X-Git-Url: https://git.saurik.com/redis.git/blobdiff_plain/74a640492fe2547ff6737d598b32896840c2bd1d..40f3ff21b7366defa4ae84ba93f4833a468669a1:/src/networking.c

diff --git a/src/networking.c b/src/networking.c
index 29eed888..edd7891d 100644
--- a/src/networking.c
+++ b/src/networking.c
@@ -14,14 +14,20 @@ redisClient *createClient(int fd) {
     redisClient *c = zmalloc(sizeof(redisClient));
     c->bufpos = 0;
 
-    anetNonBlock(NULL,fd);
-    anetTcpNoDelay(NULL,fd);
-    if (aeCreateFileEvent(server.el,fd,AE_READABLE,
-        readQueryFromClient, c) == AE_ERR)
-    {
-        close(fd);
-        zfree(c);
-        return NULL;
+    /* passing -1 as fd it is possible to create a non connected client.
+     * This is useful since all the Redis commands needs to be executed
+     * in the context of a client. When commands are executed in other
+     * contexts (for instance a Lua script) we need a non connected client. */
+    if (fd != -1) {
+        anetNonBlock(NULL,fd);
+        anetTcpNoDelay(NULL,fd);
+        if (aeCreateFileEvent(server.el,fd,AE_READABLE,
+            readQueryFromClient, c) == AE_ERR)
+        {
+            close(fd);
+            zfree(c);
+            return NULL;
+        }
     }
 
     selectDb(c,0);
@@ -30,6 +36,7 @@ redisClient *createClient(int fd) {
     c->reqtype = 0;
     c->argc = 0;
     c->argv = NULL;
+    c->cmd = NULL;
     c->multibulklen = 0;
     c->bulklen = -1;
     c->sentlen = 0;
@@ -51,7 +58,7 @@ redisClient *createClient(int fd) {
     c->pubsub_patterns = listCreate();
     listSetFreeMethod(c->pubsub_patterns,decrRefCount);
     listSetMatchMethod(c->pubsub_patterns,listMatchObjects);
-    listAddNodeTail(server.clients,c);
+    if (fd != -1) listAddNodeTail(server.clients,c);
     initClientMultiState(c);
     return c;
 }
@@ -59,6 +66,7 @@ redisClient *createClient(int fd) {
 /* Set the event loop to listen for write events on the client's socket.
  * Typically gets called every time a reply is built. */
 int _installWriteEvent(redisClient *c) {
+    if (c->flags & REDIS_LUA_CLIENT) return REDIS_OK;
     if (c->fd <= 0) return REDIS_ERR;
     if (c->bufpos == 0 && listLength(c->reply) == 0 &&
         (c->replstate == REDIS_REPL_NONE ||
@@ -239,10 +247,17 @@ void addReplyError(redisClient *c, char *err) {
 }
 
 void addReplyErrorFormat(redisClient *c, const char *fmt, ...) {
+    size_t l, j;
     va_list ap;
     va_start(ap,fmt);
     sds s = sdscatvprintf(sdsempty(),fmt,ap);
     va_end(ap);
+    /* Make sure there are no newlines in the string, otherwise invalid protocol
+     * is emitted. */
+    l = sdslen(s);
+    for (j = 0; j < l; j++) {
+        if (s[j] == '\r' || s[j] == '\n') s[j] = ' ';
+    }
     _addReplyError(c,s,sdslen(s));
     sdsfree(s);
 }
@@ -397,7 +412,7 @@ static void acceptCommonHandler(int fd) {
      * connection. Note that we create the client instead to check before
      * for this condition, since now the socket is already set in nonblocking
      * mode and we can send an error for free using the Kernel I/O */
-    if (server.maxclients && listLength(server.clients) > server.maxclients) {
+    if (listLength(server.clients) > server.maxclients) {
         char *err = "-ERR max number of clients reached\r\n";
 
