[PULL] sigsafe logging and other fixes

Peter Hutterer peter.hutterer at who-t.net
Wed Aug 22 22:09:53 PDT 2012


On Wed, Aug 22, 2012 at 01:52:16PM -0700, Keith Packard wrote:
> Peter Hutterer <peter.hutterer at who-t.net> writes:
> 
> > Two unreviewed: the win32 fsync ifdef typo, and the fix for 53568, which
> > currently crashes on the first pointer event from an Option Floating device.
> 
> I've reviewed those now.
> 
> >       mi: don't check for core events in miPointerSetPosition (#53568)
> 
> This seems very much like a 'how could that ever have been the right
> idea' kind of change to me -- setting the pointer position for a device
> shouldn't ever have depended on whether that device was delivering core
> events or not. In any case, the change seems obviously correct to me

this is largely historical, for a while coreEvents was the distinction
between "has cursor" and "doesn't have cursor". That changed in late 2007,
but stuff like this just lingers areound.

> Reviewed-by: Keith Packard <keithp at keithp.com>
> 
> However, this still looks very fragile -- fill_pointer_events does not
> check the return value from miPointerSetPosition, which is still allowed
> to return NULL in some cases. I don't think any of those are valid
> though, so I'd suggest that either miPointerSetPosition or
> positionSprite throw a fatal error instead of blindly wandering into a
> segfault.

I've sent out a patch to remove the two conditions, afaict only the second
one could even be triggered and even that would be a severe bug elsewhere.

thanks for the reviews, I'll update the pull request

Cheers,
   Peter


> 
> >       os: fix typo, fsync when WIN32 is _not_ defined
> 
> Reviewed-by: Keith Packard <keithp at keithp.com>
> 
> -- 
> keith.packard at intel.com




More information about the xorg-devel mailing list