Zou Nan hai [Thu, 18 Jan 2007 06:41:38 +0000 (14:41 +0800)]
1. Fix bug #155
2. I notice multiple ARB_occlusion_query should be able to overlap according to spec.
3. Declaring extern variables in a .c file is evil, fix it.
Keith Packard [Mon, 8 Jan 2007 07:03:01 +0000 (23:03 -0800)]
Merge branch 'master' into crestline
Keith Packard [Mon, 8 Jan 2007 07:00:48 +0000 (23:00 -0800)]
Merge branch 'origin'
Keith Packard [Mon, 8 Jan 2007 04:57:56 +0000 (20:57 -0800)]
Merge branch 'origin' into crestline
Ben Skeggs [Mon, 8 Jan 2007 00:53:45 +0000 (11:53 +1100)]
nouveau: allow the use of multiple subchannels/objects by default.
Keith Packard [Sun, 7 Jan 2007 01:14:14 +0000 (17:14 -0800)]
Merge branch 'master' into crestline
Keith Packard [Sun, 7 Jan 2007 01:13:45 +0000 (17:13 -0800)]
Various warning fixes for i965 driver.
vertex/fragment programs provided as const.
bmSetFenceLock should return bmSetFence value.
Keith Packard [Sun, 7 Jan 2007 01:13:29 +0000 (17:13 -0800)]
Initialize GL_ARB_occlusion_query only if DRM support is present.
DRM versions before 1.8 do not include the necessary ioctls to support
GL_ARB_occlusion_query, don't enable it on these versions.
Wang Zhenyu [Mon, 4 Dec 2006 07:48:04 +0000 (15:48 +0800)]
Merge branch 'master' into crestline
Conflicts:
src/mesa/drivers/dri/i965/brw_tex_layout.c
Michel Dänzer replaced the copy of the 945 mipmap layout code with that from
the 945 driver directly.
Haihao Xiang [Wed, 27 Dec 2006 00:45:02 +0000 (16:45 -0800)]
i965: Support linear format in i965.
Signed-off-by: Keith Packard <keithp@neko.keithp.com>
Zou Nan hai [Wed, 13 Dec 2006 07:27:17 +0000 (15:27 +0800)]
i965: xdemos/glxthreads get: Assertion `block->fenced' failed (9201)
Signed-off-by: Keith Packard <keithp@neko.keithp.com>
Zou Nan hai [Tue, 12 Dec 2006 07:00:27 +0000 (15:00 +0800)]
i965: Take clip rects into account when computing max prim
Signed-off-by: Keith Packard <keithp@neko.keithp.com>
Wang Zhenyu [Mon, 11 Dec 2006 08:01:56 +0000 (00:01 -0800)]
i965: ARB_occlusion_query support
Signed-off-by: Keith Packard <keithp@neko.keithp.com>
Eric Anholt [Wed, 18 Oct 2006 07:24:01 +0000 (00:24 -0700)]
i965: Avoid branch instructions while in single program flow mode.
There is an errata for Broadwater that threads don't have the instruction/loop
mask stacks initialized on thread spawn. In single program flow mode, those
stacks are not writable, so we can't initialize them. However, they do get
read during ELSE and ENDIF instructions. So, instead, replace branch
instructions in single program flow mode with predicated jumps (ADD to the ip
register), avoiding use of the more complicated branch instructions that may
fail. This is also a minor optimization as no ENDIF equivalent is necessary.
Signed-off-by: Keith Packard <keithp@neko.keithp.com>
Eric Anholt [Wed, 11 Oct 2006 19:16:09 +0000 (12:16 -0700)]
i965: Connect INTEL_DEBUG=sync up to cmd/batch ioctls.
Signed-off-by: Keith Packard <keithp@neko.keithp.com>
Brian [Sat, 6 Jan 2007 21:55:07 +0000 (14:55 -0700)]
Merge branch 'master' of git+ssh://brianp@git.freedesktop.org/git/mesa/mesa
Brian [Sat, 6 Jan 2007 19:58:28 +0000 (12:58 -0700)]
document 32-bit Z line/point fix
Brian [Sat, 6 Jan 2007 19:55:17 +0000 (12:55 -0700)]
Use GLuint instead of GLint to store intermediate Z values. Fixes problems when using 32-bit Z buffer.
Eric Anholt [Sat, 6 Jan 2007 02:19:58 +0000 (18:19 -0800)]
Add reporting of damage by DRI drivers when the extension support is available.