         /* That's a best effort error message, don't check write errors */
@@ -419,7 +434,7 @@ void acceptTcpHandler(aeEventLoop *el, int fd, void *privdata, int mask) {
 
     cfd = anetTcpAccept(server.neterr, fd, cip, &cport);
     if (cfd == AE_ERR) {
-        redisLog(REDIS_VERBOSE,"Accepting client connection: %s", server.neterr);
+        redisLog(REDIS_WARNING,"Accepting client connection: %s", server.neterr);
         return;
     }
     redisLog(REDIS_VERBOSE,"Accepted %s:%d", cip, cport);
@@ -434,7 +449,7 @@ void acceptUnixHandler(aeEventLoop *el, int fd, void *privdata, int mask) {
 
     cfd = anetUnixAccept(server.neterr, fd);
     if (cfd == AE_ERR) {
-        redisLog(REDIS_VERBOSE,"Accepting client connection: %s", server.neterr);
+        redisLog(REDIS_WARNING,"Accepting client connection: %s", server.neterr);
         return;
     }
     redisLog(REDIS_VERBOSE,"Accepted connection to %s", server.unixsocket);
@@ -447,6 +462,7 @@ static void freeClientArgv(redisClient *c) {
     for (j = 0; j < c->argc; j++)
         decrRefCount(c->argv[j]);
     c->argc = 0;
+    c->cmd = NULL;
 }
 
 void freeClient(redisClient *c) {
@@ -487,25 +503,6 @@ void freeClient(redisClient *c) {
         redisAssert(ln != NULL);
         listDelNode(server.unblocked_clients,ln);
     }
-    /* Remove from the list of clients waiting for swapped keys, or ready
-     * to be restarted, but not yet woken up again. */
-    if (c->flags & REDIS_IO_WAIT) {
-        redisAssert(server.ds_enabled);
-        if (listLength(c->io_keys) == 0) {
-            ln = listSearchKey(server.io_ready_clients,c);
-
-            /* When this client is waiting to be woken up (REDIS_IO_WAIT),
-             * it should be present in the list io_ready_clients */
-            redisAssert(ln != NULL);
-            listDelNode(server.io_ready_clients,ln);
-        } else {
-            while (listLength(c->io_keys)) {
-                ln = listFirst(c->io_keys);
-                dontWaitForSwappedKey(c,ln->value);
-            }
-        }
-        server.cache_blocked_clients--;
-    }
     listRelease(c->io_keys);
     /* Master/slave cleanup.
      * Case 1: we lost the connection with a slave. */
@@ -522,6 +519,7 @@ void freeClient(redisClient *c) {
     if (c->flags & REDIS_MASTER) {
         server.master = NULL;
         server.replstate = REDIS_REPL_CONNECT;
+        server.repl_down_since = time(NULL);
         /* Since we lost the connection with the master, we should also
          * close the connection with all our slaves if we have any, so
          * when we'll resync with the master the other slaves will sync again
@@ -612,7 +610,7 @@ void sendReplyToClient(aeEventLoop *el, int fd, void *privdata, int mask) {
         }
     }
     if (totwritten > 0) c->lastinteraction = time(NULL);
-    if (listLength(c->reply) == 0) {
+    if (c->bufpos == 0 && listLength(c->reply) == 0) {
         c->sentlen = 0;
         aeDeleteFileEvent(server.el,c->fd,AE_WRITABLE);
 
@@ -627,6 +625,8 @@ void resetClient(redisClient *c) {
     c->reqtype = 0;
     c->multibulklen = 0;
     c->bulklen = -1;
+    /* We clear the ASKING flag as well if we are not inside a MULTI. */
+    if (!(c->flags & REDIS_MULTI)) c->flags &= (~REDIS_ASKING);
 }
 
 void closeTimedoutClients(void) {
@@ -700,70 +700,85 @@ static void setProtocolError(redisClient *c, int pos) {
 
 int processMultibulkBuffer(redisClient *c) {
     char *newline = NULL;
-    char *eptr;
-    int pos = 0, tolerr;
-    long bulklen;
+    int pos = 0, ok;
+    long long ll;
 
     if (c->multibulklen == 0) {
         /* The client should have been reset */
-        redisAssert(c->argc == 0);
+        redisAssertWithInfo(c,NULL,c->argc == 0);
 
