) limit = INT_MAX;
+
+    mp_cur_init(&c,(const unsigned char *)s+offset,len-offset);
+
+    /* We loop over the decode because this could be a stream
+     * of multiple top-level values serialized together */
+    for(cnt = 0; c.left > 0 && cnt < limit; cnt++) {
+        mp_decode_to_lua_type(L,&c);
+
+        if (c.err == MP_CUR_ERROR_EOF) {
+            return luaL_error(L,"Missing bytes in input.");
+        } else if (c.err == MP_CUR_ERROR_BADFMT) {
+            return luaL_error(L,"Bad data format in input.");
+        }
     }
 
-    s = (const unsigned char*) lua_tolstring(L,-1,&len);
-    c = mp_cur_new(s,len);
-    mp_decode_to_lua_type(L,c);
-    
-    if (c->err == MP_CUR_ERROR_EOF) {
-        mp_cur_free(c);
-        lua_pushstring(L,"Missing bytes in input.");
-        lua_error(L);
-    } else if (c->err == MP_CUR_ERROR_BADFMT) {
-        mp_cur_free(c);
-        lua_pushstring(L,"Bad data format in input.");
-        lua_error(L);
-    } else if (c->left != 0) {
-        mp_cur_free(c);
-        lua_pushstring(L,"Extra bytes in input.");
-        lua_error(L);
+    if (!decode_all) {
+        /* c->left is the remaining size of the input buffer.
+         * subtract the entire buffer size from the unprocessed size
+         * to get our next start offset */
+        int offset = len - c.left;
+        /* Return offset -1 when we have have processed the entire buffer. */
+        lua_pushinteger(L, c.left == 0 ? -1 : offset);
+        /* Results are returned with the arg elements still
+         * in place. Lua takes care of only returning
+         * elements above the args for us.
+         * In this case, we have one arg on the stack
+         * for this function, so we insert our first return
+         * value at position 2. */
+        lua_insert(L, 2);
+        cnt += 1; /* increase return count by one to make room for offset */
     }
-    mp_cur_free(c);
-    return 1;
+
+    return cnt;
 }
 
-/* ---------------------------------------------------------------------------- */
+static int mp_unpack(lua_State *L) {
+    return mp_unpack_full(L, 0, 0);
+}
 
-static const struct luaL_reg thislib[] = {
+static int mp_unpack_one(lua_State *L) {
+    int offset = luaL_optinteger(L, 2, 0);
+    /* Variable pop because offset may not exist */
+    lua_pop(L, lua_gettop(L)-1);
+    return mp_unpack_full(L, 1, offset);
+}
+
+static int mp_unpack_limit(lua_State *L) {
+    int limit = luaL_checkinteger(L, 2);
+    int offset = luaL_optinteger(L, 3, 0);
+    /* Variable pop because offset may not exist */
+    lua_pop(L, lua_gettop(L)-1);
+
+    return mp_unpack_full(L, limit, offset);
+}
+
+static int mp_safe(lua_State *L) {
+    int argc, err, total_results;
+
+    argc = lua_gettop(L);
+
+    /* This adds our function to the bottom of the stack
+     * (the "call this function" position) */
+    lua_pushvalue(L, lua_upvalueindex(1));
+    lua_insert(L, 1);
+
+    err = lua_pcall(L, argc, LUA_MULTRET, 0);
+    total_results = lua_gettop(L);
+
+    if (!err) {
+        return total_results;
+    } else {
+        lua_pushnil(L);
+        lua_insert(L,-2);
+        return 2;
+    }
+}
+
+/* -------------------------------------------------------------------------- */
+static const struct luaL_Reg cmds[] = {
     {"pack", mp_pack},
     {"unpack", mp_unpack},
-    {NULL, NULL}
+    {"unpack_one", mp_unpack_one},
+    {"unpack_limit", mp_unpack_limit},
+    {0}
 };
 
-LUALIB_API int luaopen_cmsgpack (lua_State *L) {
-    luaL_register(L, "cmsgpack", thislib);
+static int luaopen_create(lua_State *L) {
+    int i;
+    /* Manually construct our module table instead of
+     * relying on _register or _newlib */
+    lua_newtable(L);
 
+    for (i = 0; i < (sizeof(cmds)/sizeof(*cmds) - 1); i++) {
+        lua_pushcfunction(L, cmds[i].func);
+        lua_setfield(L, -2, cmds[i].name);
+    }
+
+    /* Add metadata */
+    lua_pushliteral(L, LUACMSGPACK_NAME);
+    lua_setfield(L, -2, "_NAME");
     lua_pushliteral(L, LUACMSGPACK_VERSION);
     lua_setfield(L, -2, "_VERSION");
     lua_pushliteral(L, LUACMSGPACK_COPYRIGHT);
     lua_setfield(L, -2, "_COPYRIGHT");
     lua_pushliteral(L, LUACMSGPACK_DESCRIPTION);
-    lua_setfield(L, -2, "_DESCRIPTION"); 
+    lua_setfield(L, -2, "_DESCRIPTION");
+    return 1;
+}
+
+LUALIB_API int luaopen_cmsgpack(lua_State *L) {
+    luaopen_create(L);
+
+#if LUA_VERSION_NUM < 502
+    /* Register name globally for 5.1 */
+    lua_pushvalue(L, -1);
+    lua_setglobal(L, LUACMSGPACK_NAME);
+#endif
+
+    return 1;
+}
+
+LUALIB_API int luaopen_cmsgpack_safe(lua_State *L) {
+    int i;
+
+    luaopen_cmsgpack(L);
+
+    /* Wrap all functions in the safe handler */
+    for (i = 0; i < (sizeof(cmds)/sizeof(*cmds) - 1); i++) {
+        lua_getfield(L, -1, cmds[i].name);
+        lua_pushcclosure(L, mp_safe, 1);
+        lua_setfield(L, -2, cmds[i].name);
+    }
+
+#if LUA_VERSION_NUM < 502
+    /* Register name globally for 5.1 */
+    lua_pushvalue(L, -1);
+    lua_setglobal(L, LUACMSGPACK_SAFE_NAME);
+#endif
+
     return 1;
 }
 

deps/lua/src/strbuf.c

@@ -1,6 +1,6 @@
-/* strbuf - string buffer routines
+/* strbuf - String buffer routines
  *
- * Copyright (c) 2010-2011  Mark Pulford <mark@kyne.com.au>
+ * Copyright (c) 2010-2012  Mark Pulford <mark@kyne.com.au>
  *
  * Permission is hereby granted, free of charge, to any person obtaining
  * a copy of this software and associated documentation files (the
@@ -29,7 +29,7 @@
 
 #include "strbuf.h"
 
-void die(const char *fmt, ...)
+static void die(const char *fmt, ...)
 {
     va_list arg;
 

deps/lua/src/strbuf.h

@@ -1,6 +1,6 @@
 /* strbuf - String buffer routines
  *
- * Copyright (c) 2010-2011  Mark Pulford <mark@kyne.com.au>
+ * Copyright (c) 2010-2012  Mark Pulford <mark@kyne.com.au>
  *
  * Permission is hereby granted, free of charge, to any person obtaining
  * a copy of this software and associated documentation files (the
@@ -62,7 +62,9 @@ extern void strbuf_resize(strbuf_t *s, int len);
 static int strbuf_empty_length(strbuf_t *s);
 static int strbuf_length(strbuf_t *s);
 static char *strbuf_string(strbuf_t *s, int *len);
-static  void strbuf_ensure_empty_length(strbuf_t *s, int len);
+static void strbuf_ensure_empty_length(strbuf_t *s, int len);
+static char *strbuf_empty_ptr(strbuf_t *s);
+static void strbuf_extend_length(strbuf_t *s, int len);
 
 /* Update */
 extern void strbuf_append_fmt(strbuf_t *s, int len, const char *fmt, ...);
@@ -96,6 +98,16 @@ static inline void strbuf_ensure_empty_length(strbuf_t *s, int len)
         strbuf_resize(s, s->length + len);
 }
 
+static inline char *strbuf_empty_ptr(strbuf_t *s)
+{
+    return s->buf + s->length;
+}
+
+static inline void strbuf_extend_length(strbuf_t *s, int len)
+{
+    s->length += len;
+}
+
 static inline int strbuf_length(strbuf_t *s)
 {
     return s->length;

redis.conf

@@ -36,6 +36,17 @@
 # Note that Redis will write a pid file in /var/run/redis.pid when daemonized.
 daemonize no
 
+# If you run Redis from upstart or systemd, Redis can interact with your
+# supervision tree. Options:
+#   supervised no      - no supervision interaction
+#   supervised upstart - signal upstart by putting Redis into SIGSTOP mode
+#   supervised systemd - signal systemd by writing READY=1 to $NOTIFY_SOCKET
+#   supervised auto    - detect upstart or systemd method based on
+#                        UPSTART_JOB or NOTIFY_SOCKET environment variables
+# Note: these supervision methods only signal "process is ready."
+#       They do not enable continuous liveness pings back to your supervisor.
+supervised no
+
 # When running daemonized, Redis writes a pid file in /var/run/redis.pid by
 # default. You can specify a custom pid file location here.
 pidfile /var/run/redis.pid
@@ -242,6 +253,10 @@ slave-read-only yes
 
 # Replication SYNC strategy: disk or socket.
 #
+# -------------------------------------------------------
+# WARNING: DISKLESS REPLICATION IS EXPERIMENTAL CURRENTLY
+# -------------------------------------------------------
+#
 # New slaves and reconnecting slaves that are not able to continue the replication
 # process just receiving differences, need to do what is called a "full
 # synchronization". An RDB file is transmitted from the master to the slaves.
@@ -268,7 +283,7 @@ slave-read-only yes
 repl-diskless-sync no
 
 # When diskless replication is enabled, it is possible to configure the delay
-# the server waits in order to spawn the child that trnasfers the RDB via socket
+# the server waits in order to spawn the child that transfers the RDB via socket
 # to the slaves.
 #
 # This is important since once the transfer starts, it is not possible to serve
@@ -615,6 +630,12 @@ lua-time-limit 5000
 
 ################################ REDIS CLUSTER  ###############################
 #
+# ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+# WARNING EXPERIMENTAL: Redis Cluster is considered to be stable code, however
+# in order to mark it as "mature" we need to wait for a non trivial percentage
+# of users to deploy it in production.
+# ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
+#
 # Normal Redis instances can't be part of a Redis Cluster; only nodes that are
 # started as cluster nodes can. In order to start a Redis instance as a
 # cluster node enable the cluster support uncommenting the following:
@@ -756,7 +777,7 @@ slowlog-max-len 128
 # By default latency monitoring is disabled since it is mostly not needed
 # if you don't have latency issues, and collecting data has a performance
 # impact, that while very small, can be measured under big load. Latency
-# monitoring can easily be enalbed at runtime using the command
+# monitoring can easily be enabled at runtime using the command
 # "CONFIG SET latency-monitor-threshold <milliseconds>" if needed.
 latency-monitor-threshold 0
 
@@ -814,11 +835,36 @@ notify-keyspace-events ""
 hash-max-ziplist-entries 512
 hash-max-ziplist-value 64
 
-# Similarly to hashes, small lists are also encoded in a special way in order
-# to save a lot of space. The special representation is only used when
-# you are under the following limits:
-list-max-ziplist-entries 512
-list-max-ziplist-value 64
+# Lists are also encoded in a special way to save a lot of space.
+# The number of entries allowed per internal list node can be specified
+# as a fixed maximum size or a maximum number of elements.
+# For a fixed maximum size, use -5 through -1, meaning:
+# -5: max size: 64 Kb  <-- not recommended for normal workloads
+# -4: max size: 32 Kb  <-- not recommended
+# -3: max size: 16 Kb  <-- probably not recommended
+# -2: max size: 8 Kb   <-- good
+# -1: max size: 4 Kb   <-- good
+# Positive numbers mean store up to _exactly_ that number of elements
+# per list node.
+# The highest performing option is usually -2 (8 Kb size) or -1 (4 Kb size),
+# but if your use case is unique, adjust the settings as necessary.
+list-max-ziplist-size -2
+
+# Lists may also be compressed.
+# Compress depth is the number of quicklist ziplist nodes from *each* side of
+# the list to *exclude* from compression.  The head and tail of the list
+# are always uncompressed for fast push/pop operations.  Settings are:
+# 0: disable all list compression
+# 1: depth 1 means "don't start compressing until after 1 node into the list,
+#    going from either the head or tail"
+#    So: [head]->node->node->...->node->[tail]
+#    [head], [tail] will always be uncompressed; inner nodes will compress.
+# 2: [head]->[next]->node->node->...->node->[prev]->[tail]
+#    2 here means: don't compress head or head->next or tail->prev or tail,
+#    but compress all nodes between them.
+# 3: [head]->[next]->[next]->node->node->...->node->[prev]->[prev]->[tail]
+# etc.
+list-compress-depth 0
 
 # Sets have a special encoding in just one case: when a set is composed
 # of just strings that happen to be integers in radix 10 in the range

src/Makefile

@@ -18,7 +18,7 @@ OPTIMIZATION?=-O2
 DEPENDENCY_TARGETS=hiredis linenoise lua
 
 # Default settings
-STD=-std=c99 -pedantic
+STD=-std=c99 -pedantic -DREDIS_STATIC=''
 WARN=-Wall -W
 OPT=$(OPTIMIZATION)
 
@@ -46,6 +46,10 @@ ifeq ($(USE_JEMALLOC),yes)
 	MALLOC=jemalloc
 endif
 
+ifeq ($(USE_JEMALLOC),no)
+	MALLOC=libc
+endif
+
 # Override default settings if possible
 -include .make-settings
 
@@ -58,7 +62,7 @@ ifeq ($(uname_S),SunOS)
 	# SunOS
 	INSTALL=cp -pf
 	FINAL_CFLAGS+= -D__EXTENSIONS__ -D_XPG6
-	FINAL_LIBS+= -ldl -lnsl -lsocket -lresolv -lpthread
+	FINAL_LIBS+= -ldl -lnsl -lsocket -lresolv -lpthread -lrt
 else
 ifeq ($(uname_S),Darwin)
 	# Darwin (nothing to do)
@@ -113,17 +117,16 @@ endif
 
 REDIS_SERVER_NAME=redis-server
 REDIS_SENTINEL_NAME=redis-sentinel
-REDIS_SERVER_OBJ=adlist.o ae.o anet.o dict.o redis.o sds.o zmalloc.o lzf_c.o lzf_d.o pqsort.o zipmap.o sha1.o ziplist.o release.o networking.o util.o object.o db.o replication.o rdb.o t_string.o t_list.o t_set.o t_zset.o t_hash.o config.o aof.o pubsub.o multi.o debug.o sort.o intset.o syncio.o cluster.o crc16.o endianconv.o slowlog.o scripting.o bio.o rio.o rand.o memtest.o crc64.o bitops.o sentinel.o notify.o setproctitle.o blocked.o hyperloglog.o latency.o sparkline.o
+REDIS_SERVER_OBJ=adlist.o quicklist.o ae.o anet.o dict.o redis.o sds.o zmalloc.o lzf_c.o lzf_d.o pqsort.o zipmap.o sha1.o ziplist.o release.o networking.o util.o object.o db.o replication.o rdb.o t_string.o t_list.o t_set.o t_zset.o t_hash.o config.o aof.o pubsub.o multi.o debug.o sort.o intset.o syncio.o cluster.o crc16.o endianconv.o slowlog.o scripting.o bio.o rio.o rand.o memtest.o crc64.o bitops.o sentinel.o notify.o setproctitle.o blocked.o hyperloglog.o latency.o sparkline.o redis-check-rdb.o
 REDIS_CLI_NAME=redis-cli
 REDIS_CLI_OBJ=anet.o sds.o adlist.o redis-cli.o zmalloc.o release.o anet.o ae.o crc64.o
 REDIS_BENCHMARK_NAME=redis-benchmark
 REDIS_BENCHMARK_OBJ=ae.o anet.o redis-benchmark.o sds.o adlist.o zmalloc.o redis-benchmark.o
-REDIS_CHECK_DUMP_NAME=redis-check-dump
-REDIS_CHECK_DUMP_OBJ=redis-check-dump.o lzf_c.o lzf_d.o crc64.o
+REDIS_CHECK_RDB_NAME=redis-check-rdb
 REDIS_CHECK_AOF_NAME=redis-check-aof
 REDIS_CHECK_AOF_OBJ=redis-check-aof.o
 
-all: $(REDIS_SERVER_NAME) $(REDIS_SENTINEL_NAME) $(REDIS_CLI_NAME) $(REDIS_BENCHMARK_NAME) $(REDIS_CHECK_DUMP_NAME) $(REDIS_CHECK_AOF_NAME)
+all: $(REDIS_SERVER_NAME) $(REDIS_SENTINEL_NAME) $(REDIS_CLI_NAME) $(REDIS_BENCHMARK_NAME) $(REDIS_CHECK_RDB_NAME) $(REDIS_CHECK_AOF_NAME)
 	@echo ""
 	@echo "Hint: It's a good idea to run 'make test' ;)"
 	@echo ""
@@ -174,6 +177,10 @@ $(REDIS_SERVER_NAME): $(REDIS_SERVER_OBJ)
 $(REDIS_SENTINEL_NAME): $(REDIS_SERVER_NAME)
 	$(REDIS_INSTALL) $(REDIS_SERVER_NAME) $(REDIS_SENTINEL_NAME)
 
+# redis-check-rdb
+$(REDIS_CHECK_RDB_NAME): $(REDIS_SERVER_NAME)
+	$(REDIS_INSTALL) $(REDIS_SERVER_NAME) $(REDIS_CHECK_RDB_NAME)
+
 # redis-cli
 $(REDIS_CLI_NAME): $(REDIS_CLI_OBJ)
 	$(REDIS_LD) -o $@ $^ ../deps/hiredis/libhiredis.a ../deps/linenoise/linenoise.o $(FINAL_LIBS)
@@ -182,10 +189,6 @@ $(REDIS_CLI_NAME): $(REDIS_CLI_OBJ)
 $(REDIS_BENCHMARK_NAME): $(REDIS_BENCHMARK_OBJ)
 	$(REDIS_LD) -o $@ $^ ../deps/hiredis/libhiredis.a $(FINAL_LIBS)
 
-# redis-check-dump
-$(REDIS_CHECK_DUMP_NAME): $(REDIS_CHECK_DUMP_OBJ)
-	$(REDIS_LD) -o $@ $^ $(FINAL_LIBS)
-
 # redis-check-aof
 $(REDIS_CHECK_AOF_NAME): $(REDIS_CHECK_AOF_OBJ)
 	$(REDIS_LD) -o $@ $^ $(FINAL_LIBS)
@@ -197,7 +200,7 @@ $(REDIS_CHECK_AOF_NAME): $(REDIS_CHECK_AOF_OBJ)
 	$(REDIS_CC) -c $<
 
 clean:
-	rm -rf $(REDIS_SERVER_NAME) $(REDIS_SENTINEL_NAME) $(REDIS_CLI_NAME) $(REDIS_BENCHMARK_NAME) $(REDIS_CHECK_DUMP_NAME) $(REDIS_CHECK_AOF_NAME) *.o *.gcda *.gcno *.gcov redis.info lcov-html
+	rm -rf $(REDIS_SERVER_NAME) $(REDIS_SENTINEL_NAME) $(REDIS_CLI_NAME) $(REDIS_BENCHMARK_NAME) $(REDIS_CHECK_RDB_NAME) $(REDIS_CHECK_AOF_NAME) *.o *.gcda *.gcno *.gcov redis.info lcov-html
 
 .PHONY: clean
 
@@ -221,6 +224,10 @@ lcov:
 	@geninfo -o redis.info .
 	@genhtml --legend -o lcov-html redis.info
 
+test-sds: sds.c sds.h
+	$(REDIS_CC) sds.c zmalloc.c -DSDS_TEST_MAIN -o /tmp/sds_test
+	/tmp/sds_test
+
 .PHONY: lcov
 
 bench: $(REDIS_BENCHMARK_NAME)
@@ -249,5 +256,6 @@ install: all
 	$(REDIS_INSTALL) $(REDIS_SERVER_NAME) $(INSTALL_BIN)
 	$(REDIS_INSTALL) $(REDIS_BENCHMARK_NAME) $(INSTALL_BIN)
 	$(REDIS_INSTALL) $(REDIS_CLI_NAME) $(INSTALL_BIN)
-	$(REDIS_INSTALL) $(REDIS_CHECK_DUMP_NAME) $(INSTALL_BIN)
+	$(REDIS_INSTALL) $(REDIS_CHECK_RDB_NAME) $(INSTALL_BIN)
 	$(REDIS_INSTALL) $(REDIS_CHECK_AOF_NAME) $(INSTALL_BIN)
+	@ln -sf $(REDIS_SERVER_NAME) $(INSTALL_BIN)/$(REDIS_SENTINEL_NAME)

src/anet.c

@@ -589,6 +589,23 @@ error:
     return -1;
 }
 
+/* Format an IP,port pair into something easy to parse. If IP is IPv6
+ * (matches for ":"), the ip is surrounded by []. IP and port are just
+ * separated by colons. This the standard to display addresses within Redis. */
+int anetFormatAddr(char *buf, size_t buf_len, char *ip, int port) {
+    return snprintf(buf,buf_len, strchr(ip,':') ?
+           "[%s]:%d" : "%s:%d", ip, port);
+}
+
+/* Like anetFormatAddr() but extract ip and port from the socket's peer. */
+int anetFormatPeer(int fd, char *buf, size_t buf_len) {
+    char ip[INET6_ADDRSTRLEN];
+    int port;
+
+    anetPeerToString(fd,ip,sizeof(ip),&port);
+    return anetFormatAddr(buf, buf_len, ip, port);
+}
+
 int anetSockName(int fd, char *ip, size_t ip_len, int *port) {
     struct sockaddr_storage sa;
     socklen_t salen = sizeof(sa);
@@ -610,3 +627,11 @@ int anetSockName(int fd, char *ip, size_t ip_len, int *port) {
     }
     return 0;
 }
+
+int anetFormatSock(int fd, char *fmt, size_t fmt_len) {
+    char ip[INET6_ADDRSTRLEN];
+    int port;
+
+    anetSockName(fd,ip,sizeof(ip),&port);
+    return anetFormatAddr(fmt, fmt_len, ip, port);
+}

src/anet.h

@@ -70,5 +70,8 @@ int anetSendTimeout(char *err, int fd, long long ms);
 int anetPeerToString(int fd, char *ip, size_t ip_len, int *port);
 int anetKeepAlive(char *err, int fd, int interval);
 int anetSockName(int fd, char *ip, size_t ip_len, int *port);
+int anetFormatAddr(char *fmt, size_t fmt_len, char *ip, int port);
+int anetFormatPeer(int fd, char *fmt, size_t fmt_len);
+int anetFormatSock(int fd, char *fmt, size_t fmt_len);
 
 #endif

src/aof.c

@@ -770,52 +770,29 @@ int rioWriteBulkObject(rio *r, robj *obj) {
 int rewriteListObject(rio *r, robj *key, robj *o) {
     long long count = 0, items = listTypeLength(o);
 
-    if (o->encoding == REDIS_ENCODING_ZIPLIST) {
-        unsigned char *zl = o->ptr;
-        unsigned char *p = ziplistIndex(zl,0);
-        unsigned char *vstr;
-        unsigned int vlen;
-        long long vlong;
+    if (o->encoding == REDIS_ENCODING_QUICKLIST) {
+        quicklist *list = o->ptr;
+        quicklistIter *li = quicklistGetIterator(list, AL_START_HEAD);
+        quicklistEntry entry;
 
-        while(ziplistGet(p,&vstr,&vlen,&vlong)) {
+        while (quicklistNext(li,&entry)) {
             if (count == 0) {
                 int cmd_items = (items > REDIS_AOF_REWRITE_ITEMS_PER_CMD) ?
                     REDIS_AOF_REWRITE_ITEMS_PER_CMD : items;
-
                 if (rioWriteBulkCount(r,'*',2+cmd_items) == 0) return 0;
                 if (rioWriteBulkString(r,"RPUSH",5) == 0) return 0;
                 if (rioWriteBulkObject(r,key) == 0) return 0;
             }
-            if (vstr) {
-                if (rioWriteBulkString(r,(char*)vstr,vlen) == 0) return 0;
+
+            if (entry.value) {
+                if (rioWriteBulkString(r,(char*)entry.value,entry.sz) == 0) return 0;
             } else {
-                if (rioWriteBulkLongLong(r,vlong) == 0) return 0;
+                if (rioWriteBulkLongLong(r,entry.longval) == 0) return 0;
             }
-            p = ziplistNext(zl,p);
-            if (++count == REDIS_AOF_REWRITE_ITEMS_PER_CMD) count = 0;
-            items--;
-        }
-    } else if (o->encoding == REDIS_ENCODING_LINKEDLIST) {
-        list *list = o->ptr;
-        listNode *ln;
-        listIter li;
-
-        listRewind(list,&li);
-        while((ln = listNext(&li))) {
-            robj *eleobj = listNodeValue(ln);
-
-            if (count == 0) {
-                int cmd_items = (items > REDIS_AOF_REWRITE_ITEMS_PER_CMD) ?
-                    REDIS_AOF_REWRITE_ITEMS_PER_CMD : items;
-
-                if (rioWriteBulkCount(r,'*',2+cmd_items) == 0) return 0;
-                if (rioWriteBulkString(r,"RPUSH",5) == 0) return 0;
-                if (rioWriteBulkObject(r,key) == 0) return 0;
-            }
-            if (rioWriteBulkObject(r,eleobj) == 0) return 0;
             if (++count == REDIS_AOF_REWRITE_ITEMS_PER_CMD) count = 0;
             items--;
         }
+        quicklistReleaseIterator(li);
     } else {
         redisPanic("Unknown list encoding");
     }
@@ -1105,6 +1082,7 @@ int rewriteAppendOnlyFile(char *filename) {
             }
         }
         dictReleaseIterator(di);
+        di = NULL;
     }
 
     /* Do an initial slow fsync here while the parent is still sending

src/bitops.c

@@ -70,16 +70,19 @@ size_t redisPopcount(void *s, long count) {
         count--;
     }
 
-    /* Count bits 16 bytes at a time */
+    /* Count bits 28 bytes at a time */
     p4 = (uint32_t*)p;
-    while(count>=16) {
-        uint32_t aux1, aux2, aux3, aux4;
+    while(count>=28) {
+        uint32_t aux1, aux2, aux3, aux4, aux5, aux6, aux7;
 
         aux1 = *p4++;
         aux2 = *p4++;
         aux3 = *p4++;
         aux4 = *p4++;
-        count -= 16;
+        aux5 = *p4++;
+        aux6 = *p4++;
+        aux7 = *p4++;
+        count -= 28;
 
         aux1 = aux1 - ((aux1 >> 1) & 0x55555555);
         aux1 = (aux1 & 0x33333333) + ((aux1 >> 2) & 0x33333333);
@@ -89,10 +92,19 @@ size_t redisPopcount(void *s, long count) {
         aux3 = (aux3 & 0x33333333) + ((aux3 >> 2) & 0x33333333);
         aux4 = aux4 - ((aux4 >> 1) & 0x55555555);
         aux4 = (aux4 & 0x33333333) + ((aux4 >> 2) & 0x33333333);
-        bits += ((((aux1 + (aux1 >> 4)) & 0x0F0F0F0F) * 0x01010101) >> 24) +
-                ((((aux2 + (aux2 >> 4)) & 0x0F0F0F0F) * 0x01010101) >> 24) +
-                ((((aux3 + (aux3 >> 4)) & 0x0F0F0F0F) * 0x01010101) >> 24) +
-                ((((aux4 + (aux4 >> 4)) & 0x0F0F0F0F) * 0x01010101) >> 24);
+        aux5 = aux5 - ((aux5 >> 1) & 0x55555555);
+        aux5 = (aux5 & 0x33333333) + ((aux5 >> 2) & 0x33333333);
+        aux6 = aux6 - ((aux6 >> 1) & 0x55555555);
+        aux6 = (aux6 & 0x33333333) + ((aux6 >> 2) & 0x33333333);
+        aux7 = aux7 - ((aux7 >> 1) & 0x55555555);
+        aux7 = (aux7 & 0x33333333) + ((aux7 >> 2) & 0x33333333);
+        bits += ((((aux1 + (aux1 >> 4)) & 0x0F0F0F0F) +
+                    ((aux2 + (aux2 >> 4)) & 0x0F0F0F0F) +
+                    ((aux3 + (aux3 >> 4)) & 0x0F0F0F0F) +
+                    ((aux4 + (aux4 >> 4)) & 0x0F0F0F0F) +
+                    ((aux5 + (aux5 >> 4)) & 0x0F0F0F0F) +
+                    ((aux6 + (aux6 >> 4)) & 0x0F0F0F0F) +
+                    ((aux7 + (aux7 >> 4)) & 0x0F0F0F0F))* 0x01010101) >> 24;
     }
     /* Count the remaining bytes. */
     p = (unsigned char*)p4;
@@ -348,7 +360,7 @@ void bitopCommand(redisClient *c) {
          * can take a fast path that performs much better than the
          * vanilla algorithm. */
         j = 0;
-        if (minlen && numkeys <= 16) {
+        if (minlen >= sizeof(unsigned long)*4 && numkeys <= 16) {
             unsigned long *lp[16];
             unsigned long *lres = (unsigned long*) res;
 

src/cluster.c

@@ -40,6 +40,7 @@
 #include <sys/socket.h>
 #include <sys/stat.h>
 #include <sys/file.h>
+#include <math.h>
 
 /* A global reference to myself is handy to make code more clear.
  * Myself always points to server.cluster->myself, that is, the clusterNode
@@ -178,7 +179,7 @@ int clusterLoadConfig(char *filename) {
             clusterAddNode(n);
         }
         /* Address and port */
-        if ((p = strchr(argv[1],':')) == NULL) goto fmterr;
+        if ((p = strrchr(argv[1],':')) == NULL) goto fmterr;
         *p = '\0';
         memcpy(n->ip,argv[1],strlen(argv[1])+1);
         n->port = atoi(p+1);
@@ -479,6 +480,7 @@ void clusterInit(void) {
      * the IP address via MEET messages. */
     myself->port = server.port;
 
+    server.cluster->mf_end = 0;
     resetManualFailover();
 }
 
@@ -593,7 +595,7 @@ void clusterAcceptHandler(aeEventLoop *el, int fd, void *privdata, int mask) {
         if (cfd == ANET_ERR) {
             if (errno != EWOULDBLOCK)
                 redisLog(REDIS_VERBOSE,
-                    "Accepting cluster node: %s", server.neterr);
+                    "Error accepting cluster node: %s", server.neterr);
             return;
         }
         anetNonBlock(NULL,cfd);
@@ -782,8 +784,11 @@ int clusterNodeRemoveSlave(clusterNode *master, clusterNode *slave) {
 
     for (j = 0; j < master->numslaves; j++) {
         if (master->slaves[j] == slave) {
-            memmove(master->slaves+j,master->slaves+(j+1),
-                (master->numslaves-1)-j);
+            if ((j+1) < master->numslaves) {
+                int remaining_slaves = (master->numslaves - j) - 1;
+                memmove(master->slaves+j,master->slaves+(j+1),
+                        (sizeof(*master->slaves) * remaining_slaves));
+            }
             master->numslaves--;
             return REDIS_OK;
         }
@@ -818,15 +823,30 @@ int clusterCountNonFailingSlaves(clusterNode *n) {
     return okslaves;
 }
 
+/* Low level cleanup of the node structure. Only called by clusterDelNode(). */
 void freeClusterNode(clusterNode *n) {
     sds nodename;
+    int j;
 
+    /* If the node is a master with associated slaves, we have to set
+     * all the slaves->slaveof fields to NULL (unknown). */
+    if (nodeIsMaster(n)) {
+        for (j = 0; j < n->numslaves; j++)
+            n->slaves[j]->slaveof = NULL;
+    }
+
+    /* Remove this node from the list of slaves of its master. */
+    if (nodeIsSlave(n) && n->slaveof) clusterNodeRemoveSlave(n->slaveof,n);
+
+    /* Unlink from the set of nodes. */
     nodename = sdsnewlen(n->name, REDIS_CLUSTER_NAMELEN);
     redisAssert(dictDelete(server.cluster->nodes,nodename) == DICT_OK);
     sdsfree(nodename);
-    if (n->slaveof) clusterNodeRemoveSlave(n->slaveof, n);
+
+    /* Release link and associated data structures. */
     if (n->link) freeClusterLink(n->link);
     listRelease(n->fail_reports);
+    zfree(n->slaves);
     zfree(n);
 }
 
@@ -839,11 +859,16 @@ int clusterAddNode(clusterNode *node) {
     return (retval == DICT_OK) ? REDIS_OK : REDIS_ERR;
 }
 
-/* Remove a node from the cluster:
- * 1) Mark all the nodes handled by it as unassigned.
- * 2) Remove all the failure reports sent by this node.
- * 3) Free the node, that will in turn remove it from the hash table
- *    and from the list of slaves of its master, if it is a slave node.
+/* Remove a node from the cluster. The functio performs the high level
+ * cleanup, calling freeClusterNode() for the low level cleanup.
+ * Here we do the following:
+ *
+ * 1) Mark all the slots handled by it as unassigned.
+ * 2) Remove all the failure reports sent by this node and referenced by
+ *    other nodes.
+ * 3) Free the node with freeClusterNode() that will in turn remove it
+ *    from the hash table and from the list of slaves of its master, if
+ *    it is a slave node.
  */
 void clusterDelNode(clusterNode *delnode) {
     int j;
@@ -870,11 +895,7 @@ void clusterDelNode(clusterNode *delnode) {
     }
     dictReleaseIterator(di);
 
-    /* 3) Remove this node from its master's slaves if needed. */
-    if (nodeIsSlave(delnode) && delnode->slaveof)
-        clusterNodeRemoveSlave(delnode->slaveof,delnode);
-
-    /* 4) Free the node, unlinking it from the cluster. */
+    /* 3) Free the node, unlinking it from the cluster. */
     freeClusterNode(delnode);
 }
 
@@ -1118,6 +1139,7 @@ int clusterStartHandshake(char *ip, int port) {
 
     /* Set norm_ip as the normalized string representation of the node
      * IP address. */
+    memset(norm_ip,0,REDIS_IP_STR_LEN);
     if (sa.ss_family == AF_INET)
         inet_ntop(AF_INET,
             (void*)&(((struct sockaddr_in *)&sa)->sin_addr),
@@ -1232,7 +1254,7 @@ void nodeIp2String(char *buf, clusterLink *link) {
  * The function returns 0 if the node address is still the same,
  * otherwise 1 is returned. */
 int nodeUpdateAddressIfNeeded(clusterNode *node, clusterLink *link, int port) {
-    char ip[REDIS_IP_STR_LEN];
+    char ip[REDIS_IP_STR_LEN] = {0};
 
     /* We don't proceed if the link is the same as the sender link, as this
      * function is designed to see if the node link is consistent with the
@@ -1463,7 +1485,8 @@ int clusterProcessPacket(clusterLink *link) {
 
     /* Perform sanity checks */
     if (totlen < 16) return 1; /* At least signature, version, totlen, count. */
-    if (ntohs(hdr->ver) != 0) return 1; /* Can't handle versions other than 0.*/
+    if (ntohs(hdr->ver) != CLUSTER_PROTO_VER)
+        return 1; /* Can't handle versions other than the current one.*/
     if (totlen > sdslen(link->rcvbuf)) return 1;
     if (type == CLUSTERMSG_TYPE_PING || type == CLUSTERMSG_TYPE_PONG ||
         type == CLUSTERMSG_TYPE_MEET)
@@ -1482,7 +1505,8 @@ int clusterProcessPacket(clusterLink *link) {
     } else if (type == CLUSTERMSG_TYPE_PUBLISH) {
         uint32_t explen = sizeof(clusterMsg)-sizeof(union clusterMsgData);
 
-        explen += sizeof(clusterMsgDataPublish) +
+        explen += sizeof(clusterMsgDataPublish) -
+                8 +
                 ntohl(hdr->data.publish.msg.channel_len) +
                 ntohl(hdr->data.publish.msg.message_len);
         if (totlen != explen) return 1;
@@ -1533,7 +1557,7 @@ int clusterProcessPacket(clusterLink *link) {
         }
     }
 
-    /* Process packets by type. */
+    /* Initial processing of PING and MEET requests replying with a PONG. */
     if (type == CLUSTERMSG_TYPE_PING || type == CLUSTERMSG_TYPE_MEET) {
         redisLog(REDIS_DEBUG,"Ping packet received: %p", (void*)link->node);
 
@@ -1543,8 +1567,12 @@ int clusterProcessPacket(clusterLink *link) {
          * later if we changed address, and those nodes will use our
          * official address to connect to us. So by obtaining this address
          * from the socket is a simple way to discover / update our own
-         * address in the cluster without it being hardcoded in the config. */
-        if (type == CLUSTERMSG_TYPE_MEET) {
+         * address in the cluster without it being hardcoded in the config.
+         *
+         * However if we don't have an address at all, we update the address
+         * even with a normal PING packet. If it's wrong it will be fixed
+         * by MEET later. */
+        if (type == CLUSTERMSG_TYPE_MEET || myself->ip[0] == '\0') {
             char ip[REDIS_IP_STR_LEN];
 
             if (anetSockName(link->fd,ip,sizeof(ip),NULL) != -1 &&
@@ -1571,14 +1599,17 @@ int clusterProcessPacket(clusterLink *link) {
             clusterDoBeforeSleep(CLUSTER_TODO_SAVE_CONFIG);
         }
 
-        /* Get info from the gossip section */
-        clusterProcessGossipSection(hdr,link);
+        /* If this is a MEET packet from an unknown node, we still process
+         * the gossip section here since we have to trust the sender because
+         * of the message type. */
+        if (!sender && type == CLUSTERMSG_TYPE_MEET)
+            clusterProcessGossipSection(hdr,link);
 
         /* Anyway reply with a PONG */
         clusterSendPing(link,CLUSTERMSG_TYPE_PONG);
     }
 
-    /* PING or PONG: process config information. */
+    /* PING, PONG, MEET: process config information. */
     if (type == CLUSTERMSG_TYPE_PING || type == CLUSTERMSG_TYPE_PONG ||
         type == CLUSTERMSG_TYPE_MEET)
     {
@@ -1600,7 +1631,7 @@ int clusterProcessPacket(clusterLink *link) {
                     }
                     /* Free this node as we already have it. This will
                      * cause the link to be freed as well. */
-                    freeClusterNode(link->node);
+                    clusterDelNode(link->node);
                     return 0;
                 }
 
@@ -1775,7 +1806,7 @@ int clusterProcessPacket(clusterLink *link) {
         }
 
         /* Get info from the gossip section */
-        clusterProcessGossipSection(hdr,link);
+        if (sender) clusterProcessGossipSection(hdr,link);
     } else if (type == CLUSTERMSG_TYPE_FAIL) {
         clusterNode *failing;
 
@@ -2007,7 +2038,8 @@ void clusterBroadcastMessage(void *buf, size_t len) {
     dictReleaseIterator(di);
 }
 
-/* Build the message header */
+/* Build the message header. hdr must point to a buffer at least
+ * sizeof(clusterMsg) in bytes. */
 void clusterBuildMessageHdr(clusterMsg *hdr, int type) {
     int totlen = 0;
     uint64_t offset;
@@ -2021,6 +2053,7 @@ void clusterBuildMessageHdr(clusterMsg *hdr, int type) {
               myself->slaveof : myself;
 
     memset(hdr,0,sizeof(*hdr));
+    hdr->ver = htons(CLUSTER_PROTO_VER);
     hdr->sig[0] = 'R';
     hdr->sig[1] = 'C';
     hdr->sig[2] = 'm';
@@ -2067,40 +2100,90 @@ void clusterBuildMessageHdr(clusterMsg *hdr, int type) {
 /* Send a PING or PONG packet to the specified node, making sure to add enough
  * gossip informations. */
 void clusterSendPing(clusterLink *link, int type) {
-    unsigned char buf[sizeof(clusterMsg)];
-    clusterMsg *hdr = (clusterMsg*) buf;
-    int gossipcount = 0, totlen;
-    /* freshnodes is the number of nodes we can still use to populate the
-     * gossip section of the ping packet. Basically we start with the nodes
-     * we have in memory minus two (ourself and the node we are sending the
-     * message to). Every time we add a node we decrement the counter, so when
-     * it will drop to <= zero we know there is no more gossip info we can
-     * send. */
+    unsigned char *buf;
+    clusterMsg *hdr;
+    int gossipcount = 0; /* Number of gossip sections added so far. */
+    int wanted; /* Number of gossip sections we want to append if possible. */
+    int totlen; /* Total packet length. */
+    /* freshnodes is the max number of nodes we can hope to append at all:
+     * nodes available minus two (ourself and the node we are sending the
+     * message to). However practically there may be less valid nodes since
+     * nodes in handshake state, disconnected, are not considered. */
     int freshnodes = dictSize(server.cluster->nodes)-2;
 
+    /* How many gossip sections we want to add? 1/10 of the number of nodes
+     * and anyway at least 3. Why 1/10?
+     *
+     * If we have N masters, with N/10 entries, and we consider that in
+     * node_timeout we exchange with each other node at least 4 packets
+     * (we ping in the worst case in node_timeout/2 time, and we also
+     * receive two pings from the host), we have a total of 8 packets
+     * in the node_timeout*2 falure reports validity time. So we have
+     * that, for a single PFAIL node, we can expect to receive the following
+     * number of failure reports (in the specified window of time):
+     *
+     * PROB * GOSSIP_ENTRIES_PER_PACKET * TOTAL_PACKETS:
+     *
+     * PROB = probability of being featured in a single gossip entry,
+     *        which is 1 / NUM_OF_NODES.
+     * ENTRIES = 10.
+     * TOTAL_PACKETS = 2 * 4 * NUM_OF_MASTERS.
+     *
+     * If we assume we have just masters (so num of nodes and num of masters
+     * is the same), with 1/10 we always get over the majority, and specifically
+     * 80% of the number of nodes, to account for many masters failing at the
+     * same time.
+     *
+     * Since we have non-voting slaves that lower the probability of an entry
+     * to feature our node, we set the number of entires per packet as
+     * 10% of the total nodes we have. */
+    wanted = floor(dictSize(server.cluster->nodes)/10);
+    if (wanted < 3) wanted = 3;
+    if (wanted > freshnodes) wanted = freshnodes;
+
+    /* Compute the maxium totlen to allocate our buffer. We'll fix the totlen
+     * later according to the number of gossip sections we really were able
+     * to put inside the packet. */
+    totlen = sizeof(clusterMsg)-sizeof(union clusterMsgData);
+    totlen += (sizeof(clusterMsgDataGossip)*wanted);
+    /* Note: clusterBuildMessageHdr() expects the buffer to be always at least
+     * sizeof(clusterMsg) or more. */
+    if (totlen < (int)sizeof(clusterMsg)) totlen = sizeof(clusterMsg);
+    buf = zcalloc(totlen);
+    hdr = (clusterMsg*) buf;
+
+    /* Populate the header. */
     if (link->node && type == CLUSTERMSG_TYPE_PING)
         link->node->ping_sent = mstime();
     clusterBuildMessageHdr(hdr,type);
 
     /* Populate the gossip fields */
-    while(freshnodes > 0 && gossipcount < 3) {
+    int maxiterations = wanted*3;
+    while(freshnodes > 0 && gossipcount < wanted && maxiterations--) {
         dictEntry *de = dictGetRandomKey(server.cluster->nodes);
         clusterNode *this = dictGetVal(de);
         clusterMsgDataGossip *gossip;
         int j;
 
+        /* Don't include this node: the whole packet header is about us
+         * already, so we just gossip about other nodes. */
+        if (this == myself) continue;
+
+        /* Give a bias to FAIL/PFAIL nodes. */
+        if (maxiterations > wanted*2 &&
+            !(this->flags & (REDIS_NODE_PFAIL|REDIS_NODE_FAIL)))
+            continue;
+
         /* In the gossip section don't include:
-         * 1) Myself.
-         * 2) Nodes in HANDSHAKE state.
+         * 1) Nodes in HANDSHAKE state.
          * 3) Nodes with the NOADDR flag set.
          * 4) Disconnected nodes if they don't have configured slots.
          */
-        if (this == myself ||
-            this->flags & (REDIS_NODE_HANDSHAKE|REDIS_NODE_NOADDR) ||
+        if (this->flags & (REDIS_NODE_HANDSHAKE|REDIS_NODE_NOADDR) ||
             (this->link == NULL && this->numslots == 0))
         {
-                freshnodes--; /* otherwise we may loop forever. */
-                continue;
+            freshnodes--; /* Tecnically not correct, but saves CPU. */
+            continue;
         }
 
         /* Check if we already added this node */
@@ -2119,13 +2202,19 @@ void clusterSendPing(clusterLink *link, int type) {
         memcpy(gossip->ip,this->ip,sizeof(this->ip));
         gossip->port = htons(this->port);
         gossip->flags = htons(this->flags);
+        gossip->notused1 = 0;
+        gossip->notused2 = 0;
         gossipcount++;
     }
+
+    /* Ready to send... fix the totlen fiend and queue the message in the
+     * output buffer. */
     totlen = sizeof(clusterMsg)-sizeof(union clusterMsgData);
     totlen += (sizeof(clusterMsgDataGossip)*gossipcount);
     hdr->count = htons(gossipcount);
     hdr->totlen = htonl(totlen);
     clusterSendMessage(link,buf,totlen);
+    zfree(buf);
 }
 
 /* Send a PONG packet to every connected node that's not in handshake state
@@ -2181,7 +2270,7 @@ void clusterSendPublish(clusterLink *link, robj *channel, robj *message) {
 
     clusterBuildMessageHdr(hdr,CLUSTERMSG_TYPE_PUBLISH);
     totlen = sizeof(clusterMsg)-sizeof(union clusterMsgData);
-    totlen += sizeof(clusterMsgDataPublish) + channel_len + message_len;
+    totlen += sizeof(clusterMsgDataPublish) - 8 + channel_len + message_len;
 
     hdr->data.publish.msg.channel_len = htonl(channel_len);
     hdr->data.publish.msg.message_len = htonl(message_len);
@@ -2514,7 +2603,7 @@ void clusterHandleSlaveFailover(void) {
 
     /* Compute the failover timeout (the max time we have to send votes
      * and wait for replies), and the failover retry time (the time to wait
-     * before waiting again.
+     * before trying to get voted again).
      *
      * Timeout is MIN(NODE_TIMEOUT*2,2000) milliseconds.
      * Retry is two times the Timeout.
@@ -2772,6 +2861,7 @@ void clusterHandleSlaveMigration(int max_slaves) {
             }
         }
     }
+    dictReleaseIterator(di);
 
     /* Step 4: perform the migration if there is a target, and if I'm the
      * candidate. */
@@ -2893,7 +2983,7 @@ void clusterCron(void) {
         /* A Node in HANDSHAKE state has a limited lifespan equal to the
          * configured node timeout. */
         if (nodeInHandshake(node) && now - node->ctime > handshake_timeout) {
-            freeClusterNode(node);
+            clusterDelNode(node);
             continue;
         }
 
@@ -3235,7 +3325,7 @@ void clusterCloseAllSlots(void) {
 
 void clusterUpdateState(void) {
     int j, new_state;
-    int unreachable_masters = 0;
+    int reachable_masters = 0;
     static mstime_t among_minority_time;
     static mstime_t first_call_time = 0;
 
@@ -3271,8 +3361,8 @@ void clusterUpdateState(void) {
     /* Compute the cluster size, that is the number of master nodes
      * serving at least a single slot.
      *
-     * At the same time count the number of unreachable masters with
-     * at least one node. */
+     * At the same time count the number of reachable masters having
+     * at least one slot. */
     {
         dictIterator *di;
         dictEntry *de;
@@ -3284,20 +3374,19 @@ void clusterUpdateState(void) {
 
             if (nodeIsMaster(node) && node->numslots) {
                 server.cluster->size++;
-                if (node->flags & (REDIS_NODE_FAIL|REDIS_NODE_PFAIL))
-                    unreachable_masters++;
+                if ((node->flags & (REDIS_NODE_FAIL|REDIS_NODE_PFAIL)) == 0)
+                    reachable_masters++;
             }
         }
         dictReleaseIterator(di);
     }
 
-    /* If we can't reach at least half the masters, change the cluster state
-     * to FAIL, as we are not even able to mark nodes as FAIL in this side
-     * of the netsplit because of lack of majority. */
+    /* If we are in a minority partition, change the cluster state
+     * to FAIL. */
     {
         int needed_quorum = (server.cluster->size / 2) + 1;
 
-        if (unreachable_masters >= needed_quorum) {
+        if (reachable_masters < needed_quorum) {
             new_state = REDIS_CLUSTER_FAIL;
             among_minority_time = mstime();
         }
@@ -3881,10 +3970,7 @@ void clusterCommand(redisClient *c) {
             server.cluster->stats_bus_messages_sent,
             server.cluster->stats_bus_messages_received
         );
-        addReplySds(c,sdscatprintf(sdsempty(),"$%lu\r\n",
-            (unsigned long)sdslen(info)));
-        addReplySds(c,info);
-        addReply(c,shared.crlf);
+        addReplyBulkSds(c, info);
     } else if (!strcasecmp(c->argv[1]->ptr,"saveconfig") && c->argc == 2) {
         int retval = clusterSaveConfig(1);
 
@@ -4008,6 +4094,18 @@ void clusterCommand(redisClient *c) {
             addReplyBulkCString(c,ni);
             sdsfree(ni);
         }
+    } else if (!strcasecmp(c->argv[1]->ptr,"count-failure-reports") &&
+               c->argc == 3)
+    {
+        /* CLUSTER COUNT-FAILURE-REPORTS <NODE ID> */
+        clusterNode *n = clusterLookupNode(c->argv[2]->ptr);
+
+        if (!n) {
+            addReplyErrorFormat(c,"Unknown node %s", (char*)c->argv[2]->ptr);
+            return;
+        } else {
+            addReplyLongLong(c,clusterNodeFailureReportsCount(n));
+        }
     } else if (!strcasecmp(c->argv[1]->ptr,"failover") &&
                (c->argc == 2 || c->argc == 3))
     {
@@ -4305,8 +4403,8 @@ int migrateGetSocket(redisClient *c, robj *host, robj *port, long timeout) {
     }
 
     /* Create the socket */
-    fd = anetTcpNonBlockConnect(server.neterr,c->argv[1]->ptr,
-                atoi(c->argv[2]->ptr));
+    fd = anetTcpNonBlockBindConnect(server.neterr,c->argv[1]->ptr,
+                atoi(c->argv[2]->ptr),REDIS_BIND_ADDR);
     if (fd == -1) {
         sdsfree(name);
         addReplyErrorFormat(c,"Can't connect to target node: %s",
@@ -4406,7 +4504,7 @@ try_again:
     /* Check if the key is here. If not we reply with success as there is
      * nothing to migrate (for instance the key expired in the meantime), but
      * we include such information in the reply string. */
-    if ((o = lookupKeyRead(c->db,c->argv[3])) == NULL) {
+    if ((o = lookupKeyWrite(c->db,c->argv[3])) == NULL) {
         addReplySds(c,sdsnew("+NOKEY\r\n"));
         return;
     }

src/cluster.h

@@ -163,10 +163,11 @@ typedef struct {
     char nodename[REDIS_CLUSTER_NAMELEN];
     uint32_t ping_sent;
     uint32_t pong_received;
-    char ip[REDIS_IP_STR_LEN];    /* IP address last time it was seen */
-    uint16_t port;  /* port last time it was seen */
-    uint16_t flags;
-    uint32_t notused; /* for 64 bit alignment */
+    char ip[REDIS_IP_STR_LEN];  /* IP address last time it was seen */
+    uint16_t port;              /* port last time it was seen */
+    uint16_t flags;             /* node->flags copy */
+    uint16_t notused1;          /* Some room for future improvements. */
+    uint32_t notused2;
 } clusterMsgDataGossip;
 
 typedef struct {
@@ -176,7 +177,10 @@ typedef struct {
 typedef struct {
     uint32_t channel_len;
     uint32_t message_len;
-    unsigned char bulk_data[8]; /* defined as 8 just for alignment concerns. */
+    /* We can't reclare bulk_data as bulk_data[] since this structure is
+     * nested. The 8 bytes are removed from the count during the message
+     * length computation. */
+    unsigned char bulk_data[8];
 } clusterMsgDataPublish;
 
 typedef struct {
@@ -208,6 +212,7 @@ union clusterMsgData {
     } update;
 };
 
