| Message ID | 1266857480.3174.10.camel@aglitke |
|---|---|
| State | New |
| Headers | show |
On Mon, 22 Feb 2010 10:51:20 -0600 Adam Litke <agl@us.ibm.com> wrote: > On Fri, 2010-02-19 at 15:47 -0600, Anthony Liguori wrote: > > On 02/12/2010 02:55 PM, Adam Litke wrote: > > > Arghh... Adding missing S-O-B > > > > > > Hi Anthony. I wonder if there was a problem when importing my async > > > command handler patchset. Since the 'balloon' command completes > > > immediately, it must call the completion callback before returning. > > > That call was missing but is added by the patch below. > > > > > > Signed-off-by: Adam Litke<agl@us.ibm.com> > > > > > > > Applied. Thanks. > > This patch application failed. My patch adds a cb() call in > do_balloon(), but the change in git has added the cb() call to > do_info_balloon(). That is causing qemu segfaults. Applying the > following should correct the damage. Thanks. > > Fix for commit: 5c366a8a3d7ac71beda8499caa815cb3ea95eb58 > > The cb() call is needed in do_balloon(), not do_info_balloon(). It won't segfault on QMP, but will print a duplicated response on query-balloon and no success response on balloon. This patch fixes both. One question, though: > @@ -2332,6 +2331,7 @@ static int do_balloon(Monitor *mon, const QDict *params, > return -1; > } > > + cb(opaque, NULL); > return 0; > } Can't (or shouldn't) this be called by common code upon handler completion?
On Fri, 2010-02-26 at 17:26 -0300, Luiz Capitulino wrote: > This patch fixes both. One question, though: > > > @@ -2332,6 +2331,7 @@ static int do_balloon(Monitor *mon, const QDict *params, > > return -1; > > } > > > > + cb(opaque, NULL); > > return 0; > > } > > Can't (or shouldn't) this be called by common code upon handler > completion? The explicit cb() call is only needed for synchronous commands that want to use this API. For asynchronous commands, the top-half (ie. do_info_balloon) returns right away and the callback is called when the asynchronous command completes (in hw/virtio-balloon.c). Since do_balloon() is completely finished when it returns, it has to manually call cb(). If we decide to merge all commands into this new API in the future, then we could have common code call the cb() for all synchronous commands. For now, I just implemented do_balloon() this way as a proof of concept to show that synchronous commands can use the API.
On 02/22/2010 10:51 AM, Adam Litke wrote: > This patch application failed. My patch adds a cb() call in > do_balloon(), but the change in git has added the cb() call to > do_info_balloon(). That is causing qemu segfaults. Applying the > following should correct the damage. Thanks. > > Fix for commit: 5c366a8a3d7ac71beda8499caa815cb3ea95eb58 > > The cb() call is needed in do_balloon(), not do_info_balloon(). > > Signed-off-by: Adam Litke<agl@us.ibm.com> > Applied. Thanks. Regards, Anthony Liguori > diff --git a/monitor.c b/monitor.c > index b1a6edc..c7d2117 100644 > --- a/monitor.c > +++ b/monitor.c > @@ -2309,7 +2309,6 @@ static int do_info_balloon(Monitor *mon, MonitorCompletion cb, void *opaque) > return -1; > } > > - cb(opaque, NULL); > return 0; > } > > @@ -2332,6 +2331,7 @@ static int do_balloon(Monitor *mon, const QDict *params, > return -1; > } > > + cb(opaque, NULL); > return 0; > } > > > > >
diff --git a/monitor.c b/monitor.c index b1a6edc..c7d2117 100644 --- a/monitor.c +++ b/monitor.c @@ -2309,7 +2309,6 @@ static int do_info_balloon(Monitor *mon, MonitorCompletion cb, void *opaque) return -1; } - cb(opaque, NULL); return 0; } @@ -2332,6 +2331,7 @@ static int do_balloon(Monitor *mon, const QDict *params, return -1; } + cb(opaque, NULL); return 0; }