         /* Multi bulk length cannot be read without a \r\n */
-        newline = strstr(c->querybuf,"\r\n");
+        newline = strchr(c->querybuf,'\r');
         if (newline == NULL)
             return REDIS_ERR;
 
+        /* Buffer should also contain \n */
+        if (newline-(c->querybuf) > ((signed)sdslen(c->querybuf)-2))
+            return REDIS_ERR;
+
         /* We know for sure there is a whole line since newline != NULL,
          * so go ahead and find out the multi bulk length. */
-        redisAssert(c->querybuf[0] == '*');
-        c->multibulklen = strtol(c->querybuf+1,&eptr,10);
-        pos = (newline-c->querybuf)+2;
-        if (c->multibulklen <= 0) {
-            c->querybuf = sdsrange(c->querybuf,pos,-1);
-            return REDIS_OK;
-        } else if (c->multibulklen > 1024*1024) {
+        redisAssertWithInfo(c,NULL,c->querybuf[0] == '*');
+        ok = string2ll(c->querybuf+1,newline-(c->querybuf+1),&ll);
+        if (!ok || ll > 1024*1024) {
             addReplyError(c,"Protocol error: invalid multibulk length");
             setProtocolError(c,pos);
             return REDIS_ERR;
         }
 
+        pos = (newline-c->querybuf)+2;
+        if (ll <= 0) {
+            c->querybuf = sdsrange(c->querybuf,pos,-1);
+            return REDIS_OK;
+        }
+
+        c->multibulklen = ll;
+
         /* Setup argv array on client structure */
         if (c->argv) zfree(c->argv);
         c->argv = zmalloc(sizeof(robj*)*c->multibulklen);
-
-        /* Search new newline */
-        newline = strstr(c->querybuf+pos,"\r\n");
     }
 
-    redisAssert(c->multibulklen > 0);
+    redisAssertWithInfo(c,NULL,c->multibulklen > 0);
     while(c->multibulklen) {
         /* Read bulk length if unknown */
         if (c->bulklen == -1) {
-            newline = strstr(c->querybuf+pos,"\r\n");
-            if (newline != NULL) {
-                if (c->querybuf[pos] != '$') {
-                    addReplyErrorFormat(c,
-                        "Protocol error: expected '$', got '%c'",
-                        c->querybuf[pos]);
-                    setProtocolError(c,pos);
-                    return REDIS_ERR;
-                }
+            newline = strchr(c->querybuf+pos,'\r');
+            if (newline == NULL)
+                break;
 