+#define CLUSTER_PROTO_VER 0 /* Cluster bus protocol version. */
 
 typedef struct {
     char sig[4];        /* Siganture "RCmb" (Redis Cluster message bus). */

src/config.c

@@ -60,6 +60,8 @@ clientBufferLimitsConfig clientBufferLimitsDefaults[REDIS_CLIENT_TYPE_COUNT] = {
  * Config file parsing
  *----------------------------------------------------------------------------*/
 
+int supervisedToMode(const char *str);
+
 int yesnotoi(char *s) {
     if (!strcasecmp(s,"yes")) return 1;
     else if (!strcasecmp(s,"no")) return 0;
@@ -397,9 +399,13 @@ void loadServerConfigFromString(char *config) {
         } else if (!strcasecmp(argv[0],"hash-max-ziplist-value") && argc == 2) {
             server.hash_max_ziplist_value = memtoll(argv[1], NULL);
         } else if (!strcasecmp(argv[0],"list-max-ziplist-entries") && argc == 2){
-            server.list_max_ziplist_entries = memtoll(argv[1], NULL);
+            /* DEAD OPTION */
         } else if (!strcasecmp(argv[0],"list-max-ziplist-value") && argc == 2) {
-            server.list_max_ziplist_value = memtoll(argv[1], NULL);
+            /* DEAD OPTION */
+        } else if (!strcasecmp(argv[0],"list-max-ziplist-size") && argc == 2) {
+            server.list_max_ziplist_size = atoi(argv[1]);
+        } else if (!strcasecmp(argv[0],"list-compress-depth") && argc == 2) {
+            server.list_compress_depth = atoi(argv[1]);
         } else if (!strcasecmp(argv[0],"set-max-intset-entries") && argc == 2) {
             server.set_max_intset_entries = memtoll(argv[1], NULL);
         } else if (!strcasecmp(argv[0],"zset-max-ziplist-entries") && argc == 2) {
@@ -529,6 +535,15 @@ void loadServerConfigFromString(char *config) {
                 goto loaderr;
             }
             server.notify_keyspace_events = flags;
+        } else if (!strcasecmp(argv[0],"supervised") && argc == 2) {
+            int mode = supervisedToMode(argv[1]);
+
+            if (mode == -1) {
+                err = "Invalid option for 'supervised'. "
+                    "Allowed values: 'upstart', 'systemd', 'auto', or 'no'";
+                goto loaderr;
+            }
+            server.supervised_mode = mode;
         } else if (!strcasecmp(argv[0],"sentinel")) {
             /* argc == 1 is handled by main() as we need to enter the sentinel
              * mode ASAP. */
@@ -795,12 +810,12 @@ void configSetCommand(redisClient *c) {
     } else if (!strcasecmp(c->argv[2]->ptr,"hash-max-ziplist-value")) {
         if (getLongLongFromObject(o,&ll) == REDIS_ERR || ll < 0) goto badfmt;
         server.hash_max_ziplist_value = ll;
-    } else if (!strcasecmp(c->argv[2]->ptr,"list-max-ziplist-entries")) {
+    } else if (!strcasecmp(c->argv[2]->ptr,"list-max-ziplist-size")) {
         if (getLongLongFromObject(o,&ll) == REDIS_ERR || ll < 0) goto badfmt;
-        server.list_max_ziplist_entries = ll;
-    } else if (!strcasecmp(c->argv[2]->ptr,"list-max-ziplist-value")) {
+        server.list_max_ziplist_size = ll;
+    } else if (!strcasecmp(c->argv[2]->ptr,"list-compress-depth")) {
         if (getLongLongFromObject(o,&ll) == REDIS_ERR || ll < 0) goto badfmt;
-        server.list_max_ziplist_value = ll;
+        server.list_compress_depth = ll;
     } else if (!strcasecmp(c->argv[2]->ptr,"set-max-intset-entries")) {
         if (getLongLongFromObject(o,&ll) == REDIS_ERR || ll < 0) goto badfmt;
         server.set_max_intset_entries = ll;
@@ -1004,6 +1019,47 @@ badfmt: /* Bad format errors */
     } \
 } while(0);
 
+char *maxmemoryToString() {
+    char *s;
+    switch(server.maxmemory_policy) {
+    case REDIS_MAXMEMORY_VOLATILE_LRU: s = "volatile-lru"; break;
+    case REDIS_MAXMEMORY_VOLATILE_TTL: s = "volatile-ttl"; break;
+    case REDIS_MAXMEMORY_VOLATILE_RANDOM: s = "volatile-random"; break;
+    case REDIS_MAXMEMORY_ALLKEYS_LRU: s = "allkeys-lru"; break;
+    case REDIS_MAXMEMORY_ALLKEYS_RANDOM: s = "allkeys-random"; break;
+    case REDIS_MAXMEMORY_NO_EVICTION: s = "noeviction"; break;
+    default: s = "unknown"; break;
+    }
+    return s;
+}
+
+int supervisedToMode(const char *str) {
+    int mode;
+    if (!strcasecmp(str,"upstart")) {
+        mode = REDIS_SUPERVISED_UPSTART;
+    } else if (!strcasecmp(str,"systemd")) {
+        mode = REDIS_SUPERVISED_SYSTEMD;
+    } else if (!strcasecmp(str,"auto")) {
+        mode = REDIS_SUPERVISED_AUTODETECT;
+    } else if (!strcasecmp(str,"no")) {
+        mode = REDIS_SUPERVISED_NONE;
+    } else {
+        mode = -1;
+    }
+    return mode;
+}
+
+char *supervisedToString(void) {
+    char *s;
+    switch(server.supervised_mode) {
+    case REDIS_SUPERVISED_UPSTART: s = "upstart"; break;
+    case REDIS_SUPERVISED_SYSTEMD: s = "systemd"; break;
+    case REDIS_SUPERVISED_AUTODETECT: s = "auto"; break;
+    case REDIS_SUPERVISED_NONE: s = "no"; break;
+    default: s = "no"; break;
+    }
+    return s;
+}
 void configGetCommand(redisClient *c) {
     robj *o = c->argv[2];
     void *replylen = addDeferredMultiBulkLength(c);
@@ -1033,10 +1089,10 @@ void configGetCommand(redisClient *c) {
             server.hash_max_ziplist_entries);
     config_get_numerical_field("hash-max-ziplist-value",
             server.hash_max_ziplist_value);
-    config_get_numerical_field("list-max-ziplist-entries",
-            server.list_max_ziplist_entries);
-    config_get_numerical_field("list-max-ziplist-value",
-            server.list_max_ziplist_value);
+    config_get_numerical_field("list-max-ziplist-size",
+            server.list_max_ziplist_size);
+    config_get_numerical_field("list-compress-depth",
+            server.list_compress_depth);
     config_get_numerical_field("set-max-intset-entries",
             server.set_max_intset_entries);
     config_get_numerical_field("zset-max-ziplist-entries",
@@ -1112,19 +1168,8 @@ void configGetCommand(redisClient *c) {
         matches++;
     }
     if (stringmatch(pattern,"maxmemory-policy",0)) {
-        char *s;
-
-        switch(server.maxmemory_policy) {
-        case REDIS_MAXMEMORY_VOLATILE_LRU: s = "volatile-lru"; break;
-        case REDIS_MAXMEMORY_VOLATILE_TTL: s = "volatile-ttl"; break;
-        case REDIS_MAXMEMORY_VOLATILE_RANDOM: s = "volatile-random"; break;
-        case REDIS_MAXMEMORY_ALLKEYS_LRU: s = "allkeys-lru"; break;
-        case REDIS_MAXMEMORY_ALLKEYS_RANDOM: s = "allkeys-random"; break;
-        case REDIS_MAXMEMORY_NO_EVICTION: s = "noeviction"; break;
-        default: s = "unknown"; break; /* too harmless to panic */
-        }
         addReplyBulkCString(c,"maxmemory-policy");
-        addReplyBulkCString(c,s);
+        addReplyBulkCString(c,maxmemoryToString());
         matches++;
     }
     if (stringmatch(pattern,"appendfsync",0)) {
@@ -1170,6 +1215,11 @@ void configGetCommand(redisClient *c) {
         addReplyBulkCString(c,s);
         matches++;
     }
+    if (stringmatch(pattern,"supervised",0)) {
+        addReplyBulkCString(c,"supervised");
+        addReplyBulkCString(c,supervisedToString());
+        matches++;
+    }
     if (stringmatch(pattern,"client-output-buffer-limit",0)) {
         sds buf = sdsempty();
         int j;
@@ -1854,8 +1904,8 @@ int rewriteConfig(char *path) {
     rewriteConfigNotifykeyspaceeventsOption(state);
     rewriteConfigNumericalOption(state,"hash-max-ziplist-entries",server.hash_max_ziplist_entries,REDIS_HASH_MAX_ZIPLIST_ENTRIES);
     rewriteConfigNumericalOption(state,"hash-max-ziplist-value",server.hash_max_ziplist_value,REDIS_HASH_MAX_ZIPLIST_VALUE);
-    rewriteConfigNumericalOption(state,"list-max-ziplist-entries",server.list_max_ziplist_entries,REDIS_LIST_MAX_ZIPLIST_ENTRIES);
-    rewriteConfigNumericalOption(state,"list-max-ziplist-value",server.list_max_ziplist_value,REDIS_LIST_MAX_ZIPLIST_VALUE);
+    rewriteConfigNumericalOption(state,"list-max-ziplist-size",server.list_max_ziplist_size,REDIS_LIST_MAX_ZIPLIST_SIZE);
+    rewriteConfigNumericalOption(state,"list-compress-depth",server.list_compress_depth,REDIS_LIST_COMPRESS_DEPTH);
     rewriteConfigNumericalOption(state,"set-max-intset-entries",server.set_max_intset_entries,REDIS_SET_MAX_INTSET_ENTRIES);
     rewriteConfigNumericalOption(state,"zset-max-ziplist-entries",server.zset_max_ziplist_entries,REDIS_ZSET_MAX_ZIPLIST_ENTRIES);
     rewriteConfigNumericalOption(state,"zset-max-ziplist-value",server.zset_max_ziplist_value,REDIS_ZSET_MAX_ZIPLIST_VALUE);
@@ -1865,6 +1915,12 @@ int rewriteConfig(char *path) {
     rewriteConfigNumericalOption(state,"hz",server.hz,REDIS_DEFAULT_HZ);
     rewriteConfigYesNoOption(state,"aof-rewrite-incremental-fsync",server.aof_rewrite_incremental_fsync,REDIS_DEFAULT_AOF_REWRITE_INCREMENTAL_FSYNC);
     rewriteConfigYesNoOption(state,"aof-load-truncated",server.aof_load_truncated,REDIS_DEFAULT_AOF_LOAD_TRUNCATED);
+    rewriteConfigEnumOption(state,"supervised",server.supervised_mode,
+        "upstart", REDIS_SUPERVISED_UPSTART,
+        "systemd", REDIS_SUPERVISED_SYSTEMD,
+        "auto", REDIS_SUPERVISED_AUTODETECT,
+        "no", REDIS_SUPERVISED_NONE,
+        NULL, REDIS_SUPERVISED_NONE);
     if (server.sentinel_mode) rewriteConfigSentinelOption(state);
 
     /* Step 3: remove all the orphaned lines in the old file, that is, lines

src/config.h

@@ -48,6 +48,7 @@
 #define HAVE_PROC_STAT 1
 #define HAVE_PROC_MAPS 1
 #define HAVE_PROC_SMAPS 1
+#define HAVE_PROC_SOMAXCONN 1
 #endif
 
 /* Test for task_info() */
@@ -56,10 +57,15 @@
 #endif
 
 /* Test for backtrace() */
-#if defined(__APPLE__) || defined(__linux__)
+#if defined(__APPLE__) || (defined(__linux__) && defined(__GLIBC__))
 #define HAVE_BACKTRACE 1
 #endif
 
+/* MSG_NOSIGNAL. */
+#ifdef __linux__
+#define HAVE_MSG_NOSIGNAL 1
+#endif
+
 /* Test for polling API */
 #ifdef __linux__
 #define HAVE_EPOLL 1
@@ -112,7 +118,7 @@
 #define USE_SETPROCTITLE
 #endif
 
-#if (defined __linux || defined __APPLE__)
+#if ((defined __linux && defined(__GLIBC__)) || defined __APPLE__)
 #define USE_SETPROCTITLE
 #define INIT_SETPROCTITLE_REPLACEMENT
 void spt_init(int argc, char *argv[]);

src/crc64.c

@@ -181,9 +181,13 @@ uint64_t crc64(uint64_t crc, const unsigned char *s, uint64_t l) {
 }
 
 /* Test main */
-#ifdef TEST_MAIN
+#ifdef REDIS_TEST
 #include <stdio.h>
-int main(void) {
+
+#define UNUSED(x) (void)(x)
+int crc64Test(int argc, char *argv[]) {
+    UNUSED(argc);
+    UNUSED(argv);
     printf("e9c6d914c4b8d9ca == %016llx\n",
         (unsigned long long) crc64(0,(unsigned char*)"123456789",9));
     return 0;

src/crc64.h

@@ -5,4 +5,8 @@
 
 uint64_t crc64(uint64_t crc, const unsigned char *s, uint64_t l);
 
+#ifdef REDIS_TEST
+int crc64Test(int argc, char *argv[]);
+#endif
+
 #endif

src/db.c

@@ -60,7 +60,32 @@ robj *lookupKey(redisDb *db, robj *key) {
 robj *lookupKeyRead(redisDb *db, robj *key) {
     robj *val;
 
-    expireIfNeeded(db,key);
+    if (expireIfNeeded(db,key) == 1) {
+        /* Key expired. If we are in the context of a master, expireIfNeeded()
+         * returns 0 only when the key does not exist at all, so it's save
+         * to return NULL ASAP. */
+        if (server.masterhost == NULL) return NULL;
+
+        /* However if we are in the context of a slave, expireIfNeeded() will
+         * not really try to expire the key, it only returns information
+         * about the "logical" status of the key: key expiring is up to the
+         * master in order to have a consistent view of master's data set.
+         *
+         * However, if the command caller is not the master, and as additional
+         * safety measure, the command invoked is a read-only command, we can
+         * safely return NULL here, and provide a more consistent behavior
+         * to clients accessign expired values in a read-only fashion, that
+         * will say the key as non exisitng.
+         *
+         * Notably this covers GETs when slaves are used to scale reads. */
+        if (server.current_client &&
+            server.current_client != server.master &&
+            server.current_client->cmd &&
+            server.current_client->cmd->flags & REDIS_CMD_READONLY)
+        {
+            return NULL;
+        }
+    }
     val = lookupKey(db,key);
     if (val == NULL)
         server.stat_keyspace_misses++;
@@ -381,7 +406,7 @@ void scanCallback(void *privdata, const dictEntry *de) {
     } else if (o->type == REDIS_ZSET) {
         key = dictGetKey(de);
         incrRefCount(key);
-        val = createStringObjectFromLongDouble(*(double*)dictGetVal(de));
+        val = createStringObjectFromLongDouble(*(double*)dictGetVal(de),0);
     } else {
         redisPanic("Type not handled in SCAN callback.");
     }
@@ -425,8 +450,8 @@ void scanGenericCommand(redisClient *c, robj *o, unsigned long cursor) {
     list *keys = listCreate();
     listNode *node, *nextnode;
     long count = 10;
-    sds pat;
-    int patlen, use_pattern = 0;
+    sds pat = NULL;
+    int patlen = 0, use_pattern = 0;
     dict *ht;
 
     /* Object must be NULL (to iterate keys names), or the type of the object
@@ -877,7 +902,7 @@ void expireGenericCommand(redisClient *c, long long basetime, int unit) {
     when += basetime;
 
     /* No key, return zero. */
-    if (lookupKeyRead(c->db,key) == NULL) {
+    if (lookupKeyWrite(c->db,key) == NULL) {
         addReply(c,shared.czero);
         return;
     }

src/debug.c

@@ -252,6 +252,12 @@ void computeDatasetDigest(unsigned char *final) {
     }
 }
 
+void inputCatSds(void *result, const char *str) {
+    /* result is actually a (sds *), so re-cast it here */
+    sds *info = (sds *)result;
+    *info = sdscat(*info, str);
+}
+
 void debugCommand(redisClient *c) {
     if (!strcasecmp(c->argv[1]->ptr,"segfault")) {
         *((char*)-1) = 'x';
@@ -295,13 +301,46 @@ void debugCommand(redisClient *c) {
         val = dictGetVal(de);
         strenc = strEncoding(val->encoding);
 
+        char extra[128] = {0};
+        if (val->encoding == REDIS_ENCODING_QUICKLIST) {
+            char *nextra = extra;
+            int remaining = sizeof(extra);
+            quicklist *ql = val->ptr;
+            /* Add number of quicklist nodes */
+            int used = snprintf(nextra, remaining, " ql_nodes:%u", ql->len);
+            nextra += used;
+            remaining -= used;
+            /* Add average quicklist fill factor */
+            double avg = (double)ql->count/ql->len;
+            used = snprintf(nextra, remaining, " ql_avg_node:%.2f", avg);
+            nextra += used;
+            remaining -= used;
+            /* Add quicklist fill level / max ziplist size */
+            used = snprintf(nextra, remaining, " ql_ziplist_max:%d", ql->fill);
+            nextra += used;
+            remaining -= used;
+            /* Add isCompressed? */
+            int compressed = ql->compress != 0;
+            used = snprintf(nextra, remaining, " ql_compressed:%d", compressed);
+            nextra += used;
+            remaining -= used;
+            /* Add total uncompressed size */
+            unsigned long sz = 0;
+            for (quicklistNode *node = ql->head; node; node = node->next) {
+                sz += node->sz;
+            }
+            used = snprintf(nextra, remaining, " ql_uncompressed_size:%lu", sz);
+            nextra += used;
+            remaining -= used;
+        }
+
         addReplyStatusFormat(c,
             "Value at:%p refcount:%d "
             "encoding:%s serializedlength:%lld "
-            "lru:%d lru_seconds_idle:%llu",
+            "lru:%d lru_seconds_idle:%llu%s",
             (void*)val, val->refcount,
             strenc, (long long) rdbSavedObjectLen(val),
-            val->lru, estimateObjectIdleTime(val));
+            val->lru, estimateObjectIdleTime(val)/1000, extra);
     } else if (!strcasecmp(c->argv[1]->ptr,"sdslen") && c->argc == 3) {
         dictEntry *de;
         robj *val;
@@ -336,9 +375,9 @@ void debugCommand(redisClient *c) {
         dictExpand(c->db->dict,keys);
         for (j = 0; j < keys; j++) {
             snprintf(buf,sizeof(buf),"%s:%lu",
-                (c->argc == 3) ? "key" : c->argv[3]->ptr, j);
+                (c->argc == 3) ? "key" : (char*)c->argv[3]->ptr, j);
             key = createStringObject(buf,strlen(buf));
-            if (lookupKeyRead(c->db,key) != NULL) {
+            if (lookupKeyWrite(c->db,key) != NULL) {
                 decrRefCount(key);
                 continue;
             }
@@ -379,6 +418,25 @@ void debugCommand(redisClient *c) {
         errstr = sdsmapchars(errstr,"\n\r","  ",2); /* no newlines in errors. */
         errstr = sdscatlen(errstr,"\r\n",2);
         addReplySds(c,errstr);
+    } else if (!strcasecmp(c->argv[1]->ptr,"structsize") && c->argc == 2) {
+        sds sizes = sdsempty();
+        sizes = sdscatprintf(sizes,"bits:%d ", (sizeof(void*) == 8)?64:32);
+        sizes = sdscatprintf(sizes,"robj:%d ", (int)sizeof(robj));
+        sizes = sdscatprintf(sizes,"dictentry:%d ", (int)sizeof(dictEntry));
+        sizes = sdscatprintf(sizes,"sdshdr:%d", (int)sizeof(struct sdshdr));
+        addReplyBulkSds(c,sizes);
+    } else if (!strcasecmp(c->argv[1]->ptr,"jemalloc") && c->argc == 3) {
+#if defined(USE_JEMALLOC)
+        if (!strcasecmp(c->argv[2]->ptr, "info")) {
+            sds info = sdsempty();
+            je_malloc_stats_print(inputCatSds, &info, NULL);
+            addReplyBulkSds(c, info);
+        } else {
+            addReplyErrorFormat(c, "Valid jemalloc debug fields: info");
+        }
+#else
+        addReplyErrorFormat(c, "jemalloc support not available");
+#endif
     } else {
         addReplyErrorFormat(c, "Unknown DEBUG subcommand or wrong number of arguments for '%s'",
             (char*)c->argv[1]->ptr);
@@ -857,7 +915,7 @@ void sigsegvHandler(int sig, siginfo_t *info, void *secret) {
 "  Suspect RAM error? Use redis-server --test-memory to verify it.\n\n"
 );
     /* free(messages); Don't call free() with possibly corrupted memory. */
-    if (server.daemonize) unlink(server.pidfile);
+    if (server.daemonize && server.supervised == 0) unlink(server.pidfile);
 
     /* Make sure we exit with the right signal at the end. So for instance
      * the core will be dumped if enabled. */

src/dict.c

@@ -211,6 +211,9 @@ int dictExpand(dict *d, unsigned long size)
     if (dictIsRehashing(d) || d->ht[0].used > size)
         return DICT_ERR;
 
+    /* Rehashing to the same table size is not useful. */
+    if (realsize == d->ht[0].size) return DICT_ERR;
+
     /* Allocate the new hash table and initialize all pointers to NULL */
     n.size = realsize;
     n.sizemask = realsize-1;
@@ -232,27 +235,27 @@ int dictExpand(dict *d, unsigned long size)
 
 /* Performs N steps of incremental rehashing. Returns 1 if there are still
  * keys to move from the old to the new hash table, otherwise 0 is returned.
+ *
  * Note that a rehashing step consists in moving a bucket (that may have more
- * than one key as we use chaining) from the old to the new hash table. */
+ * than one key as we use chaining) from the old to the new hash table, however
+ * since part of the hash table may be composed of empty spaces, it is not
+ * guaranteed that this function will rehash even a single bucket, since it
+ * will visit at max N*10 empty buckets in total, otherwise the amount of
+ * work it does would be unbound and the function may block for a long time. */
 int dictRehash(dict *d, int n) {
+    int empty_visits = n*10; /* Max number of empty buckets to visit. */
     if (!dictIsRehashing(d)) return 0;
 
-    while(n--) {
+    while(n-- && d->ht[0].used != 0) {
         dictEntry *de, *nextde;
 
-        /* Check if we already rehashed the whole table... */
-        if (d->ht[0].used == 0) {
-            zfree(d->ht[0].table);
-            d->ht[0] = d->ht[1];
-            _dictReset(&d->ht[1]);
-            d->rehashidx = -1;
-            return 0;
-        }
-
         /* Note that rehashidx can't overflow as we are sure there are more
          * elements because ht[0].used != 0 */
         assert(d->ht[0].size > (unsigned long)d->rehashidx);
-        while(d->ht[0].table[d->rehashidx] == NULL) d->rehashidx++;
+        while(d->ht[0].table[d->rehashidx] == NULL) {
+            d->rehashidx++;
+            if (--empty_visits == 0) return 1;
+        }
         de = d->ht[0].table[d->rehashidx];
         /* Move all the keys in this bucket from the old to the new hash HT */
         while(de) {
@@ -270,6 +273,17 @@ int dictRehash(dict *d, int n) {
         d->ht[0].table[d->rehashidx] = NULL;
         d->rehashidx++;
     }
+
+    /* Check if we already rehashed the whole table... */
+    if (d->ht[0].used == 0) {
+        zfree(d->ht[0].table);
+        d->ht[0] = d->ht[1];
+        _dictReset(&d->ht[1]);
+        d->rehashidx = -1;
+        return 0;
+    }
+
+    /* More to rehash... */
     return 1;
 }
 
@@ -342,7 +356,10 @@ dictEntry *dictAddRaw(dict *d, void *key)
     if ((index = _dictKeyIndex(d, key)) == -1)
         return NULL;
 
-    /* Allocate the memory and store the new entry */
+    /* Allocate the memory and store the new entry.
+     * Insert the element in top, with the assumption that in a database
+     * system it is more likely that recently added entries are accessed
+     * more frequently. */
     ht = dictIsRehashing(d) ? &d->ht[1] : &d->ht[0];
     entry = zmalloc(sizeof(*entry));
     entry->next = ht->table[index];
@@ -616,7 +633,11 @@ dictEntry *dictGetRandomKey(dict *d)
     if (dictIsRehashing(d)) _dictRehashStep(d);
     if (dictIsRehashing(d)) {
         do {
-            h = random() % (d->ht[0].size+d->ht[1].size);
+            /* We are sure there are no elements in indexes from 0
+             * to rehashidx-1 */
+            h = d->rehashidx + (random() % (d->ht[0].size +
+                                            d->ht[1].size -
+                                            d->rehashidx));
             he = (h >= d->ht[0].size) ? d->ht[1].table[h - d->ht[0].size] :
                                       d->ht[0].table[h];
         } while(he == NULL);
@@ -643,9 +664,12 @@ dictEntry *dictGetRandomKey(dict *d)
     return he;
 }
 
-/* This is a version of dictGetRandomKey() that is modified in order to
- * return multiple entries by jumping at a random place of the hash table
- * and scanning linearly for entries.
+/* This function samples the dictionary to return a few keys from random
+ * locations.
+ *
+ * It does not guarantee to return all the keys specified in 'count', nor
+ * it does guarantee to return non-duplicated elements, however it will make
+ * some effort to do both things.
  *
  * Returned pointers to hash table entries are stored into 'des' that
  * points to an array of dictEntry pointers. The array must have room for
@@ -654,28 +678,65 @@ dictEntry *dictGetRandomKey(dict *d)
  *
  * The function returns the number of items stored into 'des', that may
  * be less than 'count' if the hash table has less than 'count' elements
- * inside.
+ * inside, or if not enough elements were found in a reasonable amount of
+ * steps.
  *
  * Note that this function is not suitable when you need a good distribution
  * of the returned items, but only when you need to "sample" a given number
  * of continuous elements to run some kind of algorithm or to produce
  * statistics. However the function is much faster than dictGetRandomKey()
- * at producing N elements, and the elements are guaranteed to be non
- * repeating. */
-unsigned int dictGetRandomKeys(dict *d, dictEntry **des, unsigned int count) {
-    int j; /* internal hash table id, 0 or 1. */
-    unsigned int stored = 0;
+ * at producing N elements. */
+unsigned int dictGetSomeKeys(dict *d, dictEntry **des, unsigned int count) {
+    unsigned int j; /* internal hash table id, 0 or 1. */
+    unsigned int tables; /* 1 or 2 tables? */
+    unsigned int stored = 0, maxsizemask;
+    unsigned int maxsteps;
 
     if (dictSize(d) < count) count = dictSize(d);
-    while(stored < count) {
-        for (j = 0; j < 2; j++) {
-            /* Pick a random point inside the hash table 0 or 1. */
-            unsigned int i = random() & d->ht[j].sizemask;
-            int size = d->ht[j].size;
+    maxsteps = count*10;
 
-            /* Make sure to visit every bucket by iterating 'size' times. */
-            while(size--) {
-                dictEntry *he = d->ht[j].table[i];
+    /* Try to do a rehashing work proportional to 'count'. */
+    for (j = 0; j < count; j++) {
+        if (dictIsRehashing(d))
+            _dictRehashStep(d);
+        else
+            break;
+    }
+
+    tables = dictIsRehashing(d) ? 2 : 1;
+    maxsizemask = d->ht[0].sizemask;
+    if (tables > 1 && maxsizemask < d->ht[1].sizemask)
+        maxsizemask = d->ht[1].sizemask;
+
+    /* Pick a random point inside the larger table. */
+    unsigned int i = random() & maxsizemask;
+    unsigned int emptylen = 0; /* Continuous empty entries so far. */
+    while(stored < count && maxsteps--) {
+        for (j = 0; j < tables; j++) {
+            /* Invariant of the dict.c rehashing: up to the indexes already
+             * visited in ht[0] during the rehashing, there are no populated
+             * buckets, so we can skip ht[0] for indexes between 0 and idx-1. */
+            if (tables == 2 && j == 0 && i < d->rehashidx) {
+                /* Moreover, if we are currently out of range in the second
+                 * table, there will be no elements in both tables up to
+                 * the current rehashing index, so we jump if possible.
+                 * (this happens when going from big to small table). */
+                if (i >= d->ht[1].size) i = d->rehashidx;
+                continue;
+            }
+            if (i >= d->ht[j].size) continue; /* Out of range for this table. */
+            dictEntry *he = d->ht[j].table[i];
+
+            /* Count contiguous empty buckets, and jump to other
+             * locations if they reach 'count' (with a minimum of 5). */
+            if (he == NULL) {
+                emptylen++;
+                if (emptylen >= 5 && emptylen > count) {
+                    i = random() & maxsizemask;
+                    emptylen = 0;
+                }
+            } else {
+                emptylen = 0;
                 while (he) {
                     /* Collect all the elements of the buckets found non
                      * empty while iterating. */
@@ -685,14 +746,11 @@ unsigned int dictGetRandomKeys(dict *d, dictEntry **des, unsigned int count) {
                     stored++;
                     if (stored == count) return stored;
                 }
-                i = (i+1) & d->ht[j].sizemask;
             }
-            /* If there is only one table and we iterated it all, we should
-             * already have 'count' elements. Assert this condition. */
-            assert(dictIsRehashing(d) != 0);
         }
+        i = (i+1) & maxsizemask;
     }
-    return stored; /* Never reached. */
+    return stored;
 }
 
 /* Function to reverse bits. Algorithm from:

src/dict.h

@@ -164,7 +164,7 @@ dictIterator *dictGetSafeIterator(dict *d);
 dictEntry *dictNext(dictIterator *iter);
 void dictReleaseIterator(dictIterator *iter);
 dictEntry *dictGetRandomKey(dict *d);
-unsigned int dictGetRandomKeys(dict *d, dictEntry **des, unsigned int count);
+unsigned int dictGetSomeKeys(dict *d, dictEntry **des, unsigned int count);
 void dictPrintStats(dict *d);
 unsigned int dictGenHashFunction(const void *key, int len);
 unsigned int dictGenCaseHashFunction(const unsigned char *buf, int len);

src/endianconv.c

@@ -101,12 +101,16 @@ uint64_t intrev64(uint64_t v) {
     return v;
 }
 
-#ifdef TESTMAIN
+#ifdef REDIS_TEST
 #include <stdio.h>
 
-int main(void) {
+#define UNUSED(x) (void)(x)
+int endianconvTest(int argc, char *argv[]) {
     char buf[32];
 
+    UNUSED(argc);
+    UNUSED(argv);
+
     sprintf(buf,"ciaoroma");
     memrev16(buf);
     printf("%s\n", buf);

src/endianconv.h

@@ -71,4 +71,8 @@ uint64_t intrev64(uint64_t v);
 #define ntohu64(v) intrev64(v)
 #endif
 
+#ifdef REDIS_TEST
+int endianconvTest(int argc, char *argv[]);
+#endif
+
 #endif

src/fmacros.h

@@ -34,6 +34,7 @@
 
 #if defined(__linux__)
 #define _GNU_SOURCE
+#define _DEFAULT_SOURCE
 #endif
 
 #if defined(_AIX)

src/help.h

@@ -651,8 +651,8 @@ struct commandHelp {
     0,
     "1.0.0" },
     { "SPOP",
-    "key",
-    "Remove and return a random member from a set",
+    "key [count]",
+    "Remove and return one or multiple random members from a set",
     3,
     "1.0.0" },
     { "SRANDMEMBER",

src/hyperloglog.c

@@ -1213,7 +1213,7 @@ void pfcountCommand(redisClient *c) {
         for (j = 1; j < c->argc; j++) {
             /* Check type and size. */
             robj *o = lookupKeyRead(c->db,c->argv[j]);
-            if (o == NULL) continue; /* Assume empty HLL for non existing var. */
+            if (o == NULL) continue; /* Assume empty HLL for non existing var.*/
             if (isHLLObjectOrReply(c,o) != REDIS_OK) return;
 
             /* Merge with this HLL with our 'max' HHL by setting max[i]
@@ -1233,7 +1233,7 @@ void pfcountCommand(redisClient *c) {
      *
      * The user specified a single key. Either return the cached value
      * or compute one and update the cache. */
-    o = lookupKeyRead(c->db,c->argv[1]);
+    o = lookupKeyWrite(c->db,c->argv[1]);
     if (o == NULL) {
         /* No key? Cardinality is zero since no element was added, otherwise
          * we would have a key as HLLADD creates it as a side effect. */
@@ -1458,7 +1458,7 @@ void pfdebugCommand(redisClient *c) {
     robj *o;
     int j;
 
-    o = lookupKeyRead(c->db,c->argv[2]);
+    o = lookupKeyWrite(c->db,c->argv[2]);
     if (o == NULL) {
         addReplyError(c,"The specified key does not exist");
         return;

src/intset.c

@@ -281,44 +281,46 @@ size_t intsetBlobLen(intset *is) {
     return sizeof(intset)+intrev32ifbe(is->length)*intrev32ifbe(is->encoding);
 }
 
-#ifdef INTSET_TEST_MAIN
+#ifdef REDIS_TEST
 #include <sys/time.h>
+#include <time.h>
 
-void intsetRepr(intset *is) {
-    int i;
-    for (i = 0; i < intrev32ifbe(is->length); i++) {
+#if 0
+static void intsetRepr(intset *is) {
+    for (uint32_t i = 0; i < intrev32ifbe(is->length); i++) {
         printf("%lld\n", (uint64_t)_intsetGet(is,i));
     }
     printf("\n");
 }
 
-void error(char *err) {
+static void error(char *err) {
     printf("%s\n", err);
     exit(1);
 }
+#endif
 
-void ok(void) {
+static void ok(void) {
     printf("OK\n");
 }
 
-long long usec(void) {
+static long long usec(void) {
     struct timeval tv;
     gettimeofday(&tv,NULL);
     return (((long long)tv.tv_sec)*1000000)+tv.tv_usec;
 }
 
 #define assert(_e) ((_e)?(void)0:(_assert(#_e,__FILE__,__LINE__),exit(1)))
-void _assert(char *estr, char *file, int line) {
+static void _assert(char *estr, char *file, int line) {
     printf("\n\n=== ASSERTION FAILED ===\n");
     printf("==> %s:%d '%s' is not true\n",file,line,estr);
 }
 
-intset *createSet(int bits, int size) {
+static intset *createSet(int bits, int size) {
     uint64_t mask = (1<<bits)-1;
-    uint64_t i, value;
+    uint64_t value;
     intset *is = intsetNew();
 
-    for (i = 0; i < size; i++) {
+    for (int i = 0; i < size; i++) {
         if (bits > 32) {
             value = (rand()*rand()) & mask;
         } else {
@@ -329,10 +331,8 @@ intset *createSet(int bits, int size) {
     return is;
 }
 
-void checkConsistency(intset *is) {
-    int i;
-
-    for (i = 0; i < (intrev32ifbe(is->length)-1); i++) {
+static void checkConsistency(intset *is) {
+    for (uint32_t i = 0; i < (intrev32ifbe(is->length)-1); i++) {
         uint32_t encoding = intrev32ifbe(is->encoding);
 
         if (encoding == INTSET_ENC_INT16) {
@@ -348,11 +348,15 @@ void checkConsistency(intset *is) {
     }
 }
 
-int main(int argc, char **argv) {
+#define UNUSED(x) (void)(x)
+int intsetTest(int argc, char **argv) {
     uint8_t success;
     int i;
     intset *is;
-    sranddev();
+    srand(time(NULL));
+
+    UNUSED(argc);
+    UNUSED(argv);
 
     printf("Value encodings: "); {
         assert(_intsetValueEncoding(-32768) == INTSET_ENC_INT16);
@@ -363,8 +367,10 @@ int main(int argc, char **argv) {
         assert(_intsetValueEncoding(+2147483647) == INTSET_ENC_INT32);
         assert(_intsetValueEncoding(-2147483649) == INTSET_ENC_INT64);
         assert(_intsetValueEncoding(+2147483648) == INTSET_ENC_INT64);
-        assert(_intsetValueEncoding(-9223372036854775808ull) == INTSET_ENC_INT64);
-        assert(_intsetValueEncoding(+9223372036854775807ull) == INTSET_ENC_INT64);
+        assert(_intsetValueEncoding(-9223372036854775808ull) ==
+                    INTSET_ENC_INT64);
+        assert(_intsetValueEncoding(+9223372036854775807ull) ==
+                    INTSET_ENC_INT64);
         ok();
     }
 
@@ -378,7 +384,7 @@ int main(int argc, char **argv) {
     }
 
     printf("Large number of random adds: "); {
-        int inserts = 0;
+        uint32_t inserts = 0;
         is = intsetNew();
         for (i = 0; i < 1024; i++) {
             is = intsetAdd(is,rand()%0x800,&success);
@@ -461,7 +467,8 @@ int main(int argc, char **argv) {
 
         start = usec();
         for (i = 0; i < num; i++) intsetSearch(is,rand() % ((1<<bits)-1),NULL);
-        printf("%ld lookups, %ld element set, %lldusec\n",num,size,usec()-start);
+        printf("%ld lookups, %ld element set, %lldusec\n",
+               num,size,usec()-start);
     }
 
     printf("Stress add+delete: "); {
@@ -479,5 +486,7 @@ int main(int argc, char **argv) {
         checkConsistency(is);
         ok();
     }
+
+    return 0;
 }
 #endif

src/intset.h

@@ -47,4 +47,8 @@ uint8_t intsetGet(intset *is, uint32_t pos, int64_t *value);
 uint32_t intsetLen(intset *is);
 size_t intsetBlobLen(intset *is);
 
+#ifdef REDIS_TEST
+int intsetTest(int argc, char *argv[]);
+#endif
+
 #endif // __INTSET_H

src/latency.c

@@ -56,6 +56,32 @@ dictType latencyTimeSeriesDictType = {
     dictVanillaFree             /* val destructor */
 };
 
+/* ------------------------- Utility functions ------------------------------ */
+
+#ifdef __linux__
+/* Returns 1 if Transparent Huge Pages support is enabled in the kernel.
+ * Otherwise (or if we are unable to check) 0 is returned. */
+int THPIsEnabled(void) {
+    char buf[1024];
+
+    FILE *fp = fopen("/sys/kernel/mm/transparent_hugepage/enabled","r");
+    if (!fp) return 0;
+    if (fgets(buf,sizeof(buf),fp) == NULL) {
+        fclose(fp);
+        return 0;
+    }
+    fclose(fp);
+    return (strstr(buf,"[never]") == NULL) ? 1 : 0;
+}
+#endif
+
+/* Report the amount of AnonHugePages in smap, in bytes. If the return
+ * value of the function is non-zero, the process is being targeted by
+ * THP support, and is likely to have memory usage / latency issues. */
+int THPGetAnonHugePagesSize(void) {
+    return zmalloc_get_smap_bytes_by_field("AnonHugePages:");
+}
+
 /* ---------------------------- Latency API --------------------------------- */
 
 /* Latency monitor initialization. We just need to create the dictionary
@@ -202,7 +228,9 @@ sds createLatencyReport(void) {
     int advise_write_load_info = 0; /* Print info about AOF and write load. */
     int advise_hz = 0;              /* Use higher HZ. */
     int advise_large_objects = 0;   /* Deletion of large objects. */
+    int advise_mass_eviction = 0;   /* Avoid mass eviction of keys. */
     int advise_relax_fsync_policy = 0; /* appendfsync always is slow. */
+    int advise_disable_thp = 0;     /* AnonHugePages detected. */
     int advices = 0;
 
     /* Return ASAP if the latency engine is disabled and it looks like it
@@ -337,18 +365,29 @@ sds createLatencyReport(void) {
         }
 
         /* Eviction cycle. */
-        if (!strcasecmp(event,"eviction-cycle")) {
+        if (!strcasecmp(event,"eviction-del")) {
             advise_large_objects = 1;
             advices++;
         }
 
+        if (!strcasecmp(event,"eviction-cycle")) {
+            advise_mass_eviction = 1;
+            advices++;
+        }
+
         report = sdscatlen(report,"\n",1);
     }
     dictReleaseIterator(di);
 
-    if (eventnum == 0) {
+    /* Add non event based advices. */
+    if (THPGetAnonHugePagesSize() > 0) {
+        advise_disable_thp = 1;
+        advices++;
+    }
+
+    if (eventnum == 0 && advices == 0) {
         report = sdscat(report,"Dave, no latency spike was observed during the lifetime of this Redis instance, not in the slightest bit. I honestly think you ought to sit down calmly, take a stress pill, and think things over.\n");
-    } else if (advices == 0) {
+    } else if (eventnum > 0 && advices == 0) {
         report = sdscat(report,"\nWhile there are latency events logged, I'm not able to suggest any easy fix. Please use the Redis community to get some help, providing this report in your help request.\n");
     } else {
         /* Add all the suggestions accumulated so far. */
@@ -418,6 +457,14 @@ sds createLatencyReport(void) {
         if (advise_large_objects) {
             report = sdscat(report,"- Deleting, expiring or evicting (because of maxmemory policy) large objects is a blocking operation. If you have very large objects that are often deleted, expired, or evicted, try to fragment those objects into multiple smaller objects.\n");
         }
+
+        if (advise_mass_eviction) {
+            report = sdscat(report,"- Sudden changes to the 'maxmemory' setting via 'CONFIG SET', or allocation of large objects via sets or sorted sets intersections, STORE option of SORT, Redis Cluster large keys migrations (RESTORE command), may create sudden memory pressure forcing the server to block trying to evict keys. \n");
+        }
+
+        if (advise_disable_thp) {
+            report = sdscat(report,"- I detected a non zero amount of anonymous huge pages used by your process. This creates very serious latency events in different conditions, especially when Redis is persisting on disk. To disable THP support use the command 'echo never > /sys/kernel/mm/transparent_hugepage/enabled', make sure to also add it into /etc/rc.local so that the command will be executed again after a reboot. Note that even if you have already disabled THP, you still need to restart the Redis process to get rid of the huge pages already created.\n");
+        }
     }
 
     return report;
@@ -475,7 +522,6 @@ sds latencyCommandGenSparkeline(char *event, struct latencyTimeSeries *ts) {
     for (j = 0; j < LATENCY_TS_LEN; j++) {
         int i = (ts->idx + j) % LATENCY_TS_LEN;
         int elapsed;
-        char *label;
         char buf[64];
 
         if (ts->samples[i].time == 0) continue;
@@ -497,8 +543,7 @@ sds latencyCommandGenSparkeline(char *event, struct latencyTimeSeries *ts) {
             snprintf(buf,sizeof(buf),"%dh",elapsed/3600);
         else
             snprintf(buf,sizeof(buf),"%dd",elapsed/(3600*24));
-        label = zstrdup(buf);
-        sparklineSequenceAddSample(seq,ts->samples[i].latency,label);
+        sparklineSequenceAddSample(seq,ts->samples[i].latency,buf);
     }
 
     graph = sdscatprintf(graph,

src/latency.h

@@ -63,6 +63,7 @@ struct latencyStats {
 
 void latencyMonitorInit(void);
 void latencyAddSample(char *event, mstime_t latency);
+int THPIsEnabled(void);
 
 /* Latency monitoring macros. */
 
@@ -85,4 +86,8 @@ void latencyAddSample(char *event, mstime_t latency);
         (var) >= server.latency_monitor_threshold) \
           latencyAddSample((event),(var));
 
+/* Remove time from a nested event. */
+#define latencyRemoveNestedEvent(event_var,nested_var) \
+    event_var += nested_var;
+
 #endif /* __LATENCY_H */

src/lzfP.h

@@ -49,7 +49,7 @@
  * the difference between 15 and 14 is very small
  * for small blocks (and 14 is usually a bit faster).
  * For a low-memory/faster configuration, use HLOG == 13;
- * For best compression, use 15 or 16 (or more, up to 23).
+ * For best compression, use 15 or 16 (or more, up to 22).
  */
 #ifndef HLOG
 # define HLOG 16
@@ -94,7 +94,7 @@
 /*
  * Avoid assigning values to errno variable? for some embedding purposes
  * (linux kernel for example), this is necessary. NOTE: this breaks
- * the documentation in lzf.h.
+ * the documentation in lzf.h. Avoiding errno has no speed impact.
  */
 #ifndef AVOID_ERRNO
 # define AVOID_ERRNO 0
@@ -121,16 +121,52 @@
 # define CHECK_INPUT 1
 #endif
 
+/*
+ * Whether to store pointers or offsets inside the hash table. On
+ * 64 bit architetcures, pointers take up twice as much space,
+ * and might also be slower. Default is to autodetect.
+ */
+/*#define LZF_USER_OFFSETS autodetect */
+
 /*****************************************************************************/
 /* nothing should be changed below */
 
+#ifdef __cplusplus
+# include <cstring>
+# include <climits>
+using namespace std;
+#else
+# include <string.h>
+# include <limits.h>
+#endif
+
+#ifndef LZF_USE_OFFSETS
+# if defined (WIN32)
+#  define LZF_USE_OFFSETS defined(_M_X64)
+# else
+#  if __cplusplus > 199711L
+#   include <cstdint>
+#  else
+#   include <stdint.h>
+#  endif
+#  define LZF_USE_OFFSETS (UINTPTR_MAX > 0xffffffffU)
+# endif
+#endif
+
 typedef unsigned char u8;
 
-typedef const u8 *LZF_STATE[1 << (HLOG)];
+#if LZF_USE_OFFSETS
+# define LZF_HSLOT_BIAS ((const u8 *)in_data)
+  typedef unsigned int LZF_HSLOT;
+#else
+# define LZF_HSLOT_BIAS 0
+  typedef const u8 *LZF_HSLOT;
+#endif
+
+typedef LZF_HSLOT LZF_STATE[1 << (HLOG)];
 
 #if !STRICT_ALIGN
 /* for unaligned accesses we need a 16 bit datatype. */
-# include <limits.h>
 # if USHRT_MAX == 65535
     typedef unsigned short u16;
 # elif UINT_MAX == 65535
@@ -142,17 +178,7 @@ typedef const u8 *LZF_STATE[1 << (HLOG)];
 #endif
 
 #if ULTRA_FAST
-# if defined(VERY_FAST)
-#  undef VERY_FAST
-# endif
-#endif
-
-#if INIT_HTAB
-# ifdef __cplusplus
-#  include <cstring>
-# else
-#  include <string.h>
-# endif
+# undef VERY_FAST
 #endif
 
 #endif

src/lzf_c.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000-2008 Marc Alexander Lehmann <schmorp@schmorp.de>
+ * Copyright (c) 2000-2010 Marc Alexander Lehmann <schmorp@schmorp.de>
  *
  * Redistribution and use in source and binary forms, with or without modifica-
  * tion, are permitted provided that the following conditions are met:
@@ -40,8 +40,8 @@
 
 /*
  * don't play with this unless you benchmark!
- * decompression is not dependent on the hash function
- * the hashing function might seem strange, just believe me
+ * the data format is not dependent on the hash function.
+ * the hash function might seem strange, just believe me,
  * it works ;)
  */
 #ifndef FRST
@@ -89,9 +89,9 @@
 /*
  * compressed format
  *
- * 000LLLLL <L+1>    ; literal
- * LLLooooo oooooooo ; backref L
- * 111ooooo LLLLLLLL oooooooo ; backref L+7
+ * 000LLLLL <L+1>    ; literal, L+1=1..33 octets
+ * LLLooooo oooooooo ; backref L+1=1..7 octets, o+1=1..4096 offset
+ * 111ooooo LLLLLLLL oooooooo ; backref L+8 octets, o+1=1..4096 offset
  *
  */
 
@@ -106,7 +106,6 @@ lzf_compress (const void *const in_data, unsigned int in_len,
 #if !LZF_STATE_ARG
   LZF_STATE htab;
 #endif
-  const u8 **hslot;
   const u8 *ip = (const u8 *)in_data;
         u8 *op = (u8 *)out_data;
   const u8 *in_end  = ip + in_len;
@@ -133,10 +132,6 @@ lzf_compress (const void *const in_data, unsigned int in_len,
 
 #if INIT_HTAB
   memset (htab, 0, sizeof (htab));
-# if 0
-  for (hslot = htab; hslot < htab + HSIZE; hslot++)
-    *hslot++ = ip;
-# endif
 #endif
 
   lit = 0; op++; /* start run */
@@ -144,24 +139,23 @@ lzf_compress (const void *const in_data, unsigned int in_len,
   hval = FRST (ip);
   while (ip < in_end - 2)
     {
+      LZF_HSLOT *hslot;
+
       hval = NEXT (hval, ip);
       hslot = htab + IDX (hval);
-      ref = *hslot; *hslot = ip;
+      ref = *hslot + LZF_HSLOT_BIAS; *hslot = ip - LZF_HSLOT_BIAS;
 
       if (1
 #if INIT_HTAB
           && ref < ip /* the next test will actually take care of this, but this is faster */
 #endif
           && (off = ip - ref - 1) < MAX_OFF
-          && ip + 4 < in_end
           && ref > (u8 *)in_data
-#if STRICT_ALIGN
-          && ref[0] == ip[0]
-          && ref[1] == ip[1]
           && ref[2] == ip[2]
+#if STRICT_ALIGN
+          && ((ref[1] << 8) | ref[0]) == ((ip[1] << 8) | ip[0])
 #else
           && *(u16 *)ref == *(u16 *)ip
-          && ref[2] == ip[2]
 #endif
         )
         {
@@ -170,12 +164,13 @@ lzf_compress (const void *const in_data, unsigned int in_len,
           unsigned int maxlen = in_end - ip - len;
           maxlen = maxlen > MAX_REF ? MAX_REF : maxlen;
 
+          if (expect_false (op + 3 + 1 >= out_end)) /* first a faster conservative test */
+            if (op - !lit + 3 + 1 >= out_end) /* second the exact but rare test */
+              return 0;
+
           op [- lit - 1] = lit - 1; /* stop run */
           op -= !lit; /* undo run if length is zero */
 
-          if (expect_false (op + 3 + 1 >= out_end))
-            return 0;
-
           for (;;)
             {
               if (expect_true (maxlen > 16))
@@ -222,6 +217,7 @@ lzf_compress (const void *const in_data, unsigned int in_len,
             }
 
           *op++ = off;
+
           lit = 0; op++; /* start run */
 
           ip += len + 1;
@@ -237,12 +233,12 @@ lzf_compress (const void *const in_data, unsigned int in_len,
           hval = FRST (ip);
 
           hval = NEXT (hval, ip);
-          htab[IDX (hval)] = ip;
+          htab[IDX (hval)] = ip - LZF_HSLOT_BIAS;
           ip++;
 
 # if VERY_FAST && !ULTRA_FAST
           hval = NEXT (hval, ip);
-          htab[IDX (hval)] = ip;
+          htab[IDX (hval)] = ip - LZF_HSLOT_BIAS;
           ip++;
 # endif
 #else
@@ -251,7 +247,7 @@ lzf_compress (const void *const in_data, unsigned int in_len,
           do
             {
               hval = NEXT (hval, ip);
-              htab[IDX (hval)] = ip;
+              htab[IDX (hval)] = ip - LZF_HSLOT_BIAS;
               ip++;
             }
           while (len--);

src/lzf_d.c

@@ -1,5 +1,5 @@
 /*
- * Copyright (c) 2000-2007 Marc Alexander Lehmann <schmorp@schmorp.de>
+ * Copyright (c) 2000-2010 Marc Alexander Lehmann <schmorp@schmorp.de>
  *
  * Redistribution and use in source and binary forms, with or without modifica-
  * tion, are permitted provided that the following conditions are met:
@@ -43,14 +43,14 @@
 # define SET_ERRNO(n) errno = (n)
 #endif
 