With this, tools like ximagesrc in gstreamer correctly see updates from GL
rendering. Support requires that the Xdamage library be current (but will be
disabled if not present) plus a new X Server with support for the new
XDamagePost request. libGL now has a new interface version, and also links
against libXdamage and libXfixes to support it, but backwards compatibility
is retained.
Currently, all drivers report damage at SwapBuffers time through common code --
front buffer rendering doesn't result in damage being reported. Also, the
damage is against the root window, as our drivers don't yet render to backing
store when they should (composited environments).
Brian [Fri, 5 Jan 2007 15:42:45 +0000 (08:42 -0700)]
Merge branch 'master' of git+ssh://brianp@git.freedesktop.org/git/mesa/mesa
Brian [Fri, 5 Jan 2007 15:41:16 +0000 (08:41 -0700)]
document VBO bug fix 9445
Brian [Fri, 5 Jan 2007 15:40:06 +0000 (08:40 -0700)]
Fix glPush/PopClientAttrib() for VBO state (bug 9445).
Ian Romanick [Fri, 5 Jan 2007 02:37:10 +0000 (18:37 -0800)]
Merge branch 'master' of ssh+git://idr@git.freedesktop.org/git/mesa/mesa
Ian Romanick [Thu, 4 Jan 2007 22:53:48 +0000 (14:53 -0800)]
Use glxbyteorder.h in server-side source files.
Ian Romanick [Thu, 4 Jan 2007 22:52:53 +0000 (14:52 -0800)]
Hack indent flags for indirect_reqsize.h to eliminate extra diffs.
Ian Romanick [Thu, 4 Jan 2007 22:47:41 +0000 (14:47 -0800)]
Add missing files to server generation list.
The files indirect_reqsize.c and indirect_reqsize.h were missing from
the list of files to be generated for the server. Add them back to
the list. Also, update the INDENT_FLAGS to prevent conversion of
'GLbyte *pc' to 'GLbyte * pc' in function prototypes.
Jerome Glisse [Thu, 4 Jan 2007 19:27:49 +0000 (20:27 +0100)]
r300: Correct bug introduced by fragprog rework.
Thx for Tilman who spoted the bugs.
Brian [Thu, 4 Jan 2007 14:53:32 +0000 (07:53 -0700)]
s/unsigned/size_t/ (Matthias Hopf)
Patrice Mandin [Tue, 2 Jan 2007 13:31:16 +0000 (14:31 +0100)]
Update spot light params
Ben Skeggs [Tue, 2 Jan 2007 04:09:10 +0000 (15:09 +1100)]
nouveau: bump drm patchlevel.
Dave Airlie [Sat, 30 Dec 2006 23:01:17 +0000 (10:01 +1100)]
fix issue with i915tex advertising visuals it can't support
This may not be the proper way to fix this but it does work easily.
Haihao Xiang [Sat, 30 Dec 2006 18:30:42 +0000 (10:30 -0800)]
Use the tiled flag in the sarea to determine region tiling.
This fixes mis-rendering if back/depth fail to get set up as tiled. While it
probably won't ever be the case now that the pitch limits are loosened, this is
still the right thing to do.
Ben Skeggs [Wed, 27 Dec 2006 12:52:40 +0000 (23:52 +1100)]
nouveau: Use bufferobj interface for fragment program uploads
Ben Skeggs [Wed, 27 Dec 2006 12:30:34 +0000 (23:30 +1100)]
nouveau: Initial buffer object support
Ben Skeggs [Wed, 27 Dec 2006 04:54:30 +0000 (15:54 +1100)]
nouveau: We'll need syncNotifier for NV_MEMORY_TO_MEMORY_FORMAT too.
Ben Skeggs [Wed, 27 Dec 2006 04:50:59 +0000 (15:50 +1100)]
nouveau: add nouveau_mem_alloc/free debugging
Ben Skeggs [Wed, 27 Dec 2006 04:39:52 +0000 (15:39 +1100)]
nouveau: record *actual* type of memory that was alloc'd, not the requested types.
Ben Skeggs [Wed, 27 Dec 2006 04:38:09 +0000 (15:38 +1100)]
nouveau: Typo
Ben Skeggs [Tue, 26 Dec 2006 13:02:38 +0000 (00:02 +1100)]
nouveau: Add simple wrapper for NV_MEMORY_TO_MEMORY_FORMAT.
