vnc server closed socket after arrow "down" keyevent

This is a rewrite for https://bugs.launchpad.net/qemu/+bug/1670377

QEMU 2.6 or later
tigervncviwer 1.6  

Once get into grub boot interface(choose boot os, or recovery mode), keep pressing arrow down button for couple times, qemu will close the connection, vnc used zrle mode.

Interesting place:
1. when stopped at grub interface, only arrow up and down key could trigger it, 
2.  only in zrle or tight mode, could work well in raw mode
2. it only triggered by remote connection, not happen if local vncviewer and vnc server


A trace is attached.

Thanks



Find the root reason: qio_channel_write is going to write 290124 data, but only wrote 238644, and for the next write 51489, it returned error, then trigger vnc_client_io_error and disconnect socket. 


ssize_t vnc_client_write_buf(VncState *vs, const uint8_t *data, size_t datalen)
{
    Error *err = NULL;
    ssize_t ret;
    ret = qio_channel_write(
        vs->ioc, (const char *)data, datalen, &err);
    VNC_DEBUG("Wrote wire %p %zd -> %ld\n", data, datalen, ret);
    return vnc_client_io_error(vs, ret, &err);
}

// log file


Write Plain: Pending output 0x5579e6bd2c60 size 524288 offset 290124. Wait SSF 0
32760@1493228975.268871:object_class_dynamic_cast_assert qio-channel-socket->qio-channel (io/channel.c:60:qio_channel_writev_full)
32760@1493228975.268876:object_dynamic_cast_assert qio-channel-socket->qio-channel-socket (io/channel-socket.c:508:qio_channel_socket_writev)
Wrote wire 0x5579e6bd2c60 290124 -> 290124



Write Plain: Pending output 0x5579e6bd2c60 size 524288 offset 290124. Wait SSF 0
32760@1493228975.731842:object_class_dynamic_cast_assert qio-channel-socket->qio-channel (io/channel.c:60:qio_channel_writev_full)
32760@1493228975.731846:object_dynamic_cast_assert qio-channel-socket->qio-channel-socket (io/channel-socket.c:508:qio_channel_socket_writev)
Wrote wire 0x5579e6bd2c60 290124 -> 238644
Write Plain: Pending output 0x5579e6bd2c60 size 65536 offset 51480. Wait SSF 0
32760@1493228975.731934:object_class_dynamic_cast_assert qio-channel-socket->qio-channel (io/channel.c:60:qio_channel_writev_full)
32760@1493228975.731937:object_dynamic_cast_assert qio-channel-socket->qio-channel-socket (io/channel-socket.c:508:qio_channel_socket_writev)
Wrote wire 0x5579e6bd2c60 51480 -> -2
vnc_set_share_mode/0x5579e7b6d730: shared -> disconnected

>  Wrote wire 0x5579e6bd2c60 51480 -> -2

That is QIO_CHANNEL_ERR_BLOCK, aka EAGAIN.

This is fixed in the 2.9.0 release with

commit 537848ee62195fc06c328b1cd64f4218f404a7f1
Author: Michael Tokarev <email address hidden>
Date:   Fri Feb 3 12:52:29 2017 +0300

    vnc: do not disconnect on EAGAIN
    
    When qemu vnc server is trying to send large update to clients,
    there might be a situation when system responds with something
    like EAGAIN, indicating that there's no system memory to send
    that much data (depending on the network speed, client and server
    and what is happening).  In this case, something like this happens
    on qemu side (from strace):
    
    sendmsg(16, {msg_name(0)=NULL,
            msg_iov(1)=[{"\244\"..., 729186}],
            msg_controllen=0, msg_flags=0}, 0) = 103950
    sendmsg(16, {msg_name(0)=NULL,
            msg_iov(1)=[{"lz\346"..., 1559618}],
            msg_controllen=0, msg_flags=0}, 0) = -1 EAGAIN
    sendmsg(-1, {msg_name(0)=NULL,
            msg_iov(1)=[{"lz\346"..., 1559618}],
            msg_controllen=0, msg_flags=0}, 0) = -1 EBADF
    
    qemu closes the socket before the retry, and obviously it gets EBADF
    when trying to send to -1.
    
    This is because there WAS a special handling for EAGAIN, but now it doesn't
    work anymore, after commit 04d2529da27db512dcbd5e99d0e26d333f16efcc, because
    now in all error-like cases we initiate vnc disconnect.
    
    This change were introduced in qemu 2.6, and caused numerous grief for many
    people, resulting in their vnc clients reporting sporadic random disconnects
    from vnc server.
    
    Fix that by doing the disconnect only when necessary, i.e. omitting this
    very case of EAGAIN.
    
    Hopefully the existing condition (comparing with QIO_CHANNEL_ERR_BLOCK)
    is sufficient, as the original code (before the above commit) were
    checking for other errno values too.
    
    Apparently there's another (semi?)bug exist somewhere here, since the
    code tries to write to fd# -1, it probably should check if the connection
    is open before. But this isn't important.
    
    Signed-off-by: Michael Tokarev <email address hidden>
    Reviewed-by: Daniel P. Berrange <email address hidden>
    Message-id: <email address hidden>
    Fixes: 04d2529da27db512dcbd5e99d0e26d333f16efcc
    Cc: Daniel P. Berrange <email address hidden>
    Cc: Gerd Hoffmann <email address hidden>
    Cc: <email address hidden>
    Signed-off-by: Gerd Hoffmann <email address hidden>