-/*
+#if USE_REP_MOVSB /* small win on amd, big loss on intel */
 #if (__i386 || __amd64) && __GNUC__ >= 3
 # define lzf_movsb(dst, src, len)                \
    asm ("rep movsb"                              \
         : "=D" (dst), "=S" (src), "=c" (len)     \
         :  "0" (dst),  "1" (src),  "2" (len));
 #endif
-*/
+#endif
 
 unsigned int
 lzf_decompress (const void *const in_data,  unsigned int in_len,
@@ -86,9 +86,17 @@ lzf_decompress (const void *const in_data,  unsigned int in_len,
 #ifdef lzf_movsb
           lzf_movsb (op, ip, ctrl);
 #else
-          do
-            *op++ = *ip++;
-          while (--ctrl);
+          switch (ctrl)
+            {
+              case 32: *op++ = *ip++; case 31: *op++ = *ip++; case 30: *op++ = *ip++; case 29: *op++ = *ip++;
+              case 28: *op++ = *ip++; case 27: *op++ = *ip++; case 26: *op++ = *ip++; case 25: *op++ = *ip++;
+              case 24: *op++ = *ip++; case 23: *op++ = *ip++; case 22: *op++ = *ip++; case 21: *op++ = *ip++;
+              case 20: *op++ = *ip++; case 19: *op++ = *ip++; case 18: *op++ = *ip++; case 17: *op++ = *ip++;
+              case 16: *op++ = *ip++; case 15: *op++ = *ip++; case 14: *op++ = *ip++; case 13: *op++ = *ip++;
+              case 12: *op++ = *ip++; case 11: *op++ = *ip++; case 10: *op++ = *ip++; case  9: *op++ = *ip++;
+              case  8: *op++ = *ip++; case  7: *op++ = *ip++; case  6: *op++ = *ip++; case  5: *op++ = *ip++;
+              case  4: *op++ = *ip++; case  3: *op++ = *ip++; case  2: *op++ = *ip++; case  1: *op++ = *ip++;
+            }
 #endif
         }
       else /* back reference */
@@ -134,12 +142,39 @@ lzf_decompress (const void *const in_data,  unsigned int in_len,
           len += 2;
           lzf_movsb (op, ref, len);
 #else
-          *op++ = *ref++;
-          *op++ = *ref++;
+          switch (len)
+            {
+              default:
+                len += 2;
 
-          do
-            *op++ = *ref++;
-          while (--len);
+                if (op >= ref + len)
+                  {
+                    /* disjunct areas */
+                    memcpy (op, ref, len);
+                    op += len;
+                  }
+                else
+                  {
+                    /* overlapping, use octte by octte copying */
+                    do
+                      *op++ = *ref++;
+                    while (--len);
+                  }
+
+                break;
+
+              case 9: *op++ = *ref++;
+              case 8: *op++ = *ref++;
+              case 7: *op++ = *ref++;
+              case 6: *op++ = *ref++;
+              case 5: *op++ = *ref++;
+              case 4: *op++ = *ref++;
+              case 3: *op++ = *ref++;
+              case 2: *op++ = *ref++;
+              case 1: *op++ = *ref++;
+              case 0: *op++ = *ref++; /* two octets more */
+                      *op++ = *ref++;
+            }
 #endif
         }
     }

src/memtest.c

@@ -35,6 +35,9 @@
 #include <errno.h>
 #include <termios.h>
 #include <sys/ioctl.h>
+#if defined(__sun)
+#include <stropts.h>
+#endif
 #include "config.h"
 
 #if (ULONG_MAX == 4294967295UL)

src/networking.c

@@ -100,6 +100,7 @@ redisClient *createClient(int fd) {
     c->ctime = c->lastinteraction = server.unixtime;
     c->authenticated = 0;
     c->replstate = REDIS_REPL_NONE;
+    c->repl_put_online_on_ack = 0;
     c->reploff = 0;
     c->repl_ack_off = 0;
     c->repl_ack_time = 0;
@@ -524,6 +525,14 @@ void addReplyBulkCBuffer(redisClient *c, void *p, size_t len) {
     addReply(c,shared.crlf);
 }
 
+/* Add sds to reply (takes ownership of sds and frees it) */
+void addReplyBulkSds(redisClient *c, sds s)  {
+    addReplySds(c,sdscatfmt(sdsempty(),"$%u\r\n",
+        (unsigned long)sdslen(s)));
+    addReplySds(c,s);
+    addReply(c,shared.crlf);
+}
+
 /* Add a C nul term string as bulk reply */
 void addReplyBulkCString(redisClient *c, char *s) {
     if (s == NULL) {
@@ -838,6 +847,7 @@ void sendReplyToClient(aeEventLoop *el, int fd, void *privdata, int mask) {
          *
          * However if we are over the maxmemory limit we ignore that and
          * just deliver as much data as it is possible to deliver. */
+        server.stat_net_output_bytes += totwritten;
         if (totwritten > REDIS_MAX_WRITE_PER_EVENT &&
             (server.maxmemory == 0 ||
              zmalloc_used_memory() < server.maxmemory)) break;
@@ -925,8 +935,10 @@ int processInlineBuffer(redisClient *c) {
     sdsrange(c->querybuf,querylen+2,-1);
 
     /* Setup argv array on client structure */
-    if (c->argv) zfree(c->argv);
-    c->argv = zmalloc(sizeof(robj*)*argc);
+    if (argc) {
+        if (c->argv) zfree(c->argv);
+        c->argv = zmalloc(sizeof(robj*)*argc);
+    }
 
     /* Create redis objects for all arguments. */
     for (c->argc = 0, j = 0; j < argc; j++) {
@@ -1178,6 +1190,7 @@ void readQueryFromClient(aeEventLoop *el, int fd, void *privdata, int mask) {
         sdsIncrLen(c->querybuf,nread);
         c->lastinteraction = server.unixtime;
         if (c->flags & REDIS_MASTER) c->reploff += nread;
+        server.stat_net_input_bytes += nread;
     } else {
         server.current_client = NULL;
         return;
@@ -1214,17 +1227,6 @@ void getClientsMaxBuffers(unsigned long *longest_output_list,
     *biggest_input_buffer = bib;
 }
 
-/* This is a helper function for genClientPeerId().
- * It writes the specified ip/port to "peerid" as a null termiated string
- * in the form ip:port if ip does not contain ":" itself, otherwise
- * [ip]:port format is used (for IPv6 addresses basically). */
-void formatPeerId(char *peerid, size_t peerid_len, char *ip, int port) {
-    if (strchr(ip,':'))
-        snprintf(peerid,peerid_len,"[%s]:%d",ip,port);
-    else
-        snprintf(peerid,peerid_len,"%s:%d",ip,port);
-}
-
 /* A Redis "Peer ID" is a colon separated ip:port pair.
  * For IPv4 it's in the form x.y.z.k:port, example: "127.0.0.1:1234".
  * For IPv6 addresses we use [] around the IP part, like in "[::1]:1234".
@@ -1233,24 +1235,17 @@ void formatPeerId(char *peerid, size_t peerid_len, char *ip, int port) {
  * A Peer ID always fits inside a buffer of REDIS_PEER_ID_LEN bytes, including
  * the null term.
  *
- * The function returns REDIS_OK on succcess, and REDIS_ERR on failure.
- *
  * On failure the function still populates 'peerid' with the "?:0" string
  * in case you want to relax error checking or need to display something
  * anyway (see anetPeerToString implementation for more info). */
-int genClientPeerId(redisClient *client, char *peerid, size_t peerid_len) {
-    char ip[REDIS_IP_STR_LEN];
-    int port;
-
+void genClientPeerId(redisClient *client, char *peerid,
+                            size_t peerid_len) {
     if (client->flags & REDIS_UNIX_SOCKET) {
         /* Unix socket client. */
         snprintf(peerid,peerid_len,"%s:0",server.unixsocket);
-        return REDIS_OK;
     } else {
         /* TCP client. */
-        int retval = anetPeerToString(client->fd,ip,sizeof(ip),&port);
-        formatPeerId(peerid,peerid_len,ip,port);
-        return (retval == -1) ? REDIS_ERR : REDIS_OK;
+        anetFormatPeer(client->fd,peerid,peerid_len);
     }
 }
 

src/object.c

@@ -109,26 +109,44 @@ robj *createStringObjectFromLongLong(long long value) {
     return o;
 }
 
-/* Note: this function is defined into object.c since here it is where it
- * belongs but it is actually designed to be used just for INCRBYFLOAT */
-robj *createStringObjectFromLongDouble(long double value) {
+/* Create a string object from a long double. If humanfriendly is non-zero
+ * it does not use exponential format and trims trailing zeroes at the end,
+ * however this results in loss of precision. Otherwise exp format is used
+ * and the output of snprintf() is not modified.
+ *
+ * The 'humanfriendly' option is used for INCRBYFLOAT and HINCRBYFLOAT. */
+robj *createStringObjectFromLongDouble(long double value, int humanfriendly) {
     char buf[256];
     int len;
 
-    /* We use 17 digits precision since with 128 bit floats that precision
-     * after rounding is able to represent most small decimal numbers in a way
-     * that is "non surprising" for the user (that is, most small decimal
-     * numbers will be represented in a way that when converted back into
-     * a string are exactly the same as what the user typed.) */
-    len = snprintf(buf,sizeof(buf),"%.17Lf", value);
-    /* Now remove trailing zeroes after the '.' */
-    if (strchr(buf,'.') != NULL) {
-        char *p = buf+len-1;
-        while(*p == '0') {
-            p--;
-            len--;
+    if (isinf(value)) {
+        /* Libc in odd systems (Hi Solaris!) will format infinite in a
+         * different way, so better to handle it in an explicit way. */
+        if (value > 0) {
+            memcpy(buf,"inf",3);
+            len = 3;
+        } else {
+            memcpy(buf,"-inf",4);
+            len = 4;
         }
-        if (*p == '.') len--;
+    } else if (humanfriendly) {
+        /* We use 17 digits precision since with 128 bit floats that precision
+         * after rounding is able to represent most small decimal numbers in a
+         * way that is "non surprising" for the user (that is, most small
+         * decimal numbers will be represented in a way that when converted
+         * back into a string are exactly the same as what the user typed.) */
+        len = snprintf(buf,sizeof(buf),"%.17Lf", value);
+        /* Now remove trailing zeroes after the '.' */
+        if (strchr(buf,'.') != NULL) {
+            char *p = buf+len-1;
+            while(*p == '0') {
+                p--;
+                len--;
+            }
+            if (*p == '.') len--;
+        }
+    } else {
+        len = snprintf(buf,sizeof(buf),"%.17Lg", value);
     }
     return createStringObject(buf,len);
 }
@@ -162,11 +180,10 @@ robj *dupStringObject(robj *o) {
     }
 }
 
-robj *createListObject(void) {
-    list *l = listCreate();
+robj *createQuicklistObject(void) {
+    quicklist *l = quicklistCreate();
     robj *o = createObject(REDIS_LIST,l);
-    listSetFreeMethod(l,decrRefCountVoid);
-    o->encoding = REDIS_ENCODING_LINKEDLIST;
+    o->encoding = REDIS_ENCODING_QUICKLIST;
     return o;
 }
 
@@ -224,11 +241,8 @@ void freeStringObject(robj *o) {
 
 void freeListObject(robj *o) {
     switch (o->encoding) {
-    case REDIS_ENCODING_LINKEDLIST:
-        listRelease((list*) o->ptr);
-        break;
-    case REDIS_ENCODING_ZIPLIST:
-        zfree(o->ptr);
+    case REDIS_ENCODING_QUICKLIST:
+        quicklistRelease(o->ptr);
         break;
     default:
         redisPanic("Unknown list encoding type");
@@ -660,7 +674,7 @@ char *strEncoding(int encoding) {
     case REDIS_ENCODING_RAW: return "raw";
     case REDIS_ENCODING_INT: return "int";
     case REDIS_ENCODING_HT: return "hashtable";
-    case REDIS_ENCODING_LINKEDLIST: return "linkedlist";
+    case REDIS_ENCODING_QUICKLIST: return "quicklist";
     case REDIS_ENCODING_ZIPLIST: return "ziplist";
     case REDIS_ENCODING_INTSET: return "intset";
     case REDIS_ENCODING_SKIPLIST: return "skiplist";

src/quicklist.h

@@ -0,0 +1,169 @@
+/* quicklist.h - A generic doubly linked quicklist implementation
+ *
+ * Copyright (c) 2014, Matt Stancliff <matt@genges.com>
+ * All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions are met:
+ *
+ *   * Redistributions of source code must retain the above copyright notice,
+ *     this quicklist of conditions and the following disclaimer.
+ *   * Redistributions in binary form must reproduce the above copyright
+ *     notice, this quicklist of conditions and the following disclaimer in the
+ *     documentation and/or other materials provided with the distribution.
+ *   * Neither the name of Redis nor the names of its contributors may be used
+ *     to endorse or promote products derived from this software without
+ *     specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+ * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
+ * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
+ * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
+ * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
+ * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
+ * POSSIBILITY OF SUCH DAMAGE.
+ */
+
+#ifndef __QUICKLIST_H__
+#define __QUICKLIST_H__
+
+/* Node, quicklist, and Iterator are the only data structures used currently. */
+
+/* quicklistNode is a 32 byte struct describing a ziplist for a quicklist.
+ * We use bit fields keep the quicklistNode at 32 bytes.
+ * count: 16 bits, max 65536 (max zl bytes is 65k, so max count actually < 32k).
+ * encoding: 2 bits, RAW=1, LZF=2.
+ * container: 2 bits, NONE=1, ZIPLIST=2.
+ * recompress: 1 bit, bool, true if node is temporarry decompressed for usage.
+ * attempted_compress: 1 bit, boolean, used for verifying during testing.
+ * extra: 12 bits, free for future use; pads out the remainder of 32 bits */
+typedef struct quicklistNode {
+    struct quicklistNode *prev;
+    struct quicklistNode *next;
+    unsigned char *zl;
+    unsigned int sz;             /* ziplist size in bytes */
+    unsigned int count : 16;     /* count of items in ziplist */
+    unsigned int encoding : 2;   /* RAW==1 or LZF==2 */
+    unsigned int container : 2;  /* NONE==1 or ZIPLIST==2 */
+    unsigned int recompress : 1; /* was this node previous compressed? */
+    unsigned int attempted_compress : 1; /* node can't compress; too small */
+    unsigned int extra : 10; /* more bits to steal for future usage */
+} quicklistNode;
+
+/* quicklistLZF is a 4+N byte struct holding 'sz' followed by 'compressed'.
+ * 'sz' is byte length of 'compressed' field.
+ * 'compressed' is LZF data with total (compressed) length 'sz'
+ * NOTE: uncompressed length is stored in quicklistNode->sz.
+ * When quicklistNode->zl is compressed, node->zl points to a quicklistLZF */
+typedef struct quicklistLZF {
+    unsigned int sz; /* LZF size in bytes*/
+    char compressed[];
+} quicklistLZF;
+
+/* quicklist is a 32 byte struct (on 64-bit systems) describing a quicklist.
+ * 'count' is the number of total entries.
+ * 'len' is the number of quicklist nodes.
+ * 'compress' is: -1 if compression disabled, otherwise it's the number
+ *                of quicklistNodes to leave uncompressed at ends of quicklist.
+ * 'fill' is the user-requested (or default) fill factor. */
+typedef struct quicklist {
+    quicklistNode *head;
+    quicklistNode *tail;
+    unsigned long count;        /* total count of all entries in all ziplists */
+    unsigned int len;           /* number of quicklistNodes */
+    int fill : 16;              /* fill factor for individual nodes */
+    unsigned int compress : 16; /* depth of end nodes not to compress;0=off */
+} quicklist;
+
+typedef struct quicklistIter {
+    const quicklist *quicklist;
+    quicklistNode *current;
+    unsigned char *zi;
+    long offset; /* offset in current ziplist */
+    int direction;
+} quicklistIter;
+
+typedef struct quicklistEntry {
+    const quicklist *quicklist;
+    quicklistNode *node;
+    unsigned char *zi;
+    unsigned char *value;
+    unsigned int sz;
+    long long longval;
+    int offset;
+} quicklistEntry;
+
+#define QUICKLIST_HEAD 0
+#define QUICKLIST_TAIL -1
+
+/* quicklist node encodings */
+#define QUICKLIST_NODE_ENCODING_RAW 1
+#define QUICKLIST_NODE_ENCODING_LZF 2
+
+/* quicklist compression disable */
+#define QUICKLIST_NOCOMPRESS 0
+
+/* quicklist container formats */
+#define QUICKLIST_NODE_CONTAINER_NONE 1
+#define QUICKLIST_NODE_CONTAINER_ZIPLIST 2
+
+#define quicklistNodeIsCompressed(node)                                        \
+    ((node)->encoding == QUICKLIST_NODE_ENCODING_LZF)
+
+/* Prototypes */
+quicklist *quicklistCreate(void);
+quicklist *quicklistNew(int fill, int compress);
+void quicklistSetCompressDepth(quicklist *quicklist, int depth);
+void quicklistSetFill(quicklist *quicklist, int fill);
+void quicklistSetOptions(quicklist *quicklist, int fill, int depth);
+void quicklistRelease(quicklist *quicklist);
+int quicklistPushHead(quicklist *quicklist, void *value, const size_t sz);
+int quicklistPushTail(quicklist *quicklist, void *value, const size_t sz);
+void quicklistPush(quicklist *quicklist, void *value, const size_t sz,
+                   int where);
+void quicklistAppendZiplist(quicklist *quicklist, unsigned char *zl);
+quicklist *quicklistAppendValuesFromZiplist(quicklist *quicklist,
+                                            unsigned char *zl);
+quicklist *quicklistCreateFromZiplist(int fill, int compress,
+                                      unsigned char *zl);
+void quicklistInsertAfter(quicklist *quicklist, quicklistEntry *node,
+                          void *value, const size_t sz);
+void quicklistInsertBefore(quicklist *quicklist, quicklistEntry *node,
+                           void *value, const size_t sz);
+void quicklistDelEntry(quicklistIter *iter, quicklistEntry *entry);
+int quicklistReplaceAtIndex(quicklist *quicklist, long index, void *data,
+                            int sz);
+int quicklistDelRange(quicklist *quicklist, const long start, const long stop);
+quicklistIter *quicklistGetIterator(const quicklist *quicklist, int direction);
+quicklistIter *quicklistGetIteratorAtIdx(const quicklist *quicklist,
+                                         int direction, const long long idx);
+int quicklistNext(quicklistIter *iter, quicklistEntry *node);
+void quicklistReleaseIterator(quicklistIter *iter);
+quicklist *quicklistDup(quicklist *orig);
+int quicklistIndex(const quicklist *quicklist, const long long index,
+                   quicklistEntry *entry);
+void quicklistRewind(quicklist *quicklist, quicklistIter *li);
+void quicklistRewindTail(quicklist *quicklist, quicklistIter *li);
+void quicklistRotate(quicklist *quicklist);
+int quicklistPopCustom(quicklist *quicklist, int where, unsigned char **data,
+                       unsigned int *sz, long long *sval,
+                       void *(*saver)(unsigned char *data, unsigned int sz));
+int quicklistPop(quicklist *quicklist, int where, unsigned char **data,
+                 unsigned int *sz, long long *slong);
+unsigned int quicklistCount(quicklist *ql);
+int quicklistCompare(unsigned char *p1, unsigned char *p2, int p2_len);
+size_t quicklistGetLzf(const quicklistNode *node, void **data);
+
+#ifdef REDIS_TEST
+int quicklistTest(int argc, char *argv[]);
+#endif
+
+/* Directions for iterators */
+#define AL_START_HEAD 0
+#define AL_START_TAIL 1
+
+#endif /* __QUICKLIST_H__ */

src/rdb.c

@@ -40,6 +40,20 @@
 #include <arpa/inet.h>
 #include <sys/stat.h>
 
+#define RDB_LOAD_NONE   0
+#define RDB_LOAD_ENC    (1<<0)
+#define RDB_LOAD_PLAIN  (1<<1)
+
+#define rdbExitReportCorruptRDB(reason) rdbCheckThenExit(reason, __LINE__);
+
+void rdbCheckThenExit(char *reason, int where) {
+    redisLog(REDIS_WARNING, "Corrupt RDB detected at rdb.c:%d (%s). "
+        "Running 'redis-check-rdb %s'",
+        where, reason, server.rdb_filename);
+    redis_check_rdb(server.rdb_filename);
+    exit(1);
+}
+
 static int rdbWriteRaw(rio *rdb, void *p, size_t len) {
     if (rdb && rioWrite(rdb,p,len) == 0)
         return -1;
@@ -161,9 +175,11 @@ int rdbEncodeInteger(long long value, unsigned char *enc) {
 }
 
 /* Loads an integer-encoded object with the specified encoding type "enctype".
- * If the "encode" argument is set the function may return an integer-encoded
- * string object, otherwise it always returns a raw string object. */
-robj *rdbLoadIntegerObject(rio *rdb, int enctype, int encode) {
+ * The returned value changes according to the flags, see
+ * rdbGenerincLoadStringObject() for more info. */
+void *rdbLoadIntegerObject(rio *rdb, int enctype, int flags) {
+    int plain = flags & RDB_LOAD_PLAIN;
+    int encode = flags & RDB_LOAD_ENC;
     unsigned char enc[4];
     long long val;
 
@@ -182,12 +198,19 @@ robj *rdbLoadIntegerObject(rio *rdb, int enctype, int encode) {
         val = (int32_t)v;
     } else {
         val = 0; /* anti-warning */
-        redisPanic("Unknown RDB integer encoding type");
+        rdbExitReportCorruptRDB("Unknown RDB integer encoding type");
     }
-    if (encode)
+    if (plain) {
+        char buf[REDIS_LONGSTR_SIZE], *p;
+        int len = ll2string(buf,sizeof(buf),val);
+        p = zmalloc(len);
+        memcpy(p,buf,len);
+        return p;
+    } else if (encode) {
         return createStringObjectFromLongLong(val);
-    else
+    } else {
         return createObject(REDIS_STRING,sdsfromlonglong(val));
+    }
 }
 
 /* String objects in the form "2391" "-100" without any space and with a
@@ -209,10 +232,33 @@ int rdbTryIntegerEncoding(char *s, size_t len, unsigned char *enc) {
     return rdbEncodeInteger(value,enc);
 }
 
-int rdbSaveLzfStringObject(rio *rdb, unsigned char *s, size_t len) {
-    size_t comprlen, outlen;
+int rdbSaveLzfBlob(rio *rdb, void *data, size_t compress_len,
+                   size_t original_len) {
     unsigned char byte;
     int n, nwritten = 0;
+
+    /* Data compressed! Let's save it on disk */
+    byte = (REDIS_RDB_ENCVAL<<6)|REDIS_RDB_ENC_LZF;
+    if ((n = rdbWriteRaw(rdb,&byte,1)) == -1) goto writeerr;
+    nwritten += n;
+
+    if ((n = rdbSaveLen(rdb,compress_len)) == -1) goto writeerr;
+    nwritten += n;
+
+    if ((n = rdbSaveLen(rdb,original_len)) == -1) goto writeerr;
+    nwritten += n;
+
+    if ((n = rdbWriteRaw(rdb,data,compress_len)) == -1) goto writeerr;
+    nwritten += n;
+
+    return nwritten;
+
+writeerr:
+    return -1;
+}
+
+int rdbSaveLzfStringObject(rio *rdb, unsigned char *s, size_t len) {
+    size_t comprlen, outlen;
     void *out;
 
     /* We require at least four bytes compression for this to be worth it */
@@ -224,29 +270,16 @@ int rdbSaveLzfStringObject(rio *rdb, unsigned char *s, size_t len) {
         zfree(out);
         return 0;
     }
-    /* Data compressed! Let's save it on disk */
-    byte = (REDIS_RDB_ENCVAL<<6)|REDIS_RDB_ENC_LZF;
-    if ((n = rdbWriteRaw(rdb,&byte,1)) == -1) goto writeerr;
-    nwritten += n;
-
-    if ((n = rdbSaveLen(rdb,comprlen)) == -1) goto writeerr;
-    nwritten += n;
-
-    if ((n = rdbSaveLen(rdb,len)) == -1) goto writeerr;
-    nwritten += n;
-
-    if ((n = rdbWriteRaw(rdb,out,comprlen)) == -1) goto writeerr;
-    nwritten += n;
-
+    size_t nwritten = rdbSaveLzfBlob(rdb, out, comprlen, len);
     zfree(out);
     return nwritten;
-
-writeerr:
-    zfree(out);
-    return -1;
 }
 
-robj *rdbLoadLzfStringObject(rio *rdb) {
+/* Load an LZF compressed string in RDB format. The returned value
+ * changes according to 'flags'. For more info check the
+ * rdbGenericLoadStringObject() function. */
+void *rdbLoadLzfStringObject(rio *rdb, int flags) {
+    int plain = flags & RDB_LOAD_PLAIN;
     unsigned int len, clen;
     unsigned char *c = NULL;
     sds val = NULL;
@@ -254,14 +287,29 @@ robj *rdbLoadLzfStringObject(rio *rdb) {
     if ((clen = rdbLoadLen(rdb,NULL)) == REDIS_RDB_LENERR) return NULL;
     if ((len = rdbLoadLen(rdb,NULL)) == REDIS_RDB_LENERR) return NULL;
     if ((c = zmalloc(clen)) == NULL) goto err;
-    if ((val = sdsnewlen(NULL,len)) == NULL) goto err;
+
+    /* Allocate our target according to the uncompressed size. */
+    if (plain) {
+        val = zmalloc(len);
+    } else {
+        if ((val = sdsnewlen(NULL,len)) == NULL) goto err;
+    }
+
+    /* Load the compressed representation and uncompress it to target. */
     if (rioRead(rdb,c,clen) == 0) goto err;
     if (lzf_decompress(c,clen,val,len) == 0) goto err;
     zfree(c);
-    return createObject(REDIS_STRING,val);
+
+    if (plain)
+        return val;
+    else
+        return createObject(REDIS_STRING,val);
 err:
     zfree(c);
-    sdsfree(val);
+    if (plain)
+        zfree(val);
+    else
+        sdsfree(val);
     return NULL;
 }
 
@@ -330,10 +378,21 @@ int rdbSaveStringObject(rio *rdb, robj *obj) {
     }
 }
 
-robj *rdbGenericLoadStringObject(rio *rdb, int encode) {
+/* Load a string object from an RDB file according to flags:
+ *
+ * RDB_LOAD_NONE (no flags): load an RDB object, unencoded.
+ * RDB_LOAD_ENC: If the returned type is a Redis object, try to
+ *               encode it in a special way to be more memory
+ *               efficient. When this flag is passed the function
+ *               no longer guarantees that obj->ptr is an SDS string.
+ * RDB_LOAD_PLAIN: Return a plain string allocated with zmalloc()
+ *                 instead of a Redis object.
+ */
+void *rdbGenericLoadStringObject(rio *rdb, int flags) {
+    int encode = flags & RDB_LOAD_ENC;
+    int plain = flags & RDB_LOAD_PLAIN;
     int isencoded;
     uint32_t len;
-    robj *o;
 
     len = rdbLoadLen(rdb,&isencoded);
     if (isencoded) {
@@ -341,30 +400,39 @@ robj *rdbGenericLoadStringObject(rio *rdb, int encode) {
         case REDIS_RDB_ENC_INT8:
         case REDIS_RDB_ENC_INT16:
         case REDIS_RDB_ENC_INT32:
-            return rdbLoadIntegerObject(rdb,len,encode);
+            return rdbLoadIntegerObject(rdb,len,flags);
         case REDIS_RDB_ENC_LZF:
-            return rdbLoadLzfStringObject(rdb);
+            return rdbLoadLzfStringObject(rdb,flags);
         default:
-            redisPanic("Unknown RDB encoding type");
+            rdbExitReportCorruptRDB("Unknown RDB encoding type");
         }
     }
 
     if (len == REDIS_RDB_LENERR) return NULL;
-    o = encode ? createStringObject(NULL,len) :
-                 createRawStringObject(NULL,len);
-    if (len && rioRead(rdb,o->ptr,len) == 0) {
-        decrRefCount(o);
-        return NULL;
+    if (!plain) {
+        robj *o = encode ? createStringObject(NULL,len) :
+                           createRawStringObject(NULL,len);
+        if (len && rioRead(rdb,o->ptr,len) == 0) {
+            decrRefCount(o);
+            return NULL;
+        }
+        return o;
+    } else {
+        void *buf = zmalloc(len);
+        if (len && rioRead(rdb,buf,len) == 0) {
+            zfree(buf);
+            return NULL;
+        }
+        return buf;
     }
-    return o;
 }
 
 robj *rdbLoadStringObject(rio *rdb) {
-    return rdbGenericLoadStringObject(rdb,0);
+    return rdbGenericLoadStringObject(rdb,RDB_LOAD_NONE);
 }
 
 robj *rdbLoadEncodedStringObject(rio *rdb) {
-    return rdbGenericLoadStringObject(rdb,1);
+    return rdbGenericLoadStringObject(rdb,RDB_LOAD_ENC);
 }
 
 /* Save a double value. Doubles are saved as strings prefixed by an unsigned
@@ -433,10 +501,8 @@ int rdbSaveObjectType(rio *rdb, robj *o) {
     case REDIS_STRING:
         return rdbSaveType(rdb,REDIS_RDB_TYPE_STRING);
     case REDIS_LIST:
-        if (o->encoding == REDIS_ENCODING_ZIPLIST)
-            return rdbSaveType(rdb,REDIS_RDB_TYPE_LIST_ZIPLIST);
-        else if (o->encoding == REDIS_ENCODING_LINKEDLIST)
-            return rdbSaveType(rdb,REDIS_RDB_TYPE_LIST);
+        if (o->encoding == REDIS_ENCODING_QUICKLIST)
+            return rdbSaveType(rdb,REDIS_RDB_TYPE_LIST_QUICKLIST);
         else
             redisPanic("Unknown list encoding");
     case REDIS_SET:
@@ -477,7 +543,7 @@ int rdbLoadObjectType(rio *rdb) {
 
 /* Save a Redis object. Returns -1 on error, number of bytes written on success. */
 int rdbSaveObject(rio *rdb, robj *o) {
-    int n, nwritten = 0;
+    int n = 0, nwritten = 0;
 
     if (o->type == REDIS_STRING) {
         /* Save a string value */
@@ -485,25 +551,24 @@ int rdbSaveObject(rio *rdb, robj *o) {
         nwritten += n;
     } else if (o->type == REDIS_LIST) {
         /* Save a list value */
-        if (o->encoding == REDIS_ENCODING_ZIPLIST) {
-            size_t l = ziplistBlobLen((unsigned char*)o->ptr);
+        if (o->encoding == REDIS_ENCODING_QUICKLIST) {
+            quicklist *ql = o->ptr;
+            quicklistNode *node = ql->head;
 
-            if ((n = rdbSaveRawString(rdb,o->ptr,l)) == -1) return -1;
-            nwritten += n;
-        } else if (o->encoding == REDIS_ENCODING_LINKEDLIST) {
-            list *list = o->ptr;
-            listIter li;
-            listNode *ln;
-
-            if ((n = rdbSaveLen(rdb,listLength(list))) == -1) return -1;
+            if ((n = rdbSaveLen(rdb,ql->len)) == -1) return -1;
             nwritten += n;
 
-            listRewind(list,&li);
-            while((ln = listNext(&li))) {
-                robj *eleobj = listNodeValue(ln);
-                if ((n = rdbSaveStringObject(rdb,eleobj)) == -1) return -1;
-                nwritten += n;
-            }
+            do {
+                if (quicklistNodeIsCompressed(node)) {
+                    void *data;
+                    size_t compress_len = quicklistGetLzf(node, &data);
+                    if ((n = rdbSaveLzfBlob(rdb,data,compress_len,node->sz)) == -1) return -1;
+                    nwritten += n;
+                } else {
+                    if ((n = rdbSaveRawString(rdb,node->zl,node->sz)) == -1) return -1;
+                    nwritten += n;
+                }
+            } while ((node = node->next));
         } else {
             redisPanic("Unknown list encoding");
         }
@@ -627,6 +692,39 @@ int rdbSaveKeyValuePair(rio *rdb, robj *key, robj *val,
     return 1;
 }
 
+/* Save an AUX field. */
+int rdbSaveAuxField(rio *rdb, void *key, size_t keylen, void *val, size_t vallen) {
+    if (rdbSaveType(rdb,REDIS_RDB_OPCODE_AUX) == -1) return -1;
+    if (rdbSaveRawString(rdb,key,keylen) == -1) return -1;
+    if (rdbSaveRawString(rdb,val,vallen) == -1) return -1;
+    return 1;
+}
+
+/* Wrapper for rdbSaveAuxField() used when key/val length can be obtained
+ * with strlen(). */
+int rdbSaveAuxFieldStrStr(rio *rdb, char *key, char *val) {
+    return rdbSaveAuxField(rdb,key,strlen(key),val,strlen(val));
+}
+
+/* Wrapper for strlen(key) + integer type (up to long long range). */
+int rdbSaveAuxFieldStrInt(rio *rdb, char *key, long long val) {
+    char buf[REDIS_LONGSTR_SIZE];
+    int vlen = ll2string(buf,sizeof(buf),val);
+    return rdbSaveAuxField(rdb,key,strlen(key),buf,vlen);
+}
+
+/* Save a few default AUX fields with information about the RDB generated. */
+int rdbSaveInfoAuxFields(rio *rdb) {
+    int redis_bits = (sizeof(void*) == 8) ? 64 : 32;
+
+    /* Add a few fields about the state when the RDB was created. */
+    if (rdbSaveAuxFieldStrStr(rdb,"redis-ver",REDIS_VERSION) == -1) return -1;
+    if (rdbSaveAuxFieldStrInt(rdb,"redis-bits",redis_bits) == -1) return -1;
+    if (rdbSaveAuxFieldStrInt(rdb,"ctime",time(NULL)) == -1) return -1;
+    if (rdbSaveAuxFieldStrInt(rdb,"used-mem",zmalloc_used_memory()) == -1) return -1;
+    return 1;
+}
+
 /* Produces a dump of the database in RDB format sending it to the specified
  * Redis I/O channel. On success REDIS_OK is returned, otherwise REDIS_ERR
  * is returned and part of the output, or all the output, can be
@@ -647,6 +745,7 @@ int rdbSaveRio(rio *rdb, int *error) {
         rdb->update_cksum = rioGenericUpdateChecksum;
     snprintf(magic,sizeof(magic),"REDIS%04d",REDIS_RDB_VERSION);
     if (rdbWriteRaw(rdb,magic,9) == -1) goto werr;
+    if (rdbSaveInfoAuxFields(rdb) == -1) goto werr;
 
     for (j = 0; j < server.dbnum; j++) {
         redisDb *db = server.db+j;
@@ -659,6 +758,21 @@ int rdbSaveRio(rio *rdb, int *error) {
         if (rdbSaveType(rdb,REDIS_RDB_OPCODE_SELECTDB) == -1) goto werr;
         if (rdbSaveLen(rdb,j) == -1) goto werr;
 
+        /* Write the RESIZE DB opcode. We trim the size to UINT32_MAX, which
+         * is currently the largest type we are able to represent in RDB sizes.
+         * However this does not limit the actual size of the DB to load since
+         * these sizes are just hints to resize the hash tables. */
+        uint32_t db_size, expires_size;
+        db_size = (dictSize(db->dict) <= UINT32_MAX) ?
+                                dictSize(db->dict) :
+                                UINT32_MAX;
+        expires_size = (dictSize(db->dict) <= UINT32_MAX) ?
+                                dictSize(db->expires) :
+                                UINT32_MAX;
+        if (rdbSaveType(rdb,REDIS_RDB_OPCODE_RESIZEDB) == -1) goto werr;
+        if (rdbSaveLen(rdb,db_size) == -1) goto werr;
+        if (rdbSaveLen(rdb,expires_size) == -1) goto werr;
+
         /* Iterate this DB writing every entry */
         while((de = dictNext(di)) != NULL) {
             sds keystr = dictGetKey(de);
@@ -720,7 +834,7 @@ int rdbSave(char *filename) {
     char tmpfile[256];
     FILE *fp;
     rio rdb;
-    int error;
+    int error = 0;
 
     snprintf(tmpfile,256,"temp-%d.rdb", (int) getpid());
     fp = fopen(tmpfile,"w");
@@ -819,7 +933,7 @@ void rdbRemoveTempFile(pid_t childpid) {
 /* Load a Redis object of the specified type from the specified file.
  * On success a newly allocated object is returned, otherwise NULL. */
 robj *rdbLoadObject(int rdbtype, rio *rdb) {
-    robj *o, *ele, *dec;
+    robj *o = NULL, *ele, *dec;
     size_t len;
     unsigned int i;
 
@@ -831,33 +945,18 @@ robj *rdbLoadObject(int rdbtype, rio *rdb) {
         /* Read list value */
         if ((len = rdbLoadLen(rdb,NULL)) == REDIS_RDB_LENERR) return NULL;
 
-        /* Use a real list when there are too many entries */
-        if (len > server.list_max_ziplist_entries) {
-            o = createListObject();
-        } else {
-            o = createZiplistObject();
-        }
+        o = createQuicklistObject();
+        quicklistSetOptions(o->ptr, server.list_max_ziplist_size,
+                            server.list_compress_depth);
 
         /* Load every single element of the list */
         while(len--) {
             if ((ele = rdbLoadEncodedStringObject(rdb)) == NULL) return NULL;
-
-            /* If we are using a ziplist and the value is too big, convert
-             * the object to a real list. */
-            if (o->encoding == REDIS_ENCODING_ZIPLIST &&
-                sdsEncodedObject(ele) &&
-                sdslen(ele->ptr) > server.list_max_ziplist_value)
-                    listTypeConvert(o,REDIS_ENCODING_LINKEDLIST);
-
-            if (o->encoding == REDIS_ENCODING_ZIPLIST) {
-                dec = getDecodedObject(ele);
-                o->ptr = ziplistPush(o->ptr,dec->ptr,sdslen(dec->ptr),REDIS_TAIL);
-                decrRefCount(dec);
-                decrRefCount(ele);
-            } else {
-                ele = tryObjectEncoding(ele);
-                listAddNodeTail(o->ptr,ele);
-            }
+            dec = getDecodedObject(ele);
+            size_t len = sdslen(dec->ptr);
+            quicklistPushTail(o->ptr, dec->ptr, len);
+            decrRefCount(dec);
+            decrRefCount(ele);
         }
     } else if (rdbtype == REDIS_RDB_TYPE_SET) {
         /* Read list/set value */
@@ -989,25 +1088,33 @@ robj *rdbLoadObject(int rdbtype, rio *rdb) {
 
             /* Add pair to hash table */
             ret = dictAdd((dict*)o->ptr, field, value);
-            redisAssert(ret == DICT_OK);
+            if (ret == DICT_ERR) {
+                rdbExitReportCorruptRDB("Duplicate keys detected");
+            }
         }
 
         /* All pairs should be read by now */
         redisAssert(len == 0);
+    } else if (rdbtype == REDIS_RDB_TYPE_LIST_QUICKLIST) {
+        if ((len = rdbLoadLen(rdb,NULL)) == REDIS_RDB_LENERR) return NULL;
+        o = createQuicklistObject();
+        quicklistSetOptions(o->ptr, server.list_max_ziplist_size,
+                            server.list_compress_depth);
 
+        while (len--) {
+            unsigned char *zl = rdbGenericLoadStringObject(rdb,RDB_LOAD_PLAIN);
+            if (zl == NULL) return NULL;
+            quicklistAppendZiplist(o->ptr, zl);
+        }
     } else if (rdbtype == REDIS_RDB_TYPE_HASH_ZIPMAP  ||
                rdbtype == REDIS_RDB_TYPE_LIST_ZIPLIST ||
                rdbtype == REDIS_RDB_TYPE_SET_INTSET   ||
                rdbtype == REDIS_RDB_TYPE_ZSET_ZIPLIST ||
                rdbtype == REDIS_RDB_TYPE_HASH_ZIPLIST)
     {
-        robj *aux = rdbLoadStringObject(rdb);
-
-        if (aux == NULL) return NULL;
-        o = createObject(REDIS_STRING,NULL); /* string is just placeholder */
-        o->ptr = zmalloc(sdslen(aux->ptr));
-        memcpy(o->ptr,aux->ptr,sdslen(aux->ptr));
-        decrRefCount(aux);
+        unsigned char *encoded = rdbGenericLoadStringObject(rdb,RDB_LOAD_PLAIN);
+        if (encoded == NULL) return NULL;
+        o = createObject(REDIS_STRING,encoded); /* Obj type fixed below. */
 
         /* Fix the object encoding, and make sure to convert the encoded
          * data type into the base type if accordingly to the current
@@ -1048,8 +1155,7 @@ robj *rdbLoadObject(int rdbtype, rio *rdb) {
             case REDIS_RDB_TYPE_LIST_ZIPLIST:
                 o->type = REDIS_LIST;
                 o->encoding = REDIS_ENCODING_ZIPLIST;
-                if (ziplistLen(o->ptr) > server.list_max_ziplist_entries)
-                    listTypeConvert(o,REDIS_ENCODING_LINKEDLIST);
+                listTypeConvert(o,REDIS_ENCODING_QUICKLIST);
                 break;
             case REDIS_RDB_TYPE_SET_INTSET:
                 o->type = REDIS_SET;
@@ -1070,11 +1176,11 @@ robj *rdbLoadObject(int rdbtype, rio *rdb) {
                     hashTypeConvert(o, REDIS_ENCODING_HT);
                 break;
             default:
-                redisPanic("Unknown encoding");
+                rdbExitReportCorruptRDB("Unknown encoding");
                 break;
         }
     } else {
-        redisPanic("Unknown object type");
+        rdbExitReportCorruptRDB("Unknown object type");
     }
     return o;
 }
@@ -1087,8 +1193,9 @@ void startLoading(FILE *fp) {
     /* Load the DB */
     server.loading = 1;
     server.loading_start_time = time(NULL);
+    server.loading_loaded_bytes = 0;
     if (fstat(fileno(fp), &sb) == -1) {
-        server.loading_total_bytes = 1; /* just to avoid division by zero */
+        server.loading_total_bytes = 0;
     } else {
         server.loading_total_bytes = sb.st_size;
     }
@@ -1162,7 +1269,12 @@ int rdbLoad(char *filename) {
 
         /* Read type. */
         if ((type = rdbLoadType(&rdb)) == -1) goto eoferr;
+
+        /* Handle special types. */
         if (type == REDIS_RDB_OPCODE_EXPIRETIME) {
+            /* EXPIRETIME: load an expire associated with the next key
+             * to load. Note that after loading an expire we need to
+             * load the actual type, and continue. */
             if ((expiretime = rdbLoadTime(&rdb)) == -1) goto eoferr;
             /* We read the time so we need to read the object type again. */
             if ((type = rdbLoadType(&rdb)) == -1) goto eoferr;
@@ -1170,27 +1282,67 @@ int rdbLoad(char *filename) {
              * into milliseconds. */
             expiretime *= 1000;
         } else if (type == REDIS_RDB_OPCODE_EXPIRETIME_MS) {
-            /* Milliseconds precision expire times introduced with RDB
-             * version 3. */
+            /* EXPIRETIME_MS: milliseconds precision expire times introduced
+             * with RDB v3. Like EXPIRETIME but no with more precision. */
             if ((expiretime = rdbLoadMillisecondTime(&rdb)) == -1) goto eoferr;
             /* We read the time so we need to read the object type again. */
             if ((type = rdbLoadType(&rdb)) == -1) goto eoferr;
-        }
-
-        if (type == REDIS_RDB_OPCODE_EOF)
+        } else if (type == REDIS_RDB_OPCODE_EOF) {
+            /* EOF: End of file, exit the main loop. */
             break;
-
-        /* Handle SELECT DB opcode as a special case */
-        if (type == REDIS_RDB_OPCODE_SELECTDB) {
+        } else if (type == REDIS_RDB_OPCODE_SELECTDB) {
+            /* SELECTDB: Select the specified database. */
             if ((dbid = rdbLoadLen(&rdb,NULL)) == REDIS_RDB_LENERR)
                 goto eoferr;
             if (dbid >= (unsigned)server.dbnum) {
-                redisLog(REDIS_WARNING,"FATAL: Data file was created with a Redis server configured to handle more than %d databases. Exiting\n", server.dbnum);
+                redisLog(REDIS_WARNING,
+                    "FATAL: Data file was created with a Redis "
+                    "server configured to handle more than %d "
+                    "databases. Exiting\n", server.dbnum);
                 exit(1);
             }
             db = server.db+dbid;
-            continue;
+            continue; /* Read type again. */
+        } else if (type == REDIS_RDB_OPCODE_RESIZEDB) {
+            /* RESIZEDB: Hint about the size of the keys in the currently
+             * selected data base, in order to avoid useless rehashing. */
+            uint32_t db_size, expires_size;
+            if ((db_size = rdbLoadLen(&rdb,NULL)) == REDIS_RDB_LENERR)
+                goto eoferr;
+            if ((expires_size = rdbLoadLen(&rdb,NULL)) == REDIS_RDB_LENERR)
+                goto eoferr;
+            dictExpand(db->dict,db_size);
+            dictExpand(db->expires,expires_size);
+            continue; /* Read type again. */
+        } else if (type == REDIS_RDB_OPCODE_AUX) {
+            /* AUX: generic string-string fields. Use to add state to RDB
+             * which is backward compatible. Implementations of RDB loading
+             * are requierd to skip AUX fields they don't understand.
+             *
+             * An AUX field is composed of two strings: key and value. */
+            robj *auxkey, *auxval;
+            if ((auxkey = rdbLoadStringObject(&rdb)) == NULL) goto eoferr;
+            if ((auxval = rdbLoadStringObject(&rdb)) == NULL) goto eoferr;
+
+            if (((char*)auxkey->ptr)[0] == '%') {
+                /* All the fields with a name staring with '%' are considered
+                 * information fields and are logged at startup with a log
+                 * level of NOTICE. */
+                redisLog(REDIS_NOTICE,"RDB '%s': %s",
+                    (char*)auxkey->ptr,
+                    (char*)auxval->ptr);
+            } else {
+                /* We ignore fields we don't understand, as by AUX field
+                 * contract. */
+                redisLog(REDIS_DEBUG,"Unrecognized RDB AUX field: '%s'",
+                    (char*)auxkey->ptr);
+            }
+
+            decrRefCount(auxkey);
+            decrRefCount(auxval);
+            continue; /* Read type again. */
         }
+
         /* Read key */
         if ((key = rdbLoadStringObject(&rdb)) == NULL) goto eoferr;
         /* Read value */
@@ -1223,7 +1375,7 @@ int rdbLoad(char *filename) {
             redisLog(REDIS_WARNING,"RDB file was saved with checksum disabled: no check performed.");
         } else if (cksum != expected) {
             redisLog(REDIS_WARNING,"Wrong RDB checksum. Aborting now.");
-            exit(1);
+            rdbExitReportCorruptRDB("RDB CRC error");
         }
     }
 
@@ -1233,7 +1385,7 @@ int rdbLoad(char *filename) {
 
 eoferr: /* unexpected end of file is handled here with a fatal exit */
     redisLog(REDIS_WARNING,"Short read or OOM loading DB. Unrecoverable error, aborting now.");
-    exit(1);
+    rdbExitReportCorruptRDB("Unexpected EOF reading RDB file");
     return REDIS_ERR; /* Just to avoid warning */
 }
 
@@ -1491,7 +1643,9 @@ int rdbSaveToSlavesSockets(void) {
             {
                 retval = REDIS_ERR;
             }
+            zfree(msg);
         }
+        zfree(clientids);
         exitFromChild((retval == REDIS_OK) ? 0 : 1);
     } else {
         /* Parent */

src/rdb.h

@@ -38,7 +38,7 @@
 
 /* The current RDB version. When the format changes in a way that is no longer
  * backward compatible this number gets incremented. */
-#define REDIS_RDB_VERSION 6
+#define REDIS_RDB_VERSION 7
 
 /* Defines related to the dump file format. To store 32 bits lengths for short
  * keys requires a lot of space, so we check the most significant 2 bits of
@@ -74,6 +74,7 @@
 #define REDIS_RDB_TYPE_SET    2
 #define REDIS_RDB_TYPE_ZSET   3
 #define REDIS_RDB_TYPE_HASH   4
+/* NOTE: WHEN ADDING NEW RDB TYPE, UPDATE rdbIsObjectType() BELOW */
 
 /* Object types for encoded objects. */
 #define REDIS_RDB_TYPE_HASH_ZIPMAP    9
@@ -81,11 +82,15 @@
 #define REDIS_RDB_TYPE_SET_INTSET    11
 #define REDIS_RDB_TYPE_ZSET_ZIPLIST  12
 #define REDIS_RDB_TYPE_HASH_ZIPLIST  13
+#define REDIS_RDB_TYPE_LIST_QUICKLIST 14
+/* NOTE: WHEN ADDING NEW RDB TYPE, UPDATE rdbIsObjectType() BELOW */
 
 /* Test if a type is an object type. */
-#define rdbIsObjectType(t) ((t >= 0 && t <= 4) || (t >= 9 && t <= 13))
+#define rdbIsObjectType(t) ((t >= 0 && t <= 4) || (t >= 9 && t <= 14))
 
 /* Special RDB opcodes (saved/loaded with rdbSaveType/rdbLoadType). */
+#define REDIS_RDB_OPCODE_AUX        250
+#define REDIS_RDB_OPCODE_RESIZEDB   251
 #define REDIS_RDB_OPCODE_EXPIRETIME_MS 252
 #define REDIS_RDB_OPCODE_EXPIRETIME 253
 #define REDIS_RDB_OPCODE_SELECTDB   254

src/redis-benchmark.c

@@ -90,9 +90,10 @@ typedef struct _client {
     long long start;        /* Start time of a request */
     long long latency;      /* Request latency */
     int pending;            /* Number of pending requests (replies to consume) */
-    int selectlen;  /* If non-zero, a SELECT of 'selectlen' bytes is currently
-                       used as a prefix of the pipline of commands. This gets
-                       discarded the first time it's sent. */
+    int prefix_pending;     /* If non-zero, number of pending prefix commands. Commands
+                               such as auth and select are prefixed to the pipeline of
+                               benchmark commands and discarded after the first send. */
+    int prefixlen;          /* Size in bytes of the pending prefix commands */
 } *client;
 
 /* Prototypes */
@@ -212,20 +213,21 @@ static void readHandler(aeEventLoop *el, int fd, void *privdata, int mask) {
                 }
 
                 freeReplyObject(reply);
-
-                if (c->selectlen) {
-                    size_t j;
-
-                    /* This is the OK from SELECT. Just discard the SELECT
-                     * from the buffer. */
+                /* This is an OK for prefix commands such as auth and select.*/
+                if (c->prefix_pending > 0) {
+                    c->prefix_pending--;
                     c->pending--;
-                    sdsrange(c->obuf,c->selectlen,-1);
-                    /* We also need to fix the pointers to the strings
-                     * we need to randomize. */
-                    for (j = 0; j < c->randlen; j++)
-                        c->randptr[j] -= c->selectlen;
-                    c->selectlen = 0;
-                    continue;
+                    /* Discard prefix commands on first response.*/
+                    if (c->prefixlen > 0) {
+                        size_t j;
+                        sdsrange(c->obuf, c->prefixlen, -1);
+                        /* We also need to fix the pointers to the strings
+                        * we need to randomize. */
+                        for (j = 0; j < c->randlen; j++)
+                            c->randptr[j] -= c->prefixlen;
+                        c->prefixlen = 0;
+                    }
+                    continue;                
                 }
 
                 if (config.requests_finished < config.requests)
@@ -299,8 +301,7 @@ static void writeHandler(aeEventLoop *el, int fd, void *privdata, int mask) {
  * 2) The offsets of the __rand_int__ elements inside the command line, used
  *    for arguments randomization.
  *
- * Even when cloning another client, the SELECT command is automatically prefixed
- * if needed. */
+ * Even when cloning another client, prefix commands are applied if needed.*/
 static client createClient(char *cmd, size_t len, client from) {
     int j;
     client c = zmalloc(sizeof(struct _client));
@@ -325,12 +326,16 @@ static client createClient(char *cmd, size_t len, client from) {
      * Queue N requests accordingly to the pipeline size, or simply clone
      * the example client buffer. */
     c->obuf = sdsempty();
-
+    /* Prefix the request buffer with AUTH and/or SELECT commands, if applicable.
+     * These commands are discarded after the first response, so if the client is
+     * reused the commands will not be used again. */
+    c->prefix_pending = 0;
     if (config.auth) {
         char *buf = NULL;
         int len = redisFormatCommand(&buf, "AUTH %s", config.auth);
         c->obuf = sdscatlen(c->obuf, buf, len);
         free(buf);
+        c->prefix_pending++;
     }
 