-                bulklen = strtol(c->querybuf+pos+1,&eptr,10);
-                tolerr = (eptr[0] != '\r');
-                if (tolerr || bulklen == LONG_MIN || bulklen == LONG_MAX ||
-                    bulklen < 0 || bulklen > 512*1024*1024)
-                {
-                    addReplyError(c,"Protocol error: invalid bulk length");
-                    setProtocolError(c,pos);
-                    return REDIS_ERR;
-                }
-                pos += eptr-(c->querybuf+pos)+2;
-                c->bulklen = bulklen;
-            } else {
-                /* No newline in current buffer, so wait for more data */
+            /* Buffer should also contain \n */
+            if (newline-(c->querybuf) > ((signed)sdslen(c->querybuf)-2))
                 break;
+
+            if (c->querybuf[pos] != '$') {
+                addReplyErrorFormat(c,
+                    "Protocol error: expected '$', got '%c'",
+                    c->querybuf[pos]);
+                setProtocolError(c,pos);
+                return REDIS_ERR;
+            }
+
+            ok = string2ll(c->querybuf+pos+1,newline-(c->querybuf+pos+1),&ll);
+            if (!ok || ll < 0 || ll > 512*1024*1024) {
+                addReplyError(c,"Protocol error: invalid bulk length");
+                setProtocolError(c,pos);
+                return REDIS_ERR;
+            }
+
+            pos += newline-(c->querybuf+pos)+2;
+            if (ll >= REDIS_MBULK_BIG_ARG) {
+                /* If we are going to read a large object from network
+                 * try to make it likely that it will start at c->querybuf
+                 * boundary so that we can optimized object creation
+                 * avoiding a large copy of data. */
+                c->querybuf = sdsrange(c->querybuf,pos,-1);
+                pos = 0;
+                /* Hint the sds library about the amount of bytes this string is
+                 * going to contain. */
+                c->querybuf = sdsMakeRoomFor(c->querybuf,ll+2);
             }
+            c->bulklen = ll;
         }
 
         /* Read bulk argument */
@@ -771,15 +786,32 @@ int processMultibulkBuffer(redisClient *c) {
             /* Not enough data (+2 == trailing \r\n) */
             break;
         } else {
-            c->argv[c->argc++] = createStringObject(c->querybuf+pos,c->bulklen);
-            pos += c->bulklen+2;
+            /* Optimization: if the buffer contanins JUST our bulk element
+             * instead of creating a new object by *copying* the sds we
+             * just use the current sds string. */
+            if (pos == 0 &&
+                c->bulklen >= REDIS_MBULK_BIG_ARG &&
+                (signed) sdslen(c->querybuf) == c->bulklen+2)
+            {
+                c->argv[c->argc++] = createObject(REDIS_STRING,c->querybuf);
+                sdsIncrLen(c->querybuf,-2); /* remove CRLF */
+                c->querybuf = sdsempty();
+                /* Assume that if we saw a fat argument we'll see another one
+                 * likely... */
+                c->querybuf = sdsMakeRoomFor(c->querybuf,c->bulklen+2);
+                pos = 0;
+            } else {
+                c->argv[c->argc++] =
+                    createStringObject(c->querybuf+pos,c->bulklen);
+                pos += c->bulklen+2;
+            }
             c->bulklen = -1;
             c->multibulklen--;
         }
     }
 
     /* Trim to pos */
-    c->querybuf = sdsrange(c->querybuf,pos,-1);
+    if (pos) c->querybuf = sdsrange(c->querybuf,pos,-1);
 
     /* We're done when c->multibulk == 0 */
     if (c->multibulklen == 0) {
@@ -792,7 +824,7 @@ void processInputBuffer(redisClient *c) {
     /* Keep processing while there is something in the input buffer */
     while(sdslen(c->querybuf)) {
         /* Immediately abort if the client is in the middle of something. */
-        if (c->flags & REDIS_BLOCKED || c->flags & REDIS_IO_WAIT) return;
+        if (c->flags & REDIS_BLOCKED) return;
 
         /* REDIS_CLOSE_AFTER_REPLY closes the connection once the reply is
          * written to the client. Make sure to not let the reply grow after
@@ -829,12 +861,29 @@ void processInputBuffer(redisClient *c) {
 
 void readQueryFromClient(aeEventLoop *el, int fd, void *privdata, int mask) {
     redisClient *c = (redisClient*) privdata;
-    char buf[REDIS_IOBUF_LEN];
-    int nread;
+    int nread, readlen;
+    size_t qblen;
     REDIS_NOTUSED(el);
     REDIS_NOTUSED(mask);
 
