[Mesa-dev] MapTextureImage patch series for review

[Eric Anholt]

On Wed, 17 Aug 2011 17:03:56 -0600, Brian Paul <brian.e.paul at gmail.com> wrote:
> On 08/17/2011 04:51 PM, Eric Anholt wrote:
> > On Tue, 16 Aug 2011 18:53:00 -0600, Brian Paul<brianp at vmware.com>  wrote:
> >> On 08/15/2011 12:53 PM, Eric Anholt wrote:
> >>> Here's my "mti-tested" branch of the map-texture-image-v4 work.  It's
> >>> not regressing on intel or softpipe, and in fact fixes a couple of
> >>> tests now on those.  It is not as complete a rework as
> >>> map-texture-image-v4, but it's quite a bit of the change and it it
> >>> should be bisectable (I've had to several times so far to work out the
> >>> regressions in map-texture-image-v4).
> >>
> >> Hmmm, yeah, a lot of the core/swrast refactoring I did isn't there.  I
> >> guess I can redo that on top of your patch series, but it'll take some
> >> time (and I've had virtually no spare time lately).
> >
> > Note that for a bunch of the driver typing for the swrast base-classing,
> > coccinelle was *very* useful for producing clean changes:
> 
> I've never heard of coccinelle so I'll have to read up on that first.

It's a very cool tool with far less documentation than it needs.  It
also has issues with header and defines setup which as far as I can tell
is to improve parser performance.  I forgot to mention that I have this
"mesa.h" file for coccinelle:

#define _GNU_SOURCE
#define PTHREADS
#define DEBUG
#define HAVE_POSIX_MEMALIGN
#define IN_DRI_DRIVER
#define HAVE_ALIAS
#define FEATURE_GL 1
#define FEATURE_ES1 1
#define FEATURE_ES2 1
#define GLX_INDIRECT_RENDERING
#define TEXTURE_FLOAT_ENABLED

#define INLINE inline
#define GLAPIENTRY
#define GLAPIENTRYP *
#define PRINTFLIKE(a, b)

#define ASSERT assert

and I use this script to apply a patch:

for i in `git ls-files src/mesa`; do
        spatch -quiet -I /home/anholt/src/mesa/src/mesa \
                -macro_file spatch/mesa.h \
                -in_place -sp_file $@ $i
done

> I need to do some testing of that yet, but I have a patch that I'll
> post very soon that refactors things a bit and fixes the
> gallium/meta/swrast dependency mess.
> 
> Which piglit tests were showing regressions?

That info is long gone now, but it was fbo-generatemipmaps at least.

> > However, if we can confirm that the format/types stuff for glGetTexImage
> > of compressed is good (we don't have tests for it), I'd be up for making
> > snorm decompression rely on snorm RTT and go ahead with the metaops.  We
> > don't support that snorm RTT in i965 right now because of span functions
> > despair.  I'd like to see a renderbuffer mapping hook like this and
> > avoid spans functions before flailing around more with that.  If
> > everything works out, we're hoping to sort that out before the next
> > release.
> 
> I was planning on tackling the renderbuffer/span mess next.

Great!  I was despairing of the RB rework because I didn't want to be
the one that killed DRI1 drivers, but then I realized that we could
abuse their old spans code to get pixels in and out of a malloced
mapping.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: not available
URL: <http://lists.freedesktop.org/archives/mesa-dev/attachments/20110818/5345c4d2/attachment.pgp>