[PATCH] radeon: proper DRI2 pending events handling when client gone. (v3)

[Michel Dänzer]

On Sam, 2010-09-18 at 20:01 +0200, Oldřich Jedlička wrote: 
> On Saturday 18 September 2010 17:29:56 Christopher James Halse Rogers wrote:
> > On Sat, 2010-09-18 at 17:10 +0200, Michel Dänzer wrote:
> > > On Sam, 2010-09-18 at 12:03 +0200, Christopher James Halse Rogers
> > > 
> > > wrote:
> > > > On Tue, 2010-09-14 at 22:55 +0200, Oldřich Jedlička wrote:
> > > > > diff --git a/src/radeon_dri2.c b/src/radeon_dri2.c
> > > > > index ed7fdd6..054b405 100644
> > > > > --- a/src/radeon_dri2.c
> > > > > +++ b/src/radeon_dri2.c
> > > > > @@ -37,6 +37,7 @@
> > > > > 
> > > > >  #include "radeon.h"
> > > > >  #include "radeon_dri2.h"
> > > > >  #include "radeon_version.h"
> > > > > 
> > > > > +#include "list.h"
> > > > 
> > > > This header was only introduced in xserver 1.8.  I presume that it's
> > > > expected that radeon can build against older X servers, so you might
> > > > need to conditionally copy this into the DDX.
> > > 
> > > That shouldn't be necessary (for this), as the whole DRI2 swap
> > > scheduling support was also only added in 1.8.
> > 
> > Ah, yeah.  So the include needs to be conditional on DRI2INFOREC_VERSION
> > >= 4 or just xserver version > 1.8, but other than that is ok.
> 
> I've looked at versions and the list.h was released with xserver 1.7.99.902, 
> but the DRI2INFOREC_VERSION=4 is since 1.7.99.901. So I will use the xorg 
> version, but it will not compile with 1.8 RC1. Is that acceptable?
> 
> That would mean
> 
> #if XORG_VERSION_CURRENT >= XORG_VERSION_NUMERIC(1,7,99,902, 0)                                                                                                                     
> #include "list.h"                                                                                                                                                                   
> #endif                                                                                                                                                                              

You could guard all the code which needs list.h (directly or indirectly)
with the same check. It would mean no swap scheduling support when
building against xserver 1.7.99.901, but I don't think that's a big
deal.


-- 
Earthling Michel Dänzer           |                http://www.vmware.com
Libre software enthusiast         |          Debian, X and DRI developer