diff options
| author | Hans de Goede <hdegoede@redhat.com> | 2013-04-25 13:53:02 +0200 |
|---|---|---|
| committer | Anthony Liguori <aliguori@us.ibm.com> | 2013-04-25 14:45:47 -0500 |
| commit | 79f2007578a0a205e25a6f299cc52d6732337089 (patch) | |
| tree | a5f6c8c3d6e8f91a1d9c4bc83fe655662b3537d2 /qemu-char.c | |
| parent | e4dcd8ace175a1a6bfb8945d8ae1906be051b36b (diff) | |
| download | qemu-79f2007578a0a205e25a6f299cc52d6732337089.tar.gz qemu-79f2007578a0a205e25a6f299cc52d6732337089.tar.bz2 qemu-79f2007578a0a205e25a6f299cc52d6732337089.zip | |
qemu-char: Set foo_tag = 0 when returning FALSE from callbacks
While reviewing some patches I found this problem where tcp_chr_accept
does not clear listen_tag when returning FALSE, leading to a double
g_source_remove of the underlying source. Not really a problem unless the id
gets re-used in between, but still something we should fix.
While at it I've also reviewed all the other code in qemu-char.c for
similar problems and found that pty_chr_timer has the same problem.
Cc: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Message-id: 1366890782-10311-1-git-send-email-hdegoede@redhat.com
Signed-off-by: Anthony Liguori <aliguori@us.ibm.com>
Diffstat (limited to 'qemu-char.c')
| -rw-r--r-- | qemu-char.c | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/qemu-char.c b/qemu-char.c index f29f9b1997..64e824d0ac 100644 --- a/qemu-char.c +++ b/qemu-char.c @@ -1077,6 +1077,7 @@ static gboolean pty_chr_timer(gpointer opaque) pty_chr_update_read_handler(chr); out: + s->timer_tag = 0; return FALSE; } @@ -2642,6 +2643,7 @@ static gboolean tcp_chr_accept(GIOChannel *channel, GIOCondition cond, void *opa } fd = qemu_accept(s->listen_fd, addr, &len); if (fd < 0 && errno != EINTR) { + s->listen_tag = 0; return FALSE; } else if (fd >= 0) { if (s->do_telnetopt) |