[Spice-devel] [PATCH] server/red_worker: death to spice_printerr, too verbose by far

[Alon Levy]

On Sat, Jul 07, 2012 at 03:26:22PM -0500, Jeremy White wrote:
> On 07/06/2012 01:47 PM, Jeremy White wrote:
> > On 07/05/2012 04:38 AM, Alon Levy wrote:
> >> Replaced mostly with spice_debug, but spice_warning & spice_error as
> >> well where appropriate.
> > 
> > Boy, I do like this patch, but oddly, it generates a startup crash for me:
> > 
> > ((null):26170): SpiceWorker-ERROR **:
> > red_worker.c:10627:dev_destroy_primary_surface: double destroy of
> > primary surface
> 
> So this is just the fact that spice_error aborts; we always got a spice_printerr earlier.

abort happens only on critical by default, not on error, so this is
strange - unless you've set SPICE_ABORT_LEVEL_DEFAULT to something else.
See http://cgit.freedesktop.org/spice/spice-common/tree/common/log.c#n84

> 
> That comes from this patch:
> commit 72a0def8114073c0051f3df880f731d3968cb344
> Author: Alon Levy <alevy at redhat.com>
> Date:   Wed May 30 13:44:40 2012 +0300
> 
>     qxl_switch_mode: don't evacuate, just recreate primary surface
>     
> Specifically, we unilaterally destroy the primary on a mode switch,
> and we always perform a mode switch at start.
> 
> Is this fix as simple as moving the qxl_io_destroy_primary(qxl) call
> up into the if (qxl->primary) block?  (That seems to work here, but I don't
> quite understand the code well enough to be sure).

oops, I have it fixed in my randr branch, I'll push a patch for it,
thanks.

> 
> Cheers,
> 
> Jeremy
> _______________________________________________
> Spice-devel mailing list
> Spice-devel at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/spice-devel