Nian Wu [Wed, 27 Dec 2006 00:46:51 +0000 (16:46 -0800)]
Merge branch 'crestline' into crestline-qa
Haihao Xiang [Wed, 27 Dec 2006 00:45:02 +0000 (16:45 -0800)]
Support linear format in i965.
Fix bug #117 #118
Jerome Glisse [Tue, 26 Dec 2006 17:39:21 +0000 (18:39 +0100)]
Deleted unused file which likely have be reintroduced during git move.
This file was deleted longtime ago, guess that git migration
created it again.
Ben Skeggs [Tue, 26 Dec 2006 11:03:12 +0000 (22:03 +1100)]
nouveau: Make use of NOUVEAU_DEBUG for shader disasm
Ben Skeggs [Tue, 26 Dec 2006 10:33:58 +0000 (21:33 +1100)]
nouveau: Make the notifier stuff actually work..
Ben Skeggs [Tue, 26 Dec 2006 10:10:38 +0000 (21:10 +1100)]
nouveau: Wait on notifier to check for completion of previous commands.
We can't wait on NV_PGRAPH_STATUS. We don't have the regs mapped, and there's
no guarantee that we'll catch PGRAPH idle when multiple channels are active.
Ben Skeggs [Tue, 26 Dec 2006 09:59:49 +0000 (20:59 +1100)]
nouveau: Add notifier support functions
Jerome Glisse [Mon, 25 Dec 2006 22:34:56 +0000 (23:34 +0100)]
Rework r300 fragprog avoid using bitfield structure.
It seems that bitfield structure lead to some strange
problem on 64bits arch, don't want to waste time debugging
strange things like that so converted pfs_reg_t structure
to a GLuint and use good old masking and shifting spell.
(cherry picked from
2a7de9d095d8e60da12b11aaa1efe664b87b11d3 commit)
Ben Skeggs [Sat, 23 Dec 2006 13:50:13 +0000 (00:50 +1100)]
nouveau: Kill some compile warnings.
Ben Skeggs [Sat, 23 Dec 2006 13:13:34 +0000 (00:13 +1100)]
nouveau: Modify span routines to use nouveau_renderbuffer instead of driRenderbuffer
Ben Skeggs [Sat, 23 Dec 2006 12:51:24 +0000 (23:51 +1100)]
nouveau: maintain numClipRects/pClipRects in context.
Ben Skeggs [Sat, 23 Dec 2006 12:03:55 +0000 (23:03 +1100)]
nouveau: Don't fill nrb->dPriv for private buffers
Ben Skeggs [Fri, 22 Dec 2006 23:56:19 +0000 (10:56 +1100)]
nouveau: get 16bpp working
George Sapountzis [Wed, 6 Dec 2006 04:40:18 +0000 (06:40 +0200)]
Drop mesa wrappers for XFree86.
Sean D'Epagnier [Thu, 21 Dec 2006 08:50:33 +0000 (01:50 -0700)]
Updated keyboard input so that glut programs can read from stdin without
problems if tty input is used. Also corrected a few stdin keycodes.
Dave Airlie [Tue, 19 Dec 2006 22:30:32 +0000 (09:30 +1100)]
Merge branch 'nouveau-import'
Nian Wu [Sun, 17 Dec 2006 18:49:43 +0000 (10:49 -0800)]
Merge branch 'crestline' into crestline-qa
Ben Skeggs [Sun, 17 Dec 2006 03:38:21 +0000 (03:38 +0000)]
Don't build passthrough shader on <NV40
Brian [Sat, 16 Dec 2006 18:17:41 +0000 (11:17 -0700)]
Don't update span->array->z[] values from SLANG_FRAGMENT_FIXED_FRAGDEPTH.
This restores the behaviour of Mesa 6.5.1, fixing a regression in 6.5.2.
See bug 9345. Revisit someday...
Brian [Sat, 16 Dec 2006 18:09:14 +0000 (11:09 -0700)]
initial release notes for 6.5.3
Brian [Sat, 16 Dec 2006 18:07:00 +0000 (11:07 -0700)]
don't allow fog when using a fragment shader (bug 9346)
Ben Skeggs [Sat, 16 Dec 2006 12:32:11 +0000 (12:32 +0000)]
Get nv10_swtcl.c working enough for glxgears on NV40.