     /* If a DB number different than zero is selected, prefix our request
@@ -340,26 +345,23 @@ static client createClient(char *cmd, size_t len, client from) {
     if (config.dbnum != 0) {
         c->obuf = sdscatprintf(c->obuf,"*2\r\n$6\r\nSELECT\r\n$%d\r\n%s\r\n",
             (int)sdslen(config.dbnumstr),config.dbnumstr);
-        c->selectlen = sdslen(c->obuf);
-    } else {
-        c->selectlen = 0;
+        c->prefix_pending++;
     }
-
+    c->prefixlen = sdslen(c->obuf);
     /* Append the request itself. */
     if (from) {
         c->obuf = sdscatlen(c->obuf,
-            from->obuf+from->selectlen,
-            sdslen(from->obuf)-from->selectlen);
+            from->obuf+from->prefixlen,
+            sdslen(from->obuf)-from->prefixlen);
     } else {
         for (j = 0; j < config.pipeline; j++)
             c->obuf = sdscatlen(c->obuf,cmd,len);
     }
 
     c->written = 0;
-    c->pending = config.pipeline;
+    c->pending = config.pipeline+c->prefix_pending;
     c->randptr = NULL;
     c->randlen = 0;
-    if (c->selectlen) c->pending++;
 
     /* Find substrings in the output buffer that need to be randomized. */
     if (config.randomkeys) {
@@ -371,7 +373,7 @@ static client createClient(char *cmd, size_t len, client from) {
             for (j = 0; j < (int)c->randlen; j++) {
                 c->randptr[j] = c->obuf + (from->randptr[j]-from->obuf);
                 /* Adjust for the different select prefix length. */
-                c->randptr[j] += c->selectlen - from->selectlen;
+                c->randptr[j] += c->prefixlen - from->prefixlen;
             }
         } else {
             char *p = c->obuf;
@@ -390,7 +392,8 @@ static client createClient(char *cmd, size_t len, client from) {
             }
         }
     }
-    aeCreateFileEvent(config.el,c->context->fd,AE_WRITABLE,writeHandler,c);
+    if (config.idlemode == 0)
+        aeCreateFileEvent(config.el,c->context->fd,AE_WRITABLE,writeHandler,c);
     listAddNodeTail(config.clients,c);
     config.liveclients++;
     return c;
@@ -555,7 +558,7 @@ usage:
 " -s <socket>        Server socket (overrides host and port)\n"
 " -a <password>      Password for Redis Auth\n"
 " -c <clients>       Number of parallel connections (default 50)\n"
-" -n <requests>      Total number of requests (default 10000)\n"
+" -n <requests>      Total number of requests (default 100000)\n"
 " -d <size>          Data size of SET/GET value in bytes (default 2)\n"
 " -dbnum <db>        SELECT the specified db number (default 0)\n"
 " -k <boolean>       1=keep alive 0=reconnect (default 1)\n"
@@ -600,8 +603,12 @@ int showThroughput(struct aeEventLoop *eventLoop, long long id, void *clientData
         fprintf(stderr,"All clients disconnected... aborting.\n");
         exit(1);
     }
-
     if (config.csv) return 250;
+    if (config.idlemode == 1) {
+        printf("clients: %d\r", config.liveclients);
+        fflush(stdout);
+	return 250;
+    }
     float dt = (float)(mstime()-config.start)/1000.0;
     float rps = (float)config.requests_finished/dt;
     printf("%s: %.2f\r", config.title, rps);
@@ -635,7 +642,7 @@ int main(int argc, const char **argv) {
     signal(SIGPIPE, SIG_IGN);
 
     config.numclients = 50;
-    config.requests = 10000;
+    config.requests = 100000;
     config.liveclients = 0;
     config.el = aeCreateEventLoop(1024*10);
     aeCreateTimeEvent(config.el,1,showThroughput,NULL,NULL);
@@ -693,8 +700,8 @@ int main(int argc, const char **argv) {
     }
 
     /* Run default benchmark suite. */
+    data = zmalloc(config.datasize+1);
     do {
-        data = zmalloc(config.datasize+1);
         memset(data,'x',config.datasize);
         data[config.datasize] = '\0';
 
@@ -731,12 +738,24 @@ int main(int argc, const char **argv) {
             free(cmd);
         }
 
+        if (test_is_selected("rpush")) {
+            len = redisFormatCommand(&cmd,"RPUSH mylist %s",data);
+            benchmark("RPUSH",cmd,len);
+            free(cmd);
+        }
+
         if (test_is_selected("lpop")) {
             len = redisFormatCommand(&cmd,"LPOP mylist");
             benchmark("LPOP",cmd,len);
             free(cmd);
         }
 
+        if (test_is_selected("rpop")) {
+            len = redisFormatCommand(&cmd,"RPOP mylist");
+            benchmark("RPOP",cmd,len);
+            free(cmd);
+        }
+
         if (test_is_selected("sadd")) {
             len = redisFormatCommand(&cmd,
                 "SADD myset element:__rand_int__");

src/redis-check-rdb.c

@@ -29,74 +29,19 @@
  */
 
 
+#include "redis.h"
+#include "rdb.h"
 #include <stdlib.h>
 #include <stdio.h>
 #include <unistd.h>
 #include <fcntl.h>
 #include <sys/stat.h>
 #include <sys/mman.h>
-#include <string.h>
-#include <arpa/inet.h>
-#include <stdint.h>
-#include <limits.h>
 #include "lzf.h"
 #include "crc64.h"
 
-/* Object types */
-#define REDIS_STRING 0
-#define REDIS_LIST 1
-#define REDIS_SET 2
-#define REDIS_ZSET 3
-#define REDIS_HASH 4
-#define REDIS_HASH_ZIPMAP 9
-#define REDIS_LIST_ZIPLIST 10
-#define REDIS_SET_INTSET 11
-#define REDIS_ZSET_ZIPLIST 12
-#define REDIS_HASH_ZIPLIST 13
-
-/* Objects encoding. Some kind of objects like Strings and Hashes can be
- * internally represented in multiple ways. The 'encoding' field of the object
- * is set to one of this fields for this object. */
-#define REDIS_ENCODING_RAW 0    /* Raw representation */
-#define REDIS_ENCODING_INT 1    /* Encoded as integer */
-#define REDIS_ENCODING_ZIPMAP 2 /* Encoded as zipmap */
-#define REDIS_ENCODING_HT 3     /* Encoded as a hash table */
-
-/* Object types only used for dumping to disk */
-#define REDIS_EXPIRETIME_MS 252
-#define REDIS_EXPIRETIME 253
-#define REDIS_SELECTDB 254
-#define REDIS_EOF 255
-
-/* Defines related to the dump file format. To store 32 bits lengths for short
- * keys requires a lot of space, so we check the most significant 2 bits of
- * the first byte to interpreter the length:
- *
- * 00|000000 => if the two MSB are 00 the len is the 6 bits of this byte
- * 01|000000 00000000 =>  01, the len is 14 byes, 6 bits + 8 bits of next byte
- * 10|000000 [32 bit integer] => if it's 01, a full 32 bit len will follow
- * 11|000000 this means: specially encoded object will follow. The six bits
- *           number specify the kind of object that follows.
- *           See the REDIS_RDB_ENC_* defines.
- *
- * Lengths up to 63 are stored using a single byte, most DB keys, and may
- * values, will fit inside. */
-#define REDIS_RDB_6BITLEN 0
-#define REDIS_RDB_14BITLEN 1
-#define REDIS_RDB_32BITLEN 2
-#define REDIS_RDB_ENCVAL 3
-#define REDIS_RDB_LENERR UINT_MAX
-
-/* When a length of a string object stored on disk has the first two bits
- * set, the remaining two bits specify a special encoding for the object
- * accordingly to the following defines: */
-#define REDIS_RDB_ENC_INT8 0        /* 8 bit signed integer */
-#define REDIS_RDB_ENC_INT16 1       /* 16 bit signed integer */
-#define REDIS_RDB_ENC_INT32 2       /* 32 bit signed integer */
-#define REDIS_RDB_ENC_LZF 3         /* string compressed with FASTLZ */
-
 #define ERROR(...) { \
-    printf(__VA_ARGS__); \
+    redisLog(REDIS_WARNING, __VA_ARGS__); \
     exit(1); \
 }
 
@@ -133,28 +78,23 @@ typedef struct {
     char success;
 } entry;
 
-/* Global vars that are actually used as constants. The following double
- * values are used for double on-disk serialization, and are initialized
- * at runtime to avoid strange compiler optimizations. */
-static double R_Zero, R_PosInf, R_NegInf, R_Nan;
-
 #define MAX_TYPES_NUM 256
 #define MAX_TYPE_NAME_LEN 16
 /* store string types for output */
 static char types[MAX_TYPES_NUM][MAX_TYPE_NAME_LEN];
 
 /* Return true if 't' is a valid object type. */
-int checkType(unsigned char t) {
+static int rdbCheckType(unsigned char t) {
     /* In case a new object type is added, update the following
      * condition as necessary. */
     return
-        (t >= REDIS_HASH_ZIPMAP && t <= REDIS_HASH_ZIPLIST) ||
-        t <= REDIS_HASH ||
-        t >= REDIS_EXPIRETIME_MS;
+        (t >= REDIS_RDB_TYPE_HASH_ZIPMAP && t <= REDIS_RDB_TYPE_HASH_ZIPLIST) ||
+        t <= REDIS_RDB_TYPE_HASH ||
+        t >= REDIS_RDB_OPCODE_EXPIRETIME_MS;
 }
 
 /* when number of bytes to read is negative, do a peek */
-int readBytes(void *target, long num) {
+static int readBytes(void *target, long num) {
     char peek = (num < 0) ? 1 : 0;
     num = (num < 0) ? -num : num;
 
@@ -173,28 +113,28 @@ int processHeader(void) {
     int dump_version;
 
     if (!readBytes(buf, 9)) {
-        ERROR("Cannot read header\n");
+        ERROR("Cannot read header");
     }
 
     /* expect the first 5 bytes to equal REDIS */
     if (memcmp(buf,"REDIS",5) != 0) {
-        ERROR("Wrong signature in header\n");
+        ERROR("Wrong signature in header");
     }
 
     dump_version = (int)strtol(buf + 5, NULL, 10);
     if (dump_version < 1 || dump_version > 6) {
-        ERROR("Unknown RDB format version: %d\n", dump_version);
+        ERROR("Unknown RDB format version: %d", dump_version);
     }
     return dump_version;
 }
 
-int loadType(entry *e) {
+static int loadType(entry *e) {
     uint32_t offset = CURR_OFFSET;
 
     /* this byte needs to qualify as type */
     unsigned char t;
     if (readBytes(&t, 1)) {
-        if (checkType(t)) {
+        if (rdbCheckType(t)) {
             e->type = t;
             return 1;
         } else {
@@ -208,18 +148,18 @@ int loadType(entry *e) {
     return 0;
 }
 
-int peekType() {
+static int peekType() {
     unsigned char t;
-    if (readBytes(&t, -1) && (checkType(t)))
+    if (readBytes(&t, -1) && (rdbCheckType(t)))
         return t;
     return -1;
 }
 
 /* discard time, just consume the bytes */
-int processTime(int type) {
+static int processTime(int type) {
     uint32_t offset = CURR_OFFSET;
     unsigned char t[8];
-    int timelen = (type == REDIS_EXPIRETIME_MS) ? 8 : 4;
+    int timelen = (type == REDIS_RDB_OPCODE_EXPIRETIME_MS) ? 8 : 4;
 
     if (readBytes(t,timelen)) {
         return 1;
@@ -231,7 +171,7 @@ int processTime(int type) {
     return 0;
 }
 
-uint32_t loadLength(int *isencoded) {
+static uint32_t loadLength(int *isencoded) {
     unsigned char buf[2];
     uint32_t len;
     int type;
@@ -257,7 +197,7 @@ uint32_t loadLength(int *isencoded) {
     }
 }
 
-char *loadIntegerObject(int enctype) {
+static char *loadIntegerObject(int enctype) {
     uint32_t offset = CURR_OFFSET;
     unsigned char enc[4];
     long long val;
@@ -284,36 +224,36 @@ char *loadIntegerObject(int enctype) {
 
     /* convert val into string */
     char *buf;
-    buf = malloc(sizeof(char) * 128);
+    buf = zmalloc(sizeof(char) * 128);
     sprintf(buf, "%lld", val);
     return buf;
 }
 
-char* loadLzfStringObject() {
+static char* loadLzfStringObject() {
     unsigned int slen, clen;
     char *c, *s;
 
     if ((clen = loadLength(NULL)) == REDIS_RDB_LENERR) return NULL;
     if ((slen = loadLength(NULL)) == REDIS_RDB_LENERR) return NULL;
 
-    c = malloc(clen);
+    c = zmalloc(clen);
     if (!readBytes(c, clen)) {
-        free(c);
+        zfree(c);
         return NULL;
     }
 
-    s = malloc(slen+1);
+    s = zmalloc(slen+1);
     if (lzf_decompress(c,clen,s,slen) == 0) {
-        free(c); free(s);
+        zfree(c); zfree(s);
         return NULL;
     }
 
-    free(c);
+    zfree(c);
     return s;
 }
 
 /* returns NULL when not processable, char* when valid */
-char* loadStringObject() {
+static char* loadStringObject() {
     uint32_t offset = CURR_OFFSET;
     int isencoded;
     uint32_t len;
@@ -336,48 +276,48 @@ char* loadStringObject() {
 
     if (len == REDIS_RDB_LENERR) return NULL;
 
-    char *buf = malloc(sizeof(char) * (len+1));
+    char *buf = zmalloc(sizeof(char) * (len+1));
     if (buf == NULL) return NULL;
     buf[len] = '\0';
     if (!readBytes(buf, len)) {
-        free(buf);
+        zfree(buf);
         return NULL;
     }
     return buf;
 }
 
-int processStringObject(char** store) {
+static int processStringObject(char** store) {
     unsigned long offset = CURR_OFFSET;
     char *key = loadStringObject();
     if (key == NULL) {
         SHIFT_ERROR(offset, "Error reading string object");
-        free(key);
+        zfree(key);
         return 0;
     }
 
     if (store != NULL) {
         *store = key;
     } else {
-        free(key);
+        zfree(key);
     }
     return 1;
 }
 
-double* loadDoubleValue() {
+static double* loadDoubleValue() {
     char buf[256];
     unsigned char len;
     double* val;
 
     if (!readBytes(&len,1)) return NULL;
 
-    val = malloc(sizeof(double));
+    val = zmalloc(sizeof(double));
     switch(len) {
     case 255: *val = R_NegInf;  return val;
     case 254: *val = R_PosInf;  return val;
     case 253: *val = R_Nan;     return val;
     default:
         if (!readBytes(buf, len)) {
-            free(val);
+            zfree(val);
             return NULL;
         }
         buf[len] = '\0';
@@ -386,24 +326,24 @@ double* loadDoubleValue() {
     }
 }
 
-int processDoubleValue(double** store) {
+static int processDoubleValue(double** store) {
     unsigned long offset = CURR_OFFSET;
     double *val = loadDoubleValue();
     if (val == NULL) {
         SHIFT_ERROR(offset, "Error reading double value");
-        free(val);
+        zfree(val);
         return 0;
     }
 
     if (store != NULL) {
         *store = val;
     } else {
-        free(val);
+        zfree(val);
     }
     return 1;
 }
 
-int loadPair(entry *e) {
+static int loadPair(entry *e) {
     uint32_t offset = CURR_OFFSET;
     uint32_t i;
 
@@ -417,10 +357,10 @@ int loadPair(entry *e) {
     }
 
     uint32_t length = 0;
-    if (e->type == REDIS_LIST ||
-        e->type == REDIS_SET  ||
-        e->type == REDIS_ZSET ||
-        e->type == REDIS_HASH) {
+    if (e->type == REDIS_RDB_TYPE_LIST ||
+        e->type == REDIS_RDB_TYPE_SET  ||
+        e->type == REDIS_RDB_TYPE_ZSET ||
+        e->type == REDIS_RDB_TYPE_HASH) {
         if ((length = loadLength(NULL)) == REDIS_RDB_LENERR) {
             SHIFT_ERROR(offset, "Error reading %s length", types[e->type]);
             return 0;
@@ -428,19 +368,19 @@ int loadPair(entry *e) {
     }
 
     switch(e->type) {
-    case REDIS_STRING:
-    case REDIS_HASH_ZIPMAP:
-    case REDIS_LIST_ZIPLIST:
-    case REDIS_SET_INTSET:
-    case REDIS_ZSET_ZIPLIST:
-    case REDIS_HASH_ZIPLIST:
+    case REDIS_RDB_TYPE_STRING:
+    case REDIS_RDB_TYPE_HASH_ZIPMAP:
+    case REDIS_RDB_TYPE_LIST_ZIPLIST:
+    case REDIS_RDB_TYPE_SET_INTSET:
+    case REDIS_RDB_TYPE_ZSET_ZIPLIST:
+    case REDIS_RDB_TYPE_HASH_ZIPLIST:
         if (!processStringObject(NULL)) {
             SHIFT_ERROR(offset, "Error reading entry value");
             return 0;
         }
     break;
-    case REDIS_LIST:
-    case REDIS_SET:
+    case REDIS_RDB_TYPE_LIST:
+    case REDIS_RDB_TYPE_SET:
         for (i = 0; i < length; i++) {
             offset = CURR_OFFSET;
             if (!processStringObject(NULL)) {
@@ -449,7 +389,7 @@ int loadPair(entry *e) {
             }
         }
     break;
-    case REDIS_ZSET:
+    case REDIS_RDB_TYPE_ZSET:
         for (i = 0; i < length; i++) {
             offset = CURR_OFFSET;
             if (!processStringObject(NULL)) {
@@ -463,7 +403,7 @@ int loadPair(entry *e) {
             }
         }
     break;
-    case REDIS_HASH:
+    case REDIS_RDB_TYPE_HASH:
         for (i = 0; i < length; i++) {
             offset = CURR_OFFSET;
             if (!processStringObject(NULL)) {
@@ -486,7 +426,7 @@ int loadPair(entry *e) {
     return 1;
 }
 
-entry loadEntry() {
+static entry loadEntry() {
     entry e = { NULL, -1, 0 };
     uint32_t length, offset[4];
 
@@ -499,7 +439,7 @@ entry loadEntry() {
     }
 
     offset[1] = CURR_OFFSET;
-    if (e.type == REDIS_SELECTDB) {
+    if (e.type == REDIS_RDB_OPCODE_SELECTDB) {
         if ((length = loadLength(NULL)) == REDIS_RDB_LENERR) {
             SHIFT_ERROR(offset[1], "Error reading database number");
             return e;
@@ -508,7 +448,7 @@ entry loadEntry() {
             SHIFT_ERROR(offset[1], "Database number out of range (%d)", length);
             return e;
         }
-    } else if (e.type == REDIS_EOF) {
+    } else if (e.type == REDIS_RDB_OPCODE_EOF) {
         if (positions[level].offset < positions[level].size) {
             SHIFT_ERROR(offset[0], "Unexpected EOF");
         } else {
@@ -517,8 +457,8 @@ entry loadEntry() {
         return e;
     } else {
         /* optionally consume expire */
-        if (e.type == REDIS_EXPIRETIME ||
-            e.type == REDIS_EXPIRETIME_MS) {
+        if (e.type == REDIS_RDB_OPCODE_EXPIRETIME ||
+            e.type == REDIS_RDB_OPCODE_EXPIRETIME_MS) {
             if (!processTime(e.type)) return e;
             if (!loadType(&e)) return e;
         }
@@ -544,31 +484,31 @@ entry loadEntry() {
     return e;
 }
 
-void printCentered(int indent, int width, char* body) {
+static void printCentered(int indent, int width, char* body) {
     char head[256], tail[256];
     memset(head, '\0', 256);
     memset(tail, '\0', 256);
 
     memset(head, '=', indent);
     memset(tail, '=', width - 2 - indent - strlen(body));
-    printf("%s %s %s\n", head, body, tail);
+    redisLog(REDIS_WARNING, "%s %s %s", head, body, tail);
 }
 
-void printValid(uint64_t ops, uint64_t bytes) {
+static void printValid(uint64_t ops, uint64_t bytes) {
     char body[80];
     sprintf(body, "Processed %llu valid opcodes (in %llu bytes)",
         (unsigned long long) ops, (unsigned long long) bytes);
     printCentered(4, 80, body);
 }
 
-void printSkipped(uint64_t bytes, uint64_t offset) {
+static void printSkipped(uint64_t bytes, uint64_t offset) {
     char body[80];
     sprintf(body, "Skipped %llu bytes (resuming at 0x%08llx)",
         (unsigned long long) bytes, (unsigned long long) offset);
     printCentered(4, 80, body);
 }
 
-void printErrorStack(entry *e) {
+static void printErrorStack(entry *e) {
     unsigned int i;
     char body[64];
 
@@ -598,20 +538,20 @@ void printErrorStack(entry *e) {
 
     /* display error stack */
     for (i = 0; i < errors.level; i++) {
-        printf("0x%08lx - %s\n",
+        redisLog(REDIS_WARNING, "0x%08lx - %s",
             (unsigned long) errors.offset[i], errors.error[i]);
     }
 }
 
 void process(void) {
     uint64_t num_errors = 0, num_valid_ops = 0, num_valid_bytes = 0;
-    entry entry;
+    entry entry = { NULL, -1, 0 };
     int dump_version = processHeader();
 
     /* Exclude the final checksum for RDB >= 5. Will be checked at the end. */
     if (dump_version >= 5) {
         if (positions[0].size < 8) {
-            printf("RDB version >= 5 but no room for checksum.\n");
+            redisLog(REDIS_WARNING, "RDB version >= 5 but no room for checksum.");
             exit(1);
         }
         positions[0].size -= 8;
@@ -660,7 +600,7 @@ void process(void) {
             /* advance position */
             positions[0] = positions[1];
         }
-        free(entry.key);
+        zfree(entry.key);
     }
 
     /* because there is another potential error,
@@ -668,7 +608,7 @@ void process(void) {
     printValid(num_valid_ops, num_valid_bytes);
 
     /* expect an eof */
-    if (entry.type != REDIS_EOF) {
+    if (entry.type != REDIS_RDB_OPCODE_EOF) {
         /* last byte should be EOF, add error */
         errors.level = 0;
         SHIFT_ERROR(positions[0].offset, "Expected EOF, got %s", types[entry.type]);
@@ -696,47 +636,40 @@ void process(void) {
         if (crc != crc2) {
             SHIFT_ERROR(positions[0].offset, "RDB CRC64 does not match.");
         } else {
-            printf("CRC64 checksum is OK\n");
+            redisLog(REDIS_WARNING, "CRC64 checksum is OK");
         }
     }
 
     /* print summary on errors */
     if (num_errors) {
-        printf("\n");
-        printf("Total unprocessable opcodes: %llu\n",
+        redisLog(REDIS_WARNING, "Total unprocessable opcodes: %llu",
             (unsigned long long) num_errors);
     }
 }
 
-int main(int argc, char **argv) {
-    /* expect the first argument to be the dump file */
-    if (argc <= 1) {
-        printf("Usage: %s <dump.rdb>\n", argv[0]);
-        exit(0);
-    }
-
+int redis_check_rdb(char *rdbfilename) {
     int fd;
     off_t size;
     struct stat stat;
     void *data;
 
-    fd = open(argv[1], O_RDONLY);
+    fd = open(rdbfilename, O_RDONLY);
     if (fd < 1) {
-        ERROR("Cannot open file: %s\n", argv[1]);
+        ERROR("Cannot open file: %s", rdbfilename);
     }
     if (fstat(fd, &stat) == -1) {
-        ERROR("Cannot stat: %s\n", argv[1]);
+        ERROR("Cannot stat: %s", rdbfilename);
     } else {
         size = stat.st_size;
     }
 
     if (sizeof(size_t) == sizeof(int32_t) && size >= INT_MAX) {
-        ERROR("Cannot check dump files >2GB on a 32-bit platform\n");
+        ERROR("Cannot check dump files >2GB on a 32-bit platform");
     }
 
     data = mmap(NULL, size, PROT_READ, MAP_SHARED, fd, 0);
     if (data == MAP_FAILED) {
-        ERROR("Cannot mmap: %s\n", argv[1]);
+        ERROR("Cannot mmap: %s", rdbfilename);
     }
 
     /* Initialize static vars */
@@ -746,22 +679,16 @@ int main(int argc, char **argv) {
     errors.level = 0;
 
     /* Object types */
-    sprintf(types[REDIS_STRING], "STRING");
-    sprintf(types[REDIS_LIST], "LIST");
-    sprintf(types[REDIS_SET], "SET");
-    sprintf(types[REDIS_ZSET], "ZSET");
-    sprintf(types[REDIS_HASH], "HASH");
+    sprintf(types[REDIS_RDB_TYPE_STRING], "STRING");
+    sprintf(types[REDIS_RDB_TYPE_LIST], "LIST");
+    sprintf(types[REDIS_RDB_TYPE_SET], "SET");
+    sprintf(types[REDIS_RDB_TYPE_ZSET], "ZSET");
+    sprintf(types[REDIS_RDB_TYPE_HASH], "HASH");
 
     /* Object types only used for dumping to disk */
-    sprintf(types[REDIS_EXPIRETIME], "EXPIRETIME");
-    sprintf(types[REDIS_SELECTDB], "SELECTDB");
-    sprintf(types[REDIS_EOF], "EOF");
-
-    /* Double constants initialization */
-    R_Zero = 0.0;
-    R_PosInf = 1.0/R_Zero;
-    R_NegInf = -1.0/R_Zero;
-    R_Nan = R_Zero/R_Zero;
+    sprintf(types[REDIS_RDB_OPCODE_EXPIRETIME], "EXPIRETIME");
+    sprintf(types[REDIS_RDB_OPCODE_SELECTDB], "SELECTDB");
+    sprintf(types[REDIS_RDB_OPCODE_EOF], "EOF");
 
     process();
 
@@ -769,3 +696,15 @@ int main(int argc, char **argv) {
     close(fd);
     return 0;
 }
+
+/* RDB check main: called form redis.c when Redis is executed with the
+ * redis-check-rdb alias. */
+int redis_check_rdb_main(char **argv, int argc) {
+    if (argc != 2) {
+        fprintf(stderr, "Usage: %s <rdb-file-name>\n", argv[0]);
+        exit(1);
+    }
+    redisLog(REDIS_WARNING, "Checking RDB file %s", argv[1]);
+    exit(redis_check_rdb(argv[1]));
+    return 0;
+}

src/redis-cli.c

@@ -44,6 +44,7 @@
 #include <assert.h>
 #include <fcntl.h>
 #include <limits.h>
+#include <math.h>
 
 #include "hiredis.h"
 #include "sds.h"
@@ -60,6 +61,8 @@
 #define OUTPUT_CSV 2
 #define REDIS_CLI_KEEPALIVE_INTERVAL 15 /* seconds */
 #define REDIS_CLI_DEFAULT_PIPE_TIMEOUT 30 /* seconds */
+#define REDIS_CLI_HISTFILE_ENV "REDISCLI_HISTFILE"
+#define REDIS_CLI_HISTFILE_DEFAULT ".rediscli_history"
 
 static redisContext *context;
 static struct config {
@@ -74,6 +77,7 @@ static struct config {
     int monitor_mode;
     int pubsub_mode;
     int latency_mode;
+    int latency_dist_mode;
     int latency_history;
     int cluster_mode;
     int cluster_reissue_command;
@@ -128,9 +132,8 @@ static void cliRefreshPrompt(void) {
         len = snprintf(config.prompt,sizeof(config.prompt),"redis %s",
                        config.hostsocket);
     else
-        len = snprintf(config.prompt,sizeof(config.prompt),
-                       strchr(config.hostip,':') ? "[%s]:%d" : "%s:%d",
-                       config.hostip, config.hostport);
+        len = anetFormatAddr(config.prompt, sizeof(config.prompt),
+                           config.hostip, config.hostport);
     /* Add [dbnum] if needed */
     if (config.dbnum != 0 && config.last_cmd_type != REDIS_REPLY_ERROR)
         len += snprintf(config.prompt+len,sizeof(config.prompt)-len,"[%d]",
@@ -138,6 +141,30 @@ static void cliRefreshPrompt(void) {
     snprintf(config.prompt+len,sizeof(config.prompt)-len,"> ");
 }
 
+static sds getHistoryPath() {
+    char *path = NULL;
+    sds historyPath = NULL;
+
+    /* check the env for a histfile override */
+    path = getenv(REDIS_CLI_HISTFILE_ENV);
+    if (path != NULL && *path != '\0') {
+        if (!strcmp("/dev/null", path)) {
+            return NULL;
+        }
+
+        /* if the env is set, return it */
+        historyPath = sdscatprintf(sdsempty(), "%s", path);
+    } else {
+        char *home = getenv("HOME");
+        if (home != NULL && *home != '\0') {
+            /* otherwise, return the default */
+            historyPath = sdscatprintf(sdsempty(), "%s/%s", home, REDIS_CLI_HISTFILE_DEFAULT);
+        }
+    }
+
+    return historyPath;
+}
+
 /*------------------------------------------------------------------------------
  * Help functions
  *--------------------------------------------------------------------------- */
@@ -302,7 +329,7 @@ static void completionCallback(const char *buf, linenoiseCompletions *lc) {
  *--------------------------------------------------------------------------- */
 
 /* Send AUTH command to the server */
-static int cliAuth() {
+static int cliAuth(void) {
     redisReply *reply;
     if (config.auth == NULL) return REDIS_OK;
 
@@ -315,7 +342,7 @@ static int cliAuth() {
 }
 
 /* Send SELECT dbnum to the server */
-static int cliSelect() {
+static int cliSelect(void) {
     redisReply *reply;
     if (config.dbnum == 0) return REDIS_OK;
 
@@ -493,7 +520,7 @@ static sds cliFormatReplyCSV(redisReply *r) {
         out = sdscatrepr(out,r->str,r->len);
     break;
     case REDIS_REPLY_NIL:
-        out = sdscat(out,"NIL\n");
+        out = sdscat(out,"NIL");
     break;
     case REDIS_REPLY_ARRAY:
         for (i = 0; i < r->elements; i++) {
@@ -524,7 +551,8 @@ static int cliReadReply(int output_raw_strings) {
         }
         if (config.interactive) {
             /* Filter cases where we should reconnect */
-            if (context->err == REDIS_ERR_IO && errno == ECONNRESET)
+            if (context->err == REDIS_ERR_IO &&
+                (errno == ECONNRESET || errno == EPIPE))
                 return REDIS_ERR;
             if (context->err == REDIS_ERR_EOF)
                 return REDIS_ERR;
@@ -603,6 +631,9 @@ static int cliSendCommand(int argc, char **argv, int repeat) {
 
     output_raw = 0;
     if (!strcasecmp(command,"info") ||
+        (argc == 3 && !strcasecmp(command,"debug") &&
+                      (!strcasecmp(argv[1],"jemalloc") &&
+                       !strcasecmp(argv[2],"info"))) ||
         (argc == 2 && !strcasecmp(command,"cluster") &&
                       (!strcasecmp(argv[1],"nodes") ||
                        !strcasecmp(argv[1],"info"))) ||
@@ -741,6 +772,8 @@ static int parseOptions(int argc, char **argv) {
             config.output = OUTPUT_CSV;
         } else if (!strcmp(argv[i],"--latency")) {
             config.latency_mode = 1;
+        } else if (!strcmp(argv[i],"--latency-dist")) {
+            config.latency_dist_mode = 1;
         } else if (!strcmp(argv[i],"--latency-history")) {
             config.latency_mode = 1;
             config.latency_history = 1;
@@ -829,9 +862,12 @@ static void usage(void) {
 "                     not a tty).\n"
 "  --no-raw           Force formatted output even when STDOUT is not a tty.\n"
 "  --csv              Output in CSV format.\n"
+"  --stat             Print rolling stats about server: mem, clients, ...\n"
 "  --latency          Enter a special mode continuously sampling latency.\n"
 "  --latency-history  Like --latency but tracking latency changes over time.\n"
 "                     Default time interval is 15 sec. Change it using -i.\n"
+"  --latency-dist     Shows latency as a spectrum, requires xterm 256 colors.\n"
+"                     Default time interval is 1 sec. Change it using -i.\n"
 "  --slave            Simulate a slave showing commands received from the master.\n"
 "  --rdb <filename>   Transfer an RDB dump from remote server to local file.\n"
 "  --pipe             Transfer raw Redis protocol from stdin to server.\n"
@@ -876,6 +912,33 @@ static char **convertToSds(int count, char** args) {
   return sds;
 }
 
+static int issueCommandRepeat(int argc, char **argv, long repeat) {
+    while (1) {
+        config.cluster_reissue_command = 0;
+        if (cliSendCommand(argc,argv,repeat) != REDIS_OK) {
+            cliConnect(1);
+
+            /* If we still cannot send the command print error.
+             * We'll try to reconnect the next time. */
+            if (cliSendCommand(argc,argv,repeat) != REDIS_OK) {
+                cliPrintContextError();
+                return REDIS_ERR;
+            }
+         }
+         /* Issue the command again if we got redirected in cluster mode */
+         if (config.cluster_mode && config.cluster_reissue_command) {
+            cliConnect(1);
+         } else {
+             break;
+        }
+    }
+    return REDIS_OK;
+}
+
+static int issueCommand(int argc, char **argv) {
+    return issueCommandRepeat(argc, argv, config.repeat);
+}
+
 static void repl(void) {
     sds historyfile = NULL;
     int history = 0;
@@ -889,10 +952,9 @@ static void repl(void) {
 
     /* Only use history when stdin is a tty. */
     if (isatty(fileno(stdin))) {
-        history = 1;
-
-        if (getenv("HOME") != NULL) {
-            historyfile = sdscatprintf(sdsempty(),"%s/.rediscli_history",getenv("HOME"));
+        historyfile = getHistoryPath();
+        if (historyfile != NULL) {
+            history = 1;
             linenoiseHistoryLoad(historyfile);
         }
     }
@@ -932,26 +994,8 @@ static void repl(void) {
                         repeat = 1;
                     }
 
-                    while (1) {
-                        config.cluster_reissue_command = 0;
-                        if (cliSendCommand(argc-skipargs,argv+skipargs,repeat)
-                            != REDIS_OK)
-                        {
-                            cliConnect(1);
+                    issueCommandRepeat(argc-skipargs, argv+skipargs, repeat);
 
-                            /* If we still cannot send the command print error.
-                             * We'll try to reconnect the next time. */
-                            if (cliSendCommand(argc-skipargs,argv+skipargs,repeat)
-                                != REDIS_OK)
-                                cliPrintContextError();
-                        }
-                        /* Issue the command again if we got redirected in cluster mode */
-                        if (config.cluster_mode && config.cluster_reissue_command) {
-                            cliConnect(1);
-                        } else {
-                            break;
-                        }
-                    }
                     elapsed = mstime()-start_time;
                     if (elapsed >= 500) {
                         printf("(%.2fs)\n",(double)elapsed/1000);
@@ -972,10 +1016,9 @@ static int noninteractive(int argc, char **argv) {
     if (config.stdinarg) {
         argv = zrealloc(argv, (argc+1)*sizeof(char*));
         argv[argc] = readArgFromStdin();
-        retval = cliSendCommand(argc+1, argv, config.repeat);
+        retval = issueCommand(argc+1, argv);
     } else {
-        /* stdin is probably a tty, can be tested with S_ISCHR(s.st_mode) */
-        retval = cliSendCommand(argc, argv, config.repeat);
+        retval = issueCommand(argc, argv);
     }
     return retval;
 }
@@ -1019,7 +1062,7 @@ static int evalMode(int argc, char **argv) {
     argv2[2] = sdscatprintf(sdsempty(),"%d",keys);
 
     /* Call it */
-    return cliSendCommand(argc+3-got_comma, argv2, config.repeat);
+    return issueCommand(argc+3-got_comma, argv2);
 }
 
 /*------------------------------------------------------------------------------
@@ -1070,6 +1113,155 @@ static void latencyMode(void) {
     }
 }
 
+/*------------------------------------------------------------------------------
+ * Latency distribution mode -- requires 256 colors xterm
+ *--------------------------------------------------------------------------- */
+
+#define LATENCY_DIST_DEFAULT_INTERVAL 1000 /* milliseconds. */
+#define LATENCY_DIST_MIN_GRAY 233 /* Less than that is too hard to see gray. */
+#define LATENCY_DIST_MAX_GRAY 255
+#define LATENCY_DIST_GRAYS (LATENCY_DIST_MAX_GRAY-LATENCY_DIST_MIN_GRAY+1)
+
+/* Gray palette. */
+int spectrum_palette_size = 24;
+int spectrum_palette[] = {0, 233,234,235,236,237,238,239,240,241,242,243,244,245,246,247,248,249,250,251,252,253,254,255};
+
+/* Structure to store samples distribution. */
+struct distsamples {
+    long long max;   /* Max latency to fit into this interval (usec). */
+    long long count; /* Number of samples in this interval. */
+    int character;   /* Associated character in visualization. */
+};
+
+/* Helper function for latencyDistMode(). Performs the spectrum visualization
+ * of the collected samples targeting an xterm 256 terminal.
+ *
+ * Takes an array of distsamples structures, ordered from smaller to bigger
+ * 'max' value. Last sample max must be 0, to mean that it olds all the
+ * samples greater than the previous one, and is also the stop sentinel.
+ *
+ * "tot' is the total number of samples in the different buckets, so it
+ * is the SUM(samples[i].conut) for i to 0 up to the max sample.
+ *
+ * As a side effect the function sets all the buckets count to 0. */
+void showLatencyDistSamples(struct distsamples *samples, long long tot) {
+    int j;
+
+     /* We convert samples into a index inside the palette
+     * proportional to the percentage a given bucket represents.
+     * This way intensity of the different parts of the spectrum
+     * don't change relative to the number of requests, which avoids to
+     * pollute the visualization with non-latency related info. */
+    printf("\033[38;5;0m"); /* Set foreground color to black. */
+    for (j = 0; ; j++) {
+        int coloridx =
+            ceil((float) samples[j].count / tot * (spectrum_palette_size-1));
+        int color = spectrum_palette[coloridx];
+        printf("\033[48;5;%dm%c", (int)color, samples[j].character);
+        samples[j].count = 0;
+        if (samples[j].max == 0) break; /* Last sample. */
+    }
+    printf("\033[0m\n");
+    fflush(stdout);
+}
+
+/* Show the legend: different buckets values and colors meaning, so
+ * that the spectrum is more easily readable. */
+void showLatencyDistLegend(void) {
+    int j;
+
+    printf("---------------------------------------------\n");
+    printf(". - * #          .01 .125 .25 .5 milliseconds\n");
+    printf("1,2,3,...,9      from 1 to 9     milliseconds\n");
+    printf("A,B,C,D,E        10,20,30,40,50  milliseconds\n");
+    printf("F,G,H,I,J        .1,.2,.3,.4,.5       seconds\n");
+    printf("K,L,M,N,O,P,Q,?  1,2,4,8,16,30,60,>60 seconds\n");
+    printf("From 0 to 100%%: ");
+    for (j = 0; j < spectrum_palette_size; j++) {
+        printf("\033[48;5;%dm ", spectrum_palette[j]);
+    }
+    printf("\033[0m\n");
+    printf("---------------------------------------------\n");
+}
+
+static void latencyDistMode(void) {
+    redisReply *reply;
+    long long start, latency, count = 0;
+    long long history_interval =
+        config.interval ? config.interval/1000 :
+                          LATENCY_DIST_DEFAULT_INTERVAL;
+    long long history_start = ustime();
+    int j, outputs = 0;
+
+    struct distsamples samples[] = {
+        /* We use a mostly logarithmic scale, with certain linear intervals
+         * which are more interesting than others, like 1-10 milliseconds
+         * range. */
+        {10,0,'.'},         /* 0.01 ms */
+        {125,0,'-'},        /* 0.125 ms */
+        {250,0,'*'},        /* 0.25 ms */
+        {500,0,'#'},        /* 0.5 ms */
+        {1000,0,'1'},       /* 1 ms */
+        {2000,0,'2'},       /* 2 ms */
+        {3000,0,'3'},       /* 3 ms */
+        {4000,0,'4'},       /* 4 ms */
+        {5000,0,'5'},       /* 5 ms */
+        {6000,0,'6'},       /* 6 ms */
+        {7000,0,'7'},       /* 7 ms */
+        {8000,0,'8'},       /* 8 ms */
+        {9000,0,'9'},       /* 9 ms */
+        {10000,0,'A'},      /* 10 ms */
+        {20000,0,'B'},      /* 20 ms */
+        {30000,0,'C'},      /* 30 ms */
+        {40000,0,'D'},      /* 40 ms */
+        {50000,0,'E'},      /* 50 ms */
+        {100000,0,'F'},     /* 0.1 s */
+        {200000,0,'G'},     /* 0.2 s */
+        {300000,0,'H'},     /* 0.3 s */
+        {400000,0,'I'},     /* 0.4 s */
+        {500000,0,'J'},     /* 0.5 s */
+        {1000000,0,'K'},    /* 1 s */
+        {2000000,0,'L'},    /* 2 s */
+        {4000000,0,'M'},    /* 4 s */
+        {8000000,0,'N'},    /* 8 s */
+        {16000000,0,'O'},   /* 16 s */
+        {30000000,0,'P'},   /* 30 s */
+        {60000000,0,'Q'},   /* 1 minute */
+        {0,0,'?'},          /* > 1 minute */
+    };
+
+    if (!context) exit(1);
+    while(1) {
+        start = ustime();
+        reply = redisCommand(context,"PING");
+        if (reply == NULL) {
+            fprintf(stderr,"\nI/O error\n");
+            exit(1);
+        }
+        latency = ustime()-start;
+        freeReplyObject(reply);
+        count++;
+
+        /* Populate the relevant bucket. */
+        for (j = 0; ; j++) {
+            if (samples[j].max == 0 || latency <= samples[j].max) {
+                samples[j].count++;
+                break;
+            }
+        }
+
+        /* From time to time show the spectrum. */
+        if (count && (ustime()-history_start)/1000 > history_interval) {
+            if ((outputs++ % 20) == 0)
+                showLatencyDistLegend();
+            showLatencyDistSamples(samples,count);
+            history_start = ustime();
+            count = 0;
+        }
+        usleep(LATENCY_SAMPLE_RATE * 1000);
+    }
+}
+
 /*------------------------------------------------------------------------------
  * Slave mode
  *--------------------------------------------------------------------------- */
@@ -1511,7 +1703,7 @@ static void findBigKeys(void) {
     for(i=0;i<TYPE_NONE; i++) {
         maxkeys[i] = sdsempty();
         if(!maxkeys[i]) {
-            fprintf(stderr, "Failed to allocate memory for largest key names!");
+            fprintf(stderr, "Failed to allocate memory for largest key names!\n");
             exit(1);
         }
     }
@@ -1743,6 +1935,7 @@ static void statMode(void) {
         /* Children */
         aux = getLongInfoField(reply->str,"bgsave_in_progress");
         aux |= getLongInfoField(reply->str,"aof_rewrite_in_progress") << 1;
+        aux |= getLongInfoField(reply->str,"loading") << 2;
         switch(aux) {
         case 0: break;
         case 1:
@@ -1754,6 +1947,9 @@ static void statMode(void) {
         case 3:
             printf("SAVE+AOF");
             break;
+        case 4:
+            printf("LOAD");
+            break;
         }
 
         printf("\n");
@@ -1886,6 +2082,7 @@ int main(int argc, char **argv) {
     config.monitor_mode = 0;
     config.pubsub_mode = 0;
     config.latency_mode = 0;
+    config.latency_dist_mode = 0;
     config.latency_history = 0;
     config.cluster_mode = 0;
     config.slave_mode = 0;
@@ -1920,6 +2117,12 @@ int main(int argc, char **argv) {
         latencyMode();
     }
 
+    /* Latency distribution mode */
+    if (config.latency_dist_mode) {
+        if (cliConnect(0) == REDIS_ERR) exit(1);
+        latencyDistMode();
+    }
+
     /* Slave mode */
     if (config.slave_mode) {
         if (cliConnect(0) == REDIS_ERR) exit(1);
@@ -1962,6 +2165,9 @@ int main(int argc, char **argv) {
 
     /* Start interactive mode when no command is provided */
     if (argc == 0 && !config.eval) {
+        /* Ignore SIGPIPE in interactive mode to force a reconnect */
+        signal(SIGPIPE, SIG_IGN);
+
         /* Note that in repl mode we don't abort on connection error.
          * A new attempt will be performed for every command send. */
         cliConnect(0);

src/redis-trib.rb

@@ -50,14 +50,16 @@ end
 class ClusterNode
     def initialize(addr)
         s = addr.split(":")
-        if s.length != 2
-            puts "Invalid node name #{addr}"
-            exit 1
+        if s.length < 2
+           puts "Invalid IP or Port (given as #{addr}) - use IP:Port format"
+           exit 1
         end
+        port = s.pop # removes port from split array
+        ip = s.join(":") # if s.length > 1 here, it's IPv6, so restore address
         @r = nil
         @info = {}
-        @info[:host] = s[0]
-        @info[:port] = s[1]
+        @info[:host] = ip
+        @info[:port] = port
         @info[:slots] = {}
         @info[:migrating] = {}
         @info[:importing] = {}
@@ -70,7 +72,7 @@ class ClusterNode
         @friends
     end
 
-    def slots 
+    def slots
         @info[:slots]
     end
 
@@ -152,7 +154,7 @@ class ClusterNode
                     end
                 } if slots
                 @dirty = false
-                @r.cluster("info").split("\n").each{|e|    
+                @r.cluster("info").split("\n").each{|e|
                     k,v=e.split(":")
                     k = k.to_sym
                     v.chop!
@@ -211,7 +213,7 @@ class ClusterNode
         #
         # Note: this could be easily written without side effects,
         # we use 'slots' just to split the computation into steps.
-        
+
         # First step: we want an increasing array of integers
         # for instance: [1,2,3,4,5,8,9,20,21,22,23,24,25,30]
         slots = @info[:slots].keys.sort
@@ -271,7 +273,7 @@ class ClusterNode
     def info
         @info
     end
-    
+
     def is_dirty?
         @dirty
     end
@@ -538,7 +540,6 @@ class RedisTrib
         nodes_count = @nodes.length
         masters_count = @nodes.length / (@replicas+1)
         masters = []
-        slaves = []
 
         # The first step is to split instances by IP. This is useful as
         # we'll try to allocate master nodes in different physical machines
@@ -556,16 +557,31 @@ class RedisTrib
 
         # Select master instances
         puts "Using #{masters_count} masters:"
-        while masters.length < masters_count
-            ips.each{|ip,nodes_list|
-                next if nodes_list.length == 0
-                masters << nodes_list.shift
-                puts masters[-1]
-                nodes_count -= 1
-                break if masters.length == masters_count
-            }
+        interleaved = []
+        stop = false
+        while not stop do
+            # Take one node from each IP until we run out of nodes
+            # across every IP.
+            ips.each do |ip,nodes|
+                if nodes.empty?
+                    # if this IP has no remaining nodes, check for termination
+                    if interleaved.length == nodes_count
+                        # stop when 'interleaved' has accumulated all nodes
+                        stop = true
+                        next
+                    end
+                else
+                    # else, move one node from this IP to 'interleaved'
+                    interleaved.push nodes.shift
+                end
+            end
         end
 
+        masters = interleaved.slice!(0, masters_count)
+        nodes_count -= masters.length
+
+        masters.each{|m| puts m}
+
         # Alloc slots on masters
         slots_per_node = ClusterHashSlots.to_f / masters_count
         first = 0
@@ -592,8 +608,8 @@ class RedisTrib
         # all nodes will be used.
         assignment_verbose = false
 
-        [:requested,:unused].each{|assign|
-            masters.each{|m|
+        [:requested,:unused].each do |assign|
+            masters.each do |m|
                 assigned_replicas = 0
                 while assigned_replicas < @replicas
                     break if nodes_count == 0
@@ -607,21 +623,33 @@ class RedisTrib
                                  "role too (#{nodes_count} remaining)."
                         end
                     end
-                    ips.each{|ip,nodes_list|
-                        next if nodes_list.length == 0
-                        # Skip instances with the same IP as the master if we
-                        # have some more IPs available.
-                        next if ip == m.info[:host] && nodes_count > nodes_list.length
-                        slave = nodes_list.shift
-                        slave.set_as_replica(m.info[:name])
-                        nodes_count -= 1
-                        assigned_replicas += 1
-                        puts "Adding replica #{slave} to #{m}"
-                        break
-                    }
+
+                    # Return the first node not matching our current master
+                    node = interleaved.find{|n| n.info[:host] != m.info[:host]}
+
+                    # If we found a node, use it as a best-first match.
+                    # Otherwise, we didn't find a node on a different IP, so we
+                    # go ahead and use a same-IP replica.
+                    if node
+                        slave = node
+                        interleaved.delete node
+                    else
+                        slave = interleaved.shift
+                    end
+                    slave.set_as_replica(m.info[:name])
+                    nodes_count -= 1
+                    assigned_replicas += 1
+                    puts "Adding replica #{slave} to #{m}"
+
+                    # If we are in the "assign extra nodes" loop,
+                    # we want to assign one extra replica to each
+                    # master before repeating masters.
+                    # This break lets us assign extra replicas to masters
+                    # in a round-robin way.
+                    break if assign == :unused
                 end
-            }
-        }
+            end
+        end
     end
 
     def flush_nodes_config
@@ -761,7 +789,7 @@ class RedisTrib
 
     # Move slots between source and target nodes using MIGRATE.
     #
-    # Options: 
+    # Options:
     # :verbose -- Print a dot for every moved key.
     # :fix     -- We are moving in the context of a fix. Use REPLACE.
     # :cold    -- Move keys without opening / reconfiguring the nodes.
@@ -1137,7 +1165,7 @@ class RedisTrib
         # right node as needed.
         cursor = nil
         while cursor != 0
-            cursor,keys = source.scan(cursor,:count,1000)
+            cursor,keys = source.scan(cursor, :count => 1000)
             cursor = cursor.to_i
             keys.each{|k|
                 # Migrate keys using the MIGRATE command.
@@ -1204,7 +1232,7 @@ end
 
 #################################################################################
 # Libraries
-# 
+#
 # We try to don't depend on external libs since this is a critical part
 # of Redis Cluster.
 #################################################################################

src/redis.c

@@ -31,6 +31,7 @@
 #include "cluster.h"
 #include "slowlog.h"
 #include "bio.h"
+#include "latency.h"
 
 #include <time.h>
 #include <signal.h>
@@ -45,12 +46,14 @@
 #include <sys/time.h>
 #include <sys/resource.h>
 #include <sys/uio.h>
+#include <sys/un.h>
 #include <limits.h>
 #include <float.h>
 #include <math.h>
 #include <sys/resource.h>
 #include <sys/utsname.h>
 #include <locale.h>
+#include <sys/sysctl.h>
 
