Hi On Tue, Aug 3, 2021 at 12:41 PM Daniel P. Berrangé wrote: > On Fri, Jul 23, 2021 at 02:28:22PM +0400, marcandre.lureau@redhat.com > wrote: > > From: Marc-André Lureau > > > > The "file" chardev may call qemu_chr_open_fd() with fd_in=-1. This may > > cause invalid system calls, as the QIOChannel is assumed to be properly > > initialized later on. > > > > Signed-off-by: Marc-André Lureau > > --- > > chardev/char-fd.c | 22 +++++++++++++--------- > > 1 file changed, 13 insertions(+), 9 deletions(-) > > > > diff --git a/chardev/char-fd.c b/chardev/char-fd.c > > index 1cd62f2779..ee85a52c06 100644 > > --- a/chardev/char-fd.c > > +++ b/chardev/char-fd.c > > @@ -133,15 +133,19 @@ void qemu_chr_open_fd(Chardev *chr, > > FDChardev *s = FD_CHARDEV(chr); > > char *name; > > > > - s->ioc_in = QIO_CHANNEL(qio_channel_file_new_fd(fd_in)); > > - name = g_strdup_printf("chardev-file-in-%s", chr->label); > > - qio_channel_set_name(QIO_CHANNEL(s->ioc_in), name); > > - g_free(name); > > - s->ioc_out = QIO_CHANNEL(qio_channel_file_new_fd(fd_out)); > > - name = g_strdup_printf("chardev-file-out-%s", chr->label); > > - qio_channel_set_name(QIO_CHANNEL(s->ioc_out), name); > > - g_free(name); > > - qemu_set_nonblock(fd_out); > > + if (fd_in >= 0) { > > + s->ioc_in = QIO_CHANNEL(qio_channel_file_new_fd(fd_in)); > > + name = g_strdup_printf("chardev-file-in-%s", chr->label); > > + qio_channel_set_name(QIO_CHANNEL(s->ioc_in), name); > > + g_free(name); > > + } > > + if (fd_out >= 0) { > > + s->ioc_out = QIO_CHANNEL(qio_channel_file_new_fd(fd_out)); > > + name = g_strdup_printf("chardev-file-out-%s", chr->label); > > + qio_channel_set_name(QIO_CHANNEL(s->ioc_out), name); > > + g_free(name); > > + qemu_set_nonblock(fd_out); > > + } > > Other code in this file assumes ioc_out is non-NULL, so this looks > like an incomplete fix. > > Right, I am adding a patch to correct the source watch creation, and fixing this one handling !ioc_out condition. thanks -- Marc-André Lureau