Re: [Qemu-devel] [PATCH] vhost-user: fix watcher need be removed when vhost-user hotplug

[Michael S. Tsirkin]

On Sat, Jul 22, 2017 at 09:24:27AM +0000, Marc-André Lureau wrote:
> 
> 
> On Sat, Jul 22, 2017 at 2:35 AM Michael S. Tsirkin <address@hidden> wrote:
> 
>     On Fri, Jul 21, 2017 at 11:19:04AM +0000, Marc-André Lureau wrote:
>     > Hi
>     >
>     > On Fri, Jul 21, 2017 at 7:18 AM w00273186 <address@hidden> wrote:
>     >
>     >     From: Yunjian Wang <address@hidden>
>     >
>     >     "nc" is freed after hotplug vhost-user, but the watcher don't be
>     removed.
>     >     The QEMU crash when the watcher access the "nc" on socket 
> disconnect.
>     >
>     >
>     >
>     > This is actually your 3rd iteration on the patch
>     >
>     > Could your describe your changes since:
>     > "[PATCH v2] vhost-user: fix watcher need be removed when vhost-user
>     hotplug"
>     >
>     > Thanks
> 
>     Yes but it's a 3-liner. That's way below the limit where you need
>     detailed change history. Does the patch make sense to you?
> 
> 
> 
> That's not all, the fact that he didn't come up with the same solution in the
> first place, and I didn't notice a problem either with the previous approach 
> is
> enough to ask from some clarification on which approach is best, and I bet
> there is something to say.

I'm rather confused.  Looks like you were the one who asked for the change.
Really we want to attract new contributors and a small bugfix like this
seems like a very good way to start contributing. Changelog is already
3 times the size of the patch here. So I think we should just get the patch
reviewed and applied if correct. Do you plan to review it?

> Furthermore, we would really benefit from having repeatable cases for this 
> kind
> of fixes.

I agree disconnect path is but tested adequately but I don't think we
are at a point where we should be asking for testcases for every use
after free bug that gets fixed.

>  
> 
>     >
>     >         Program received signal SIGSEGV, Segmentation fault.
>     >         #0  object_get_class (address@hidden) at qom/object.c:750
>     >         #1  0x00007f9bb4180da1 in qemu_chr_fe_disconnect (be=<optimized
>     out>)
>     >     at chardev/char-fe.c:372
>     >         #2  0x00007f9bb40d1100 in net_vhost_user_watch (chan=<optimized
>     out>,
>     >     cond=<optimized out>, opaque=<optimized out>) at 
> net/vhost-user.c:188
>     >         #3  0x00007f9baf97f99a in g_main_context_dispatch () from /usr/
>     lib64/
>     >     libglib-2.0.so.0
>     >         #4  0x00007f9bb41d7ebc in glib_pollfds_poll () at util/
>     main-loop.c:213
>     >         #5  os_host_main_loop_wait (timeout=<optimized out>) at util/
>     >     main-loop.c:261
>     >         #6  main_loop_wait (address@hidden) at util/
>     >     main-loop.c:515
>     >         #7  0x00007f9bb3e266a7 in main_loop () at vl.c:1917
>     >         #8  main (argc=<optimized out>, argv=<optimized out>, envp=
>     <optimized
>     >     out>) at vl.c:4786
>     >
>     >     Signed-off-by: Yunjian Wang <address@hidden>
>     >     ---
>     >      net/vhost-user.c | 4 ++++
>     >      1 file changed, 4 insertions(+)
>     >
>     >     diff --git a/net/vhost-user.c b/net/vhost-user.c
>     >     index 36f32a2..c23927c 100644
>     >     --- a/net/vhost-user.c
>     >     +++ b/net/vhost-user.c
>     >     @@ -151,6 +151,10 @@ static void vhost_user_cleanup(NetClientState
>     *nc)
>     >              s->vhost_net = NULL;
>     >          }
>     >          if (nc->queue_index == 0) {
>     >     +        if (s->watch) {
>     >     +            g_source_remove(s->watch);
>     >     +            s->watch = 0;
>     >     +        }
>     >              qemu_chr_fe_deinit(&s->chr, true);
>     >          }
>     >
>     >     --
>     >     1.8.3.1
>     >
>     >
>     >
>     >
>     > --
>     > Marc-André Lureau
> 
> --
> Marc-André Lureau


Why do you even bother including the patch if you use a client that
corrupts both the patch and the commit log formatting? It's not a good
example to give to new contributors and it doesn't align well
with nit-picking about same commit log, in my eyes.

-- 
MST