 /* Our shared "common" objects */
 
@@ -159,7 +162,7 @@ struct redisCommand redisCommandTable[] = {
     {"smove",smoveCommand,4,"wF",0,NULL,1,2,1,0,0},
     {"sismember",sismemberCommand,3,"rF",0,NULL,1,1,1,0,0},
     {"scard",scardCommand,2,"rF",0,NULL,1,1,1,0,0},
-    {"spop",spopCommand,2,"wRsF",0,NULL,1,1,1,0,0},
+    {"spop",spopCommand,-2,"wRsF",0,NULL,1,1,1,0,0},
     {"srandmember",srandmemberCommand,-2,"rR",0,NULL,1,1,1,0,0},
     {"sinter",sinterCommand,-2,"rS",0,NULL,1,-1,1,0,0},
     {"sinterstore",sinterstoreCommand,-3,"wm",0,NULL,1,-1,1,0,0},
@@ -246,7 +249,7 @@ struct redisCommand redisCommandTable[] = {
     {"pttl",pttlCommand,2,"rF",0,NULL,1,1,1,0,0},
     {"persist",persistCommand,2,"wF",0,NULL,1,1,1,0,0},
     {"slaveof",slaveofCommand,3,"ast",0,NULL,0,0,0,0,0},
-    {"role",roleCommand,1,"last",0,NULL,0,0,0,0,0},
+    {"role",roleCommand,1,"lst",0,NULL,0,0,0,0,0},
     {"debug",debugCommand,-2,"as",0,NULL,0,0,0,0,0},
     {"config",configCommand,-2,"art",0,NULL,0,0,0,0,0},
     {"subscribe",subscribeCommand,-2,"rpslt",0,NULL,0,0,0,0,0},
@@ -258,19 +261,19 @@ struct redisCommand redisCommandTable[] = {
     {"watch",watchCommand,-2,"rsF",0,NULL,1,-1,1,0,0},
     {"unwatch",unwatchCommand,1,"rsF",0,NULL,0,0,0,0,0},
     {"cluster",clusterCommand,-2,"ar",0,NULL,0,0,0,0,0},
-    {"restore",restoreCommand,-4,"awm",0,NULL,1,1,1,0,0},
-    {"restore-asking",restoreCommand,-4,"awmk",0,NULL,1,1,1,0,0},
-    {"migrate",migrateCommand,-6,"aw",0,NULL,0,0,0,0,0},
+    {"restore",restoreCommand,-4,"wm",0,NULL,1,1,1,0,0},
+    {"restore-asking",restoreCommand,-4,"wmk",0,NULL,1,1,1,0,0},
+    {"migrate",migrateCommand,-6,"w",0,NULL,0,0,0,0,0},
     {"asking",askingCommand,1,"r",0,NULL,0,0,0,0,0},
     {"readonly",readonlyCommand,1,"rF",0,NULL,0,0,0,0,0},
     {"readwrite",readwriteCommand,1,"rF",0,NULL,0,0,0,0,0},
-    {"dump",dumpCommand,2,"ar",0,NULL,1,1,1,0,0},
+    {"dump",dumpCommand,2,"r",0,NULL,1,1,1,0,0},
     {"object",objectCommand,3,"r",0,NULL,2,2,2,0,0},
-    {"client",clientCommand,-2,"ars",0,NULL,0,0,0,0,0},
+    {"client",clientCommand,-2,"rs",0,NULL,0,0,0,0,0},
     {"eval",evalCommand,-3,"s",0,evalGetKeys,0,0,0,0,0},
     {"evalsha",evalShaCommand,-3,"s",0,evalGetKeys,0,0,0,0,0},
     {"slowlog",slowlogCommand,-2,"r",0,NULL,0,0,0,0,0},
-    {"script",scriptCommand,-2,"ras",0,NULL,0,0,0,0,0},
+    {"script",scriptCommand,-2,"rs",0,NULL,0,0,0,0,0},
     {"time",timeCommand,1,"rRF",0,NULL,0,0,0,0,0},
     {"bitop",bitopCommand,-4,"wm",0,NULL,2,-1,1,0,0},
     {"bitcount",bitcountCommand,-2,"r",0,NULL,1,1,1,0,0},
@@ -279,7 +282,7 @@ struct redisCommand redisCommandTable[] = {
     {"command",commandCommand,0,"rlt",0,NULL,0,0,0,0,0},
     {"pfselftest",pfselftestCommand,1,"r",0,NULL,0,0,0,0,0},
     {"pfadd",pfaddCommand,-2,"wmF",0,NULL,1,1,1,0,0},
-    {"pfcount",pfcountCommand,-2,"w",0,NULL,1,1,1,0,0},
+    {"pfcount",pfcountCommand,-2,"r",0,NULL,1,1,1,0,0},
     {"pfmerge",pfmergeCommand,-2,"wm",0,NULL,1,-1,1,0,0},
     {"pfdebug",pfdebugCommand,-3,"w",0,NULL,0,0,0,0,0},
     {"latency",latencyCommand,-2,"arslt",0,NULL,0,0,0,0,0}
@@ -875,27 +878,30 @@ unsigned int getLRUClock(void) {
 }
 
 /* Add a sample to the operations per second array of samples. */
-void trackOperationsPerSecond(void) {
-    long long t = mstime() - server.ops_sec_last_sample_time;
-    long long ops = server.stat_numcommands - server.ops_sec_last_sample_ops;
+void trackInstantaneousMetric(int metric, long long current_reading) {
+    long long t = mstime() - server.inst_metric[metric].last_sample_time;
+    long long ops = current_reading -
+                    server.inst_metric[metric].last_sample_count;
     long long ops_sec;
 
     ops_sec = t > 0 ? (ops*1000/t) : 0;
 
-    server.ops_sec_samples[server.ops_sec_idx] = ops_sec;
-    server.ops_sec_idx = (server.ops_sec_idx+1) % REDIS_OPS_SEC_SAMPLES;
-    server.ops_sec_last_sample_time = mstime();
-    server.ops_sec_last_sample_ops = server.stat_numcommands;
+    server.inst_metric[metric].samples[server.inst_metric[metric].idx] =
+        ops_sec;
+    server.inst_metric[metric].idx++;
+    server.inst_metric[metric].idx %= REDIS_METRIC_SAMPLES;
+    server.inst_metric[metric].last_sample_time = mstime();
+    server.inst_metric[metric].last_sample_count = current_reading;
 }
 
 /* Return the mean of all the samples. */
-long long getOperationsPerSecond(void) {
+long long getInstantaneousMetric(int metric) {
     int j;
     long long sum = 0;
 
-    for (j = 0; j < REDIS_OPS_SEC_SAMPLES; j++)
-        sum += server.ops_sec_samples[j];
-    return sum / REDIS_OPS_SEC_SAMPLES;
+    for (j = 0; j < REDIS_METRIC_SAMPLES; j++)
+        sum += server.inst_metric[metric].samples[j];
+    return sum / REDIS_METRIC_SAMPLES;
 }
 
 /* Check for timeouts. Returns non-zero if the client was terminated */
@@ -1067,7 +1073,13 @@ int serverCron(struct aeEventLoop *eventLoop, long long id, void *clientData) {
     /* Update the time cache. */
     updateCachedTime();
 
-    run_with_period(100) trackOperationsPerSecond();
+    run_with_period(100) {
+        trackInstantaneousMetric(REDIS_METRIC_COMMAND,server.stat_numcommands);
+        trackInstantaneousMetric(REDIS_METRIC_NET_INPUT,
+                server.stat_net_input_bytes);
+        trackInstantaneousMetric(REDIS_METRIC_NET_OUTPUT,
+                server.stat_net_output_bytes);
+    }
 
     /* We have just REDIS_LRU_BITS bits per object for LRU information.
      * So we use an (eventually wrapping) LRU clock.
@@ -1403,6 +1415,8 @@ void initServerConfig(void) {
     server.syslog_ident = zstrdup(REDIS_DEFAULT_SYSLOG_IDENT);
     server.syslog_facility = LOG_LOCAL0;
     server.daemonize = REDIS_DEFAULT_DAEMONIZE;
+    server.supervised = 0;
+    server.supervised_mode = REDIS_SUPERVISED_NONE;
     server.aof_state = REDIS_AOF_OFF;
     server.aof_fsync = REDIS_DEFAULT_AOF_FSYNC;
     server.aof_no_fsync_on_rewrite = REDIS_DEFAULT_AOF_NO_FSYNC_ON_REWRITE;
@@ -1420,7 +1434,7 @@ void initServerConfig(void) {
     server.aof_flush_postponed_start = 0;
     server.aof_rewrite_incremental_fsync = REDIS_DEFAULT_AOF_REWRITE_INCREMENTAL_FSYNC;
     server.aof_load_truncated = REDIS_DEFAULT_AOF_LOAD_TRUNCATED;
-    server.pidfile = zstrdup(REDIS_DEFAULT_PID_FILE);
+    server.pidfile = NULL;
     server.rdb_filename = zstrdup(REDIS_DEFAULT_RDB_FILENAME);
     server.aof_filename = zstrdup(REDIS_DEFAULT_AOF_FILENAME);
     server.requirepass = NULL;
@@ -1436,8 +1450,8 @@ void initServerConfig(void) {
     server.maxmemory_samples = REDIS_DEFAULT_MAXMEMORY_SAMPLES;
     server.hash_max_ziplist_entries = REDIS_HASH_MAX_ZIPLIST_ENTRIES;
     server.hash_max_ziplist_value = REDIS_HASH_MAX_ZIPLIST_VALUE;
-    server.list_max_ziplist_entries = REDIS_LIST_MAX_ZIPLIST_ENTRIES;
-    server.list_max_ziplist_value = REDIS_LIST_MAX_ZIPLIST_VALUE;
+    server.list_max_ziplist_size = REDIS_LIST_MAX_ZIPLIST_SIZE;
+    server.list_compress_depth = REDIS_LIST_COMPRESS_DEPTH;
     server.set_max_intset_entries = REDIS_SET_MAX_INTSET_ENTRIES;
     server.zset_max_ziplist_entries = REDIS_ZSET_MAX_ZIPLIST_ENTRIES;
     server.zset_max_ziplist_value = REDIS_ZSET_MAX_ZIPLIST_VALUE;
@@ -1515,6 +1529,7 @@ void initServerConfig(void) {
     server.lpushCommand = lookupCommandByCString("lpush");
     server.lpopCommand = lookupCommandByCString("lpop");
     server.rpopCommand = lookupCommandByCString("rpop");
+    server.sremCommand = lookupCommandByCString("srem");
 
     /* Slow log */
     server.slowlog_log_slower_than = REDIS_SLOWLOG_LOG_SLOWER_THAN;
@@ -1553,33 +1568,33 @@ void adjustOpenFilesLimit(void) {
         /* Set the max number of files if the current limit is not enough
          * for our needs. */
         if (oldlimit < maxfiles) {
-            rlim_t f;
+            rlim_t bestlimit;
             int setrlimit_error = 0;
 
             /* Try to set the file limit to match 'maxfiles' or at least
              * to the higher value supported less than maxfiles. */
-            f = maxfiles;
-            while(f > oldlimit) {
+            bestlimit = maxfiles;
+            while(bestlimit > oldlimit) {
                 rlim_t decr_step = 16;
 
-                limit.rlim_cur = f;
-                limit.rlim_max = f;
+                limit.rlim_cur = bestlimit;
+                limit.rlim_max = bestlimit;
                 if (setrlimit(RLIMIT_NOFILE,&limit) != -1) break;
                 setrlimit_error = errno;
 
-                /* We failed to set file limit to 'f'. Try with a
+                /* We failed to set file limit to 'bestlimit'. Try with a
                  * smaller limit decrementing by a few FDs per iteration. */
-                if (f < decr_step) break;
-                f -= decr_step;
+                if (bestlimit < decr_step) break;
+                bestlimit -= decr_step;
             }
 
             /* Assume that the limit we get initially is still valid if
              * our last try was even lower. */
-            if (f < oldlimit) f = oldlimit;
+            if (bestlimit < oldlimit) bestlimit = oldlimit;
 
-            if (f != maxfiles) {
+            if (bestlimit < maxfiles) {
                 int old_maxclients = server.maxclients;
-                server.maxclients = f-REDIS_MIN_RESERVED_FDS;
+                server.maxclients = bestlimit-REDIS_MIN_RESERVED_FDS;
                 if (server.maxclients < 1) {
                     redisLog(REDIS_WARNING,"Your current 'ulimit -n' "
                         "of %llu is not enough for Redis to start. "
@@ -1600,7 +1615,7 @@ void adjustOpenFilesLimit(void) {
                     "maxclients has been reduced to %d to compensate for "
                     "low ulimit. "
                     "If you need higher maxclients increase 'ulimit -n'.",
-                    (unsigned long long) oldlimit, server.maxclients);
+                    (unsigned long long) bestlimit, server.maxclients);
             } else {
                 redisLog(REDIS_NOTICE,"Increased maximum number of open files "
                     "to %llu (it was originally set to %llu).",
@@ -1611,6 +1626,23 @@ void adjustOpenFilesLimit(void) {
     }
 }
 
+/* Check that server.tcp_backlog can be actually enforced in Linux according
+ * to the value of /proc/sys/net/core/somaxconn, or warn about it. */
+void checkTcpBacklogSettings(void) {
+#ifdef HAVE_PROC_SOMAXCONN
+    FILE *fp = fopen("/proc/sys/net/core/somaxconn","r");
+    char buf[1024];
+    if (!fp) return;
+    if (fgets(buf,sizeof(buf),fp) != NULL) {
+        int somaxconn = atoi(buf);
+        if (somaxconn > 0 && somaxconn < server.tcp_backlog) {
+            redisLog(REDIS_WARNING,"WARNING: The TCP backlog setting of %d cannot be enforced because /proc/sys/net/core/somaxconn is set to the lower value of %d.", server.tcp_backlog, somaxconn);
+        }
+    }
+    fclose(fp);
+#endif
+}
+
 /* Initialize a set of file descriptors to listen to the specified 'port'
  * binding the addresses specified in the Redis server configuration.
  *
@@ -1681,6 +1713,8 @@ int listenToPort(int port, int *fds, int *count) {
  * to reset via CONFIG RESETSTAT. The function is also used in order to
  * initialize these fields in initServer() at server startup. */
 void resetServerStats(void) {
+    int j;
+
     server.stat_numcommands = 0;
     server.stat_numconnections = 0;
     server.stat_expiredkeys = 0;
@@ -1693,10 +1727,15 @@ void resetServerStats(void) {
     server.stat_sync_full = 0;
     server.stat_sync_partial_ok = 0;
     server.stat_sync_partial_err = 0;
-    memset(server.ops_sec_samples,0,sizeof(server.ops_sec_samples));
-    server.ops_sec_idx = 0;
-    server.ops_sec_last_sample_time = mstime();
-    server.ops_sec_last_sample_ops = 0;
+    for (j = 0; j < REDIS_METRIC_COUNT; j++) {
+        server.inst_metric[j].idx = 0;
+        server.inst_metric[j].last_sample_time = mstime();
+        server.inst_metric[j].last_sample_count = 0;
+        memset(server.inst_metric[j].samples,0,
+            sizeof(server.inst_metric[j].samples));
+    }
+    server.stat_net_input_bytes = 0;
+    server.stat_net_output_bytes = 0;
 }
 
 void initServer(void) {
@@ -1723,6 +1762,7 @@ void initServer(void) {
     server.clients_waiting_acks = listCreate();
     server.get_ack_from_slaves = 0;
     server.clients_paused = 0;
+    server.system_memory_size = zmalloc_get_memory_size();
 
     createSharedObjects();
     adjustOpenFilesLimit();
@@ -1962,6 +2002,9 @@ struct redisCommand *lookupCommandOrOriginal(sds name) {
  * + REDIS_PROPAGATE_NONE (no propagation of command at all)
  * + REDIS_PROPAGATE_AOF (propagate into the AOF file if is enabled)
  * + REDIS_PROPAGATE_REPL (propagate into the replication link)
+ *
+ * This should not be used inside commands implementation. Use instead
+ * alsoPropagate(), preventCommandPropagation(), forceCommandPropagation().
  */
 void propagate(struct redisCommand *cmd, int dbid, robj **argv, int argc,
                int flags)
@@ -1973,11 +2016,31 @@ void propagate(struct redisCommand *cmd, int dbid, robj **argv, int argc,
 }
 
 /* Used inside commands to schedule the propagation of additional commands
- * after the current command is propagated to AOF / Replication. */
+ * after the current command is propagated to AOF / Replication.
+ *
+ * 'cmd' must be a pointer to the Redis command to replicate, dbid is the
+ * database ID the command should be propagated into.
+ * Arguments of the command to propagte are passed as an array of redis
+ * objects pointers of len 'argc', using the 'argv' vector.
+ *
+ * The function does not take a reference to the passed 'argv' vector,
+ * so it is up to the caller to release the passed argv (but it is usually
+ * stack allocated).  The function autoamtically increments ref count of
+ * passed objects, so the caller does not need to. */
 void alsoPropagate(struct redisCommand *cmd, int dbid, robj **argv, int argc,
                    int target)
 {
-    redisOpArrayAppend(&server.also_propagate,cmd,dbid,argv,argc,target);
+    robj **argvcopy;
+    int j;
+
+    if (server.loading) return; /* No propagation during loading. */
+
+    argvcopy = zmalloc(sizeof(robj*)*argc);
+    for (j = 0; j < argc; j++) {
+        argvcopy[j] = argv[j];
+        incrRefCount(argv[j]);
+    }
+    redisOpArrayAppend(&server.also_propagate,cmd,dbid,argvcopy,argc,target);
 }
 
 /* It is possible to call the function forceCommandPropagation() inside a
@@ -1988,6 +2051,13 @@ void forceCommandPropagation(redisClient *c, int flags) {
     if (flags & REDIS_PROPAGATE_AOF) c->flags |= REDIS_FORCE_AOF;
 }
 
+/* Avoid that the executed command is propagated at all. This way we
+ * are free to just propagate what we want using the alsoPropagate()
+ * API. */
+void preventCommandPropagation(redisClient *c) {
+    c->flags |= REDIS_PREVENT_PROP;
+}
+
 /* Call() is the core of Redis execution of a command */
 void call(redisClient *c, int flags) {
     long long dirty, start, duration;
@@ -1997,7 +2067,7 @@ void call(redisClient *c, int flags) {
      * not generated from reading an AOF. */
     if (listLength(server.monitors) &&
         !server.loading &&
-        !(c->cmd->flags & REDIS_CMD_SKIP_MONITOR))
+        !(c->cmd->flags & (REDIS_CMD_SKIP_MONITOR|REDIS_CMD_ADMIN)))
     {
         replicationFeedMonitors(c,server.monitors,c->db->id,c->argv,c->argc);
     }
@@ -2041,7 +2111,7 @@ void call(redisClient *c, int flags) {
     }
 
     /* Propagate the command into the AOF and replication link */
-    if (flags & REDIS_CALL_PROPAGATE) {
+    if (flags & REDIS_CALL_PROPAGATE && (c->flags & REDIS_PREVENT_PROP) == 0) {
         int flags = REDIS_PROPAGATE_NONE;
 
         if (c->flags & REDIS_FORCE_REPL) flags |= REDIS_PROPAGATE_REPL;
@@ -2052,20 +2122,24 @@ void call(redisClient *c, int flags) {
             propagate(c->cmd,c->db->id,c->argv,c->argc,flags);
     }
 
-    /* Restore the old FORCE_AOF/REPL flags, since call can be executed
+    /* Restore the old replication flags, since call can be executed
      * recursively. */
-    c->flags &= ~(REDIS_FORCE_AOF|REDIS_FORCE_REPL);
-    c->flags |= client_old_flags & (REDIS_FORCE_AOF|REDIS_FORCE_REPL);
+    c->flags &= ~(REDIS_FORCE_AOF|REDIS_FORCE_REPL|REDIS_PREVENT_PROP);
+    c->flags |= client_old_flags &
+        (REDIS_FORCE_AOF|REDIS_FORCE_REPL|REDIS_PREVENT_PROP);
 
     /* Handle the alsoPropagate() API to handle commands that want to propagate
-     * multiple separated commands. */
+     * multiple separated commands. Note that alsoPropagate() is not affected
+     * by REDIS_PREVENT_PROP flag. */
     if (server.also_propagate.numops) {
         int j;
         redisOp *rop;
 
-        for (j = 0; j < server.also_propagate.numops; j++) {
-            rop = &server.also_propagate.ops[j];
-            propagate(rop->cmd, rop->dbid, rop->argv, rop->argc, rop->target);
+        if (flags & REDIS_CALL_PROPAGATE) {
+            for (j = 0; j < server.also_propagate.numops; j++) {
+                rop = &server.also_propagate.ops[j];
+                propagate(rop->cmd,rop->dbid,rop->argv,rop->argc,rop->target);
+            }
         }
         redisOpArrayFree(&server.also_propagate);
     }
@@ -2336,7 +2410,7 @@ int prepareForShutdown(int flags) {
             return REDIS_ERR;
         }
     }
-    if (server.daemonize) {
+    if (server.daemonize || server.pidfile) {
         redisLog(REDIS_NOTICE,"Removing the pid file.");
         unlink(server.pidfile);
     }
@@ -2443,7 +2517,6 @@ void timeCommand(redisClient *c) {
     addReplyBulkLongLong(c,tv.tv_usec);
 }
 
-
 /* Helper function for addReplyCommand() to output flags. */
 int addReplyCommandFlag(redisClient *c, struct redisCommand *cmd, int f, char *reply) {
     if (cmd->flags & f) {
@@ -2662,7 +2735,10 @@ sds genRedisInfoString(char *section) {
     if (allsections || defsections || !strcasecmp(section,"memory")) {
         char hmem[64];
         char peak_hmem[64];
+        char total_system_hmem[64];
         size_t zmalloc_used = zmalloc_used_memory();
+        size_t total_system_mem = server.system_memory_size;
+        char *evict_policy = maxmemoryToString();
 
         /* Peak memory is updated from time to time by serverCron() so it
          * may happen that the instantaneous value is slightly bigger than
@@ -2673,6 +2749,8 @@ sds genRedisInfoString(char *section) {
 
         bytesToHuman(hmem,zmalloc_used);
         bytesToHuman(peak_hmem,server.stat_peak_memory);
+        bytesToHuman(total_system_hmem,total_system_mem);
+
         if (sections++) info = sdscat(info,"\r\n");
         info = sdscatprintf(info,
             "# Memory\r\n"
@@ -2681,17 +2759,23 @@ sds genRedisInfoString(char *section) {
             "used_memory_rss:%zu\r\n"
             "used_memory_peak:%zu\r\n"
             "used_memory_peak_human:%s\r\n"
+            "total_system_memory:%lu\r\n"
+            "total_system_memory_human:%s\r\n"
             "used_memory_lua:%lld\r\n"
             "mem_fragmentation_ratio:%.2f\r\n"
-            "mem_allocator:%s\r\n",
+            "mem_allocator:%s\r\n"
+            "maxmemory_policy:%s\r\n",
             zmalloc_used,
             hmem,
             server.resident_set_size,
             server.stat_peak_memory,
             peak_hmem,
+            (unsigned long)total_system_mem,
+            total_system_hmem,
             ((long long)lua_gc(server.lua,LUA_GCCOUNT,0))*1024LL,
             zmalloc_get_fragmentation_ratio(server.resident_set_size),
-            ZMALLOC_LIB
+            ZMALLOC_LIB,
+            evict_policy
             );
     }
 
@@ -2756,14 +2840,14 @@ sds genRedisInfoString(char *section) {
                                     server.loading_loaded_bytes;
 
             perc = ((double)server.loading_loaded_bytes /
-                   server.loading_total_bytes) * 100;
+                   (server.loading_total_bytes+1)) * 100;
 
-            elapsed = server.unixtime-server.loading_start_time;
+            elapsed = time(NULL)-server.loading_start_time;
             if (elapsed == 0) {
                 eta = 1; /* A fake 1 second figure if we don't have
                             enough info */
             } else {
-                eta = (elapsed*remaining_bytes)/server.loading_loaded_bytes;
+                eta = (elapsed*remaining_bytes)/(server.loading_loaded_bytes+1);
             }
 
             info = sdscatprintf(info,
@@ -2789,6 +2873,10 @@ sds genRedisInfoString(char *section) {
             "total_connections_received:%lld\r\n"
             "total_commands_processed:%lld\r\n"
             "instantaneous_ops_per_sec:%lld\r\n"
+            "total_net_input_bytes:%lld\r\n"
+            "total_net_output_bytes:%lld\r\n"
+            "instantaneous_input_kbps:%.2f\r\n"
+            "instantaneous_output_kbps:%.2f\r\n"
             "rejected_connections:%lld\r\n"
             "sync_full:%lld\r\n"
             "sync_partial_ok:%lld\r\n"
@@ -2803,7 +2891,11 @@ sds genRedisInfoString(char *section) {
             "migrate_cached_sockets:%ld\r\n",
             server.stat_numconnections,
             server.stat_numcommands,
-            getOperationsPerSecond(),
+            getInstantaneousMetric(REDIS_METRIC_COMMAND),
+            server.stat_net_input_bytes,
+            server.stat_net_output_bytes,
+            (float)getInstantaneousMetric(REDIS_METRIC_NET_INPUT)/1024,
+            (float)getInstantaneousMetric(REDIS_METRIC_NET_OUTPUT)/1024,
             server.stat_rejected_conn,
             server.stat_sync_full,
             server.stat_sync_partial_ok,
@@ -3002,11 +3094,7 @@ void infoCommand(redisClient *c) {
         addReply(c,shared.syntaxerr);
         return;
     }
-    sds info = genRedisInfoString(section);
-    addReplySds(c,sdscatprintf(sdsempty(),"$%lu\r\n",
-        (unsigned long)sdslen(info)));
-    addReplySds(c,info);
-    addReply(c,shared.crlf);
+    addReplyBulkSds(c, genRedisInfoString(section));
 }
 
 void monitorCommand(redisClient *c) {
@@ -3094,13 +3182,7 @@ void evictionPoolPopulate(dict *sampledict, dict *keydict, struct evictionPoolEn
         samples = zmalloc(sizeof(samples[0])*server.maxmemory_samples);
     }
 
-#if 1 /* Use bulk get by default. */
-    count = dictGetRandomKeys(sampledict,samples,server.maxmemory_samples);
-#else
-    count = server.maxmemory_samples;
-    for (j = 0; j < count; j++) samples[j] = dictGetRandomKey(sampledict);
-#endif
-
+    count = dictGetSomeKeys(sampledict,samples,server.maxmemory_samples);
     for (j = 0; j < count; j++) {
         unsigned long long idle;
         sds key;
@@ -3155,7 +3237,7 @@ void evictionPoolPopulate(dict *sampledict, dict *keydict, struct evictionPoolEn
 int freeMemoryIfNeeded(void) {
     size_t mem_used, mem_tofree, mem_freed;
     int slaves = listLength(server.slaves);
-    mstime_t latency;
+    mstime_t latency, eviction_latency;
 
     /* Remove the size of slaves output buffers and AOF buffer from the
      * count of used memory. */
@@ -3286,7 +3368,11 @@ int freeMemoryIfNeeded(void) {
                  * AOF and Output buffer memory will be freed eventually so
                  * we only care about memory used by the key space. */
                 delta = (long long) zmalloc_used_memory();
+                latencyStartMonitor(eviction_latency);
                 dbDelete(db,keyobj);
+                latencyEndMonitor(eviction_latency);
+                latencyAddSampleIfNeeded("eviction-del",eviction_latency);
+                latencyRemoveNestedEvent(latency,eviction_latency);
                 delta -= (long long) zmalloc_used_memory();
                 mem_freed += delta;
                 server.stat_evictedkeys++;
@@ -3330,14 +3416,21 @@ int linuxOvercommitMemoryValue(void) {
     return atoi(buf);
 }
 
-void linuxOvercommitMemoryWarning(void) {
+void linuxMemoryWarnings(void) {
     if (linuxOvercommitMemoryValue() == 0) {
         redisLog(REDIS_WARNING,"WARNING overcommit_memory is set to 0! Background save may fail under low memory condition. To fix this issue add 'vm.overcommit_memory = 1' to /etc/sysctl.conf and then reboot or run the command 'sysctl vm.overcommit_memory=1' for this to take effect.");
     }
+    if (THPIsEnabled()) {
+        redisLog(REDIS_WARNING,"WARNING you have Transparent Huge Pages (THP) support enabled in your kernel. This will create latency and memory usage issues with Redis. To fix this issue run the command 'echo never > /sys/kernel/mm/transparent_hugepage/enabled' as root, and add it to your /etc/rc.local in order to retain the setting after a reboot. Redis must be restarted after THP is disabled.");
+    }
 }
 #endif /* __linux__ */
 
 void createPidFile(void) {
+    /* If pidfile requested, but no pidfile defined, use
+     * default pidfile path */
+    if (!server.pidfile) server.pidfile = zstrdup(REDIS_DEFAULT_PID_FILE);
+
     /* Try to write the pid file in a best-effort way. */
     FILE *fp = fopen(server.pidfile,"w");
     if (fp) {
@@ -3400,15 +3493,27 @@ void redisAsciiArt(void) {
     else if (server.sentinel_mode) mode = "sentinel";
     else mode = "standalone";
 
-    snprintf(buf,1024*16,ascii_logo,
-        REDIS_VERSION,
-        redisGitSHA1(),
-        strtol(redisGitDirty(),NULL,10) > 0,
-        (sizeof(long) == 8) ? "64" : "32",
-        mode, server.port,
-        (long) getpid()
-    );
-    redisLogRaw(REDIS_NOTICE|REDIS_LOG_RAW,buf);
+    if (server.syslog_enabled) {
+        redisLog(REDIS_NOTICE,
+            "Redis %s (%s/%d) %s bit, %s mode, port %d, pid %ld ready to start.",
+            REDIS_VERSION,
+            redisGitSHA1(),
+            strtol(redisGitDirty(),NULL,10) > 0,
+            (sizeof(long) == 8) ? "64" : "32",
+            mode, server.port,
+            (long) getpid()
+        );
+    } else {
+        snprintf(buf,1024*16,ascii_logo,
+            REDIS_VERSION,
+            redisGitSHA1(),
+            strtol(redisGitDirty(),NULL,10) > 0,
+            (sizeof(long) == 8) ? "64" : "32",
+            mode, server.port,
+            (long) getpid()
+        );
+        redisLogRaw(REDIS_NOTICE|REDIS_LOG_RAW,buf);
+    }
     zfree(buf);
 }
 
@@ -3517,9 +3622,131 @@ void redisSetProcTitle(char *title) {
 #endif
 }
 
+/*
+ * Check whether systemd or upstart have been used to start redis.
+ */
+
+int redisSupervisedUpstart(void) {
+    const char *upstart_job = getenv("UPSTART_JOB");
+
+    if (!upstart_job) {
+        redisLog(REDIS_WARNING,
+                "upstart supervision requested, but UPSTART_JOB not found");
+        return 0;
+    }
+
+    redisLog(REDIS_NOTICE, "supervised by upstart, will stop to signal readyness");
+    raise(SIGSTOP);
+    unsetenv("UPSTART_JOB");
+    return 1;
+}
+
+int redisSupervisedSystemd(void) {
+    const char *notify_socket = getenv("NOTIFY_SOCKET");
+    int fd = 1;
+    struct sockaddr_un su;
+    struct iovec iov;
+    struct msghdr hdr;
+    int sendto_flags = 0;
+
+    if (!notify_socket) {
+        redisLog(REDIS_WARNING,
+                "systemd supervision requested, but NOTIFY_SOCKET not found");
+        return 0;
+    }
+
+    if ((strchr("@/", notify_socket[0])) == NULL || strlen(notify_socket) < 2) {
+        return 0;
+    }
+
+    redisLog(REDIS_NOTICE, "supervised by systemd, will signal readyness");
+    if ((fd = socket(AF_UNIX, SOCK_DGRAM, 0)) == -1) {
+        redisLog(REDIS_WARNING,
+                "Can't connect to systemd socket %s", notify_socket);
+        return 0;
+    }
+
+    memset(&su, 0, sizeof(su));
+    su.sun_family = AF_UNIX;
+    strncpy (su.sun_path, notify_socket, sizeof(su.sun_path) -1);
+    su.sun_path[sizeof(su.sun_path) - 1] = '\0';
+
+    if (notify_socket[0] == '@')
+        su.sun_path[0] = '\0';
+
+    memset(&iov, 0, sizeof(iov));
+    iov.iov_base = "READY=1";
+    iov.iov_len = strlen("READY=1");
+
+    memset(&hdr, 0, sizeof(hdr));
+    hdr.msg_name = &su;
+    hdr.msg_namelen = offsetof(struct sockaddr_un, sun_path) +
+        strlen(notify_socket);
+    hdr.msg_iov = &iov;
+    hdr.msg_iovlen = 1;
+
+    unsetenv("NOTIFY_SOCKET");
+#ifdef HAVE_MSG_NOSIGNAL
+    sendto_flags |= MSG_NOSIGNAL;
+#endif
+    if (sendmsg(fd, &hdr, sendto_flags) < 0) {
+        redisLog(REDIS_WARNING, "Can't send notification to systemd");
+        close(fd);
+        return 0;
+    }
+    close(fd);
+    return 1;
+}
+
+int redisIsSupervised(int mode) {
+    if (mode == REDIS_SUPERVISED_AUTODETECT) {
+        const char *upstart_job = getenv("UPSTART_JOB");
+        const char *notify_socket = getenv("NOTIFY_SOCKET");
+
+        if (upstart_job) {
+            redisSupervisedUpstart();
+        } else if (notify_socket) {
+            redisSupervisedSystemd();
+        }
+    } else if (mode == REDIS_SUPERVISED_UPSTART) {
+        return redisSupervisedUpstart();
+    } else if (mode == REDIS_SUPERVISED_SYSTEMD) {
+        return redisSupervisedSystemd();
+    }
+
+    return 0;
+}
+
+
 int main(int argc, char **argv) {
     struct timeval tv;
 
+#ifdef REDIS_TEST
+    if (argc == 3 && !strcasecmp(argv[1], "test")) {
+        if (!strcasecmp(argv[2], "ziplist")) {
+            return ziplistTest(argc, argv);
+        } else if (!strcasecmp(argv[2], "quicklist")) {
+            quicklistTest(argc, argv);
+        } else if (!strcasecmp(argv[2], "intset")) {
+            return intsetTest(argc, argv);
+        } else if (!strcasecmp(argv[2], "zipmap")) {
+            return zipmapTest(argc, argv);
+        } else if (!strcasecmp(argv[2], "sha1test")) {
+            return sha1Test(argc, argv);
+        } else if (!strcasecmp(argv[2], "util")) {
+            return utilTest(argc, argv);
+        } else if (!strcasecmp(argv[2], "sds")) {
+            return sdsTest(argc, argv);
+        } else if (!strcasecmp(argv[2], "endianconv")) {
+            return endianconvTest(argc, argv);
+        } else if (!strcasecmp(argv[2], "crc64")) {
+            return crc64Test(argc, argv);
+        }
+
+        return -1; /* test not found */
+    }
+#endif
+
     /* We need to initialize our libraries, and the server configuration. */
 #ifdef INIT_SETPROCTITLE_REPLACEMENT
     spt_init(argc, argv);
@@ -3541,6 +3768,12 @@ int main(int argc, char **argv) {
         initSentinel();
     }
 
+    /* Check if we need to start in redis-check-rdb mode. We just execute
+     * the program main. However the program is part of the Redis executable
+     * so that we can easily execute an RDB check on loading errors. */
+    if (strstr(argv[0],"redis-check-rdb") != NULL)
+        exit(redis_check_rdb_main(argv,argc));
+
     if (argc >= 2) {
         int j = 1; /* First option to parse in argv[] */
         sds options = sdsempty();
@@ -3572,6 +3805,11 @@ int main(int argc, char **argv) {
         while(j != argc) {
             if (argv[j][0] == '-' && argv[j][1] == '-') {
                 /* Option name */
+                if (!strcmp(argv[j], "--check-rdb")) {
+                    /* Argument has no options, need to skip for parsing. */
+                    j++;
+                    continue;
+                }
                 if (sdslen(options)) options = sdscat(options,"\n");
                 options = sdscat(options,argv[j]+2);
                 options = sdscat(options," ");
@@ -3596,9 +3834,13 @@ int main(int argc, char **argv) {
     } else {
         redisLog(REDIS_WARNING, "Warning: no config file specified, using the default config. In order to specify a config file use %s /path/to/%s.conf", argv[0], server.sentinel_mode ? "sentinel" : "redis");
     }
-    if (server.daemonize) daemonize();
+
+    server.supervised = redisIsSupervised(server.supervised_mode);
+    int background = server.daemonize && !server.supervised;
+    if (background) daemonize();
+
     initServer();
-    if (server.daemonize) createPidFile();
+    if (background || server.pidfile) createPidFile();
     redisSetProcTitle(argv[0]);
     redisAsciiArt();
 
@@ -3606,8 +3848,9 @@ int main(int argc, char **argv) {
         /* Things not needed when running in Sentinel mode. */
         redisLog(REDIS_WARNING,"Server started, Redis version " REDIS_VERSION);
     #ifdef __linux__
-        linuxOvercommitMemoryWarning();
+        linuxMemoryWarnings();
     #endif
+        checkTcpBacklogSettings();
         loadDataFromDisk();
         if (server.cluster_enabled) {
             if (verifyClusterConfigWithData() == REDIS_ERR) {

src/redis.h

@@ -32,10 +32,7 @@
 
 #include "fmacros.h"
 #include "config.h"
-
-#if defined(__sun)
 #include "solarisfixes.h"
-#endif
 
 #include <stdio.h>
 #include <stdlib.h>
@@ -65,6 +62,13 @@ typedef long long mstime_t; /* millisecond time type. */
 #include "util.h"    /* Misc functions useful in many places */
 #include "latency.h" /* Latency monitor API */
 #include "sparkline.h" /* ASII graphs API */
+#include "quicklist.h"
+
+/* Following includes allow test functions to be called from Redis main() */
+#include "zipmap.h"
+#include "sha1.h"
+#include "endianconv.h"
+#include "crc64.h"
 
 /* Error codes */
 #define REDIS_OK                0
@@ -97,7 +101,6 @@ typedef long long mstime_t; /* millisecond time type. */
 #define REDIS_REPL_PING_SLAVE_PERIOD 10
 #define REDIS_RUN_ID_SIZE 40
 #define REDIS_EOF_MARK_SIZE 40
-#define REDIS_OPS_SEC_SAMPLES 16
 #define REDIS_DEFAULT_REPL_BACKLOG_SIZE (1024*1024)    /* 1mb */
 #define REDIS_DEFAULT_REPL_BACKLOG_TIME_LIMIT (60*60)  /* 1 hour */
 #define REDIS_REPL_BACKLOG_MIN_SIZE (1024*16)          /* 16k */
@@ -128,7 +131,7 @@ typedef long long mstime_t; /* millisecond time type. */
 #define REDIS_DEFAULT_AOF_REWRITE_INCREMENTAL_FSYNC 1
 #define REDIS_DEFAULT_MIN_SLAVES_TO_WRITE 0
 #define REDIS_DEFAULT_MIN_SLAVES_MAX_LAG 10
-#define REDIS_IP_STR_LEN INET6_ADDRSTRLEN
+#define REDIS_IP_STR_LEN 46 /* INET6_ADDRSTRLEN is 46, but we need to be sure */
 #define REDIS_PEER_ID_LEN (REDIS_IP_STR_LEN+32) /* Must be enough for ip:port */
 #define REDIS_BINDADDR_MAX 16
 #define REDIS_MIN_RESERVED_FDS 32
@@ -140,6 +143,13 @@ typedef long long mstime_t; /* millisecond time type. */
 #define ACTIVE_EXPIRE_CYCLE_SLOW 0
 #define ACTIVE_EXPIRE_CYCLE_FAST 1
 
+/* Instantaneous metrics tracking. */
+#define REDIS_METRIC_SAMPLES 16     /* Number of samples per metric. */
+#define REDIS_METRIC_COMMAND 0      /* Number of commands executed. */
+#define REDIS_METRIC_NET_INPUT 1    /* Bytes read to network .*/
+#define REDIS_METRIC_NET_OUTPUT 2   /* Bytes written to network. */
+#define REDIS_METRIC_COUNT 3
+
 /* Protocol and I/O related defines */
 #define REDIS_MAX_QUERYBUF_LEN  (1024*1024*1024) /* 1GB max query buffer. */
 #define REDIS_IOBUF_LEN         (1024*16)  /* Generic I/O buffer size */
@@ -192,6 +202,7 @@ typedef long long mstime_t; /* millisecond time type. */
 #define REDIS_ENCODING_INTSET 6  /* Encoded as intset */
 #define REDIS_ENCODING_SKIPLIST 7  /* Encoded as skiplist */
 #define REDIS_ENCODING_EMBSTR 8  /* Embedded sds string encoding */
+#define REDIS_ENCODING_QUICKLIST 9 /* Encoded as linked list of ziplists */
 
 /* Defines related to the dump file format. To store 32 bits lengths for short
  * keys requires a lot of space, so we check the most significant 2 bits of
@@ -246,6 +257,7 @@ typedef long long mstime_t; /* millisecond time type. */
 #define REDIS_PRE_PSYNC (1<<16)   /* Instance don't understand PSYNC. */
 #define REDIS_READONLY (1<<17)    /* Cluster client is in read-only state. */
 #define REDIS_PUBSUB (1<<18)      /* Client is in Pub/Sub mode. */
+#define REDIS_PREVENT_PROP (1<<19)  /* Don't propagate to AOF / Slaves. */
 
 /* Client block type (btype field in client structure)
  * if REDIS_BLOCKED flag is set. */
@@ -302,6 +314,12 @@ typedef long long mstime_t; /* millisecond time type. */
 #define REDIS_LOG_RAW (1<<10) /* Modifier to log without timestamp */
 #define REDIS_DEFAULT_VERBOSITY REDIS_NOTICE
 
+/* Supervision options */
+#define REDIS_SUPERVISED_NONE 0
+#define REDIS_SUPERVISED_AUTODETECT 1
+#define REDIS_SUPERVISED_SYSTEMD 2
+#define REDIS_SUPERVISED_UPSTART 3
+
 /* Anti-warning macro... */
 #define REDIS_NOTUSED(V) ((void) V)
 
@@ -317,12 +335,14 @@ typedef long long mstime_t; /* millisecond time type. */
 /* Zip structure related defaults */
 #define REDIS_HASH_MAX_ZIPLIST_ENTRIES 512
 #define REDIS_HASH_MAX_ZIPLIST_VALUE 64
-#define REDIS_LIST_MAX_ZIPLIST_ENTRIES 512
-#define REDIS_LIST_MAX_ZIPLIST_VALUE 64
 #define REDIS_SET_MAX_INTSET_ENTRIES 512
 #define REDIS_ZSET_MAX_ZIPLIST_ENTRIES 128
 #define REDIS_ZSET_MAX_ZIPLIST_VALUE 64
 
+/* List defaults */
+#define REDIS_LIST_MAX_ZIPLIST_SIZE -2
+#define REDIS_LIST_COMPRESS_DEPTH 0
+
 /* HyperLogLog defines */
 #define REDIS_DEFAULT_HLL_SPARSE_MAX_BYTES 3000
 
@@ -532,6 +552,7 @@ typedef struct redisClient {
     int flags;              /* REDIS_SLAVE | REDIS_MONITOR | REDIS_MULTI ... */
     int authenticated;      /* when requirepass is non-NULL */
     int replstate;          /* replication state if this is a slave */
+    int repl_put_online_on_ack; /* Install slave write handler on ACK. */
     int repldbfd;           /* replication DB file descriptor */
     off_t repldboff;        /* replication DB file offset */
     off_t repldbsize;       /* replication DB file size */
@@ -688,7 +709,7 @@ struct redisServer {
     off_t loading_process_events_interval_bytes;
     /* Fast pointers to often looked up command */
     struct redisCommand *delCommand, *multiCommand, *lpushCommand, *lpopCommand,
-                        *rpopCommand;
+                        *rpopCommand, *sremCommand;
     /* Fields used only for stats */
     time_t stat_starttime;          /* Server start time */
     long long stat_numcommands;     /* Number of processed commands */
@@ -709,12 +730,16 @@ struct redisServer {
     long long slowlog_log_slower_than; /* SLOWLOG time limit (to get logged) */
     unsigned long slowlog_max_len;     /* SLOWLOG max number of items logged */
     size_t resident_set_size;       /* RSS sampled in serverCron(). */
-    /* The following two are used to track instantaneous "load" in terms
-     * of operations per second. */
-    long long ops_sec_last_sample_time; /* Timestamp of last sample (in ms) */
-    long long ops_sec_last_sample_ops;  /* numcommands in last sample */
-    long long ops_sec_samples[REDIS_OPS_SEC_SAMPLES];
-    int ops_sec_idx;
+    long long stat_net_input_bytes; /* Bytes read from network. */
+    long long stat_net_output_bytes; /* Bytes written to network. */
+    /* The following two are used to track instantaneous metrics, like
+     * number of operations per second, network traffic. */
+    struct {
+        long long last_sample_time; /* Timestamp of last sample in ms */
+        long long last_sample_count;/* Count in last sample */
+        long long samples[REDIS_METRIC_SAMPLES];
+        int idx;
+    } inst_metric[REDIS_METRIC_COUNT];
     /* Configuration */
     int verbosity;                  /* Loglevel in redis.conf */
     int maxidletime;                /* Client timeout in seconds */
@@ -722,6 +747,8 @@ struct redisServer {
     int active_expire_enabled;      /* Can be disabled for testing purposes. */
     size_t client_max_querybuf_len; /* Limit for client query buffer length */
     int dbnum;                      /* Total number of configured DBs */
+    int supervised;                 /* 1 if supervised, 0 otherwise. */
+    int supervised_mode;            /* See REDIS_SUPERVISED_* */
     int daemonize;                  /* True if running as a daemon */
     clientBufferLimitsConfig client_obuf_limits[REDIS_CLIENT_TYPE_COUNT];
     /* AOF persistence */
@@ -851,12 +878,14 @@ struct redisServer {
     /* Zip structure config, see redis.conf for more information  */
     size_t hash_max_ziplist_entries;
     size_t hash_max_ziplist_value;
-    size_t list_max_ziplist_entries;
-    size_t list_max_ziplist_value;
     size_t set_max_intset_entries;
     size_t zset_max_ziplist_entries;
     size_t zset_max_ziplist_value;
     size_t hll_sparse_max_bytes;
+    /* List parameters */
+    int list_max_ziplist_size;
+    int list_compress_depth;
+    /* time cache */
     time_t unixtime;        /* Unix time sampled every cron cycle. */
     long long mstime;       /* Like 'unixtime' but with milliseconds resolution. */
     /* Pubsub */
@@ -896,6 +925,8 @@ struct redisServer {
     int assert_line;
     int bug_report_start; /* True if bug report header was already logged. */
     int watchdog_period;  /* Software watchdog period in ms. 0 = off */
+    /* System hardware info */
+    size_t system_memory_size;  /* Total memory in system as reported by OS */
 };
 
 typedef struct pubsubPattern {
@@ -944,15 +975,13 @@ typedef struct {
     robj *subject;
     unsigned char encoding;
     unsigned char direction; /* Iteration direction */
-    unsigned char *zi;
-    listNode *ln;
+    quicklistIter *iter;
 } listTypeIterator;
 
 /* Structure for an entry while iterating over a list. */
 typedef struct {
     listTypeIterator *li;
-    unsigned char *zi;  /* Entry in ziplist */
-    listNode *ln;       /* Entry in linked list */
+    quicklistEntry entry; /* Entry in quicklist */
 } listTypeEntry;
 
 /* Structure to hold set iteration abstraction. */
@@ -1029,6 +1058,7 @@ void addReplyBulkCBuffer(redisClient *c, void *p, size_t len);
 void addReplyBulkLongLong(redisClient *c, long long ll);
 void addReply(redisClient *c, robj *obj);
 void addReplySds(redisClient *c, sds s);
+void addReplyBulkSds(redisClient *c, sds s);
 void addReplyError(redisClient *c, char *err);
 void addReplyStatus(redisClient *c, char *status);
 void addReplyDouble(redisClient *c, double d);
@@ -1078,7 +1108,7 @@ int listTypeNext(listTypeIterator *li, listTypeEntry *entry);
 robj *listTypeGet(listTypeEntry *entry);
 void listTypeInsert(listTypeEntry *entry, robj *value, int where);
 int listTypeEqual(listTypeEntry *entry, robj *o);
-void listTypeDelete(listTypeEntry *entry);
+void listTypeDelete(listTypeIterator *iter, listTypeEntry *entry);
 void listTypeConvert(robj *subject, int enc);
 void unblockClientWaitingData(redisClient *c);
 void handleClientsBlockedOnLists(void);
@@ -1115,8 +1145,8 @@ robj *tryObjectEncoding(robj *o);
 robj *getDecodedObject(robj *o);
 size_t stringObjectLen(robj *o);
 robj *createStringObjectFromLongLong(long long value);
-robj *createStringObjectFromLongDouble(long double value);
-robj *createListObject(void);
+robj *createStringObjectFromLongDouble(long double value, int humanfriendly);
+robj *createQuicklistObject(void);
 robj *createZiplistObject(void);
 robj *createSetObject(void);
 robj *createIntsetObject(void);
@@ -1223,6 +1253,7 @@ void call(redisClient *c, int flags);
 void propagate(struct redisCommand *cmd, int dbid, robj **argv, int argc, int flags);
 void alsoPropagate(struct redisCommand *cmd, int dbid, robj **argv, int argc, int target);
 void forceCommandPropagation(redisClient *c, int flags);
+void preventCommandPropagation(redisClient *c);
 int prepareForShutdown();
 #ifdef __GNUC__
 void redisLog(int level, const char *fmt, ...)
@@ -1243,6 +1274,7 @@ void closeListeningSockets(int unlink_unix_socket);
 void updateCachedTime(void);
 void resetServerStats(void);
 unsigned int getLRUClock(void);
+char *maxmemoryToString(void);
 
 /* Set data type */
 robj *setTypeCreate(robj *value);
@@ -1254,6 +1286,7 @@ void setTypeReleaseIterator(setTypeIterator *si);
 int setTypeNext(setTypeIterator *si, robj **objele, int64_t *llele);
 robj *setTypeNextObject(setTypeIterator *si);
 int setTypeRandomElement(robj *setobj, robj **objele, int64_t *llele);
+unsigned long setTypeRandomElements(robj *set, unsigned long count, robj *aux_set);
 unsigned long setTypeSize(robj *subject);
 void setTypeConvert(robj *subject, int enc);
 
@@ -1349,6 +1382,10 @@ void sentinelTimer(void);
 char *sentinelHandleConfiguration(char **argv, int argc);
 void sentinelIsRunning(void);
 
+/* redis-check-rdb */
+int redis_check_rdb(char *rdbfilename);
+int redis_check_rdb_main(char **argv, int argc);
+
 /* Scripting */
 void scriptingInit(void);
 

src/replication.c

@@ -40,6 +40,7 @@
 void replicationDiscardCachedMaster(void);
 void replicationResurrectCachedMaster(int newfd);
 void replicationSendAck(void);
+void putSlaveOnline(redisClient *slave);
 
 /* --------------------------- Utility functions ---------------------------- */
 
@@ -55,7 +56,7 @@ char *replicationGetSlaveName(redisClient *c) {
     buf[0] = '\0';
     if (anetPeerToString(c->fd,ip,sizeof(ip),NULL) != -1) {
         if (c->slave_listening_port)
-            snprintf(buf,sizeof(buf),"%s:%d",ip,c->slave_listening_port);
+            anetFormatAddr(buf,sizeof(buf),ip,c->slave_listening_port);
         else
             snprintf(buf,sizeof(buf),"%s:<unknown-slave-port>",ip);
     } else {
@@ -398,6 +399,7 @@ int masterTryPartialResynchronization(redisClient *c) {
     c->flags |= REDIS_SLAVE;
     c->replstate = REDIS_REPL_ONLINE;
     c->repl_ack_time = server.unixtime;
+    c->repl_put_online_on_ack = 0;
     listAddNodeTail(server.slaves,c);
     /* We can't use the connection buffers since they are used to accumulate
      * new commands at this stage. But we are sure the socket send buffer is
@@ -623,6 +625,11 @@ void replconfCommand(redisClient *c) {
             if (offset > c->repl_ack_off)
                 c->repl_ack_off = offset;
             c->repl_ack_time = server.unixtime;
+            /* If this was a diskless replication, we need to really put
+             * the slave online when the first ACK is received (which
+             * confirms slave is online and ready to get more data). */
+            if (c->repl_put_online_on_ack && c->replstate == REDIS_REPL_ONLINE)
+                putSlaveOnline(c);
             /* Note: this command does not reply anything! */
             return;
         } else if (!strcasecmp(c->argv[j]->ptr,"getack")) {
@@ -652,6 +659,7 @@ void replconfCommand(redisClient *c) {
  * 3) Update the count of good slaves. */
 void putSlaveOnline(redisClient *slave) {
     slave->replstate = REDIS_REPL_ONLINE;
+    slave->repl_put_online_on_ack = 0;
     slave->repl_ack_time = server.unixtime;
     if (aeCreateFileEvent(server.el, slave->fd, AE_WRITABLE,
         sendReplyToClient, slave) == AE_ERR) {
@@ -660,6 +668,8 @@ void putSlaveOnline(redisClient *slave) {
         return;
     }
     refreshGoodSlavesCount();
+    redisLog(REDIS_NOTICE,"Synchronization with slave %s succeeded",
+        replicationGetSlaveName(slave));
 }
 
 void sendBulkToSlave(aeEventLoop *el, int fd, void *privdata, int mask) {
@@ -680,6 +690,7 @@ void sendBulkToSlave(aeEventLoop *el, int fd, void *privdata, int mask) {
             freeClient(slave);
             return;
         }
+        server.stat_net_output_bytes += nwritten;
         sdsrange(slave->replpreamble,nwritten,-1);
         if (sdslen(slave->replpreamble) == 0) {
             sdsfree(slave->replpreamble);
@@ -708,12 +719,12 @@ void sendBulkToSlave(aeEventLoop *el, int fd, void *privdata, int mask) {
         return;
     }
     slave->repldboff += nwritten;
+    server.stat_net_output_bytes += nwritten;
     if (slave->repldboff == slave->repldbsize) {
         close(slave->repldbfd);
         slave->repldbfd = -1;
         aeDeleteFileEvent(server.el,slave->fd,AE_WRITABLE);
         putSlaveOnline(slave);
-        redisLog(REDIS_NOTICE,"Synchronization with slave succeeded (disk)");
     }
 }
 