Gary Wong [Fri, 15 Dec 2006 22:18:36 +0000 (17:18 -0500)]
Fix copy-and-paste mistake in comment ("scaling" should read "translation").
Before _mesa_Translatef().
Ben Skeggs [Fri, 15 Dec 2006 21:02:23 +0000 (21:02 +0000)]
NV1x/2x hw_func stubs.
Brian [Thu, 14 Dec 2006 22:20:15 +0000 (15:20 -0700)]
Merge branch 'master' of git+ssh://brianp@git.freedesktop.org/git/mesa/mesa
Brian [Thu, 14 Dec 2006 21:27:05 +0000 (14:27 -0700)]
Replace __extension__ with LONGSTRING.
Brian [Thu, 14 Dec 2006 16:38:06 +0000 (09:38 -0700)]
Check in a long-misplaced spec update.
Michel Dänzer [Thu, 14 Dec 2006 11:57:59 +0000 (12:57 +0100)]
Fix copy-and-paste-o of my e-mail address.
Michel Dänzer [Thu, 14 Dec 2006 11:47:44 +0000 (12:47 +0100)]
Avoid failing assertion in intel_miptree_set_image_offset() with cube maps.
Cube maps still aren't working quite correctly though.
Michel Dänzer [Thu, 14 Dec 2006 11:42:51 +0000 (12:42 +0100)]
Fix some corner cases in i945_miptree_layout_2d().
Based on a patch from Keith Whitwell, with some further fixes.
Michel Dänzer [Thu, 14 Dec 2006 11:32:41 +0000 (12:32 +0100)]
Share code to lay out >= 945 style 2D mipmaps between i915tex and i965 drivers.
Use the i965 version as it has some fixes over the i915tex version.
Michel Dänzer [Thu, 14 Dec 2006 09:56:10 +0000 (10:56 +0100)]
intel_finalize_mipmap_tree: Add more conditions for rebuilding mipmap trees.
These are taken from the i965 driver and fix corruption of some mipmap levels
under some circumsances with 945 chipsets at least.
Also flush the batchbuffer after copying data between trees, or some apps fail
an assertion elsewhere.
Michel Dänzer [Thu, 14 Dec 2006 09:49:26 +0000 (10:49 +0100)]
intel_batchbuffer_flush: Don't assert cliprects when lock is not held.
This is a legitimate situation when copying texture data between mipmap trees.
Michel Dänzer [Thu, 14 Dec 2006 09:24:09 +0000 (10:24 +0100)]
intelTexSubimage: Fix last parameter for intel_miptree_image_map().
Michel Dänzer [Thu, 14 Dec 2006 09:01:43 +0000 (10:01 +0100)]
_mesa_swizzle_ubyte_image: Only use single swizzle_copy call when strides match.
This fixes texture data corruption with glTexSubimage (and probably glTexImage
under some circumstances) with the texstore swizzle path.
Michel Dänzer [Tue, 12 Dec 2006 08:23:53 +0000 (09:23 +0100)]
mipmap_limits: Fix display of current texture filtering mode.
Ben Skeggs [Thu, 14 Dec 2006 04:34:38 +0000 (04:34 +0000)]
Improve SwapBuffers a bit.
Ben Skeggs [Thu, 14 Dec 2006 04:12:05 +0000 (04:12 +0000)]
0x4497 doesn't have NV30_TCL_PRIMITIVE_3D_NORMALIZE_ENABLE
Ben Skeggs [Thu, 14 Dec 2006 03:24:57 +0000 (03:24 +0000)]
Some more voodoo to get 3D going with a minimal initial context.
Roland Scheidegger [Wed, 13 Dec 2006 23:34:44 +0000 (00:34 +0100)]
submit vertex weights to make World of Warcraft maybe happy (bug 8250)
submit the vertex weights to hw, which will enable broken vertex programs
errorneously using them to work. Note however that this will only work
if glWeight is used, there is no code in mesa at all to deal with weight
vertex array (glWeightPointerARB).
Brian [Wed, 13 Dec 2006 22:31:14 +0000 (15:31 -0700)]
Move all the code for computing ctx->_TriangleCaps into state.c.
ctx->_TriangleCaps should probably go away altogether someday...
Nian Wu [Wed, 13 Dec 2006 21:49:00 +0000 (13:49 -0800)]
Merge git://proxy01.pd.intel.com:9419/git/mesa/mesa into crestline
Zou Nan hai [Wed, 13 Dec 2006 07:22:45 +0000 (15:22 +0800)]
Fix bug #93, i965 driver not thread safe.