-    nread = read(fd, buf, REDIS_IOBUF_LEN);
+    readlen = REDIS_IOBUF_LEN;
+    /* If this is a multi bulk request, and we are processing a bulk reply
+     * that is large enough, try to maximize the probabilty that the query
+     * buffer contains excatly the SDS string representing the object, even
+     * at the risk of requring more read(2) calls. This way the function
+     * processMultiBulkBuffer() can avoid copying buffers to create the
+     * Redis Object representing the argument. */
+    if (c->reqtype == REDIS_REQ_MULTIBULK && c->multibulklen && c->bulklen != -1
+        && c->bulklen >= REDIS_MBULK_BIG_ARG)
+    {
+        int remaining = (unsigned)(c->bulklen+2)-sdslen(c->querybuf);
+
+        if (remaining < readlen) readlen = remaining;
+    }
+
+    qblen = sdslen(c->querybuf);
+    c->querybuf = sdsMakeRoomFor(c->querybuf, readlen);
+    nread = read(fd, c->querybuf+qblen, readlen);
     if (nread == -1) {
         if (errno == EAGAIN) {
             nread = 0;
@@ -849,7 +898,7 @@ void readQueryFromClient(aeEventLoop *el, int fd, void *privdata, int mask) {
         return;
     }
     if (nread) {
-        c->querybuf = sdscatlen(c->querybuf,buf,nread);
+        sdsIncrLen(c->querybuf,nread);
         c->lastinteraction = time(NULL);
     } else {
         return;
@@ -902,7 +951,6 @@ void clientCommand(redisClient *c) {
             if (p == flags) *p++ = 'N';
             if (client->flags & REDIS_MULTI) *p++ = 'x';
             if (client->flags & REDIS_BLOCKED) *p++ = 'b';
-            if (client->flags & REDIS_IO_WAIT) *p++ = 'i';
             if (client->flags & REDIS_DIRTY_CAS) *p++ = 'd';
             if (client->flags & REDIS_CLOSE_AFTER_REPLY) *p++ = 'c';
             if (client->flags & REDIS_UNBLOCKED) *p++ = 'u';
@@ -942,3 +990,51 @@ void clientCommand(redisClient *c) {
         addReplyError(c, "Syntax error, try CLIENT (LIST | KILL ip:port)");
     }
 }
+
+/* Rewrite the command vector of the client. All the new objects ref count
+ * is incremented. The old command vector is freed, and the old objects
+ * ref count is decremented. */
+void rewriteClientCommandVector(redisClient *c, int argc, ...) {
+    va_list ap;
+    int j;
+    robj **argv; /* The new argument vector */
+
+    argv = zmalloc(sizeof(robj*)*argc);
+    va_start(ap,argc);
+    for (j = 0; j < argc; j++) {
+        robj *a;
+        
+        a = va_arg(ap, robj*);
+        argv[j] = a;
+        incrRefCount(a);
+    }
+    /* We free the objects in the original vector at the end, so we are
+     * sure that if the same objects are reused in the new vector the
+     * refcount gets incremented before it gets decremented. */
+    for (j = 0; j < c->argc; j++) decrRefCount(c->argv[j]);
+    zfree(c->argv);
+    /* Replace argv and argc with our new versions. */
+    c->argv = argv;
+    c->argc = argc;
+    c->cmd = lookupCommand(c->argv[0]->ptr);
+    redisAssertWithInfo(c,NULL,c->cmd != NULL);
+    va_end(ap);
+}
+
+/* Rewrite a single item in the command vector.
+ * The new val ref count is incremented, and the old decremented. */
+void rewriteClientCommandArgument(redisClient *c, int i, robj *newval) {
+    robj *oldval;
+   
+    redisAssertWithInfo(c,NULL,i < c->argc);
+    oldval = c->argv[i];
+    c->argv[i] = newval;
+    incrRefCount(newval);
+    decrRefCount(oldval);
+
+    /* If this is the command name make sure to fix c->cmd. */
+    if (i == 0) {
+        c->cmd = lookupCommand(c->argv[0]->ptr);
+        redisAssertWithInfo(c,NULL,c->cmd != NULL);
+    }
+}