@@ -752,10 +763,16 @@ void updateSlavesWaitingBgsave(int bgsaveerr, int type) {
              * diskless replication, our work is trivial, we can just put
              * the slave online. */
             if (type == REDIS_RDB_CHILD_TYPE_SOCKET) {
-                putSlaveOnline(slave);
                 redisLog(REDIS_NOTICE,
-                    "Synchronization with slave %s succeeded (socket)",
+                    "Streamed RDB transfer with slave %s succeeded (socket). Waiting for REPLCONF ACK from slave to enable streaming",
                         replicationGetSlaveName(slave));
+                /* Note: we wait for a REPLCONF ACK message from slave in
+                 * order to really put it online (install the write handler
+                 * so that the accumulated data can be transfered). However
+                 * we change the replication state ASAP, since our slave
+                 * is technically online now. */
+                slave->replstate = REDIS_REPL_ONLINE;
+                slave->repl_put_online_on_ack = 1;
             } else {
                 if (bgsaveerr != REDIS_OK) {
                     freeClient(slave);
@@ -837,6 +854,23 @@ void replicationEmptyDbCallback(void *privdata) {
     replicationSendNewlineToMaster();
 }
 
+/* Once we have a link with the master and the synchroniziation was
+ * performed, this function materializes the master client we store
+ * at server.master, starting from the specified file descriptor. */
+void replicationCreateMasterClient(int fd) {
+    server.master = createClient(fd);
+    server.master->flags |= REDIS_MASTER;
+    server.master->authenticated = 1;
+    server.repl_state = REDIS_REPL_CONNECTED;
+    server.master->reploff = server.repl_master_initial_offset;
+    memcpy(server.master->replrunid, server.repl_master_runid,
+        sizeof(server.repl_master_runid));
+    /* If master offset is set to -1, this master is old and is not
+     * PSYNC capable, so we flag it accordingly. */
+    if (server.master->reploff == -1)
+        server.master->flags |= REDIS_PRE_PSYNC;
+}
+
 /* Asynchronously read the SYNC payload we receive from a master */
 #define REPL_MAX_WRITTEN_BEFORE_FSYNC (1024*1024*8) /* 8 MB */
 void readSyncBulkPayload(aeEventLoop *el, int fd, void *privdata, int mask) {
@@ -923,13 +957,14 @@ void readSyncBulkPayload(aeEventLoop *el, int fd, void *privdata, int mask) {
         replicationAbortSyncTransfer();
         return;
     }
+    server.stat_net_input_bytes += nread;
 
     /* When a mark is used, we want to detect EOF asap in order to avoid
      * writing the EOF mark into the file... */
     int eof_reached = 0;
 
     if (usemark) {
-        /* Update the last bytes array, and check if it matches our delimiter. */
+        /* Update the last bytes array, and check if it matches our delimiter.*/
         if (nread >= REDIS_RUN_ID_SIZE) {
             memcpy(lastbytes,buf+nread-REDIS_RUN_ID_SIZE,REDIS_RUN_ID_SIZE);
         } else {
@@ -999,17 +1034,7 @@ void readSyncBulkPayload(aeEventLoop *el, int fd, void *privdata, int mask) {
         /* Final setup of the connected slave <- master link */
         zfree(server.repl_transfer_tmpfile);
         close(server.repl_transfer_fd);
-        server.master = createClient(server.repl_transfer_s);
-        server.master->flags |= REDIS_MASTER;
-        server.master->authenticated = 1;
-        server.repl_state = REDIS_REPL_CONNECTED;
-        server.master->reploff = server.repl_master_initial_offset;
-        memcpy(server.master->replrunid, server.repl_master_runid,
-            sizeof(server.repl_master_runid));
-        /* If master offset is set to -1, this master is old and is not
-         * PSYNC capable, so we flag it accordingly. */
-        if (server.master->reploff == -1)
-            server.master->flags |= REDIS_PRE_PSYNC;
+        replicationCreateMasterClient(server.repl_transfer_s);
         redisLog(REDIS_NOTICE, "MASTER <-> SLAVE sync: Finished with success");
         /* Restart the AOF subsystem now that we finished the sync. This
          * will trigger an AOF rewrite, and when done will start appending
@@ -1357,7 +1382,8 @@ error:
 int connectWithMaster(void) {
     int fd;
 
-    fd = anetTcpNonBlockConnect(NULL,server.masterhost,server.masterport);
+    fd = anetTcpNonBlockBindConnect(NULL,
+        server.masterhost,server.masterport,REDIS_BIND_ADDR);
     if (fd == -1) {
         redisLog(REDIS_WARNING,"Unable to connect to MASTER: %s",
             strerror(errno));
@@ -2002,8 +2028,9 @@ void replicationCron(void) {
             if (slave->flags & REDIS_PRE_PSYNC) continue;
             if ((server.unixtime - slave->repl_ack_time) > server.repl_timeout)
             {
-                    redisLog(REDIS_WARNING, "Disconnecting timedout slave: %s",
-                        replicationGetSlaveName(slave));
+                redisLog(REDIS_WARNING, "Disconnecting timedout slave: %s",
+                    replicationGetSlaveName(slave));
+                freeClient(slave);
             }
         }
     }

src/scripting.c

@@ -30,6 +30,7 @@
 #include "redis.h"
 #include "sha1.h"
 #include "rand.h"
+#include "cluster.h"
 
 #include <lua.h>
 #include <lauxlib.h>
@@ -213,11 +214,27 @@ int luaRedisGenericCommand(lua_State *lua, int raise_error) {
     static int argv_size = 0;
     static robj *cached_objects[LUA_CMD_OBJCACHE_SIZE];
     static size_t cached_objects_len[LUA_CMD_OBJCACHE_SIZE];
+    static int inuse = 0;   /* Recursive calls detection. */
+
+    /* By using Lua debug hooks it is possible to trigger a recursive call
+     * to luaRedisGenericCommand(), which normally should never happen.
+     * To make this function reentrant is futile and makes it slower, but
+     * we should at least detect such a misuse, and abort. */
+    if (inuse) {
+        char *recursion_warning =
+            "luaRedisGenericCommand() recursive call detected. "
+            "Are you doing funny stuff with Lua debug hooks?";
+        redisLog(REDIS_WARNING,"%s",recursion_warning);
+        luaPushError(lua,recursion_warning);
+        return 1;
+    }
+    inuse++;
 
     /* Require at least one argument */
     if (argc == 0) {
         luaPushError(lua,
             "Please specify at least one argument for redis.call()");
+        inuse--;
         return 1;
     }
 
@@ -272,6 +289,7 @@ int luaRedisGenericCommand(lua_State *lua, int raise_error) {
         }
         luaPushError(lua,
             "Lua redis() command arguments must be strings or integers");
+        inuse--;
         return 1;
     }
 
@@ -291,6 +309,7 @@ int luaRedisGenericCommand(lua_State *lua, int raise_error) {
             luaPushError(lua,"Unknown Redis command called from Lua script");
         goto cleanup;
     }
+    c->cmd = cmd;
 
     /* There are commands that are not allowed inside scripts. */
     if (cmd->flags & REDIS_CMD_NOSCRIPT) {
@@ -337,8 +356,23 @@ int luaRedisGenericCommand(lua_State *lua, int raise_error) {
     if (cmd->flags & REDIS_CMD_RANDOM) server.lua_random_dirty = 1;
     if (cmd->flags & REDIS_CMD_WRITE) server.lua_write_dirty = 1;
 
+    /* If this is a Redis Cluster node, we need to make sure Lua is not
+     * trying to access non-local keys. */
+    if (server.cluster_enabled) {
+        /* Duplicate relevant flags in the lua client. */
+        c->flags &= ~(REDIS_READONLY|REDIS_ASKING);
+        c->flags |= server.lua_caller->flags & (REDIS_READONLY|REDIS_ASKING);
+        if (getNodeByQuery(c,c->cmd,c->argv,c->argc,NULL,NULL) !=
+                           server.cluster->myself)
+        {
+            luaPushError(lua,
+                "Lua script attempted to access a non local key in a "
+                "cluster node");
+            goto cleanup;
+        }
+    }
+
     /* Run the command */
-    c->cmd = cmd;
     call(c,REDIS_CALL_SLOWLOG | REDIS_CALL_STATS);
 
     /* Convert the result of the Redis command into a suitable Lua type.
@@ -409,8 +443,10 @@ cleanup:
          * return the plain error. */
         lua_pushstring(lua,"err");
         lua_gettable(lua,-2);
+        inuse--;
         return lua_error(lua);
     }
+    inuse--;
     return 1;
 }
 
@@ -537,6 +573,7 @@ void luaLoadLib(lua_State *lua, const char *libname, lua_CFunction luafunc) {
 LUALIB_API int (luaopen_cjson) (lua_State *L);
 LUALIB_API int (luaopen_struct) (lua_State *L);
 LUALIB_API int (luaopen_cmsgpack) (lua_State *L);
+LUALIB_API int (luaopen_bit) (lua_State *L);
 
 void luaLoadLibraries(lua_State *lua) {
     luaLoadLib(lua, "", luaopen_base);
@@ -547,6 +584,7 @@ void luaLoadLibraries(lua_State *lua) {
     luaLoadLib(lua, "cjson", luaopen_cjson);
     luaLoadLib(lua, "struct", luaopen_struct);
     luaLoadLib(lua, "cmsgpack", luaopen_cmsgpack);
+    luaLoadLib(lua, "bit", luaopen_bit);
 
 #if 0 /* Stuff that we don't load currently, for sandboxing concerns. */
     luaLoadLib(lua, LUA_LOADLIBNAME, luaopen_package);

src/sds.c

@@ -295,7 +295,7 @@ sds sdscpy(sds s, const char *t) {
  * conversion. 's' must point to a string with room for at least
  * SDS_LLSTR_SIZE bytes.
  *
- * The function returns the lenght of the null-terminated string
+ * The function returns the length of the null-terminated string
  * representation stored at 's'. */
 #define SDS_LLSTR_SIZE 21
 int sdsll2str(char *s, long long value) {
@@ -369,7 +369,7 @@ sds sdsfromlonglong(long long value) {
     return sdsnewlen(buf,len);
 }
 
-/* Like sdscatpritf() but gets va_list instead of being variadic. */
+/* Like sdscatprintf() but gets va_list instead of being variadic. */
 sds sdscatvprintf(sds s, const char *fmt, va_list ap) {
     va_list cpy;
     char staticbuf[1024], *buf = staticbuf, *t;
@@ -390,7 +390,7 @@ sds sdscatvprintf(sds s, const char *fmt, va_list ap) {
         buf[buflen-2] = '\0';
         va_copy(cpy,ap);
         vsnprintf(buf, buflen, fmt, cpy);
-        va_end(ap);
+        va_end(cpy);
         if (buf[buflen-2] != '\0') {
             if (buf != staticbuf) zfree(buf);
             buflen *= 2;
@@ -415,7 +415,7 @@ sds sdscatvprintf(sds s, const char *fmt, va_list ap) {
  *
  * Example:
  *
- * s = sdsempty("Sum is: ");
+ * s = sdsnew("Sum is: ");
  * s = sdscatprintf(s,"%d+%d = %d",a,b,a+b).
  *
  * Often you need to create a string from scratch with the printf-alike
@@ -570,7 +570,7 @@ sds sdstrim(sds s, const char *cset) {
     sp = start = s;
     ep = end = s+sdslen(s)-1;
     while(sp <= end && strchr(cset, *sp)) sp++;
-    while(ep > start && strchr(cset, *ep)) ep--;
+    while(ep > sp && strchr(cset, *ep)) ep--;
     len = (sp > ep) ? 0 : ((ep-sp)+1);
     if (sh->buf != sp) memmove(sh->buf, sp, len);
     sh->buf[len] = '\0';
@@ -643,8 +643,8 @@ void sdstoupper(sds s) {
  *
  * Return value:
  *
- *     1 if s1 > s2.
- *    -1 if s1 < s2.
+ *     positive if s1 > s2.
+ *     negative if s1 < s2.
  *     0 if s1 and s2 are exactly the same binary string.
  *
  * If two strings share exactly the same prefix, but one of the two has
@@ -962,12 +962,15 @@ sds sdsjoin(char **argv, int argc, char *sep) {
     return join;
 }
 
-#ifdef SDS_TEST_MAIN
+#if defined(REDIS_TEST) || defined(SDS_TEST_MAIN)
 #include <stdio.h>
 #include "testhelp.h"
 #include "limits.h"
 
-int main(void) {
+#define UNUSED(x) (void)(x)
+int sdsTest(int argc, char *argv[]) {
+    UNUSED(argc);
+    UNUSED(argv);
     {
         struct sdshdr *sh;
         sds x = sdsnew("foo"), y;
@@ -1013,6 +1016,18 @@ int main(void) {
             sdslen(x) == 35 &&
             memcmp(x,"--4294967295,18446744073709551615--",35) == 0)
 
+        sdsfree(x);
+        x = sdsnew(" x ");
+        sdstrim(x," x");
+        test_cond("sdstrim() works when all chars match",
+            sdslen(x) == 0)
+
+        sdsfree(x);
+        x = sdsnew(" x ");
+        sdstrim(x," ");
+        test_cond("sdstrim() works when a single char remains",
+            sdslen(x) == 1 && x[0] == 'x')
+
         sdsfree(x);
         x = sdsnew("xxciaoyyy");
         sdstrim(x,"xy");
@@ -1080,7 +1095,7 @@ int main(void) {
             memcmp(y,"\"\\a\\n\\x00foo\\r\"",15) == 0)
 
         {
-            int oldfree;
+            unsigned int oldfree;
 
             sdsfree(x);
             x = sdsnew("0");
@@ -1101,3 +1116,9 @@ int main(void) {
     return 0;
 }
 #endif
+
+#ifdef SDS_TEST_MAIN
+int main(void) {
+    return sdsTest();
+}
+#endif

src/sds.h

@@ -98,4 +98,8 @@ void sdsIncrLen(sds s, int incr);
 sds sdsRemoveFreeSpace(sds s);
 size_t sdsAllocSize(sds s);
 
+#ifdef REDIS_TEST
+int sdsTest(int argc, char *argv[]);
+#endif
+
 #endif

src/sentinel.c

@@ -190,6 +190,7 @@ typedef struct sentinelRedisInstance {
      * are set to NULL no script is executed. */
     char *notification_script;
     char *client_reconfig_script;
+    sds info; /* cached INFO output */
 } sentinelRedisInstance;
 
 /* Main state. */
@@ -576,7 +577,7 @@ void sentinelEvent(int level, char *type, sentinelRedisInstance *ri,
     if (level == REDIS_WARNING && ri != NULL) {
         sentinelRedisInstance *master = (ri->flags & SRI_MASTER) ?
                                          ri : ri->master;
-        if (master->notification_script) {
+        if (master && master->notification_script) {
             sentinelScheduleScriptExecution(master->notification_script,
                 type,msg,NULL);
         }
@@ -896,7 +897,7 @@ sentinelRedisInstance *createSentinelRedisInstance(char *name, int flags, char *
     sentinelRedisInstance *ri;
     sentinelAddr *addr;
     dict *table = NULL;
-    char slavename[128], *sdsname;
+    char slavename[REDIS_PEER_ID_LEN], *sdsname;
 
     redisAssert(flags & (SRI_MASTER|SRI_SLAVE|SRI_SENTINEL));
     redisAssert((flags & SRI_MASTER) || master != NULL);
@@ -907,9 +908,7 @@ sentinelRedisInstance *createSentinelRedisInstance(char *name, int flags, char *
 
     /* For slaves and sentinel we use ip:port as name. */
     if (flags & (SRI_SLAVE|SRI_SENTINEL)) {
-        snprintf(slavename,sizeof(slavename),
-            strchr(hostname,':') ? "[%s]:%d" : "%s:%d",
-            hostname,port);
+        anetFormatAddr(slavename, sizeof(slavename), hostname, port);
         name = slavename;
     }
 
@@ -983,6 +982,7 @@ sentinelRedisInstance *createSentinelRedisInstance(char *name, int flags, char *
     ri->promoted_slave = NULL;
     ri->notification_script = NULL;
     ri->client_reconfig_script = NULL;
+    ri->info = NULL;
 
     /* Role */
     ri->role_reported = ri->flags & (SRI_MASTER|SRI_SLAVE);
@@ -1015,6 +1015,7 @@ void releaseSentinelRedisInstance(sentinelRedisInstance *ri) {
     sdsfree(ri->slave_master_host);
     sdsfree(ri->leader);
     sdsfree(ri->auth_pass);
+    sdsfree(ri->info);
     releaseSentinelAddr(ri->addr);
 
     /* Clear state into the master if needed. */
@@ -1030,11 +1031,11 @@ sentinelRedisInstance *sentinelRedisInstanceLookupSlave(
 {
     sds key;
     sentinelRedisInstance *slave;
+    char buf[REDIS_PEER_ID_LEN];
 
     redisAssert(ri->flags & SRI_MASTER);
-    key = sdscatprintf(sdsempty(),
-        strchr(ip,':') ? "[%s]:%d" : "%s:%d",
-        ip,port);
+    anetFormatAddr(buf,sizeof(buf),ip,port);
+    key = sdsnew(buf);
     slave = dictFetchValue(ri->slaves,key);
     sdsfree(key);
     return slave;
@@ -1785,6 +1786,10 @@ void sentinelRefreshInstanceInfo(sentinelRedisInstance *ri, const char *info) {
     int numlines, j;
     int role = 0;
 
+    /* cache full INFO output for instance */
+    sdsfree(ri->info);
+    ri->info = sdsnew(info);
+
     /* The following fields must be reset to a given value in the case they
      * are not found at all in the INFO output. */
     ri->master_link_down_time = 0;
@@ -2777,6 +2782,67 @@ void sentinelCommand(redisClient *c) {
     } else if (!strcasecmp(c->argv[1]->ptr,"set")) {
         if (c->argc < 3 || c->argc % 2 == 0) goto numargserr;
         sentinelSetCommand(c);
+    } else if (!strcasecmp(c->argv[1]->ptr,"info-cache")) {
+        if (c->argc < 2) goto numargserr;
+        mstime_t now = mstime();
+
+        /* Create an ad-hoc dictionary type so that we can iterate
+         * a dictionary composed of just the master groups the user
+         * requested. */
+        dictType copy_keeper = instancesDictType;
+        copy_keeper.valDestructor = NULL;
+        dict *masters_local = sentinel.masters;
+        if (c->argc > 2) {
+            masters_local = dictCreate(&copy_keeper, NULL);
+
+            for (int i = 2; i < c->argc; i++) {
+                sentinelRedisInstance *ri;
+                ri = sentinelGetMasterByName(c->argv[i]->ptr);
+                if (!ri) continue; /* ignore non-existing names */
+                dictAdd(masters_local, ri->name, ri);
+            }
+        }
+
+        /* Reply format:
+         *   1.) master name
+         *   2.) 1.) info from master
+         *       2.) info from replica
+         *       ...
+         *   3.) other master name
+         *   ...
+         */
+        addReplyMultiBulkLen(c,dictSize(masters_local) * 2);
+
+        dictIterator  *di;
+        dictEntry *de;
+        di = dictGetIterator(masters_local);
+        while ((de = dictNext(di)) != NULL) {
+            sentinelRedisInstance *ri = dictGetVal(de);
+            addReplyBulkCBuffer(c,ri->name,strlen(ri->name));
+            addReplyMultiBulkLen(c,dictSize(ri->slaves) + 1); /* +1 for self */
+            addReplyMultiBulkLen(c,2);
+            addReplyLongLong(c, now - ri->info_refresh);
+            if (ri->info)
+                addReplyBulkCBuffer(c,ri->info,sdslen(ri->info));
+            else
+                addReply(c,shared.nullbulk);
+
+            dictIterator *sdi;
+            dictEntry *sde;
+            sdi = dictGetIterator(ri->slaves);
+            while ((sde = dictNext(sdi)) != NULL) {
+                sentinelRedisInstance *sri = dictGetVal(sde);
+                addReplyMultiBulkLen(c,2);
+                addReplyLongLong(c, now - sri->info_refresh);
+                if (sri->info)
+                    addReplyBulkCBuffer(c,sri->info,sdslen(sri->info));
+                else
+                    addReply(c,shared.nullbulk);
+            }
+            dictReleaseIterator(sdi);
+        }
+        dictReleaseIterator(di);
+        if (masters_local != sentinel.masters) dictRelease(masters_local);
     } else {
         addReplyErrorFormat(c,"Unknown sentinel subcommand '%s'",
                                (char*)c->argv[1]->ptr);
@@ -2842,10 +2908,7 @@ void sentinelInfoCommand(redisClient *c) {
         dictReleaseIterator(di);
     }
 
-    addReplySds(c,sdscatprintf(sdsempty(),"$%lu\r\n",
-        (unsigned long)sdslen(info)));
-    addReplySds(c,info);
-    addReply(c,shared.crlf);
+    addReplyBulkSds(c, info);
 }
 
 /* Implements Sentinel verison of the ROLE command. The output is

src/sha1.c

@@ -24,9 +24,7 @@ A million repetitions of "a"
 #include <stdio.h>
 #include <string.h>
 #include <sys/types.h>	/* for u_int*_t */
-#if defined(__sun)
 #include "solarisfixes.h"
-#endif
 #include "sha1.h"
 #include "config.h"
 
@@ -199,16 +197,19 @@ void SHA1Final(unsigned char digest[20], SHA1_CTX* context)
 }
 /* ================ end of sha1.c ================ */
 
-#if 0
+#ifdef REDIS_TEST
 #define BUFSIZE 4096
 
-int
-main(int argc, char **argv)
+#define UNUSED(x) (void)(x)
+int sha1Test(int argc, char **argv)
 {
     SHA1_CTX ctx;
     unsigned char hash[20], buf[BUFSIZE];
     int i;
 
+    UNUSED(argc);
+    UNUSED(argv);
+
     for(i=0;i<BUFSIZE;i++)
         buf[i] = i;
 
@@ -223,6 +224,4 @@ main(int argc, char **argv)
     printf("\n");
     return 0;
 }
-
 #endif
-

src/sha1.h

@@ -1,3 +1,5 @@
+#ifndef SHA1_H
+#define SHA1_H
 /* ================ sha1.h ================ */
 /*
 SHA-1 in C
@@ -15,3 +17,8 @@ void SHA1Transform(u_int32_t state[5], const unsigned char buffer[64]);
 void SHA1Init(SHA1_CTX* context);
 void SHA1Update(SHA1_CTX* context, const unsigned char* data, u_int32_t len);
 void SHA1Final(unsigned char digest[20], SHA1_CTX* context);
+
+#ifdef REDIS_TEST
+int sha1Test(int argc, char **argv);
+#endif
+#endif

src/solarisfixes.h

@@ -28,6 +28,8 @@
  * POSSIBILITY OF SUCH DAMAGE.
  */
 
+#if defined(__sun)
+
 #if defined(__GNUC__)
 #include <math.h>
 #undef isnan
@@ -48,3 +50,5 @@
 #define u_int uint
 #define u_int32_t uint32_t
 #endif /* __GNUC__ */
+
+#endif /* __sun */

src/sort.c

@@ -220,7 +220,7 @@ void sortCommand(redisClient *c) {
     if (sortval)
         incrRefCount(sortval);
     else
-        sortval = createListObject();
+        sortval = createQuicklistObject();
 
     /* The SORT command has an SQL-alike syntax, parse it */
     while(j < c->argc) {
@@ -285,16 +285,15 @@ void sortCommand(redisClient *c) {
         return;
     }
 
-    /* For the STORE option, or when SORT is called from a Lua script,
-     * we want to force a specific ordering even when no explicit ordering
-     * was asked (SORT BY nosort). This guarantees that replication / AOF
-     * is deterministic.
+    /* When sorting a set with no sort specified, we must sort the output
+     * so the result is consistent across scripting and replication.
      *
-     * However in the case 'dontsort' is true, but the type to sort is a
-     * sorted set, we don't need to do anything as ordering is guaranteed
-     * in this special case. */
-    if ((storekey || c->flags & REDIS_LUA_CLIENT) &&
-        (dontsort && sortval->type != REDIS_ZSET))
+     * The other types (list, sorted set) will retain their native order
+     * even if no sort order is requested, so they remain stable across
+     * scripting and replication. */
+    if (dontsort &&
+        sortval->type == REDIS_SET &&
+        (storekey || c->flags & REDIS_LUA_CLIENT))
     {
         /* Force ALPHA sorting */
         dontsort = 0;
@@ -323,17 +322,17 @@ void sortCommand(redisClient *c) {
     }
     if (end >= vectorlen) end = vectorlen-1;
 
-    /* Optimization:
+    /* Whenever possible, we load elements into the output array in a more
+     * direct way. This is possible if:
      *
-     * 1) if the object to sort is a sorted set.
+     * 1) The object to sort is a sorted set or a list (internally sorted).
      * 2) There is nothing to sort as dontsort is true (BY <constant string>).
-     * 3) We have a LIMIT option that actually reduces the number of elements
-     *    to fetch.
      *
-     * In this case to load all the objects in the vector is a huge waste of
-     * resources. We just allocate a vector that is big enough for the selected
-     * range length, and make sure to load just this part in the vector. */
-    if (sortval->type == REDIS_ZSET &&
+     * In this special case, if we have a LIMIT option that actually reduces
+     * the number of elements to fetch, we also optimize to just load the
+     * range we are interested in and allocating a vector that is big enough
+     * for the selected range length. */
+    if ((sortval->type == REDIS_ZSET || sortval->type == REDIS_LIST) &&
         dontsort &&
         (start != 0 || end != vectorlen-1))
     {
@@ -344,7 +343,32 @@ void sortCommand(redisClient *c) {
     vector = zmalloc(sizeof(redisSortObject)*vectorlen);
     j = 0;
 
-    if (sortval->type == REDIS_LIST) {
+    if (sortval->type == REDIS_LIST && dontsort) {
+        /* Special handling for a list, if 'dontsort' is true.
+         * This makes sure we return elements in the list original
+         * ordering, accordingly to DESC / ASC options.
+         *
+         * Note that in this case we also handle LIMIT here in a direct
+         * way, just getting the required range, as an optimization. */
+        if (end >= start) {
+            listTypeIterator *li;
+            listTypeEntry entry;
+            li = listTypeInitIterator(sortval,
+                    desc ? (long)(listTypeLength(sortval) - start - 1) : start,
+                    desc ? REDIS_HEAD : REDIS_TAIL);
+
+            while(j < vectorlen && listTypeNext(li,&entry)) {
+                vector[j].obj = listTypeGet(&entry);
+                vector[j].u.score = 0;
+                vector[j].u.cmpobj = NULL;
+                j++;
+            }
+            listTypeReleaseIterator(li);
+            /* Fix start/end: output code is not aware of this optimization. */
+            end -= start;
+            start = 0;
+        }
+    } else if (sortval->type == REDIS_LIST) {
         listTypeIterator *li = listTypeInitIterator(sortval,0,REDIS_TAIL);
         listTypeEntry entry;
         while(listTypeNext(li,&entry)) {
@@ -400,10 +424,7 @@ void sortCommand(redisClient *c) {
             j++;
             ln = desc ? ln->backward : ln->level[0].forward;
         }
-        /* The code producing the output does not know that in the case of
-         * sorted set, 'dontsort', and LIMIT, we are able to get just the
-         * range, already sorted, so we need to adjust "start" and "end"
-         * to make sure start is set to 0. */
+        /* Fix start/end: output code is not aware of this optimization. */
         end -= start;
         start = 0;
     } else if (sortval->type == REDIS_ZSET) {
@@ -510,7 +531,7 @@ void sortCommand(redisClient *c) {
             }
         }
     } else {
-        robj *sobj = createZiplistObject();
+        robj *sobj = createQuicklistObject();
 
         /* STORE option specified, set the sorting result as a List object */
         for (j = start; j <= end; j++) {

src/sparkline.c

@@ -49,7 +49,7 @@ static int label_margin_top = 1;
  * sparklineSequenceAddSample(seq, 10, NULL);
  * sparklineSequenceAddSample(seq, 20, NULL);
  * sparklineSequenceAddSample(seq, 30, "last sample label");
- * sds output = sparklineRender(seq, 80, 4);
+ * sds output = sparklineRender(sdsempty(), seq, 80, 4, SPARKLINE_FILL);
  * freeSparklineSequence(seq);
  * ------------------------------------------------------------------------- */
 
@@ -63,6 +63,7 @@ struct sequence *createSparklineSequence(void) {
 
 /* Add a new sample into a sequence. */
 void sparklineSequenceAddSample(struct sequence *seq, double value, char *label) {
+    label = (label == NULL || label[0] == '\0') ? NULL : zstrdup(label);
     if (seq->length == 0) {
         seq->min = seq->max = value;
     } else {

src/t_hash.c

@@ -565,7 +565,7 @@ void hincrbyfloatCommand(redisClient *c) {
     }
 
     value += incr;
-    new = createStringObjectFromLongDouble(value);
+    new = createStringObjectFromLongDouble(value,1);
     hashTypeTryObjectEncoding(o,&c->argv[2],NULL);
     hashTypeSet(o,c->argv[2],new);
     addReplyBulk(c,new);

src/t_list.c

@@ -33,75 +33,37 @@
  * List API
  *----------------------------------------------------------------------------*/
 
-/* Check the argument length to see if it requires us to convert the ziplist
- * to a real list. Only check raw-encoded objects because integer encoded
- * objects are never too long. */
-void listTypeTryConversion(robj *subject, robj *value) {
-    if (subject->encoding != REDIS_ENCODING_ZIPLIST) return;
-    if (sdsEncodedObject(value) &&
-        sdslen(value->ptr) > server.list_max_ziplist_value)
-            listTypeConvert(subject,REDIS_ENCODING_LINKEDLIST);
-}
-
 /* The function pushes an element to the specified list object 'subject',
  * at head or tail position as specified by 'where'.
  *
  * There is no need for the caller to increment the refcount of 'value' as
  * the function takes care of it if needed. */
 void listTypePush(robj *subject, robj *value, int where) {
-    /* Check if we need to convert the ziplist */
-    listTypeTryConversion(subject,value);
-    if (subject->encoding == REDIS_ENCODING_ZIPLIST &&
-        ziplistLen(subject->ptr) >= server.list_max_ziplist_entries)
-            listTypeConvert(subject,REDIS_ENCODING_LINKEDLIST);
-
-    if (subject->encoding == REDIS_ENCODING_ZIPLIST) {
-        int pos = (where == REDIS_HEAD) ? ZIPLIST_HEAD : ZIPLIST_TAIL;
+    if (subject->encoding == REDIS_ENCODING_QUICKLIST) {
+        int pos = (where == REDIS_HEAD) ? QUICKLIST_HEAD : QUICKLIST_TAIL;
         value = getDecodedObject(value);
-        subject->ptr = ziplistPush(subject->ptr,value->ptr,sdslen(value->ptr),pos);
+        size_t len = sdslen(value->ptr);
+        quicklistPush(subject->ptr, value->ptr, len, pos);
         decrRefCount(value);
-    } else if (subject->encoding == REDIS_ENCODING_LINKEDLIST) {
-        if (where == REDIS_HEAD) {
-            listAddNodeHead(subject->ptr,value);
-        } else {
-            listAddNodeTail(subject->ptr,value);
-        }
-        incrRefCount(value);
     } else {
         redisPanic("Unknown list encoding");
     }
 }
 
+void *listPopSaver(unsigned char *data, unsigned int sz) {
+    return createStringObject((char*)data,sz);
+}
+
 robj *listTypePop(robj *subject, int where) {
+    long long vlong;
     robj *value = NULL;
-    if (subject->encoding == REDIS_ENCODING_ZIPLIST) {
-        unsigned char *p;
-        unsigned char *vstr;
-        unsigned int vlen;
-        long long vlong;
-        int pos = (where == REDIS_HEAD) ? 0 : -1;
-        p = ziplistIndex(subject->ptr,pos);
-        if (ziplistGet(p,&vstr,&vlen,&vlong)) {
-            if (vstr) {
-                value = createStringObject((char*)vstr,vlen);
-            } else {
+
+    int ql_where = where == REDIS_HEAD ? QUICKLIST_HEAD : QUICKLIST_TAIL;
+    if (subject->encoding == REDIS_ENCODING_QUICKLIST) {
+        if (quicklistPopCustom(subject->ptr, ql_where, (unsigned char **)&value,
+                               NULL, &vlong, listPopSaver)) {
+            if (!value)
                 value = createStringObjectFromLongLong(vlong);
-            }
-            /* We only need to delete an element when it exists */
-            subject->ptr = ziplistDelete(subject->ptr,&p);
-        }
-    } else if (subject->encoding == REDIS_ENCODING_LINKEDLIST) {
-        list *list = subject->ptr;
-        listNode *ln;
-        if (where == REDIS_HEAD) {
-            ln = listFirst(list);
-        } else {
-            ln = listLast(list);
-        }
-        if (ln != NULL) {
-            value = listNodeValue(ln);
-            incrRefCount(value);
-            listDelNode(list,ln);
         }
     } else {
         redisPanic("Unknown list encoding");
@@ -110,25 +72,28 @@ robj *listTypePop(robj *subject, int where) {
 }
 
 unsigned long listTypeLength(robj *subject) {
-    if (subject->encoding == REDIS_ENCODING_ZIPLIST) {
-        return ziplistLen(subject->ptr);
-    } else if (subject->encoding == REDIS_ENCODING_LINKEDLIST) {
-        return listLength((list*)subject->ptr);
+    if (subject->encoding == REDIS_ENCODING_QUICKLIST) {
+        return quicklistCount(subject->ptr);
     } else {
         redisPanic("Unknown list encoding");
     }
 }
 
 /* Initialize an iterator at the specified index. */
-listTypeIterator *listTypeInitIterator(robj *subject, long index, unsigned char direction) {
+listTypeIterator *listTypeInitIterator(robj *subject, long index,
+                                       unsigned char direction) {
     listTypeIterator *li = zmalloc(sizeof(listTypeIterator));
     li->subject = subject;
     li->encoding = subject->encoding;
     li->direction = direction;
-    if (li->encoding == REDIS_ENCODING_ZIPLIST) {
-        li->zi = ziplistIndex(subject->ptr,index);
-    } else if (li->encoding == REDIS_ENCODING_LINKEDLIST) {
-        li->ln = listIndex(subject->ptr,index);
+    li->iter = NULL;
+    /* REDIS_HEAD means start at TAIL and move *towards* head.
+     * REDIS_TAIL means start at HEAD and move *towards tail. */
+    int iter_direction =
+        direction == REDIS_HEAD ? AL_START_TAIL : AL_START_HEAD;
+    if (li->encoding == REDIS_ENCODING_QUICKLIST) {
+        li->iter = quicklistGetIteratorAtIdx(li->subject->ptr,
+                                             iter_direction, index);
     } else {
         redisPanic("Unknown list encoding");
     }
@@ -137,6 +102,7 @@ listTypeIterator *listTypeInitIterator(robj *subject, long index, unsigned char
 
 /* Clean up the iterator. */
 void listTypeReleaseIterator(listTypeIterator *li) {
+    zfree(li->iter);
     zfree(li);
 }
 
@@ -148,24 +114,8 @@ int listTypeNext(listTypeIterator *li, listTypeEntry *entry) {
     redisAssert(li->subject->encoding == li->encoding);
 
     entry->li = li;
-    if (li->encoding == REDIS_ENCODING_ZIPLIST) {
-        entry->zi = li->zi;
-        if (entry->zi != NULL) {
-            if (li->direction == REDIS_TAIL)
-                li->zi = ziplistNext(li->subject->ptr,li->zi);
-            else
-                li->zi = ziplistPrev(li->subject->ptr,li->zi);
-            return 1;
-        }
-    } else if (li->encoding == REDIS_ENCODING_LINKEDLIST) {
-        entry->ln = li->ln;
-        if (entry->ln != NULL) {
-            if (li->direction == REDIS_TAIL)
-                li->ln = li->ln->next;
-            else
-                li->ln = li->ln->prev;
-            return 1;
-        }
+    if (li->encoding == REDIS_ENCODING_QUICKLIST) {
+        return quicklistNext(li->iter, &entry->entry);
     } else {
         redisPanic("Unknown list encoding");
     }
@@ -174,24 +124,14 @@ int listTypeNext(listTypeIterator *li, listTypeEntry *entry) {
 
 /* Return entry or NULL at the current position of the iterator. */
 robj *listTypeGet(listTypeEntry *entry) {
-    listTypeIterator *li = entry->li;
     robj *value = NULL;
-    if (li->encoding == REDIS_ENCODING_ZIPLIST) {
-        unsigned char *vstr;
-        unsigned int vlen;
-        long long vlong;
-        redisAssert(entry->zi != NULL);
-        if (ziplistGet(entry->zi,&vstr,&vlen,&vlong)) {
-            if (vstr) {
-                value = createStringObject((char*)vstr,vlen);
-            } else {
-                value = createStringObjectFromLongLong(vlong);
-            }
+    if (entry->li->encoding == REDIS_ENCODING_QUICKLIST) {
+        if (entry->entry.value) {
+            value = createStringObject((char *)entry->entry.value,
+                                       entry->entry.sz);
+        } else {
+            value = createStringObjectFromLongLong(entry->entry.longval);
         }
-    } else if (li->encoding == REDIS_ENCODING_LINKEDLIST) {
-        redisAssert(entry->ln != NULL);
-        value = listNodeValue(entry->ln);
-        incrRefCount(value);
     } else {
         redisPanic("Unknown list encoding");
     }
@@ -199,30 +139,18 @@ robj *listTypeGet(listTypeEntry *entry) {
 }
 
 void listTypeInsert(listTypeEntry *entry, robj *value, int where) {
-    robj *subject = entry->li->subject;
-    if (entry->li->encoding == REDIS_ENCODING_ZIPLIST) {
+    if (entry->li->encoding == REDIS_ENCODING_QUICKLIST) {
         value = getDecodedObject(value);
+        sds str = value->ptr;
+        size_t len = sdslen(str);
         if (where == REDIS_TAIL) {
-            unsigned char *next = ziplistNext(subject->ptr,entry->zi);
-
-            /* When we insert after the current element, but the current element
-             * is the tail of the list, we need to do a push. */
-            if (next == NULL) {
-                subject->ptr = ziplistPush(subject->ptr,value->ptr,sdslen(value->ptr),REDIS_TAIL);
-            } else {
-                subject->ptr = ziplistInsert(subject->ptr,next,value->ptr,sdslen(value->ptr));
-            }
-        } else {
-            subject->ptr = ziplistInsert(subject->ptr,entry->zi,value->ptr,sdslen(value->ptr));
+            quicklistInsertAfter((quicklist *)entry->entry.quicklist,
+                                 &entry->entry, str, len);
+        } else if (where == REDIS_HEAD) {
+            quicklistInsertBefore((quicklist *)entry->entry.quicklist,
+                                  &entry->entry, str, len);
         }
         decrRefCount(value);
-    } else if (entry->li->encoding == REDIS_ENCODING_LINKEDLIST) {
-        if (where == REDIS_TAIL) {
-            listInsertNode(subject->ptr,entry->ln,value,AL_START_TAIL);
-        } else {
-            listInsertNode(subject->ptr,entry->ln,value,AL_START_HEAD);
-        }
-        incrRefCount(value);
     } else {
         redisPanic("Unknown list encoding");
     }
@@ -230,59 +158,33 @@ void listTypeInsert(listTypeEntry *entry, robj *value, int where) {
 
 /* Compare the given object with the entry at the current position. */
 int listTypeEqual(listTypeEntry *entry, robj *o) {
-    listTypeIterator *li = entry->li;
-    if (li->encoding == REDIS_ENCODING_ZIPLIST) {
+    if (entry->li->encoding == REDIS_ENCODING_QUICKLIST) {
         redisAssertWithInfo(NULL,o,sdsEncodedObject(o));
-        return ziplistCompare(entry->zi,o->ptr,sdslen(o->ptr));
-    } else if (li->encoding == REDIS_ENCODING_LINKEDLIST) {
-        return equalStringObjects(o,listNodeValue(entry->ln));
+        return quicklistCompare(entry->entry.zi,o->ptr,sdslen(o->ptr));
     } else {
         redisPanic("Unknown list encoding");
     }
 }
 
 /* Delete the element pointed to. */
-void listTypeDelete(listTypeEntry *entry) {
-    listTypeIterator *li = entry->li;
-    if (li->encoding == REDIS_ENCODING_ZIPLIST) {
-        unsigned char *p = entry->zi;
-        li->subject->ptr = ziplistDelete(li->subject->ptr,&p);
-
-        /* Update position of the iterator depending on the direction */
-        if (li->direction == REDIS_TAIL)
-            li->zi = p;
-        else
-            li->zi = ziplistPrev(li->subject->ptr,p);
-    } else if (entry->li->encoding == REDIS_ENCODING_LINKEDLIST) {
-        listNode *next;
-        if (li->direction == REDIS_TAIL)
-            next = entry->ln->next;
-        else
-            next = entry->ln->prev;
-        listDelNode(li->subject->ptr,entry->ln);
-        li->ln = next;
+void listTypeDelete(listTypeIterator *iter, listTypeEntry *entry) {
+    if (entry->li->encoding == REDIS_ENCODING_QUICKLIST) {
+        quicklistDelEntry(iter->iter, &entry->entry);
     } else {
         redisPanic("Unknown list encoding");
     }
 }
 
+/* Create a quicklist from a single ziplist */
 void listTypeConvert(robj *subject, int enc) {
-    listTypeIterator *li;
-    listTypeEntry entry;
-    redisAssertWithInfo(NULL,subject,subject->type == REDIS_LIST);
+    redisAssertWithInfo(NULL,subject,subject->type==REDIS_LIST);
+    redisAssertWithInfo(NULL,subject,subject->encoding==REDIS_ENCODING_ZIPLIST);
 
-    if (enc == REDIS_ENCODING_LINKEDLIST) {
-        list *l = listCreate();
-        listSetFreeMethod(l,decrRefCountVoid);
-
-        /* listTypeGet returns a robj with incremented refcount */
-        li = listTypeInitIterator(subject,0,REDIS_TAIL);
-        while (listTypeNext(li,&entry)) listAddNodeTail(l,listTypeGet(&entry));
-        listTypeReleaseIterator(li);
-
-        subject->encoding = REDIS_ENCODING_LINKEDLIST;
-        zfree(subject->ptr);
-        subject->ptr = l;
+    if (enc == REDIS_ENCODING_QUICKLIST) {
+        size_t zlen = server.list_max_ziplist_size;
+        int depth = server.list_compress_depth;
+        subject->ptr = quicklistCreateFromZiplist(zlen, depth, subject->ptr);
+        subject->encoding = REDIS_ENCODING_QUICKLIST;
     } else {
         redisPanic("Unsupported list conversion");
     }
@@ -304,7 +206,9 @@ void pushGenericCommand(redisClient *c, int where) {
     for (j = 2; j < c->argc; j++) {
         c->argv[j] = tryObjectEncoding(c->argv[j]);
         if (!lobj) {
-            lobj = createZiplistObject();
+            lobj = createQuicklistObject();
+            quicklistSetOptions(lobj->ptr, server.list_max_ziplist_size,
+                                server.list_compress_depth);
             dbAdd(c->db,c->argv[1],lobj);
         }
         listTypePush(lobj,c->argv[j],where);
@@ -334,17 +238,10 @@ void pushxGenericCommand(redisClient *c, robj *refval, robj *val, int where) {
     listTypeEntry entry;
     int inserted = 0;
 
-    if ((subject = lookupKeyReadOrReply(c,c->argv[1],shared.czero)) == NULL ||
+    if ((subject = lookupKeyWriteOrReply(c,c->argv[1],shared.czero)) == NULL ||
         checkType(c,subject,REDIS_LIST)) return;
 
     if (refval != NULL) {
-        /* We're not sure if this value can be inserted yet, but we cannot
-         * convert the list inside the iterator. We don't want to loop over
-         * the list twice (once to see if the value can be inserted and once
-         * to do the actual insert), so we assume this value can be inserted
-         * and convert the ziplist to a regular list if necessary. */
-        listTypeTryConversion(subject,val);
-
         /* Seek refval from head to tail */
         iter = listTypeInitIterator(subject,0,REDIS_TAIL);
         while (listTypeNext(iter,&entry)) {
@@ -357,10 +254,6 @@ void pushxGenericCommand(redisClient *c, robj *refval, robj *val, int where) {
         listTypeReleaseIterator(iter);
 
         if (inserted) {
-            /* Check if the length exceeds the ziplist length threshold. */
-            if (subject->encoding == REDIS_ENCODING_ZIPLIST &&
-                ziplistLen(subject->ptr) > server.list_max_ziplist_entries)
-                    listTypeConvert(subject,REDIS_ENCODING_LINKEDLIST);
             signalModifiedKey(c->db,c->argv[1]);
             notifyKeyspaceEvent(REDIS_NOTIFY_LIST,"linsert",
                                 c->argv[1],c->db->id);
@@ -418,31 +311,19 @@ void lindexCommand(redisClient *c) {
     if ((getLongFromObjectOrReply(c, c->argv[2], &index, NULL) != REDIS_OK))
         return;
 
-    if (o->encoding == REDIS_ENCODING_ZIPLIST) {
-        unsigned char *p;
-        unsigned char *vstr;
-        unsigned int vlen;
-        long long vlong;
-        p = ziplistIndex(o->ptr,index);
-        if (ziplistGet(p,&vstr,&vlen,&vlong)) {
-            if (vstr) {
-                value = createStringObject((char*)vstr,vlen);
+    if (o->encoding == REDIS_ENCODING_QUICKLIST) {
+        quicklistEntry entry;
+        if (quicklistIndex(o->ptr, index, &entry)) {
+            if (entry.value) {
+                value = createStringObject((char*)entry.value,entry.sz);
             } else {
-                value = createStringObjectFromLongLong(vlong);
+                value = createStringObjectFromLongLong(entry.longval);
             }
             addReplyBulk(c,value);
             decrRefCount(value);
         } else {
             addReply(c,shared.nullbulk);
         }
-    } else if (o->encoding == REDIS_ENCODING_LINKEDLIST) {
-        listNode *ln = listIndex(o->ptr,index);
-        if (ln != NULL) {
-            value = listNodeValue(ln);
-            addReplyBulk(c,value);
-        } else {
-            addReply(c,shared.nullbulk);
-        }
     } else {
         redisPanic("Unknown list encoding");
     }
@@ -452,35 +333,18 @@ void lsetCommand(redisClient *c) {
     robj *o = lookupKeyWriteOrReply(c,c->argv[1],shared.nokeyerr);
     if (o == NULL || checkType(c,o,REDIS_LIST)) return;
     long index;
-    robj *value = (c->argv[3] = tryObjectEncoding(c->argv[3]));
+    robj *value = c->argv[3];
 
     if ((getLongFromObjectOrReply(c, c->argv[2], &index, NULL) != REDIS_OK))
         return;
 
-    listTypeTryConversion(o,value);
-    if (o->encoding == REDIS_ENCODING_ZIPLIST) {
-        unsigned char *p, *zl = o->ptr;
-        p = ziplistIndex(zl,index);
-        if (p == NULL) {
+    if (o->encoding == REDIS_ENCODING_QUICKLIST) {
+        quicklist *ql = o->ptr;
+        int replaced = quicklistReplaceAtIndex(ql, index,
+                                               value->ptr, sdslen(value->ptr));
+        if (!replaced) {
             addReply(c,shared.outofrangeerr);
         } else {
-            o->ptr = ziplistDelete(o->ptr,&p);
-            value = getDecodedObject(value);
-            o->ptr = ziplistInsert(o->ptr,p,value->ptr,sdslen(value->ptr));
-            decrRefCount(value);
-            addReply(c,shared.ok);
-            signalModifiedKey(c->db,c->argv[1]);
-            notifyKeyspaceEvent(REDIS_NOTIFY_LIST,"lset",c->argv[1],c->db->id);
-            server.dirty++;
-        }
-    } else if (o->encoding == REDIS_ENCODING_LINKEDLIST) {
-        listNode *ln = listIndex(o->ptr,index);
-        if (ln == NULL) {
-            addReply(c,shared.outofrangeerr);
-        } else {
-            decrRefCount((robj*)listNodeValue(ln));
-            listNodeValue(ln) = value;
-            incrRefCount(value);
             addReply(c,shared.ok);
             signalModifiedKey(c->db,c->argv[1]);
             notifyKeyspaceEvent(REDIS_NOTIFY_LIST,"lset",c->argv[1],c->db->id);
@@ -549,43 +413,28 @@ void lrangeCommand(redisClient *c) {
 
     /* Return the result in form of a multi-bulk reply */
     addReplyMultiBulkLen(c,rangelen);
-    if (o->encoding == REDIS_ENCODING_ZIPLIST) {
-        unsigned char *p = ziplistIndex(o->ptr,start);
-        unsigned char *vstr;
-        unsigned int vlen;
-        long long vlong;
+    if (o->encoding == REDIS_ENCODING_QUICKLIST) {
+        listTypeIterator *iter = listTypeInitIterator(o, start, REDIS_TAIL);
 
         while(rangelen--) {
-            ziplistGet(p,&vstr,&vlen,&vlong);
-            if (vstr) {
-                addReplyBulkCBuffer(c,vstr,vlen);
+            listTypeEntry entry;
+            listTypeNext(iter, &entry);
+            quicklistEntry *qe = &entry.entry;
+            if (qe->value) {
+                addReplyBulkCBuffer(c,qe->value,qe->sz);
             } else {
-                addReplyBulkLongLong(c,vlong);
+                addReplyBulkLongLong(c,qe->longval);
             }
-            p = ziplistNext(o->ptr,p);
-        }
-    } else if (o->encoding == REDIS_ENCODING_LINKEDLIST) {
-        listNode *ln;
-
-        /* If we are nearest to the end of the list, reach the element
-         * starting from tail and going backward, as it is faster. */
-        if (start > llen/2) start -= llen;
-        ln = listIndex(o->ptr,start);
-
-        while(rangelen--) {
-            addReplyBulk(c,ln->value);
-            ln = ln->next;
         }
+        listTypeReleaseIterator(iter);
     } else {
-        redisPanic("List encoding is not LINKEDLIST nor ZIPLIST!");
+        redisPanic("List encoding is not QUICKLIST!");
     }
 }
 
 void ltrimCommand(redisClient *c) {
     robj *o;
-    long start, end, llen, j, ltrim, rtrim;
-    list *list;
-    listNode *ln;
+    long start, end, llen, ltrim, rtrim;
 
     if ((getLongFromObjectOrReply(c, c->argv[2], &start, NULL) != REDIS_OK) ||
         (getLongFromObjectOrReply(c, c->argv[3], &end, NULL) != REDIS_OK)) return;
@@ -612,19 +461,9 @@ void ltrimCommand(redisClient *c) {
     }
 
     /* Remove list elements to perform the trim */
-    if (o->encoding == REDIS_ENCODING_ZIPLIST) {
-        o->ptr = ziplistDeleteRange(o->ptr,0,ltrim);
-        o->ptr = ziplistDeleteRange(o->ptr,-rtrim,rtrim);
-    } else if (o->encoding == REDIS_ENCODING_LINKEDLIST) {
-        list = o->ptr;
-        for (j = 0; j < ltrim; j++) {
-            ln = listFirst(list);
-            listDelNode(list,ln);
-        }
-        for (j = 0; j < rtrim; j++) {
-            ln = listLast(list);
-            listDelNode(list,ln);
-        }
+    if (o->encoding == REDIS_ENCODING_QUICKLIST) {
+        quicklistDelRange(o->ptr,0,ltrim);
+        quicklistDelRange(o->ptr,-rtrim,rtrim);
     } else {
         redisPanic("Unknown list encoding");
     }
@@ -641,10 +480,9 @@ void ltrimCommand(redisClient *c) {
 
 void lremCommand(redisClient *c) {
     robj *subject, *obj;
-    obj = c->argv[3] = tryObjectEncoding(c->argv[3]);
+    obj = c->argv[3];
     long toremove;
     long removed = 0;
-    listTypeEntry entry;
 
     if ((getLongFromObjectOrReply(c, c->argv[2], &toremove, NULL) != REDIS_OK))
         return;
@@ -652,10 +490,6 @@ void lremCommand(redisClient *c) {
     subject = lookupKeyWriteOrReply(c,c->argv[1],shared.czero);
     if (subject == NULL || checkType(c,subject,REDIS_LIST)) return;
 