I am not confident of it is 100% thread safe now.
bufmgr_fake.c need a total rewrite later
(cherry picked from
606632ca27558ee1335be2f4a5906f2baa240a6a commit)
Zou Nan hai [Tue, 12 Dec 2006 06:56:59 +0000 (14:56 +0800)]
fix bug #99.
prim_count overflow when there is more than 1 cliprect
(cherry picked from
84b958d66fe7d3fe03ed12b493e3f3197f656531 commit)
Michel Dänzer [Mon, 11 Dec 2006 16:45:06 +0000 (17:45 +0100)]
minstall: Pass correct destination file path to $RM regardless of source path.
(cherry picked from
26626c0052cd5442f609659aa9ed25ac78e65b55 commit)
Michel Dänzer [Mon, 11 Dec 2006 16:36:35 +0000 (17:36 +0100)]
minstall: Always remove destination file before (re-)creating it.
This avoids issues with overwriting files that are being used.
(cherry picked from
d71a5647a3ed4aadd46edfa8a031ffc87d88c5f7 commit)
Zou Nan hai [Wed, 13 Dec 2006 21:25:12 +0000 (13:25 -0800)]
ARB_occlusion_query support
Zou Nan hai [Wed, 13 Dec 2006 21:24:35 +0000 (13:24 -0800)]
if (tex width < 4), mipmap calculation will be out of range
Brian [Wed, 13 Dec 2006 15:57:06 +0000 (08:57 -0700)]
Remove the xdemo.c program from PROGS since it doesn't work with GLX/DRI.
Brian [Wed, 13 Dec 2006 15:30:26 +0000 (08:30 -0700)]
Use XDisplayName() when reporting errors (bug 8079).
Zou Nan hai [Wed, 13 Dec 2006 07:27:17 +0000 (15:27 +0800)]
Fix bug #93
George Sapountzis [Thu, 7 Dec 2006 16:01:13 +0000 (18:01 +0200)]
Bug 7260: mach64 texture memory mng cleanup
mach64 uses its own set of texture memory management routines which are buggy,
running a second DRI client kills the first one. This patch ports mach64 code
to the stock dri texture managment code.
George Sapountzis [Thu, 7 Dec 2006 15:56:47 +0000 (17:56 +0200)]
Bug 7861: mach64 with render acceleration should restore texture state
RENDER acceleration uses texturing, thus when RENDER acceleration is enabled,
the mach64 DRI driver should restore texture state when acquiring the DRI lock.
George Sapountzis [Thu, 7 Dec 2006 15:53:59 +0000 (17:53 +0200)]
Bug 7790: Polygons incorrectly clipped by mach64 driver
un-break strict-aliasing rules
Zou Nan hai [Tue, 12 Dec 2006 07:00:27 +0000 (15:00 +0800)]
fix for bug #99
Eric Anholt [Mon, 11 Dec 2006 18:50:25 +0000 (10:50 -0800)]
Merge branch 'origin' into crestline
Michel Dänzer [Mon, 11 Dec 2006 16:45:06 +0000 (17:45 +0100)]
minstall: Pass correct destination file path to $RM regardless of source path.
Michel Dänzer [Mon, 11 Dec 2006 16:36:35 +0000 (17:36 +0100)]
minstall: Always remove destination file before (re-)creating it.
This avoids issues with overwriting files that are being used.
Wang Zhenyu [Mon, 11 Dec 2006 08:01:56 +0000 (00:01 -0800)]
ARB_occlusion_query support
Wang Zhenyu [Mon, 11 Dec 2006 08:00:51 +0000 (00:00 -0800)]
if (tex width < 4), mipmap calculation will be out of range
Eric Anholt [Wed, 18 Oct 2006 07:24:01 +0000 (00:24 -0700)]
Avoid branch instructions while in single program flow mode.
There is an errata for Broadwater that threads don't have the instruction/loop
mask stacks initialized on thread spawn. In single program flow mode, those
stacks are not writable, so we can't initialize them. However, they do get
read during ELSE and ENDIF instructions. So, instead, replace branch
instructions in single program flow mode with predicated jumps (ADD to the ip
register), avoiding use of the more complicated branch instructions that may
fail. This is also a minor optimization as no ENDIF equivalent is necessary.