postgresql socket读数据返回-1

如下所示：

    {
        n = secure_raw_read(port, ptr, len);   // pg的socket读是非阻塞读，所以返回-1不影响，后面等到socket可读之后继续读。见下文socket返回值解释
        waitfor = WL_SOCKET_READABLE;
    }

    /* In blocking mode, wait until the socket is ready */
    if (n < 0 && !port->noblock && (errno == EWOULDBLOCK || errno == EAGAIN))   // 非阻塞读，且errono==EAGAIN，所以就是每次处理完一个SQL语句后，会继续读下一个SQL，发现读不到，进到这里，通过latch设置等待客户端可读事件
    {
        WaitEvent    event;

        Assert(waitfor);

        ModifyWaitEvent(FeBeWaitSet, 0, waitfor, NULL);

        WaitEventSetWait(FeBeWaitSet, -1 /* no timeout */ , &event, 1,
                         WAIT_EVENT_CLIENT_READ);

        /*
         * If the postmaster has died, it's not safe to continue running,
         * because it is the postmaster's job to kill us if some other backend
         * exits uncleanly.  Moreover, we won't run very well in this state;
         * helper processes like walwriter and the bgwriter will exit, so
         * performance may be poor.  Finally, if we don't exit, pg_ctl will be
         * unable to restart the postmaster without manual intervention, so no
         * new connections can be accepted.  Exiting clears the deck for a
         * postmaster restart.
         *
         * (Note that we only make this check when we would otherwise sleep on
         * our latch.  We might still continue running for a while if the
         * postmaster is killed in mid-query, or even through multiple queries
         * if we never have to wait for read.  We don't want to burn too many
         * cycles checking for this very rare condition, and this should cause
         * us to exit quickly in most cases.)
         */
        if (event.events & WL_POSTMASTER_DEATH)
            ereport(FATAL,
                    (errcode(ERRCODE_ADMIN_SHUTDOWN),
                     errmsg("terminating connection due to unexpected postmaster exit")));

        /* Handle interrupt. */
        if (event.events & WL_LATCH_SET)
        {
            ResetLatch(MyLatch);
            ProcessClientReadInterrupt(true);

            /*
             * We'll retry the read. Most likely it will return immediately
             * because there's still no data available, and we'll wait for the
             * socket to become ready again.
             */
        }
        goto retry;
    }

在底层，linux内核 4.x平台中，等待socket可读是通过epoll_wait实现的。

#0  0x00007fa94fd880bb in epoll_wait () from /lib64/libc.so.6
#1  0x00000000007916de in WaitEventSetWaitBlock (nevents=1, occurred_events=0x7fffb042fd40, cur_timeout=-1, set=0x2bf7e58) at latch.c:1295
#2  WaitEventSetWait (set=0x2bf7e58, timeout=timeout@entry=-1, occurred_events=occurred_events@entry=0x7fffb042fd40, nevents=nevents@entry=1, wait_event_info=wait_event_info@entry=100663296) at latch.c:1247
#3  0x0000000000688233 in secure_read (port=0x2c5b970, ptr=0xdc7500 <PqRecvBuffer>, len=8192) at be-secure.c:184
#4  0x000000000068eb7b in pq_recvbuf () at pqcomm.c:947
#5  pq_recvbuf () at pqcomm.c:923
#6  0x000000000068f985 in pq_getbyte () at pqcomm.c:990
#7  0x00000000007b531e in SocketBackend (inBuf=0x7fffb042ff40) at postgres.c:357
#8  ReadCommand (inBuf=0x7fffb042ff40) at postgres.c:530
#9  PostgresMain (argc=<optimized out>, argv=argv@entry=0x2c64490, dbname=<optimized out>, username=<optimized out>) at postgres.c:4598
#10 0x000000000073549d in BackendRun (port=0x2c5b970, port=0x2c5b970) at postmaster.c:5063

3.x内核版本中我记得是poll而非poll。

socket读写返回不同值的总结

在调用socket读写函数read(),write()时，都会有返回值。如果没有正确处理返回值，就可能引入一些问题

1当read()或者write()函数返回值大于0时，表示实际从缓冲区读取或者写入的字节数目

2当read()函数返回值为0时，表示对端已经关闭了 socket，这时候也要关闭这个socket，否则会导致socket泄露。netstat命令查看下，如果有closewait状态的socket,就是socket泄露了

当write()函数返回0时，表示当前写缓冲区已满，是正常情况，下次再来写就行了。

3当read()或者write()返回-1时，一般要判断errno

如果errno == EINTR,表示系统当前中断了，直接忽略

如果errno == EAGAIN或者EWOULDBLOCK，非阻塞socket直接忽略;如果是阻塞的socket,一般是读写操作超时了，还未返回。这个超时是指socket的SO_RCVTIMEO与SO_SNDTIMEO两个属性。所以在使用阻塞socket时，不要将超时时间设置的过小。不然返回了-1，你也不知道是socket连接是真的断开了，还是正常的网络抖动。一般情况下，阻塞的socket返回了-1，都需要关闭重新连接。

4.另外，对于非阻塞的connect,可能返回-1.这时需要判断errno，如果 errno == EINPROGRESS，表示正在处理中，否则表示连接出错了，需要关闭重连。之后使用select，检测到该socket的可写事件时，要判断getsockopt(c->fd, SOL_SOCKET, SO_ERROR, &err, &errlen)，看socket是否出错了。如果err值为0,则表示connect成功；否则也应该关闭重连

5 在使用epoll时，有ET与LT两种模式。ET模式下，socket需要read或者write到返回-1为止。对于非阻塞的socket没有问题，但是如果是阻塞的socket，正如第三条中所说的，只有超时才会返回。所以在ET模式下千万不要使用阻塞的socket。那么LT模式为什么没问题呢？一般情况下，使用LT模式，我们只要调用一次read或者write函数，如果没有读完或者没有写完，下次再来就是了。由于已经返回了可读或者可写事件，所以可以保证调用一次read或者write会正常返回。

注：除了read/write外，send/recv也可以读写socket，差别是多了个选项，但是在pg中用的是send/recv，第四个选项标志位均没有使用。第四个标志位的含义可以参考https://www.cnblogs.com/haichun/p/3519232.html。