-    /* Make sure obj is raw when we're dealing with a ziplist */
-    if (subject->encoding == REDIS_ENCODING_ZIPLIST)
-        obj = getDecodedObject(obj);
-
     listTypeIterator *li;
     if (toremove < 0) {
         toremove = -toremove;
@@ -664,9 +498,10 @@ void lremCommand(redisClient *c) {
         li = listTypeInitIterator(subject,0,REDIS_TAIL);
     }
 
+    listTypeEntry entry;
     while (listTypeNext(li,&entry)) {
         if (listTypeEqual(&entry,obj)) {
-            listTypeDelete(&entry);
+            listTypeDelete(li, &entry);
             server.dirty++;
             removed++;
             if (toremove && removed == toremove) break;
@@ -674,11 +509,10 @@ void lremCommand(redisClient *c) {
     }
     listTypeReleaseIterator(li);
 
-    /* Clean up raw encoded object */
-    if (subject->encoding == REDIS_ENCODING_ZIPLIST)
-        decrRefCount(obj);
+    if (listTypeLength(subject) == 0) {
+        dbDelete(c->db,c->argv[1]);
+    }
 
-    if (listTypeLength(subject) == 0) dbDelete(c->db,c->argv[1]);
     addReplyLongLong(c,removed);
     if (removed) signalModifiedKey(c->db,c->argv[1]);
 }
@@ -702,7 +536,9 @@ void lremCommand(redisClient *c) {
 void rpoplpushHandlePush(redisClient *c, robj *dstkey, robj *dstobj, robj *value) {
     /* Create the list if the key does not exist */
     if (!dstobj) {
-        dstobj = createZiplistObject();
+        dstobj = createQuicklistObject();
+        quicklistSetOptions(dstobj->ptr, server.list_max_ziplist_size,
+                            server.list_compress_depth);
         dbAdd(c->db,dstkey,dstobj);
     }
     signalModifiedKey(c->db,dstkey);
@@ -1010,7 +846,9 @@ void handleClientsBlockedOnLists(void) {
                     }
                 }
 
-                if (listTypeLength(o) == 0) dbDelete(rl->db,rl->key);
+                if (listTypeLength(o) == 0) {
+                    dbDelete(rl->db,rl->key);
+                }
                 /* We don't call signalModifiedKey() as it was already called
                  * when an element was pushed on the list. */
             }

src/t_set.c

@@ -33,7 +33,8 @@
  * Set Commands
  *----------------------------------------------------------------------------*/
 
-void sunionDiffGenericCommand(redisClient *c, robj **setkeys, int setnum, robj *dstkey, int op);
+void sunionDiffGenericCommand(redisClient *c, robj **setkeys, int setnum,
+                              robj *dstkey, int op);
 
 /* Factory method to return a set that *can* hold "value". When the object has
  * an integer-encodable value, an intset will be returned. Otherwise a regular
@@ -44,6 +45,11 @@ robj *setTypeCreate(robj *value) {
     return createSetObject();
 }
 
+/* Add the specified value into a set. The function takes care of incrementing
+ * the reference count of the object if needed in order to retain a copy.
+ *
+ * If the value was already member of the set, nothing is done and 0 is
+ * returned, otherwise the new element is added and 1 is returned. */
 int setTypeAdd(robj *subject, robj *value) {
     long long llval;
     if (subject->encoding == REDIS_ENCODING_HT) {
@@ -68,7 +74,8 @@ int setTypeAdd(robj *subject, robj *value) {
 
             /* The set *was* an intset and this value is not integer
              * encodable, so dictAdd should always work. */
-            redisAssertWithInfo(NULL,value,dictAdd(subject->ptr,value,NULL) == DICT_OK);
+            redisAssertWithInfo(NULL,value,
+                                dictAdd(subject->ptr,value,NULL) == DICT_OK);
             incrRefCount(value);
             return 1;
         }
@@ -235,7 +242,8 @@ void setTypeConvert(robj *setobj, int enc) {
         si = setTypeInitIterator(setobj);
         while (setTypeNext(si,NULL,&intele) != -1) {
             element = createStringObjectFromLongLong(intele);
-            redisAssertWithInfo(NULL,element,dictAdd(d,element,NULL) == DICT_OK);
+            redisAssertWithInfo(NULL,element,
+                                dictAdd(d,element,NULL) == DICT_OK);
         }
         setTypeReleaseIterator(si);
 
@@ -377,15 +385,185 @@ void scardCommand(redisClient *c) {
     addReplyLongLong(c,setTypeSize(o));
 }
 
+/* Handle the "SPOP key <count>" variant. The normal version of the
+ * command is handled by the spopCommand() function itself. */
+
+/* How many times bigger should be the set compared to the remaining size
+ * for us to use the "create new set" strategy? Read later in the
+ * implementation for more info. */
+#define SPOP_MOVE_STRATEGY_MUL 5
+
+void spopWithCountCommand(redisClient *c) {
+    long l;
+    unsigned long count, size;
+    robj *set;
+
+    /* Get the count argument */
+    if (getLongFromObjectOrReply(c,c->argv[2],&l,NULL) != REDIS_OK) return;
+    if (l >= 0) {
+        count = (unsigned) l;
+    } else {
+        addReply(c,shared.outofrangeerr);
+        return;
+    }
+
+    /* Make sure a key with the name inputted exists, and that it's type is
+     * indeed a set. Otherwise, return nil */
+    if ((set = lookupKeyReadOrReply(c,c->argv[1],shared.emptymultibulk))
+        == NULL || checkType(c,set,REDIS_SET)) return;
+
+    /* If count is zero, serve an empty multibulk ASAP to avoid special
+     * cases later. */
+    if (count == 0) {
+        addReply(c,shared.emptymultibulk);
+        return;
+    }
+
+    size = setTypeSize(set);
+
+    /* Generate an SPOP keyspace notification */
+    notifyKeyspaceEvent(REDIS_NOTIFY_SET,"spop",c->argv[1],c->db->id);
+    server.dirty += count;
+
+    /* CASE 1:
+     * The number of requested elements is greater than or equal to
+     * the number of elements inside the set: simply return the whole set. */
+    if (count >= size) {
+        /* We just return the entire set */
+        sunionDiffGenericCommand(c,c->argv+1,1,NULL,REDIS_OP_UNION);
+
+        /* Delete the set as it is now empty */
+        dbDelete(c->db,c->argv[1]);
+        notifyKeyspaceEvent(REDIS_NOTIFY_GENERIC,"del",c->argv[1],c->db->id);
+
+        /* Propagate this command as an DEL operation */
+        rewriteClientCommandVector(c,2,shared.del,c->argv[1]);
+        signalModifiedKey(c->db,c->argv[1]);
+        server.dirty++;
+        return;
+    }
+
+    /* Case 2 and 3 require to replicate SPOP as a set of SERM commands.
+     * Prepare our replication argument vector. Also send the array length
+     * which is common to both the code paths. */
+    robj *propargv[3];
+    propargv[0] = createStringObject("SREM",4);
+    propargv[1] = c->argv[1];
+    addReplyMultiBulkLen(c,count);
+
+    /* Common iteration vars. */
+    robj *objele;
+    int encoding;
+    int64_t llele;
+    unsigned long remaining = size-count; /* Elements left after SPOP. */
+
+    /* If we are here, the number of requested elements is less than the
+     * number of elements inside the set. Also we are sure that count < size.
+     * Use two different strategies.
+     *
+     * CASE 2: The number of elements to return is small compared to the
+     * set size. We can just extract random elements and return them to
+     * the set. */
+    if (remaining*SPOP_MOVE_STRATEGY_MUL > count) {
+        while(count--) {
+            encoding = setTypeRandomElement(set,&objele,&llele);
+            if (encoding == REDIS_ENCODING_INTSET) {
+                objele = createStringObjectFromLongLong(llele);
+            } else {
+                incrRefCount(objele);
+            }
+
+            /* Return the element to the client and remove from the set. */
+            addReplyBulk(c,objele);
+            setTypeRemove(set,objele);
+
+            /* Replicate/AOF this command as an SREM operation */
+            propargv[2] = objele;
+            alsoPropagate(server.sremCommand,c->db->id,propargv,3,
+                REDIS_PROPAGATE_AOF|REDIS_PROPAGATE_REPL);
+            decrRefCount(objele);
+        }
+    } else {
+    /* CASE 3: The number of elements to return is very big, approaching
+     * the size of the set itself. After some time extracting random elements
+     * from such a set becomes computationally expensive, so we use
+     * a different strategy, we extract random elements that we don't
+     * want to return (the elements that will remain part of the set),
+     * creating a new set as we do this (that will be stored as the original
+     * set). Then we return the elements left in the original set and
+     * release it. */
+        robj *newset = NULL;
+
+        /* Create a new set with just the remaining elements. */
+        while(remaining--) {
+            encoding = setTypeRandomElement(set,&objele,&llele);
+            if (encoding == REDIS_ENCODING_INTSET) {
+                objele = createStringObjectFromLongLong(llele);
+            } else {
+                incrRefCount(objele);
+            }
+            if (!newset) newset = setTypeCreate(objele);
+            setTypeAdd(newset,objele);
+            setTypeRemove(set,objele);
+            decrRefCount(objele);
+        }
+
+        /* Assign the new set as the key value. */
+        incrRefCount(set); /* Protect the old set value. */
+        dbOverwrite(c->db,c->argv[1],newset);
+
+        /* Tranfer the old set to the client and release it. */
+        setTypeIterator *si;
+        si = setTypeInitIterator(set);
+        while((encoding = setTypeNext(si,&objele,&llele)) != -1) {
+            if (encoding == REDIS_ENCODING_INTSET) {
+                objele = createStringObjectFromLongLong(llele);
+            } else {
+                incrRefCount(objele);
+            }
+            addReplyBulk(c,objele);
+
+            /* Replicate/AOF this command as an SREM operation */
+            propargv[2] = objele;
+            alsoPropagate(server.sremCommand,c->db->id,propargv,3,
+                REDIS_PROPAGATE_AOF|REDIS_PROPAGATE_REPL);
+
+            decrRefCount(objele);
+        }
+        setTypeReleaseIterator(si);
+        decrRefCount(set);
+    }
+
+    /* Don't propagate the command itself even if we incremented the
+     * dirty counter. We don't want to propagate an SPOP command since
+     * we propagated the command as a set of SREMs operations using
+     * the alsoPropagate() API. */
+    decrRefCount(propargv[0]);
+    preventCommandPropagation(c);
+}
+
 void spopCommand(redisClient *c) {
     robj *set, *ele, *aux;
     int64_t llele;
     int encoding;
 
+    if (c->argc == 3) {
+        spopWithCountCommand(c);
+        return;
+    } else if (c->argc > 3) {
+        addReply(c,shared.syntaxerr);
+        return;
+    }
+
+    /* Make sure a key with the name inputted exists, and that it's type is
+     * indeed a set */
     if ((set = lookupKeyWriteOrReply(c,c->argv[1],shared.nullbulk)) == NULL ||
         checkType(c,set,REDIS_SET)) return;
 
+    /* Get a random element from the set */
     encoding = setTypeRandomElement(set,&ele,&llele);
+
+    /* Remove the element from the set */
     if (encoding == REDIS_ENCODING_INTSET) {
         ele = createStringObjectFromLongLong(llele);
         set->ptr = intsetRemove(set->ptr,llele,NULL);
@@ -393,6 +571,7 @@ void spopCommand(redisClient *c) {
         incrRefCount(ele);
         setTypeRemove(set,ele);
     }
+
     notifyKeyspaceEvent(REDIS_NOTIFY_SET,"spop",c->argv[1],c->db->id);
 
     /* Replicate/AOF this command as an SREM operation */
@@ -401,11 +580,16 @@ void spopCommand(redisClient *c) {
     decrRefCount(ele);
     decrRefCount(aux);
 
+    /* Add the element to the reply */
     addReplyBulk(c,ele);
+
+    /* Delete the set if it's empty */
     if (setTypeSize(set) == 0) {
         dbDelete(c->db,c->argv[1]);
         notifyKeyspaceEvent(REDIS_NOTIFY_GENERIC,"del",c->argv[1],c->db->id);
     }
+
+    /* Set has been modified */
     signalModifiedKey(c->db,c->argv[1]);
     server.dirty++;
 }
@@ -587,7 +771,8 @@ int qsortCompareSetsByRevCardinality(const void *s1, const void *s2) {
     return  (o2 ? setTypeSize(o2) : 0) - (o1 ? setTypeSize(o1) : 0);
 }
 
-void sinterGenericCommand(redisClient *c, robj **setkeys, unsigned long setnum, robj *dstkey) {
+void sinterGenericCommand(redisClient *c, robj **setkeys,
+                          unsigned long setnum, robj *dstkey) {
     robj **sets = zmalloc(sizeof(robj*)*setnum);
     setTypeIterator *si;
     robj *eleobj, *dstset = NULL;
@@ -734,7 +919,8 @@ void sinterstoreCommand(redisClient *c) {
 #define REDIS_OP_DIFF 1
 #define REDIS_OP_INTER 2
 
-void sunionDiffGenericCommand(redisClient *c, robj **setkeys, int setnum, robj *dstkey, int op) {
+void sunionDiffGenericCommand(redisClient *c, robj **setkeys, int setnum,
+                              robj *dstkey, int op) {
     robj **sets = zmalloc(sizeof(robj*)*setnum);
     setTypeIterator *si;
     robj *ele, *dstset = NULL;

src/t_string.c

@@ -61,6 +61,8 @@ static int checkStringLength(redisClient *c, long long size) {
 #define REDIS_SET_NO_FLAGS 0
 #define REDIS_SET_NX (1<<0)     /* Set if key not exists. */
 #define REDIS_SET_XX (1<<1)     /* Set if key exists. */
+#define REDIS_SET_EX (1<<2)     /* Set if time in seconds is given */
+#define REDIS_SET_PX (1<<3)     /* Set if time in ms in given */
 
 void setGenericCommand(redisClient *c, int flags, robj *key, robj *val, robj *expire, int unit, robj *ok_reply, robj *abort_reply) {
     long long milliseconds = 0; /* initialized to avoid any harmness warning */
@@ -102,18 +104,28 @@ void setCommand(redisClient *c) {
         robj *next = (j == c->argc-1) ? NULL : c->argv[j+1];
 
         if ((a[0] == 'n' || a[0] == 'N') &&
-            (a[1] == 'x' || a[1] == 'X') && a[2] == '\0') {
+            (a[1] == 'x' || a[1] == 'X') && a[2] == '\0' &&
+            !(flags & REDIS_SET_XX))
+        {
             flags |= REDIS_SET_NX;
         } else if ((a[0] == 'x' || a[0] == 'X') &&
-                   (a[1] == 'x' || a[1] == 'X') && a[2] == '\0') {
+                   (a[1] == 'x' || a[1] == 'X') && a[2] == '\0' &&
+                   !(flags & REDIS_SET_NX))
+        {
             flags |= REDIS_SET_XX;
         } else if ((a[0] == 'e' || a[0] == 'E') &&
-                   (a[1] == 'x' || a[1] == 'X') && a[2] == '\0' && next) {
+                   (a[1] == 'x' || a[1] == 'X') && a[2] == '\0' &&
+                   !(flags & REDIS_SET_PX) && next)
+        {
+            flags |= REDIS_SET_EX;
             unit = UNIT_SECONDS;
             expire = next;
             j++;
         } else if ((a[0] == 'p' || a[0] == 'P') &&
-                   (a[1] == 'x' || a[1] == 'X') && a[2] == '\0' && next) {
+                   (a[1] == 'x' || a[1] == 'X') && a[2] == '\0' &&
+                   !(flags & REDIS_SET_EX) && next)
+        {
+            flags |= REDIS_SET_PX;
             unit = UNIT_MILLISECONDS;
             expire = next;
             j++;
@@ -397,7 +409,7 @@ void incrbyfloatCommand(redisClient *c) {
         addReplyError(c,"increment would produce NaN or Infinity");
         return;
     }
-    new = createStringObjectFromLongDouble(value);
+    new = createStringObjectFromLongDouble(value,1);
     if (o)
         dbOverwrite(c->db,c->argv[1],new);
     else

src/t_zset.c

@@ -1382,7 +1382,7 @@ void zremrangeGenericCommand(redisClient *c, int rangetype) {
     robj *key = c->argv[1];
     robj *zobj;
     int keyremoved = 0;
-    unsigned long deleted;
+    unsigned long deleted = 0;
     zrangespec range;
     zlexrangespec lexrange;
     long start, end, llen;

src/util.c

@@ -40,6 +40,7 @@
 #include <stdint.h>
 
 #include "util.h"
+#include "sha1.h"
 
 /* Glob-style pattern matching. */
 int stringmatchlen(const char *pattern, int patternLen,
@@ -428,11 +429,44 @@ int d2string(char *buf, size_t len, double value) {
  * having run_id == A, and you reconnect and it has run_id == B, you can be
  * sure that it is either a different instance or it was restarted. */
 void getRandomHexChars(char *p, unsigned int len) {
-    FILE *fp = fopen("/dev/urandom","r");
     char *charset = "0123456789abcdef";
     unsigned int j;
 
-    if (fp == NULL || fread(p,len,1,fp) == 0) {
+    /* Global state. */
+    static int seed_initialized = 0;
+    static unsigned char seed[20]; /* The SHA1 seed, from /dev/urandom. */
+    static uint64_t counter = 0; /* The counter we hash with the seed. */
+
+    if (!seed_initialized) {
+        /* Initialize a seed and use SHA1 in counter mode, where we hash
+         * the same seed with a progressive counter. For the goals of this
+         * function we just need non-colliding strings, there are no
+         * cryptographic security needs. */
+        FILE *fp = fopen("/dev/urandom","r");
+        if (fp && fread(seed,sizeof(seed),1,fp) == 1)
+            seed_initialized = 1;
+        if (fp) fclose(fp);
+    }
+
+    if (seed_initialized) {
+        while(len) {
+            unsigned char digest[20];
+            SHA1_CTX ctx;
+            unsigned int copylen = len > 20 ? 20 : len;
+
+            SHA1Init(&ctx);
+            SHA1Update(&ctx, seed, sizeof(seed));
+            SHA1Update(&ctx, (unsigned char*)&counter,sizeof(counter));
+            SHA1Final(digest, &ctx);
+            counter++;
+
+            memcpy(p,digest,copylen);
+            /* Convert to hex digits. */
+            for (j = 0; j < copylen; j++) p[j] = charset[p[j] & 0x0F];
+            len -= copylen;
+            p += copylen;
+        }
+    } else {
         /* If we can't read from /dev/urandom, do some reasonable effort
          * in order to create some entropy, since this function is used to
          * generate run_id and cluster instance IDs */
@@ -459,14 +493,12 @@ void getRandomHexChars(char *p, unsigned int len) {
             x += sizeof(pid);
         }
         /* Finally xor it with rand() output, that was already seeded with
-         * time() at startup. */
-        for (j = 0; j < len; j++)
+         * time() at startup, and convert to hex digits. */
+        for (j = 0; j < len; j++) {
             p[j] ^= rand();
+            p[j] = charset[p[j] & 0x0F];
+        }
     }
-    /* Turn it into hex digits taking just 4 bits out of 8 for every byte. */
-    for (j = 0; j < len; j++)
-        p[j] = charset[p[j] & 0x0F];
-    if (fp) fclose(fp);
 }
 
 /* Given the filename, return the absolute path as an SDS string, or NULL
@@ -529,10 +561,10 @@ int pathIsBaseName(char *path) {
     return strchr(path,'/') == NULL && strchr(path,'\\') == NULL;
 }
 
-#ifdef UTIL_TEST_MAIN
+#ifdef REDIS_TEST
 #include <assert.h>
 
-void test_string2ll(void) {
+static void test_string2ll(void) {
     char buf[32];
     long long v;
 
@@ -587,7 +619,7 @@ void test_string2ll(void) {
     assert(string2ll(buf,strlen(buf),&v) == 0);
 }
 
-void test_string2l(void) {
+static void test_string2l(void) {
     char buf[32];
     long v;
 
@@ -636,9 +668,55 @@ void test_string2l(void) {
 #endif
 }
 
-int main(int argc, char **argv) {
+static void test_ll2string(void) {
+    char buf[32];
+    long long v;
+    int sz;
+
+    v = 0;
+    sz = ll2string(buf, sizeof buf, v);
+    assert(sz == 1);
+    assert(!strcmp(buf, "0"));
+
+    v = -1;
+    sz = ll2string(buf, sizeof buf, v);
+    assert(sz == 2);
+    assert(!strcmp(buf, "-1"));
+
+    v = 99;
+    sz = ll2string(buf, sizeof buf, v);
+    assert(sz == 2);
+    assert(!strcmp(buf, "99"));
+
+    v = -99;
+    sz = ll2string(buf, sizeof buf, v);
+    assert(sz == 3);
+    assert(!strcmp(buf, "-99"));
+
+    v = -2147483648;
+    sz = ll2string(buf, sizeof buf, v);
+    assert(sz == 11);
+    assert(!strcmp(buf, "-2147483648"));
+
+    v = LLONG_MIN;
+    sz = ll2string(buf, sizeof buf, v);
+    assert(sz == 20);
+    assert(!strcmp(buf, "-9223372036854775808"));
+
+    v = LLONG_MAX;
+    sz = ll2string(buf, sizeof buf, v);
+    assert(sz == 19);
+    assert(!strcmp(buf, "9223372036854775807"));
+}
+
+#define UNUSED(x) (void)(x)
+int utilTest(int argc, char **argv) {
+    UNUSED(argc);
+    UNUSED(argv);
+
     test_string2ll();
     test_string2l();
+    test_ll2string();
     return 0;
 }
 #endif

src/util.h

@@ -42,4 +42,8 @@ int d2string(char *buf, size_t len, double value);
 sds getAbsolutePath(char *filename);
 int pathIsBaseName(char *path);
 
+#ifdef REDIS_TEST
+int utilTest(int argc, char **argv);
+#endif
+
 #endif

src/ziplist.c

@@ -143,6 +143,7 @@
 #define ZIPLIST_TAIL_OFFSET(zl) (*((uint32_t*)((zl)+sizeof(uint32_t))))
 #define ZIPLIST_LENGTH(zl)      (*((uint16_t*)((zl)+sizeof(uint32_t)*2)))
 #define ZIPLIST_HEADER_SIZE     (sizeof(uint32_t)*2+sizeof(uint16_t))
+#define ZIPLIST_END_SIZE        (sizeof(uint8_t))
 #define ZIPLIST_ENTRY_HEAD(zl)  ((zl)+ZIPLIST_HEADER_SIZE)
 #define ZIPLIST_ENTRY_TAIL(zl)  ((zl)+intrev32ifbe(ZIPLIST_TAIL_OFFSET(zl)))
 #define ZIPLIST_ENTRY_END(zl)   ((zl)+intrev32ifbe(ZIPLIST_BYTES(zl))-1)
@@ -162,6 +163,13 @@ typedef struct zlentry {
     unsigned char *p;
 } zlentry;
 
+#define ZIPLIST_ENTRY_ZERO(zle) { \
+    (zle)->prevrawlensize = (zle)->prevrawlen = 0; \
+    (zle)->lensize = (zle)->len = (zle)->headersize = 0; \
+    (zle)->encoding = 0; \
+    (zle)->p = NULL; \
+}
+
 /* Extract the encoding from the byte pointed by 'ptr' and set it into
  * 'encoding'. */
 #define ZIP_ENTRY_ENCODING(ptr, encoding) do {  \
@@ -169,6 +177,8 @@ typedef struct zlentry {
     if ((encoding) < ZIP_STR_MASK) (encoding) &= ZIP_STR_MASK; \
 } while(0)
 
+void ziplistRepr(unsigned char *zl);
+
 /* Return bytes needed to store integer encoded by 'encoding' */
 static unsigned int zipIntSize(unsigned char encoding) {
     switch(encoding) {
@@ -404,14 +414,12 @@ static int64_t zipLoadInteger(unsigned char *p, unsigned char encoding) {
 }
 
 /* Return a struct with all information about an entry. */
-static zlentry zipEntry(unsigned char *p) {
-    zlentry e;
+static void zipEntry(unsigned char *p, zlentry *e) {
 
-    ZIP_DECODE_PREVLEN(p, e.prevrawlensize, e.prevrawlen);
-    ZIP_DECODE_LENGTH(p + e.prevrawlensize, e.encoding, e.lensize, e.len);
-    e.headersize = e.prevrawlensize + e.lensize;
-    e.p = p;
-    return e;
+    ZIP_DECODE_PREVLEN(p, e->prevrawlensize, e->prevrawlen);
+    ZIP_DECODE_LENGTH(p + e->prevrawlensize, e->encoding, e->lensize, e->len);
+    e->headersize = e->prevrawlensize + e->lensize;
+    e->p = p;
 }
 
 /* Create a new empty ziplist. */
@@ -460,13 +468,13 @@ static unsigned char *__ziplistCascadeUpdate(unsigned char *zl, unsigned char *p
     zlentry cur, next;
 
     while (p[0] != ZIP_END) {
-        cur = zipEntry(p);
+        zipEntry(p, &cur);
         rawlen = cur.headersize + cur.len;
         rawlensize = zipPrevEncodeLength(NULL,rawlen);
 
         /* Abort if there is no next entry. */
         if (p[rawlen] == ZIP_END) break;
-        next = zipEntry(p+rawlen);
+        zipEntry(p+rawlen, &next);
 
         /* Abort when "prevlen" has not changed. */
         if (next.prevrawlen == rawlen) break;
@@ -521,7 +529,7 @@ static unsigned char *__ziplistDelete(unsigned char *zl, unsigned char *p, unsig
     int nextdiff = 0;
     zlentry first, tail;
 
-    first = zipEntry(p);
+    zipEntry(p, &first);
     for (i = 0; p[0] != ZIP_END && i < num; i++) {
         p += zipRawEntryLength(p);
         deleted++;
@@ -545,7 +553,7 @@ static unsigned char *__ziplistDelete(unsigned char *zl, unsigned char *p, unsig
             /* When the tail contains more than one entry, we need to take
              * "nextdiff" in account as well. Otherwise, a change in the
              * size of prevlen doesn't have an effect on the *tail* offset. */
-            tail = zipEntry(p);
+            zipEntry(p, &tail);
             if (p[tail.headersize+tail.len] != ZIP_END) {
                 ZIPLIST_TAIL_OFFSET(zl) =
                    intrev32ifbe(intrev32ifbe(ZIPLIST_TAIL_OFFSET(zl))+nextdiff);
@@ -635,7 +643,7 @@ static unsigned char *__ziplistInsert(unsigned char *zl, unsigned char *p, unsig
         /* When the tail contains more than one entry, we need to take
          * "nextdiff" in account as well. Otherwise, a change in the
          * size of prevlen doesn't have an effect on the *tail* offset. */
-        tail = zipEntry(p+reqlen);
+        zipEntry(p+reqlen, &tail);
         if (p[reqlen+tail.headersize+tail.len] != ZIP_END) {
             ZIPLIST_TAIL_OFFSET(zl) =
                 intrev32ifbe(intrev32ifbe(ZIPLIST_TAIL_OFFSET(zl))+nextdiff);
@@ -665,6 +673,121 @@ static unsigned char *__ziplistInsert(unsigned char *zl, unsigned char *p, unsig
     return zl;
 }
 
+/* Merge ziplists 'first' and 'second' by appending 'second' to 'first'.
+ *
+ * NOTE: The larger ziplist is reallocated to contain the new merged ziplist.
+ * Either 'first' or 'second' can be used for the result.  The parameter not
+ * used will be free'd and set to NULL.
+ *
+ * After calling this function, the input parameters are no longer valid since
+ * they are changed and free'd in-place.
+ *
+ * The result ziplist is the contents of 'first' followed by 'second'.
+ *
+ * On failure: returns NULL if the merge is impossible.
+ * On success: returns the merged ziplist (which is expanded version of either
+ * 'first' or 'second', also frees the other unused input ziplist, and sets the
+ * input ziplist argument equal to newly reallocated ziplist return value. */
+unsigned char *ziplistMerge(unsigned char **first, unsigned char **second) {
+    /* If any params are null, we can't merge, so NULL. */
+    if (first == NULL || *first == NULL || second == NULL || *second == NULL)
+        return NULL;
+
+    /* Can't merge same list into itself. */
+    if (*first == *second)
+        return NULL;
+
+    size_t first_bytes = intrev32ifbe(ZIPLIST_BYTES(*first));
+    size_t first_len = intrev16ifbe(ZIPLIST_LENGTH(*first));
+
+    size_t second_bytes = intrev32ifbe(ZIPLIST_BYTES(*second));
+    size_t second_len = intrev16ifbe(ZIPLIST_LENGTH(*second));
+
+    int append;
+    unsigned char *source, *target;
+    size_t target_bytes, source_bytes;
+    /* Pick the largest ziplist so we can resize easily in-place.
+     * We must also track if we are now appending or prepending to
+     * the target ziplist. */
+    if (first_len >= second_len) {
+        /* retain first, append second to first. */
+        target = *first;
+        target_bytes = first_bytes;
+        source = *second;
+        source_bytes = second_bytes;
+        append = 1;
+    } else {
+        /* else, retain second, prepend first to second. */
+        target = *second;
+        target_bytes = second_bytes;
+        source = *first;
+        source_bytes = first_bytes;
+        append = 0;
+    }
+
+    /* Calculate final bytes (subtract one pair of metadata) */
+    size_t zlbytes = first_bytes + second_bytes -
+                     ZIPLIST_HEADER_SIZE - ZIPLIST_END_SIZE;
+    size_t zllength = first_len + second_len;
+
+    /* Combined zl length should be limited within UINT16_MAX */
+    zllength = zllength < UINT16_MAX ? zllength : UINT16_MAX;
+
+    /* Save offset positions before we start ripping memory apart. */
+    size_t first_offset = intrev32ifbe(ZIPLIST_TAIL_OFFSET(*first));
+    size_t second_offset = intrev32ifbe(ZIPLIST_TAIL_OFFSET(*second));
+
+    /* Extend target to new zlbytes then append or prepend source. */
+    target = zrealloc(target, zlbytes);
+    if (append) {
+        /* append == appending to target */
+        /* Copy source after target (copying over original [END]):
+         *   [TARGET - END, SOURCE - HEADER] */
+        memcpy(target + target_bytes - ZIPLIST_END_SIZE,
+               source + ZIPLIST_HEADER_SIZE,
+               source_bytes - ZIPLIST_HEADER_SIZE);
+    } else {
+        /* !append == prepending to target */
+        /* Move target *contents* exactly size of (source - [END]),
+         * then copy source into vacataed space (source - [END]):
+         *   [SOURCE - END, TARGET - HEADER] */
+        memmove(target + source_bytes - ZIPLIST_END_SIZE,
+                target + ZIPLIST_HEADER_SIZE,
+                target_bytes - ZIPLIST_HEADER_SIZE);
+        memcpy(target, source, source_bytes - ZIPLIST_END_SIZE);
+    }
+
+    /* Update header metadata. */
+    ZIPLIST_BYTES(target) = intrev32ifbe(zlbytes);
+    ZIPLIST_LENGTH(target) = intrev16ifbe(zllength);
+    /* New tail offset is:
+     *   + N bytes of first ziplist
+     *   - 1 byte for [END] of first ziplist
+     *   + M bytes for the offset of the original tail of the second ziplist
+     *   - J bytes for HEADER because second_offset keeps no header. */
+    ZIPLIST_TAIL_OFFSET(target) = intrev32ifbe(
+                                   (first_bytes - ZIPLIST_END_SIZE) +
+                                   (second_offset - ZIPLIST_HEADER_SIZE));
+
+    /* __ziplistCascadeUpdate just fixes the prev length values until it finds a
+     * correct prev length value (then it assumes the rest of the list is okay).
+     * We tell CascadeUpdate to start at the first ziplist's tail element to fix
+     * the merge seam. */
+    target = __ziplistCascadeUpdate(target, target+first_offset);
+
+    /* Now free and NULL out what we didn't realloc */
+    if (append) {
+        zfree(*second);
+        *second = NULL;
+        *first = target;
+    } else {
+        zfree(*first);
+        *first = NULL;
+        *second = target;
+    }
+    return target;
+}
+
 unsigned char *ziplistPush(unsigned char *zl, unsigned char *s, unsigned int slen, int where) {
     unsigned char *p;
     p = (where == ZIPLIST_HEAD) ? ZIPLIST_ENTRY_HEAD(zl) : ZIPLIST_ENTRY_END(zl);
@@ -748,7 +871,7 @@ unsigned int ziplistGet(unsigned char *p, unsigned char **sstr, unsigned int *sl
     if (p == NULL || p[0] == ZIP_END) return 0;
     if (sstr) *sstr = NULL;
 
-    entry = zipEntry(p);
+    zipEntry(p, &entry);
     if (ZIP_IS_STR(entry.encoding)) {
         if (sstr) {
             *slen = entry.len;
@@ -783,7 +906,7 @@ unsigned char *ziplistDelete(unsigned char *zl, unsigned char **p) {
 }
 
 /* Delete a range of entries from the ziplist. */
-unsigned char *ziplistDeleteRange(unsigned char *zl, unsigned int index, unsigned int num) {
+unsigned char *ziplistDeleteRange(unsigned char *zl, int index, unsigned int num) {
     unsigned char *p = ziplistIndex(zl,index);
     return (p == NULL) ? zl : __ziplistDelete(zl,p,num);
 }
@@ -796,7 +919,7 @@ unsigned int ziplistCompare(unsigned char *p, unsigned char *sstr, unsigned int
     long long zval, sval;
     if (p[0] == ZIP_END) return 0;
 
-    entry = zipEntry(p);
+    zipEntry(p, &entry);
     if (ZIP_IS_STR(entry.encoding)) {
         /* Raw compare */
         if (entry.len == slen) {
@@ -913,7 +1036,7 @@ void ziplistRepr(unsigned char *zl) {
         intrev32ifbe(ZIPLIST_TAIL_OFFSET(zl)));
     p = ZIPLIST_ENTRY_HEAD(zl);
     while(*p != ZIP_END) {
-        entry = zipEntry(p);
+        zipEntry(p, &entry);
         printf(
             "{"
                 "addr 0x%08lx, "
@@ -952,14 +1075,14 @@ void ziplistRepr(unsigned char *zl) {
     printf("{end}\n\n");
 }
 
-#ifdef ZIPLIST_TEST_MAIN
+#ifdef REDIS_TEST
 #include <sys/time.h>
 #include "adlist.h"
 #include "sds.h"
 
 #define debug(f, ...) { if (DEBUG) printf(f, __VA_ARGS__); }
 
-unsigned char *createList() {
+static unsigned char *createList() {
     unsigned char *zl = ziplistNew();
     zl = ziplistPush(zl, (unsigned char*)"foo", 3, ZIPLIST_TAIL);
     zl = ziplistPush(zl, (unsigned char*)"quux", 4, ZIPLIST_TAIL);
@@ -968,7 +1091,7 @@ unsigned char *createList() {
     return zl;
 }
 
-unsigned char *createIntList() {
+static unsigned char *createIntList() {
     unsigned char *zl = ziplistNew();
     char buf[32];
 
@@ -987,13 +1110,13 @@ unsigned char *createIntList() {
     return zl;
 }
 
-long long usec(void) {
+static long long usec(void) {
     struct timeval tv;
     gettimeofday(&tv,NULL);
     return (((long long)tv.tv_sec)*1000000)+tv.tv_usec;
 }
 
-void stress(int pos, int num, int maxsize, int dnum) {
+static void stress(int pos, int num, int maxsize, int dnum) {
     int i,j,k;
     unsigned char *zl;
     char posstr[2][5] = { "HEAD", "TAIL" };
@@ -1016,7 +1139,7 @@ void stress(int pos, int num, int maxsize, int dnum) {
     }
 }
 
-void pop(unsigned char *zl, int where) {
+static unsigned char *pop(unsigned char *zl, int where) {
     unsigned char *p, *vstr;
     unsigned int vlen;
     long long vlong;
@@ -1028,20 +1151,22 @@ void pop(unsigned char *zl, int where) {
         else
             printf("Pop tail: ");
 
-        if (vstr)
+        if (vstr) {
             if (vlen && fwrite(vstr,vlen,1,stdout) == 0) perror("fwrite");
-        else
+        }
+        else {
             printf("%lld", vlong);
+        }
 
         printf("\n");
-        ziplistDeleteRange(zl,-1,1);
+        return ziplistDelete(zl,&p);
     } else {
         printf("ERROR: Could not pop\n");
         exit(1);
     }
 }
 
-int randstring(char *target, unsigned int min, unsigned int max) {
+static int randstring(char *target, unsigned int min, unsigned int max) {
     int p = 0;
     int len = min+rand()%(max-min+1);
     int minval, maxval;
@@ -1067,23 +1192,24 @@ int randstring(char *target, unsigned int min, unsigned int max) {
     return len;
 }
 
-void verify(unsigned char *zl, zlentry *e) {
-    int i;
+static void verify(unsigned char *zl, zlentry *e) {
     int len = ziplistLen(zl);
     zlentry _e;
 
-    for (i = 0; i < len; i++) {
+    ZIPLIST_ENTRY_ZERO(&_e);
+
+    for (int i = 0; i < len; i++) {
         memset(&e[i], 0, sizeof(zlentry));
-        e[i] = zipEntry(ziplistIndex(zl, i));
+        zipEntry(ziplistIndex(zl, i), &e[i]);
 
         memset(&_e, 0, sizeof(zlentry));
-        _e = zipEntry(ziplistIndex(zl, -len+i));
+        zipEntry(ziplistIndex(zl, -len+i), &_e);
 
         assert(memcmp(&e[i], &_e, sizeof(zlentry)) == 0);
     }
 }
 
-int main(int argc, char **argv) {
+int ziplistTest(int argc, char **argv) {
     unsigned char *zl, *p;
     unsigned char *entry;
     unsigned int elen;
@@ -1096,21 +1222,25 @@ int main(int argc, char **argv) {
     zl = createIntList();
     ziplistRepr(zl);
 
+    zfree(zl);
+
     zl = createList();
     ziplistRepr(zl);
 
-    pop(zl,ZIPLIST_TAIL);
+    zl = pop(zl,ZIPLIST_TAIL);
     ziplistRepr(zl);
 
-    pop(zl,ZIPLIST_HEAD);
+    zl = pop(zl,ZIPLIST_HEAD);
     ziplistRepr(zl);
 
-    pop(zl,ZIPLIST_TAIL);
+    zl = pop(zl,ZIPLIST_TAIL);
     ziplistRepr(zl);
 
-    pop(zl,ZIPLIST_TAIL);
+    zl = pop(zl,ZIPLIST_TAIL);
     ziplistRepr(zl);
 
+    zfree(zl);
+
     printf("Get element at index 3:\n");
     {
         zl = createList();
@@ -1126,6 +1256,7 @@ int main(int argc, char **argv) {
             printf("%lld\n", value);
         }
         printf("\n");
+        zfree(zl);
     }
 
     printf("Get element at index 4 (out of range):\n");
@@ -1139,6 +1270,7 @@ int main(int argc, char **argv) {
             return 1;
         }
         printf("\n");
+        zfree(zl);
     }
 
     printf("Get element at index -1 (last element):\n");
@@ -1156,6 +1288,7 @@ int main(int argc, char **argv) {
             printf("%lld\n", value);
         }
         printf("\n");
+        zfree(zl);
     }
 
     printf("Get element at index -4 (first element):\n");
@@ -1173,6 +1306,7 @@ int main(int argc, char **argv) {
             printf("%lld\n", value);
         }
         printf("\n");
+        zfree(zl);
     }
 
     printf("Get element at index -5 (reverse out of range):\n");
@@ -1186,6 +1320,7 @@ int main(int argc, char **argv) {
             return 1;
         }
         printf("\n");
+        zfree(zl);
     }
 
     printf("Iterate list from 0 to end:\n");
@@ -1203,6 +1338,7 @@ int main(int argc, char **argv) {
             printf("\n");
         }
         printf("\n");
+        zfree(zl);
     }
 
     printf("Iterate list from 1 to end:\n");
@@ -1220,6 +1356,7 @@ int main(int argc, char **argv) {
             printf("\n");
         }
         printf("\n");
+        zfree(zl);
     }
 
     printf("Iterate list from 2 to end:\n");
@@ -1237,6 +1374,7 @@ int main(int argc, char **argv) {
             printf("\n");
         }
         printf("\n");
+        zfree(zl);
     }
 
     printf("Iterate starting out of range:\n");
@@ -1249,6 +1387,7 @@ int main(int argc, char **argv) {
             printf("ERROR\n");
         }
         printf("\n");
+        zfree(zl);
     }
 
     printf("Iterate from back to front:\n");
@@ -1266,6 +1405,7 @@ int main(int argc, char **argv) {
             printf("\n");
         }
         printf("\n");
+        zfree(zl);
     }
 
     printf("Iterate from back to front, deleting all items:\n");
@@ -1284,6 +1424,7 @@ int main(int argc, char **argv) {
             printf("\n");
         }
         printf("\n");
+        zfree(zl);
     }
 
     printf("Delete inclusive range 0,0:\n");
@@ -1291,6 +1432,7 @@ int main(int argc, char **argv) {
         zl = createList();
         zl = ziplistDeleteRange(zl, 0, 1);
         ziplistRepr(zl);
+        zfree(zl);
     }
 
     printf("Delete inclusive range 0,1:\n");
@@ -1298,6 +1440,7 @@ int main(int argc, char **argv) {
         zl = createList();
         zl = ziplistDeleteRange(zl, 0, 2);
         ziplistRepr(zl);
+        zfree(zl);
     }
 
     printf("Delete inclusive range 1,2:\n");
@@ -1305,6 +1448,7 @@ int main(int argc, char **argv) {
         zl = createList();
         zl = ziplistDeleteRange(zl, 1, 2);
         ziplistRepr(zl);
+        zfree(zl);
     }
 
     printf("Delete with start index out of range:\n");
@@ -1312,6 +1456,7 @@ int main(int argc, char **argv) {
         zl = createList();
         zl = ziplistDeleteRange(zl, 5, 1);
         ziplistRepr(zl);
+        zfree(zl);
     }
 
     printf("Delete with num overflow:\n");
@@ -1319,6 +1464,7 @@ int main(int argc, char **argv) {
         zl = createList();
         zl = ziplistDeleteRange(zl, 1, 5);
         ziplistRepr(zl);
+        zfree(zl);
     }
 
     printf("Delete foo while iterating:\n");
@@ -1343,11 +1489,12 @@ int main(int argc, char **argv) {
         }
         printf("\n");
         ziplistRepr(zl);
+        zfree(zl);
     }
 
     printf("Regression test for >255 byte strings:\n");
     {
-        char v1[257],v2[257];
+        char v1[257] = {0}, v2[257] = {0};
         memset(v1,'x',256);
         memset(v2,'y',256);
         zl = ziplistNew();
@@ -1362,13 +1509,15 @@ int main(int argc, char **argv) {
         assert(ziplistGet(p,&entry,&elen,&value));
         assert(strncmp(v2,(char*)entry,elen) == 0);
         printf("SUCCESS\n\n");
+        zfree(zl);
     }
 
     printf("Regression test deleting next to last entries:\n");
     {
-        char v[3][257];
-        zlentry e[3];
-        int i;
+        char v[3][257] = {{0}};
+        zlentry e[3] = {{.prevrawlensize = 0, .prevrawlen = 0, .lensize = 0,
+                         .len = 0, .headersize = 0, .encoding = 0, .p = NULL}};
+        size_t i;
 
         for (i = 0; i < (sizeof(v)/sizeof(v[0])); i++) {
             memset(v[i], 'a' + i, sizeof(v[0]));
@@ -1399,6 +1548,7 @@ int main(int argc, char **argv) {
         assert(e[1].prevrawlensize == 5);
 
         printf("SUCCESS\n\n");
+        zfree(zl);
     }
 
     printf("Create long list and check indices:\n");
@@ -1420,6 +1570,7 @@ int main(int argc, char **argv) {
             assert(999-i == value);
         }
         printf("SUCCESS\n\n");
+        zfree(zl);
     }
 
     printf("Compare strings with ziplist entries:\n");
@@ -1445,6 +1596,82 @@ int main(int argc, char **argv) {
             return 1;
         }
         printf("SUCCESS\n\n");
+        zfree(zl);
+    }
+
+    printf("Merge test:\n");
+    {
+        /* create list gives us: [hello, foo, quux, 1024] */
+        zl = createList();
+        unsigned char *zl2 = createList();
+
+        unsigned char *zl3 = ziplistNew();
+        unsigned char *zl4 = ziplistNew();
+
+        if (ziplistMerge(&zl4, &zl4)) {
+            printf("ERROR: Allowed merging of one ziplist into itself.\n");
+            return 1;
+        }
+
+        /* Merge two empty ziplists, get empty result back. */
+        zl4 = ziplistMerge(&zl3, &zl4);
+        ziplistRepr(zl4);
+        if (ziplistLen(zl4)) {
+            printf("ERROR: Merging two empty ziplists created entries.\n");
+            return 1;
+        }
+        zfree(zl4);
+
+        zl2 = ziplistMerge(&zl, &zl2);
+        /* merge gives us: [hello, foo, quux, 1024, hello, foo, quux, 1024] */
+        ziplistRepr(zl2);
+
+        if (ziplistLen(zl2) != 8) {
+            printf("ERROR: Merged length not 8, but: %u\n", ziplistLen(zl2));
+            return 1;
+        }
+
+        p = ziplistIndex(zl2,0);
+        if (!ziplistCompare(p,(unsigned char*)"hello",5)) {
+            printf("ERROR: not \"hello\"\n");
+            return 1;
+        }
+        if (ziplistCompare(p,(unsigned char*)"hella",5)) {
+            printf("ERROR: \"hella\"\n");
+            return 1;
+        }
+
+        p = ziplistIndex(zl2,3);
+        if (!ziplistCompare(p,(unsigned char*)"1024",4)) {
+            printf("ERROR: not \"1024\"\n");
+            return 1;
+        }
+        if (ziplistCompare(p,(unsigned char*)"1025",4)) {
+            printf("ERROR: \"1025\"\n");
+            return 1;
+        }
+
+        p = ziplistIndex(zl2,4);
+        if (!ziplistCompare(p,(unsigned char*)"hello",5)) {
+            printf("ERROR: not \"hello\"\n");
+            return 1;
+        }
+        if (ziplistCompare(p,(unsigned char*)"hella",5)) {
+            printf("ERROR: \"hella\"\n");
+            return 1;
+        }
+
+        p = ziplistIndex(zl2,7);
+        if (!ziplistCompare(p,(unsigned char*)"1024",4)) {
+            printf("ERROR: not \"1024\"\n");
+            return 1;
+        }
+        if (ziplistCompare(p,(unsigned char*)"1025",4)) {
+            printf("ERROR: \"1025\"\n");
+            return 1;
+        }
+        printf("SUCCESS\n\n");
+        zfree(zl);
     }
 
     printf("Stress with random payloads of different encoding:\n");
@@ -1464,7 +1691,7 @@ int main(int argc, char **argv) {
         for (i = 0; i < 20000; i++) {
             zl = ziplistNew();
             ref = listCreate();
-            listSetFreeMethod(ref,sdsfree);
+            listSetFreeMethod(ref,(void (*)(void*))sdsfree);
             len = rand() % 256;
 
             /* Create lists */
@@ -1532,5 +1759,4 @@ int main(int argc, char **argv) {
 
     return 0;
 }
-
 #endif

src/ziplist.h

@@ -32,6 +32,7 @@
 #define ZIPLIST_TAIL 1
 
 unsigned char *ziplistNew(void);
+unsigned char *ziplistMerge(unsigned char **first, unsigned char **second);
 unsigned char *ziplistPush(unsigned char *zl, unsigned char *s, unsigned int slen, int where);
 unsigned char *ziplistIndex(unsigned char *zl, int index);
 unsigned char *ziplistNext(unsigned char *zl, unsigned char *p);
@@ -39,8 +40,12 @@ unsigned char *ziplistPrev(unsigned char *zl, unsigned char *p);
 unsigned int ziplistGet(unsigned char *p, unsigned char **sval, unsigned int *slen, long long *lval);
 unsigned char *ziplistInsert(unsigned char *zl, unsigned char *p, unsigned char *s, unsigned int slen);
 unsigned char *ziplistDelete(unsigned char *zl, unsigned char **p);
-unsigned char *ziplistDeleteRange(unsigned char *zl, unsigned int index, unsigned int num);
+unsigned char *ziplistDeleteRange(unsigned char *zl, int index, unsigned int num);
 unsigned int ziplistCompare(unsigned char *p, unsigned char *s, unsigned int slen);
 unsigned char *ziplistFind(unsigned char *p, unsigned char *vstr, unsigned int vlen, unsigned int skip);
 unsigned int ziplistLen(unsigned char *zl);
 size_t ziplistBlobLen(unsigned char *zl);
+
+#ifdef REDIS_TEST
+int ziplistTest(int argc, char *argv[]);
+#endif

src/zipmap.c

@@ -51,10 +51,9 @@
  * <len> is the length of the following string (key or value).
  * <len> lengths are encoded in a single value or in a 5 bytes value.
  * If the first byte value (as an unsigned 8 bit value) is between 0 and
- * 252, it's a single-byte length. If it is 253 then a four bytes unsigned
+ * 253, it's a single-byte length. If it is 254 then a four bytes unsigned
  * integer follows (in the host byte ordering). A value of 255 is used to
- * signal the end of the hash. The special value 254 is used to mark
- * empty space that can be used to add new key/value pairs.
+ * signal the end of the hash.
  *
  * <free> is the number of free unused bytes after the string, resulting
  * from modification of values associated to a key. For instance if "foo"
@@ -371,8 +370,8 @@ size_t zipmapBlobLen(unsigned char *zm) {
     return totlen;
 }
 
-#ifdef ZIPMAP_TEST_MAIN
-void zipmapRepr(unsigned char *p) {
+#ifdef REDIS_TEST
+static void zipmapRepr(unsigned char *p) {
     unsigned int l;
 
     printf("{status %u}",*p++);
@@ -405,9 +404,13 @@ void zipmapRepr(unsigned char *p) {
     printf("\n");
 }
 
-int main(void) {
+#define UNUSED(x) (void)(x)
+int zipmapTest(int argc, char *argv[]) {
     unsigned char *zm;
 
+    UNUSED(argc);
+    UNUSED(argv);
+
     zm = zipmapNew();
 
     zm = zipmapSet(zm,(unsigned char*) "name",4, (unsigned char*) "foo",3,NULL);

src/zipmap.h

@@ -46,4 +46,8 @@ unsigned int zipmapLen(unsigned char *zm);
 size_t zipmapBlobLen(unsigned char *zm);
 void zipmapRepr(unsigned char *p);
 
+#ifdef REDIS_TEST
+int zipmapTest(int argc, char *argv[]);
+#endif
+
 #endif

src/zmalloc.c

@@ -328,27 +328,96 @@ float zmalloc_get_fragmentation_ratio(size_t rss) {
     return (float)rss/zmalloc_used_memory();
 }
 
+/* Get the sum of the specified field (converted form kb to bytes) in
+ * /proc/self/smaps. The field must be specified with trailing ":" as it
+ * apperas in the smaps output.
+ *
+ * Example: zmalloc_get_smap_bytes_by_field("Rss:");
+ */
 #if defined(HAVE_PROC_SMAPS)
-size_t zmalloc_get_private_dirty(void) {
+size_t zmalloc_get_smap_bytes_by_field(char *field) {
     char line[1024];
-    size_t pd = 0;
+    size_t bytes = 0;
     FILE *fp = fopen("/proc/self/smaps","r");
+    int flen = strlen(field);
 
     if (!fp) return 0;
     while(fgets(line,sizeof(line),fp) != NULL) {
-        if (strncmp(line,"Private_Dirty:",14) == 0) {
+        if (strncmp(line,field,flen) == 0) {
             char *p = strchr(line,'k');
             if (p) {
                 *p = '\0';
-                pd += strtol(line+14,NULL,10) * 1024;
+                bytes += strtol(line+flen,NULL,10) * 1024;
             }
         }
     }
     fclose(fp);
-    return pd;
+    return bytes;
 }
 #else
-size_t zmalloc_get_private_dirty(void) {
+size_t zmalloc_get_smap_bytes_by_field(char *field) {
+    ((void) field);
     return 0;
 }
 #endif
+
+size_t zmalloc_get_private_dirty(void) {
+    return zmalloc_get_smap_bytes_by_field("Private_Dirty:");
+}
+
+/* Returns the size of physical memory (RAM) in bytes.
+ * It looks ugly, but this is the cleanest way to achive cross platform results.
+ * Cleaned up from:
+ *
+ * http://nadeausoftware.com/articles/2012/09/c_c_tip_how_get_physical_memory_size_system
+ *
+ * Note that this function:
+ * 1) Was released under the following CC attribution license:
+ *    http://creativecommons.org/licenses/by/3.0/deed.en_US.
+ * 2) Was originally implemented by David Robert Nadeau.
+ * 3) Was modified for Redis by Matt Stancliff.
+ * 4) This note exists in order to comply with the original license.
+ */
+size_t zmalloc_get_memory_size(void) {
+#if defined(__unix__) || defined(__unix) || defined(unix) || \
+    (defined(__APPLE__) && defined(__MACH__))
+#if defined(CTL_HW) && (defined(HW_MEMSIZE) || defined(HW_PHYSMEM64))
+    int mib[2];
+    mib[0] = CTL_HW;
+#if defined(HW_MEMSIZE)
+    mib[1] = HW_MEMSIZE;            /* OSX. --------------------- */
+#elif defined(HW_PHYSMEM64)
+    mib[1] = HW_PHYSMEM64;          /* NetBSD, OpenBSD. --------- */
+#endif
+    int64_t size = 0;               /* 64-bit */
+    size_t len = sizeof(size);
+    if (sysctl( mib, 2, &size, &len, NULL, 0) == 0)
+        return (size_t)size;
+    return 0L;          /* Failed? */
+
+#elif defined(_SC_PHYS_PAGES) && defined(_SC_PAGESIZE)
+    /* FreeBSD, Linux, OpenBSD, and Solaris. -------------------- */
+    return (size_t)sysconf(_SC_PHYS_PAGES) * (size_t)sysconf(_SC_PAGESIZE);
+
+#elif defined(CTL_HW) && (defined(HW_PHYSMEM) || defined(HW_REALMEM))
+    /* DragonFly BSD, FreeBSD, NetBSD, OpenBSD, and OSX. -------- */
+    int mib[2];
+    mib[0] = CTL_HW;
+#if defined(HW_REALMEM)
+    mib[1] = HW_REALMEM;        /* FreeBSD. ----------------- */
+#elif defined(HW_PYSMEM)
+    mib[1] = HW_PHYSMEM;        /* Others. ------------------ */
+#endif
+    unsigned int size = 0;      /* 32-bit */
+    size_t len = sizeof(size);
+    if (sysctl(mib, 2, &size, &len, NULL, 0) == 0)
+        return (size_t)size;
+    return 0L;          /* Failed? */
+#endif /* sysctl and sysconf variants */
+
+#else
+    return 0L;          /* Unknown OS. */
+#endif
+}
+
+

src/zmalloc.h

@@ -76,6 +76,8 @@ void zmalloc_set_oom_handler(void (*oom_handler)(size_t));
 float zmalloc_get_fragmentation_ratio(size_t rss);
 size_t zmalloc_get_rss(void);
 size_t zmalloc_get_private_dirty(void);
+size_t zmalloc_get_smap_bytes_by_field(char *field);
+size_t zmalloc_get_memory_size(void);
 void zlibc_free(void *ptr);
 
 #ifndef HAVE_MALLOC_SIZE

tests/cluster/run.tcl

@@ -21,6 +21,7 @@ proc main {} {
 
 if {[catch main e]} {
     puts $::errorInfo
+    if {$::pause_on_error} pause_on_error
     cleanup
     exit 1
 }

tests/cluster/tests/04-resharding.tcl

@@ -53,19 +53,31 @@ test "Cluster consistency during live resharding" {
             puts -nonewline "...Starting resharding..."
             flush stdout
             set target [dict get [get_myself [randomInt 5]] id]
-            set tribpid [exec \
+            set tribpid [lindex [exec \
                 ../../../src/redis-trib.rb reshard \
                 --from all \
                 --to $target \
                 --slots 100 \
                 --yes \
-                127.0.0.1:[get_instance_attrib redis 0 port] &]
+                127.0.0.1:[get_instance_attrib redis 0 port] \
+                | [info nameofexecutable] \
+                ../tests/helpers/onlydots.tcl \
+                &] 0]
         }
 
         # Write random data to random list.
-        set key "key:[randomInt $numkeys]"
+        set listid [randomInt $numkeys]
+        set key "key:$listid"
         set ele [randomValue]
-        $cluster rpush $key $ele
+        # We write both with Lua scripts and with plain commands.
+        # This way we are able to stress Lua -> Redis command invocation
+        # as well, that has tests to prevent Lua to write into wrong
+        # hash slots.
+        if {$listid % 2} {
+            $cluster rpush $key $ele
+        } else {
+            $cluster eval {redis.call("rpush",KEYS[1],ARGV[1])} 1 $key $ele
+        }
         lappend content($key) $ele
 
         if {($j % 1000) == 0} {

tests/cluster/tests/helpers/onlydots.tcl

@@ -0,0 +1,16 @@
+# Read the standard input and only shows dots in the output, filtering out
+# all the other characters. Designed to avoid bufferization so that when
+# we get the output of redis-trib and want to show just the dots, we'll see
+# the dots as soon as redis-trib will output them.
+
+fconfigure stdin -buffering none
+
+while 1 {
+    set c [read stdin 1]
+    if {$c eq {}} {
+        exit 0; # EOF
+    } elseif {$c eq {.}} {
+        puts -nonewline .
+        flush stdout
+    }
+}

tests/cluster/tests/includes/init-tests.tcl

@@ -27,10 +27,17 @@ test "Cluster nodes are reachable" {
 
 test "Cluster nodes hard reset" {
     foreach_redis_id id {
+        if {$::valgrind} {
+            set node_timeout 10000
+        } else {
+            set node_timeout 3000
+        }
         catch {R $id flushall} ; # May fail for readonly slaves.
+        R $id MULTI
         R $id cluster reset hard
         R $id cluster set-config-epoch [expr {$id+1}]
-        R $id config set cluster-node-timeout 3000
+        R $id EXEC
+        R $id config set cluster-node-timeout $node_timeout
         R $id config set cluster-slave-validity-factor 10
         R $id config rewrite
     }

tests/instances.tcl

@@ -16,6 +16,7 @@ source ../support/server.tcl
 source ../support/test.tcl
 
 set ::verbose 0
+set ::valgrind 0
 set ::pause_on_error 0
 set ::simulate_error 0
 set ::sentinel_instances {}
@@ -32,6 +33,25 @@ if {[catch {cd tmp}]} {
     exit 1
 }
 
+# Execute the specified instance of the server specified by 'type', using
+# the provided configuration file. Returns the PID of the process.
+proc exec_instance {type cfgfile} {
+    if {$type eq "redis"} {
+        set prgname redis-server
+    } elseif {$type eq "sentinel"} {
+        set prgname redis-sentinel
+    } else {
+        error "Unknown instance type."
+    }
+
+    if {$::valgrind} {
+        set pid [exec valgrind --track-origins=yes --suppressions=../../../src/valgrind.sup --show-reachable=no --show-possibly-lost=no --leak-check=full ../../../src/${prgname} $cfgfile &]
+    } else {
+        set pid [exec ../../../src/${prgname} $cfgfile &]
+    }
+    return $pid
+}
+
 # Spawn a redis or sentinel instance, depending on 'type'.
 proc spawn_instance {type base_port count {conf {}}} {
     for {set j 0} {$j < $count} {incr j} {
@@ -58,14 +78,7 @@ proc spawn_instance {type base_port count {conf {}}} {
         close $cfg
 
         # Finally exec it and remember the pid for later cleanup.
-        if {$type eq "redis"} {
-            set prgname redis-server
-        } elseif {$type eq "sentinel"} {
-            set prgname redis-sentinel
-        } else {
-            error "Unknown instance type."
-        }
-        set pid [exec ../../../src/${prgname} $cfgfile &]
+        set pid [exec_instance $type $cfgfile]
         lappend ::pids $pid
 
         # Check availability
@@ -98,6 +111,7 @@ proc cleanup {} {
 proc abort_sentinel_test msg {
     puts "WARNING: Aborting the test."
     puts ">>>>>>>> $msg"
+    if {$::pause_on_error} pause_on_error
     cleanup
     exit 1
 }
@@ -113,6 +127,8 @@ proc parse_options {} {
             set ::pause_on_error 1
         } elseif {$opt eq "--fail"} {
             set ::simulate_error 1
+        } elseif {$opt eq {--valgrind}} {
+            set ::valgrind 1
         } elseif {$opt eq "--help"} {
             puts "Hello, I'm sentinel.tcl and I run Sentinel unit tests."
             puts "\nOptions:"
@@ -360,15 +376,31 @@ proc get_instance_id_by_port {type port} {
 # The instance can be restarted with restart-instance.
 proc kill_instance {type id} {
     set pid [get_instance_attrib $type $id pid]
+    set port [get_instance_attrib $type $id port]
+
     if {$pid == -1} {
         error "You tried to kill $type $id twice."
     }
+
     exec kill -9 $pid
     set_instance_attrib $type $id pid -1
     set_instance_attrib $type $id link you_tried_to_talk_with_killed_instance
 
     # Remove the PID from the list of pids to kill at exit.
     set ::pids [lsearch -all -inline -not -exact $::pids $pid]
+
+    # Wait for the port it was using to be available again, so that's not
+    # an issue to start a new server ASAP with the same port.
+    set retry 10
+    while {[incr retry -1]} {
+        set port_is_free [catch {set s [socket 127.0.01 $port]}]
+        if {$port_is_free} break
+        catch {close $s}
+        after 1000
+    }
+    if {$retry == 0} {
+        error "Port $port does not return available after killing instance."
+    }
 }
 
 # Return true of the instance of the specified type/id is killed.
@@ -385,12 +417,7 @@ proc restart_instance {type id} {
 
     # Execute the instance with its old setup and append the new pid
     # file for cleanup.
-    if {$type eq "redis"} {
-        set prgname redis-server
-    } else {
-        set prgname redis-sentinel
-    }
-    set pid [exec ../../../src/${prgname} $cfgfile &]
+    set pid [exec_instance $type $cfgfile]
     set_instance_attrib $type $id pid $pid
     lappend ::pids $pid
 

tests/integration/aof.tcl

@@ -204,6 +204,30 @@ tags {"aof"} {
         }
     }
 
+    ## Uses the alsoPropagate() API.
+    create_aof {
+        append_to_aof [formatCommand sadd set foo]
+        append_to_aof [formatCommand sadd set bar]
+        append_to_aof [formatCommand sadd set gah]
+        append_to_aof [formatCommand spop set 2]
+    }
+
+    start_server_aof [list dir $server_path] {
+        test "AOF+SPOP: Server should have been started" {
+            assert_equal 1 [is_alive $srv]
+        }
+
+        test "AOF+SPOP: Set should have 1 member" {
+            set client [redis [dict get $srv host] [dict get $srv port]]
+            wait_for_condition 50 100 {
+                [catch {$client ping} e] == 0
+            } else {
+                fail "Loading DB is taking too much time."
+            }
+            assert_equal 1 [$client scard set]
+        }
+    }
+
     ## Test that EXPIREAT is loaded correctly
     create_aof {
         append_to_aof [formatCommand rpush list foo]

tests/integration/rdb.tcl

@@ -90,7 +90,7 @@ start_server_and_kill_it [list "dir" $server_path] {
     test {Server should not start if RDB is corrupted} {
         wait_for_condition 50 100 {
             [string match {*RDB checksum*} \
-                [exec tail -n1 < [dict get $srv stdout]]]
+                [exec tail -n10 < [dict get $srv stdout]]]
         } else {
             fail "Server started even if RDB was corrupted!"
         }

tests/integration/replication-4.tcl

@@ -132,5 +132,24 @@ start_server {tags {"repl"}} {
             }
             assert {[$master dbsize] > 0}
         }
+
+        test {Replication of SPOP command -- alsoPropagate() API} {
+            $master del myset
+            set size [randomInt 100]
+            set content {}
+            for {set j 0} {$j < $size} {incr j} {
+                lappend content [randomValue]
+            }
+            $master sadd myset {*}$content
+
+            set count [randomInt 100]
+            set result [$master spop myset $count]
+
+            wait_for_condition 50 100 {
+                [$master debug digest] eq [$slave debug digest]
+            } else {
+                fail "SPOP replication inconsistency"
+            }
+        }
     }
 }

tests/integration/replication.tcl

@@ -118,7 +118,8 @@ foreach dl {no yes} {
                         [lindex $slaves 1] slaveof $master_host $master_port
                         [lindex $slaves 2] slaveof $master_host $master_port
 
-                        # Wait for all the three slaves to reach the "online" state
+                        # Wait for all the three slaves to reach the "online"
+                        # state from the POV of the master.
                         set retry 500
                         while {$retry} {
                             set info [r -3 info]
@@ -133,6 +134,17 @@ foreach dl {no yes} {
                             error "assertion:Slaves not correctly synchronized"
                         }
 
+                        # Wait that slaves acknowledge they are online so
+                        # we are sure that DBSIZE and DEBUG DIGEST will not
+                        # fail because of timing issues.
+                        wait_for_condition 500 100 {
+                            [lindex [[lindex $slaves 0] role] 3] eq {connected} &&
+                            [lindex [[lindex $slaves 1] role] 3] eq {connected} &&
+                            [lindex [[lindex $slaves 2] role] 3] eq {connected}
+                        } else {
+                            fail "Slaves still not connected after some time"
+                        }
+
                         # Stop the write load
                         stop_write_load $load_handle0
                         stop_write_load $load_handle1
@@ -140,16 +152,8 @@ foreach dl {no yes} {
                         stop_write_load $load_handle3
                         stop_write_load $load_handle4
 
-                        # Wait that slaves exit the "loading" state
-                        wait_for_condition 500 100 {
-                            ![string match {*loading:1*} [[lindex $slaves 0] info]] &&
-                            ![string match {*loading:1*} [[lindex $slaves 1] info]] &&
-                            ![string match {*loading:1*} [[lindex $slaves 2] info]]
-                        } else {
-                            fail "Slaves still loading data after too much time"
-                        }
-
-                        # Make sure that slaves and master have same number of keys
+                        # Make sure that slaves and master have same
+                        # number of keys
                         wait_for_condition 500 100 {
                             [$master dbsize] == [[lindex $slaves 0] dbsize] &&
                             [$master dbsize] == [[lindex $slaves 1] dbsize] &&

tests/support/cluster.tcl

@@ -226,6 +226,8 @@ proc ::redis_cluster::get_keys_from_command {cmd argv} {
     # Special handling for other commands
     switch -exact $cmd {
         mget {return $argv}
+        eval {return [lrange $argv 2 1+[lindex $argv 1]]}
+        evalsha {return [lrange $argv 2 1+[lindex $argv 1]]}
     }
 
     # All the remaining commands are not handled.

tests/support/server.tcl

@@ -70,6 +70,9 @@ proc kill_server config {
     if {$::valgrind} {
         check_valgrind_errors [dict get $config stderr]
     }
+
+    # Remove this pid from the set of active pids in the test server.
+    send_data_packet $::test_server_fd server-killed $pid
 }
 
 proc is_alive config {
@@ -204,11 +207,14 @@ proc start_server {options {code undefined}} {
     set stderr [format "%s/%s" [dict get $config "dir"] "stderr"]
 
     if {$::valgrind} {
-        exec valgrind --suppressions=src/valgrind.sup --show-reachable=no --show-possibly-lost=no --leak-check=full src/redis-server $config_file > $stdout 2> $stderr &
+        set pid [exec valgrind --track-origins=yes --suppressions=src/valgrind.sup --show-reachable=no --show-possibly-lost=no --leak-check=full src/redis-server $config_file > $stdout 2> $stderr &]
     } else {
-        exec src/redis-server $config_file > $stdout 2> $stderr &
+        set pid [exec src/redis-server $config_file > $stdout 2> $stderr &]
     }
 
+    # Tell the test server about this new instance.
+    send_data_packet $::test_server_fd server-spawned $pid
+
     # check that the server actually started
     # ugly but tries to be as fast as possible...
     if {$::valgrind} {set retrynum 1000} else {set retrynum 100}
@@ -234,9 +240,9 @@ proc start_server {options {code undefined}} {
         return
     }
 
-    # find out the pid
-    while {![info exists pid]} {
-        regexp {PID:\s(\d+)} [exec cat $stdout] _ pid
+    # Wait for actual startup
+    while {![info exists _pid]} {
+        regexp {PID:\s(\d+)} [exec cat $stdout] _ _pid
         after 100
     }
 

tests/support/test.tcl

@@ -19,9 +19,12 @@ proc assert_match {pattern value} {
     }
 }
 
-proc assert_equal {expected value} {
+proc assert_equal {expected value {detail ""}} {
     if {$expected ne $value} {
-        error "assertion:Expected '$value' to be equal to '$expected'"
+        if {$detail ne ""} {
+            set detail " (detail: $detail)"
+        }
+        error "assertion:Expected '$value' to be equal to '$expected'$detail"
     }
 }
 

tests/test_helper.tcl

@@ -65,6 +65,9 @@ set ::file ""; # If set, runs only the tests in this comma separated list
 set ::curfile ""; # Hold the filename of the current suite
 set ::accurate 0; # If true runs fuzz tests with more iterations
 set ::force_failure 0
+set ::timeout 600; # 10 minutes without progresses will quit the test.
+set ::last_progress [clock seconds]
+set ::active_servers {} ; # Pids of active Redis instances.
 
 # Set to 1 when we are running in client mode. The Redis test uses a
 # server-client model to run tests simultaneously. The server instance
@@ -200,11 +203,19 @@ proc test_server_main {} {
     vwait forever
 }
 
-# This function gets called 10 times per second, for now does nothing but
-# may be used in the future in order to detect test clients taking too much
-# time to execute the task.
+# This function gets called 10 times per second.
 proc test_server_cron {} {
-    # Do some work here.
+    set elapsed [expr {[clock seconds]-$::last_progress}]
+
+    if {$elapsed > $::timeout} {
+        set err "\[[colorstr red TIMEOUT]\]: clients state report follows."
+        puts $err
+        show_clients_state
+        kill_clients
+        force_kill_all_servers
+        the_end
+    }
+
     after 100 test_server_cron
 }
 
@@ -230,6 +241,8 @@ proc read_from_test_client fd {
     set bytes [gets $fd]
     set payload [read $fd $bytes]
     foreach {status data} $payload break
+    set ::last_progress [clock seconds]
+
     if {$status eq {ready}} {
         if {!$::quiet} {
             puts "\[$status\]: $data"
@@ -256,12 +269,15 @@ proc read_from_test_client fd {
         set ::active_clients_task($fd) "(ERR) $data"
     } elseif {$status eq {exception}} {
         puts "\[[colorstr red $status]\]: $data"
-        foreach p $::clients_pids {
-            catch {exec kill -9 $p}
-        }
+        kill_clients
+        force_kill_all_servers
         exit 1
     } elseif {$status eq {testing}} {
         set ::active_clients_task($fd) "(IN PROGRESS) $data"
+    } elseif {$status eq {server-spawned}} {
+        lappend ::active_servers $data
+    } elseif {$status eq {server-killed}} {
+        set ::active_servers [lsearch -all -inline -not -exact $::active_servers $data]
     } else {
         if {!$::quiet} {
             puts "\[$status\]: $data"
@@ -269,6 +285,31 @@ proc read_from_test_client fd {
     }
 }
 
+proc show_clients_state {} {
+    # The following loop is only useful for debugging tests that may
+    # enter an infinite loop. Commented out normally.
+    foreach x $::active_clients {
+        if {[info exist ::active_clients_task($x)]} {
+            puts "$x => $::active_clients_task($x)"
+        } else {
+            puts "$x => ???"
+        }
+    }
+}
+
+proc kill_clients {} {
+    foreach p $::clients_pids {
+        catch {exec kill $p}
+    }
+}
+
+proc force_kill_all_servers {} {
+    foreach p $::active_servers {
+        puts "Killing still running Redis server $p"
+        catch {exec kill -9 $p}
+    }
+}
+
 # A new client is idle. Remove it from the list of active clients and
 # if there are still test units to run, launch them.
 proc signal_idle_client fd {
@@ -276,17 +317,7 @@ proc signal_idle_client fd {
     set ::active_clients \
         [lsearch -all -inline -not -exact $::active_clients $fd]
 
-    if 0 {
-        # The following loop is only useful for debugging tests that may
-        # enter an infinite loop. Commented out normally.
-        foreach x $::active_clients {
-            if {[info exist ::active_clients_task($x)]} {
-                puts "$x => $::active_clients_task($x)"
-            } else {
-                puts "$x => ???"
-            }
-        }
-    }
+    if 0 {show_clients_state}
 
     # New unit to process?
     if {$::next_test != [llength $::all_tests]} {
@@ -361,7 +392,8 @@ proc print_help_screen {} {
         "--quiet            Don't show individual tests."
         "--single <unit>    Just execute the specified unit (see next option)."
         "--list-tests       List all the available test units."
-        "--clients <num>    Number of test clients (16)."
+        "--clients <num>    Number of test clients (default 16)."
+        "--timeout <sec>    Test timeout in seconds (default 10 min)."
         "--force-failure    Force the execution of a test that always fails."
         "--help             Print this help screen."
     } "\n"]
@@ -410,6 +442,9 @@ for {set j 0} {$j < [llength $argv]} {incr j} {
     } elseif {$opt eq {--clients}} {
         set ::numclients $arg
         incr j
+    } elseif {$opt eq {--timeout}} {
+        set ::timeout $arg
+        incr j
     } elseif {$opt eq {--help}} {
         print_help_screen
         exit 0

tests/unit/aofrw.tcl

@@ -77,10 +77,10 @@ start_server {tags {"aofrw"}} {
     }
 
     foreach d {string int} {
-        foreach e {ziplist linkedlist} {
+        foreach e {quicklist} {
             test "AOF rewrite of list with $e encoding, $d data" {
                 r flushall
-                if {$e eq {ziplist}} {set len 10} else {set len 1000}
+                set len 1000
                 for {set j 0} {$j < $len} {incr j} {
                     if {$d eq {string}} {
                         set data [randstring 0 16 alpha]

tests/unit/dump.tcl

@@ -157,7 +157,7 @@ start_server {tags {"dump"}} {
     test {MIGRATE can correctly transfer large values} {
         set first [srv 0 client]
         r del key
-        for {set j 0} {$j < 5000} {incr j} {
+        for {set j 0} {$j < 40000} {incr j} {
             r rpush key 1 2 3 4 5 6 7 8 9 10
             r rpush key "item 1" "item 2" "item 3" "item 4" "item 5" \
                         "item 6" "item 7" "item 8" "item 9" "item 10"
@@ -175,7 +175,7 @@ start_server {tags {"dump"}} {
             assert {[$first exists key] == 0}
             assert {[$second exists key] == 1}
             assert {[$second ttl key] == -1}
-            assert {[$second llen key] == 5000*20}
+            assert {[$second llen key] == 40000*20}
         }
     }
 

tests/unit/scan.tcl

@@ -226,4 +226,14 @@ start_server {tags {"scan"}} {
         set res [r zscan mykey 0 MATCH foo* COUNT 10000]
         lsort -unique [lindex $res 1]
     }
+
+    test "ZSCAN scores: regression test for issue #2175" {
+        r del mykey
+        for {set j 0} {$j < 500} {incr j} {
+            r zadd mykey 9.8813129168249309e-323 $j
+        }
+        set res [lindex [r zscan mykey 0] 1]
+        set first_score [lindex $res 1]
+        assert {$first_score != 0}
+    }
 }

tests/unit/scripting.tcl

@@ -184,6 +184,94 @@ start_server {tags {"scripting"}} {
         set e
     } {*against a key*}
 
+    test {EVAL - JSON numeric decoding} {
+        # We must return the table as a string because otherwise
+        # Redis converts floats to ints and we get 0 and 1023 instead
+        # of 0.0003 and 1023.2 as the parsed output.
+        r eval {return
+                 table.concat(
+                   cjson.decode(
+                    "[0.0, -5e3, -1, 0.3e-3, 1023.2, 0e10]"), " ")
+        } 0
+    } {0 -5000 -1 0.0003 1023.2 0}
+
+    test {EVAL - JSON string decoding} {
+        r eval {local decoded = cjson.decode('{"keya": "a", "keyb": "b"}')
+                return {decoded.keya, decoded.keyb}
+        } 0
+    } {a b}
+
+    test {EVAL - cmsgpack can pack double?} {
+        r eval {local encoded = cmsgpack.pack(0.1)
+                local h = ""
+                for i = 1, #encoded do
+                    h = h .. string.format("%02x",string.byte(encoded,i))
+                end
+                return h
+        } 0
+    } {cb3fb999999999999a}
+
+    test {EVAL - cmsgpack can pack negative int64?} {
+        r eval {local encoded = cmsgpack.pack(-1099511627776)
+                local h = ""
+                for i = 1, #encoded do
+                    h = h .. string.format("%02x",string.byte(encoded,i))
+                end
+                return h
+        } 0
+    } {d3ffffff0000000000}
+
+    test {EVAL - cmsgpack can pack and unpack circular references?} {
+        r eval {local a = {x=nil,y=5}
+                local b = {x=a}
+                a['x'] = b
+                local encoded = cmsgpack.pack(a)
+                local h = ""
+                -- cmsgpack encodes to a depth of 16, but can't encode
+                -- references, so the encoded object has a deep copy recusive
+                -- depth of 16.
+                for i = 1, #encoded do
+                    h = h .. string.format("%02x",string.byte(encoded,i))
+                end
+                -- when unpacked, re.x.x != re because the unpack creates
+                -- individual tables down to a depth of 16.
+                -- (that's why the encoded output is so large)
+                local re = cmsgpack.unpack(encoded)
+                assert(re)
+                assert(re.x)
+                assert(re.x.x.y == re.y)
+                assert(re.x.x.x.x.y == re.y)
+                assert(re.x.x.x.x.x.x.y == re.y)
+                assert(re.x.x.x.x.x.x.x.x.x.x.y == re.y)
+                -- maximum working depth:
+                assert(re.x.x.x.x.x.x.x.x.x.x.x.x.x.x.y == re.y)
+                -- now the last x would be b above and has no y
+                assert(re.x.x.x.x.x.x.x.x.x.x.x.x.x.x.x)
+                -- so, the final x.x is at the depth limit and was assigned nil
+                assert(re.x.x.x.x.x.x.x.x.x.x.x.x.x.x.x.x == nil)
+                return {h, re.x.x.x.x.x.x.x.x.y == re.y, re.y == 5}
+        } 0
+    } {82a17905a17881a17882a17905a17881a17882a17905a17881a17882a17905a17881a17882a17905a17881a17882a17905a17881a17882a17905a17881a17882a17905a17881a178c0 1 1}
+
+    test {EVAL - Numerical sanity check from bitop} {
+        r eval {assert(0x7fffffff == 2147483647, "broken hex literals");
+                assert(0xffffffff == -1 or 0xffffffff == 2^32-1,
+                    "broken hex literals");
+                assert(tostring(-1) == "-1", "broken tostring()");
+                assert(tostring(0xffffffff) == "-1" or
+                    tostring(0xffffffff) == "4294967295",
+                    "broken tostring()")
+        } 0
+    } {}
+
+    test {EVAL - Verify minimal bitop functionality} {
+        r eval {assert(bit.tobit(1) == 1);
+                assert(bit.band(1) == 1);
+                assert(bit.bxor(1,2) == 3);
+                assert(bit.bor(1,2,4,8,16,32,64,128) == 255)
+        } 0
+    } {}
+
     test {SCRIPTING FLUSH - is able to clear the scripts cache?} {
         r set mykey myval
         set v [r evalsha fd758d1589d044dd850a6f05d52f2eefd27f033f 1 mykey]
@@ -325,7 +413,7 @@ start_server {tags {"scripting"}} {
         r sadd myset a b c
         r mset a 1 b 2 c 3 d 4
         assert {[r spop myset] ne {}}
-        assert {[r spop myset] ne {}}
+        assert {[r spop myset 1] ne {}}
         assert {[r spop myset] ne {}}
         assert {[r mget a b c d] eq {1 2 3 4}}
         assert {[r spop myset] eq {}}

tests/unit/sort.tcl

@@ -1,8 +1,7 @@
 start_server {
     tags {"sort"}
     overrides {
-        "list-max-ziplist-value" 16
-        "list-max-ziplist-entries" 32
+        "list-max-ziplist-size" 32
         "set-max-intset-entries" 32
     }
 } {
@@ -36,9 +35,9 @@ start_server {
     }
 
     foreach {num cmd enc title} {
-        16 lpush ziplist "Ziplist"
-        1000 lpush linkedlist "Linked list"
-        10000 lpush linkedlist "Big Linked list"
+        16 lpush quicklist "Old Ziplist"
+        1000 lpush quicklist "Old Linked list"
+        10000 lpush quicklist "Old Big Linked list"
         16 sadd intset "Intset"
         1000 sadd hashtable "Hash table"
         10000 sadd hashtable "Big Hash table"
@@ -85,14 +84,14 @@ start_server {
         r sort tosort BY weight_* store sort-res
         assert_equal $result [r lrange sort-res 0 -1]
         assert_equal 16 [r llen sort-res]
-        assert_encoding ziplist sort-res
+        assert_encoding quicklist sort-res
     }
 
     test "SORT BY hash field STORE" {
         r sort tosort BY wobj_*->weight store sort-res
         assert_equal $result [r lrange sort-res 0 -1]
         assert_equal 16 [r llen sort-res]
-        assert_encoding ziplist sort-res
+        assert_encoding quicklist sort-res
     }
 
     test "SORT extracts STORE correctly" {
@@ -246,6 +245,24 @@ start_server {
         r sort mylist by num get x:*->
     } {100}
 
+    test "SORT by nosort retains native order for lists" {
+        r del testa
+        r lpush testa 2 1 4 3 5
+        r sort testa by nosort
+    } {5 3 4 1 2}
+
+    test "SORT by nosort plus store retains native order for lists" {
+        r del testa
+        r lpush testa 2 1 4 3 5
+        r sort testa by nosort store testb
+        r lrange testb 0 -1
+    } {5 3 4 1 2}
+
+    test "SORT by nosort with limit returns based on original list order" {
+        r sort testa by nosort limit 0 3 store testb
+        r lrange testb 0 -1
+    } {5 3 4}
+
     tags {"slow"} {
         set num 100
         set res [create_random_dataset $num lpush]

tests/unit/type/list-2.tcl

@@ -1,8 +1,7 @@
 start_server {
     tags {"list"}
     overrides {
-        "list-max-ziplist-value" 16
-        "list-max-ziplist-entries" 256
+        "list-max-ziplist-size" 4
     }
 } {
     source "tests/unit/type/list-common.tcl"
@@ -28,14 +27,18 @@ start_server {
                 for {set i 0} {$i < 1000} {incr i} {
                     set min [expr {int(rand()*$startlen)}]
                     set max [expr {$min+int(rand()*$startlen)}]
+                    set before_len [llength $mylist]
+                    set before_len_r [r llen mylist]
                     set mylist [lrange $mylist $min $max]
                     r ltrim mylist $min $max
-                    assert_equal $mylist [r lrange mylist 0 -1]
+                    assert_equal $mylist [r lrange mylist 0 -1] "failed trim"
 
+                    set starting [r llen mylist]
                     for {set j [r llen mylist]} {$j < $startlen} {incr j} {
                         set str [randomInt 9223372036854775807]
                         r rpush mylist $str
                         lappend mylist $str
+                        assert_equal $mylist [r lrange mylist 0 -1] "failed append match"
                     }
                 }
             }

tests/unit/type/list-3.tcl

@@ -1,8 +1,7 @@
 start_server {
     tags {list ziplist}
     overrides {
-        "list-max-ziplist-value" 200000
-        "list-max-ziplist-entries" 256
+        "list-max-ziplist-size" 16
     }
 } {
     test {Explicit regression for a list bug} {

tests/unit/type/list.tcl

@@ -1,25 +1,24 @@
 start_server {
     tags {"list"}
     overrides {
-        "list-max-ziplist-value" 16
-        "list-max-ziplist-entries" 256
+        "list-max-ziplist-size" 5
     }
 } {
     source "tests/unit/type/list-common.tcl"
 
     test {LPUSH, RPUSH, LLENGTH, LINDEX, LPOP - ziplist} {
         # first lpush then rpush
-        assert_equal 1 [r lpush myziplist1 a]
-        assert_equal 2 [r rpush myziplist1 b]
-        assert_equal 3 [r rpush myziplist1 c]
+        assert_equal 1 [r lpush myziplist1 aa]
+        assert_equal 2 [r rpush myziplist1 bb]
+        assert_equal 3 [r rpush myziplist1 cc]
         assert_equal 3 [r llen myziplist1]
-        assert_equal a [r lindex myziplist1 0]
-        assert_equal b [r lindex myziplist1 1]
-        assert_equal c [r lindex myziplist1 2]
+        assert_equal aa [r lindex myziplist1 0]
+        assert_equal bb [r lindex myziplist1 1]
+        assert_equal cc [r lindex myziplist1 2]
         assert_equal {} [r lindex myziplist2 3]
-        assert_equal c [r rpop myziplist1]
-        assert_equal a [r lpop myziplist1]
-        assert_encoding ziplist myziplist1
+        assert_equal cc [r rpop myziplist1]
+        assert_equal aa [r lpop myziplist1]
+        assert_encoding quicklist myziplist1
 
         # first rpush then lpush
         assert_equal 1 [r rpush myziplist2 a]
@@ -32,13 +31,13 @@ start_server {
         assert_equal {} [r lindex myziplist2 3]
         assert_equal a [r rpop myziplist2]
         assert_equal c [r lpop myziplist2]
-        assert_encoding ziplist myziplist2
+        assert_encoding quicklist myziplist2
     }
 
     test {LPUSH, RPUSH, LLENGTH, LINDEX, LPOP - regular list} {
         # first lpush then rpush
         assert_equal 1 [r lpush mylist1 $largevalue(linkedlist)]
-        assert_encoding linkedlist mylist1
+        assert_encoding quicklist mylist1
         assert_equal 2 [r rpush mylist1 b]
         assert_equal 3 [r rpush mylist1 c]
         assert_equal 3 [r llen mylist1]
@@ -51,7 +50,7 @@ start_server {
 
         # first rpush then lpush
         assert_equal 1 [r rpush mylist2 $largevalue(linkedlist)]
-        assert_encoding linkedlist mylist2
+        assert_encoding quicklist mylist2
         assert_equal 2 [r lpush mylist2 b]
         assert_equal 3 [r lpush mylist2 c]
         assert_equal 3 [r llen mylist2]
@@ -74,34 +73,22 @@ start_server {
         assert_equal {d c b a 0 1 2 3} [r lrange mylist 0 -1]
     }
 
-    test {DEL a list - ziplist} {
-        assert_equal 1 [r del myziplist2]
-        assert_equal 0 [r exists myziplist2]
-        assert_equal 0 [r llen myziplist2]
-    }
-
-    test {DEL a list - regular list} {
+    test {DEL a list} {
         assert_equal 1 [r del mylist2]
         assert_equal 0 [r exists mylist2]
         assert_equal 0 [r llen mylist2]
     }
 
-    proc create_ziplist {key entries} {
+    proc create_list {key entries} {
         r del $key
         foreach entry $entries { r rpush $key $entry }
-        assert_encoding ziplist $key
-    }
-
-    proc create_linkedlist {key entries} {
-        r del $key
-        foreach entry $entries { r rpush $key $entry }
-        assert_encoding linkedlist $key
+        assert_encoding quicklist $key
     }
 
     foreach {type large} [array get largevalue] {
         test "BLPOP, BRPOP: single existing list - $type" {
             set rd [redis_deferring_client]
-            create_$type blist "a b $large c d"
+            create_list blist "a b $large c d"
 
             $rd blpop blist 1
             assert_equal {blist a} [$rd read]
@@ -116,8 +103,8 @@ start_server {
 
         test "BLPOP, BRPOP: multiple existing lists - $type" {
             set rd [redis_deferring_client]
-            create_$type blist1 "a $large c"
-            create_$type blist2 "d $large f"
+            create_list blist1 "a $large c"
+            create_list blist2 "d $large f"
 
             $rd blpop blist1 blist2 1
             assert_equal {blist1 a} [$rd read]
@@ -137,7 +124,7 @@ start_server {
         test "BLPOP, BRPOP: second list has an entry - $type" {
             set rd [redis_deferring_client]
             r del blist1
-            create_$type blist2 "d $large f"
+            create_list blist2 "d $large f"
 
             $rd blpop blist1 blist2 1
             assert_equal {blist2 d} [$rd read]
@@ -151,7 +138,7 @@ start_server {
             r del target
 
             set rd [redis_deferring_client]
-            create_$type blist "a b $large c d"
+            create_list blist "a b $large c d"
 
             $rd brpoplpush blist target 1
             assert_equal d [$rd read]
@@ -517,28 +504,28 @@ start_server {
 
     foreach {type large} [array get largevalue] {
         test "LPUSHX, RPUSHX - $type" {
-            create_$type xlist "$large c"
+            create_list xlist "$large c"
             assert_equal 3 [r rpushx xlist d]
             assert_equal 4 [r lpushx xlist a]
             assert_equal "a $large c d" [r lrange xlist 0 -1]
         }
 
         test "LINSERT - $type" {
-            create_$type xlist "a $large c d"
-            assert_equal 5 [r linsert xlist before c zz]
-            assert_equal "a $large zz c d" [r lrange xlist 0 10]
-            assert_equal 6 [r linsert xlist after c yy]
-            assert_equal "a $large zz c yy d" [r lrange xlist 0 10]
-            assert_equal 7 [r linsert xlist after d dd]
-            assert_equal -1 [r linsert xlist after bad ddd]
-            assert_equal "a $large zz c yy d dd" [r lrange xlist 0 10]
-            assert_equal 8 [r linsert xlist before a aa]
-            assert_equal -1 [r linsert xlist before bad aaa]
-            assert_equal "aa a $large zz c yy d dd" [r lrange xlist 0 10]
+            create_list xlist "a $large c d"
+            assert_equal 5 [r linsert xlist before c zz] "before c"
+            assert_equal "a $large zz c d" [r lrange xlist 0 10] "lrangeA"
+            assert_equal 6 [r linsert xlist after c yy] "after c"
+            assert_equal "a $large zz c yy d" [r lrange xlist 0 10] "lrangeB"
+            assert_equal 7 [r linsert xlist after d dd] "after d"
+            assert_equal -1 [r linsert xlist after bad ddd] "after bad"
+            assert_equal "a $large zz c yy d dd" [r lrange xlist 0 10] "lrangeC"
+            assert_equal 8 [r linsert xlist before a aa] "before a"
+            assert_equal -1 [r linsert xlist before bad aaa] "before bad"
+            assert_equal "aa a $large zz c yy d dd" [r lrange xlist 0 10] "lrangeD"
 
             # check inserting integer encoded value
-            assert_equal 9 [r linsert xlist before aa 42]
-            assert_equal 42 [r lrange xlist 0 0]
+            assert_equal 9 [r linsert xlist before aa 42] "before aa"
+            assert_equal 42 [r lrange xlist 0 0] "lrangeE"
         }
     }
 
@@ -547,55 +534,7 @@ start_server {
         set e
     } {*ERR*syntax*error*}
 
-    test {LPUSHX, RPUSHX convert from ziplist to list} {
-        set large $largevalue(linkedlist)
-
-        # convert when a large value is pushed
-        create_ziplist xlist a
-        assert_equal 2 [r rpushx xlist $large]
-        assert_encoding linkedlist xlist
-        create_ziplist xlist a
-        assert_equal 2 [r lpushx xlist $large]
-        assert_encoding linkedlist xlist
-
-        # convert when the length threshold is exceeded
-        create_ziplist xlist [lrepeat 256 a]
-        assert_equal 257 [r rpushx xlist b]
-        assert_encoding linkedlist xlist
-        create_ziplist xlist [lrepeat 256 a]
-        assert_equal 257 [r lpushx xlist b]
-        assert_encoding linkedlist xlist
-    }
-
-    test {LINSERT convert from ziplist to list} {
-        set large $largevalue(linkedlist)
-
-        # convert when a large value is inserted
-        create_ziplist xlist a
-        assert_equal 2 [r linsert xlist before a $large]
-        assert_encoding linkedlist xlist
-        create_ziplist xlist a
-        assert_equal 2 [r linsert xlist after a $large]
-        assert_encoding linkedlist xlist
-
-        # convert when the length threshold is exceeded
-        create_ziplist xlist [lrepeat 256 a]
-        assert_equal 257 [r linsert xlist before a a]
-        assert_encoding linkedlist xlist
-        create_ziplist xlist [lrepeat 256 a]
-        assert_equal 257 [r linsert xlist after a a]
-        assert_encoding linkedlist xlist
-
-        # don't convert when the value could not be inserted
-        create_ziplist xlist [lrepeat 256 a]
-        assert_equal -1 [r linsert xlist before foo a]
-        assert_encoding ziplist xlist
-        create_ziplist xlist [lrepeat 256 a]
-        assert_equal -1 [r linsert xlist after foo a]
-        assert_encoding ziplist xlist
-    }
-
-    foreach {type num} {ziplist 250 linkedlist 500} {
+    foreach {type num} {quicklist 250 quicklist 500} {
         proc check_numbered_list_consistency {key} {
             set len [r llen $key]
             for {set i 0} {$i < $len} {incr i} {
@@ -664,16 +603,16 @@ start_server {
     foreach {type large} [array get largevalue] {
         test "RPOPLPUSH base case - $type" {
             r del mylist1 mylist2
-            create_$type mylist1 "a $large c d"
+            create_list mylist1 "a $large c d"
             assert_equal d [r rpoplpush mylist1 mylist2]
             assert_equal c [r rpoplpush mylist1 mylist2]
             assert_equal "a $large" [r lrange mylist1 0 -1]
             assert_equal "c d" [r lrange mylist2 0 -1]
-            assert_encoding ziplist mylist2
+            assert_encoding quicklist mylist2
         }
 
         test "RPOPLPUSH with the same list as src and dst - $type" {
-            create_$type mylist "a $large c"
+            create_list mylist "a $large c"
             assert_equal "a $large c" [r lrange mylist 0 -1]
             assert_equal c [r rpoplpush mylist mylist]
             assert_equal "c a $large" [r lrange mylist 0 -1]
@@ -681,8 +620,8 @@ start_server {
 
         foreach {othertype otherlarge} [array get largevalue] {
             test "RPOPLPUSH with $type source and existing target $othertype" {
-                create_$type srclist "a b c $large"
-                create_$othertype dstlist "$otherlarge"
+                create_list srclist "a b c $large"
+                create_list dstlist "$otherlarge"
                 assert_equal $large [r rpoplpush srclist dstlist]
                 assert_equal c [r rpoplpush srclist dstlist]
                 assert_equal "a b" [r lrange srclist 0 -1]
@@ -691,7 +630,7 @@ start_server {
                 # When we rpoplpush'ed a large value, dstlist should be
                 # converted to the same encoding as srclist.
                 if {$type eq "linkedlist"} {
-                    assert_encoding linkedlist dstlist
+                    assert_encoding quicklist dstlist
                 }
             }
         }
@@ -713,7 +652,7 @@ start_server {
     }
 
     test {RPOPLPUSH against non list dst key} {
-        create_ziplist srclist {a b c d}
+        create_list srclist {a b c d}
         r set dstlist x
         assert_error WRONGTYPE* {r rpoplpush srclist dstlist}
         assert_type string dstlist
@@ -727,7 +666,7 @@ start_server {
 
     foreach {type large} [array get largevalue] {
         test "Basic LPOP/RPOP - $type" {
-            create_$type mylist "$large 1 2"
+            create_list mylist "$large 1 2"
             assert_equal $large [r lpop mylist]
             assert_equal 2 [r rpop mylist]
             assert_equal 1 [r lpop mylist]
@@ -745,7 +684,7 @@ start_server {
         assert_error WRONGTYPE* {r rpop notalist}
     }
 
-    foreach {type num} {ziplist 250 linkedlist 500} {
+    foreach {type num} {quicklist 250 quicklist 500} {
         test "Mass RPOP/LPOP - $type" {
             r del mylist
             set sum1 0
@@ -765,24 +704,24 @@ start_server {
 
     foreach {type large} [array get largevalue] {
         test "LRANGE basics - $type" {
-            create_$type mylist "$large 1 2 3 4 5 6 7 8 9"
+            create_list mylist "$large 1 2 3 4 5 6 7 8 9"
             assert_equal {1 2 3 4 5 6 7 8} [r lrange mylist 1 -2]
             assert_equal {7 8 9} [r lrange mylist -3 -1]
             assert_equal {4} [r lrange mylist 4 4]
         }
 
         test "LRANGE inverted indexes - $type" {
-            create_$type mylist "$large 1 2 3 4 5 6 7 8 9"
+            create_list mylist "$large 1 2 3 4 5 6 7 8 9"
             assert_equal {} [r lrange mylist 6 2]
         }
 
         test "LRANGE out of range indexes including the full list - $type" {
-            create_$type mylist "$large 1 2 3"
+            create_list mylist "$large 1 2 3"
             assert_equal "$large 1 2 3" [r lrange mylist -1000 1000]
         }
 
         test "LRANGE out of range negative end index - $type" {
-            create_$type mylist "$large 1 2 3"
+            create_list mylist "$large 1 2 3"
             assert_equal $large [r lrange mylist 0 -4]
             assert_equal {} [r lrange mylist 0 -5]
         }
@@ -796,7 +735,7 @@ start_server {
         proc trim_list {type min max} {
             upvar 1 large large
             r del mylist
-            create_$type mylist "1 2 3 4 $large"
+            create_list mylist "1 2 3 4 $large"
             r ltrim mylist $min $max
             r lrange mylist 0 -1
         }
@@ -825,7 +764,7 @@ start_server {
 
     foreach {type large} [array get largevalue] {
         test "LSET - $type" {
-            create_$type mylist "99 98 $large 96 95"
+            create_list mylist "99 98 $large 96 95"
             r lset mylist 1 foo
             r lset mylist -1 bar
             assert_equal "99 foo $large 96 bar" [r lrange mylist 0 -1]
@@ -847,7 +786,7 @@ start_server {
 
     foreach {type e} [array get largevalue] {
         test "LREM remove all the occurrences - $type" {
-            create_$type mylist "$e foo bar foobar foobared zap bar test foo"
+            create_list mylist "$e foo bar foobar foobared zap bar test foo"
             assert_equal 2 [r lrem mylist 0 bar]
             assert_equal "$e foo foobar foobared zap test foo" [r lrange mylist 0 -1]
         }
@@ -863,7 +802,7 @@ start_server {
         }
 
         test "LREM starting from tail with negative count - $type" {
-            create_$type mylist "$e foo bar foobar foobared zap bar test foo foo"
+            create_list mylist "$e foo bar foobar foobared zap bar test foo foo"
             assert_equal 1 [r lrem mylist -1 bar]
             assert_equal "$e foo bar foobar foobared zap test foo foo" [r lrange mylist 0 -1]
         }
@@ -874,7 +813,7 @@ start_server {
         }
 
         test "LREM deleting objects that may be int encoded - $type" {
-            create_$type myotherlist "$e 1 2 3"
+            create_list myotherlist "$e 1 2 3"
             assert_equal 1 [r lrem myotherlist 1 2]
             assert_equal 3 [r llen myotherlist]
         }

tests/unit/type/set.tcl

@@ -293,6 +293,13 @@ start_server {
             assert_equal 0 [r scard myset]
         }
 
+        test "SPOP with <count>=1 - $type" {
+            create_set myset $contents
+            assert_encoding $type myset
+            assert_equal $contents [lsort [list [r spop myset 1] [r spop myset 1] [r spop myset 1]]]
+            assert_equal 0 [r scard myset]
+        }
+
         test "SRANDMEMBER - $type" {
             create_set myset $contents
             unset -nocomplain myset
@@ -304,6 +311,68 @@ start_server {
         }
     }
 
+    foreach {type contents} {
+        hashtable {a b c d e f g h i j k l m n o p q r s t u v w x y z} 
+        intset {1 10 11 12 13 14 15 16 17 18 19 2 20 21 22 23 24 25 26 3 4 5 6 7 8 9}
+    } {
+        test "SPOP with <count>" {
+            create_set myset $contents
+            assert_encoding $type myset
+            assert_equal $contents [lsort [concat [r spop myset 11] [r spop myset 9] [r spop myset 0] [r spop myset 4] [r spop myset 1] [r spop myset 0] [r spop myset 1] [r spop myset 0]]]
+            assert_equal 0 [r scard myset]
+        }
+    }
+
+    # As seen in intsetRandomMembers
+    test "SPOP using integers, testing Knuth's and Floyd's algorithm" {
+        create_set myset {1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20}
+        assert_encoding intset myset
+        assert_equal 20 [r scard myset]
+        r spop myset 1
+        assert_equal 19 [r scard myset]
+        r spop myset 2
+        assert_equal 17 [r scard myset]
+        r spop myset 3
+        assert_equal 14 [r scard myset]
+        r spop myset 10
+        assert_equal 4 [r scard myset]
+        r spop myset 10
+        assert_equal 0 [r scard myset]
+        r spop myset 1
+        assert_equal 0 [r scard myset]
+    } {}
+
+    test "SPOP using integers with Knuth's algorithm" {
+        r spop nonexisting_key 100
+    } {}
+
+    test "SPOP new implementation: code path #1" {
+        set content {1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20}
+        create_set myset $content
+        set res [r spop myset 30]
+        assert {[lsort $content] eq [lsort $res]}
+    }
+
+    test "SPOP new implementation: code path #2" {
+        set content {1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20}
+        create_set myset $content
+        set res [r spop myset 2]
+        assert {[llength $res] == 2}
+        assert {[r scard myset] == 18}
+        set union [concat [r smembers myset] $res]
+        assert {[lsort $union] eq [lsort $content]}
+    }
+
+    test "SPOP new implementation: code path #3" {
+        set content {1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20}
+        create_set myset $content
+        set res [r spop myset 18]
+        assert {[llength $res] == 18}
+        assert {[r scard myset] == 2}
+        set union [concat [r smembers myset] $res]
+        assert {[lsort $union] eq [lsort $content]}
+    }
+
     test "SRANDMEMBER with <count> against non existing key" {
         r srandmember nonexisting_key 100
     } {}

utils/cluster_fail_time.tcl

@@ -0,0 +1,50 @@
+# This simple script is used in order to estimate the average PFAIL->FAIL
+# state switch after a failure.
+
+set ::sleep_time 10     ; # How much to sleep to trigger PFAIL.
+set ::fail_port 30016   ; # Node to put in sleep.
+set ::other_port 30001  ; # Node to use to monitor the flag switch.
+
+proc avg vector {
+    set sum 0.0
+    foreach x $vector {
+        set sum [expr {$sum+$x}]
+    }
+    expr {$sum/[llength $vector]}
+}
+
+set samples {}
+while 1 {
+    exec redis-cli -p $::fail_port debug sleep $::sleep_time > /dev/null &
+
+    # Wait for fail? to appear.
+    while 1 {
+        set output [exec redis-cli -p $::other_port cluster nodes]
+        if {[string match {*fail\?*} $output]} break
+        after 100
+    }
+
+    puts "FAIL?"
+    set start [clock milliseconds]
+
+    # Wait for fail? to disappear.
+    while 1 {
+        set output [exec redis-cli -p $::other_port cluster nodes]
+        if {![string match {*fail\?*} $output]} break
+        after 100
+    }
+
+    puts "FAIL"
+    set now [clock milliseconds]
+    set elapsed [expr {$now-$start}]
+    puts $elapsed
+    lappend samples $elapsed
+
+    puts "AVG([llength $samples]): [avg $samples]"
+
+    # Wait for the instance to be available again.
+    exec redis-cli -p $::fail_port ping
+
+    # Wait for the fail flag to be cleared.
+    after 2000
+}

utils/create-cluster/.gitignore

@@ -0,0 +1 @@
+config.sh