Brian Paul [Thu, 9 Jul 2009 14:05:56 +0000 (08:05 -0600)]
Merge branch 'mesa_7_5_branch'
Brian Paul [Thu, 9 Jul 2009 14:04:07 +0000 (08:04 -0600)]
docs: document gl_TextureMatrix[i][j] array indexing bug fix
Brian Paul [Thu, 9 Jul 2009 13:58:50 +0000 (07:58 -0600)]
glsl: do const parameter optimization for array element actual parameters
When a function parameter is const-qualified we can avoid making a copy
of the actual parameter (we basically do a search/replace when inlining).
This is now done for array element params too, resulting in better code
(fewer MOV instructions).
We should allow some other types of function arguments here but let's be
conservative for the moment.
Brian Paul [Thu, 9 Jul 2009 13:57:29 +0000 (07:57 -0600)]
glsl: fix incorrect indexing for gl_TextureMatrix[i][j]
The two indexes were mixed up when accessing a row of a matrix in an array
of matrices.
Brian Paul [Wed, 8 Jul 2009 20:14:03 +0000 (14:14 -0600)]
docs: document glMaterial/glShadeModel display list optimization
Brian Paul [Tue, 5 May 2009 22:21:20 +0000 (16:21 -0600)]
mesa: only use fallback texture when using shaders, not fixed-function (take two)
The semantics are a little different for shaders vs. fixed-function when
trying to use an incomplete texture. The fallback texture returning
(0,0,0,1) should only be used with shaders. For fixed function, the texture
unit is truly disabled/ignored.
Fixes glean fbo test regression.
(cherry picked from commit
01e16befd4809e3e93ee7855a4a5d7df9fe010d9)
(cherry picked from commit
51325f44d1e7e62b47795a79f8038d10dc5bc30b)
[anholt: squashed these two together from master, skipping the mess in between]
Zack Rusin [Tue, 7 Jul 2009 19:48:50 +0000 (12:48 -0700)]
gallium: fixup register indexes for fog/frontface/point coord
Michel Dänzer [Tue, 7 Jul 2009 12:49:52 +0000 (14:49 +0200)]
gallium: Fixes for clobbering stencil values in combined depth/stencil textures.
Also fix one case where a 32 bit depth value was incorrectly converted to a
combined depth/stencil value.
Michel Dänzer [Tue, 7 Jul 2009 11:52:35 +0000 (13:52 +0200)]
GLX/DRI1: Mark GLX visuals with depth != screen depth non-conformant.
Such visuals are subject to automatic compositing in the X server, so DRI1
can't render to them properly.
Michel Dänzer [Tue, 7 Jul 2009 10:17:04 +0000 (12:17 +0200)]
gallium: Only set FRONT_STATUS_COPY_OF_BACK if there is a back buffer.
Fixes potential crash when SwapBuffers is called but there's no back buffer.
Zack Rusin [Tue, 7 Jul 2009 01:33:56 +0000 (21:33 -0400)]
gallium: more fog extraction fixes
fix the cases when fog coord/front face/point coord are used in the same
shader.
Ian Romanick [Mon, 6 Jul 2009 18:36:17 +0000 (11:36 -0700)]
demos: Fix vsync checking in glxgears
Of course glXGetVideoSyncSGI doesn't return the swap interval. The feature
only exists in the Mesa extension... which is the whole reason I created the
Mesa extension! Note that the Mesa extension allows drivers to default to a
swap interval of 0. If the Mesa extension exists, use its value. Only
consider the SGI extension when the Mesa extension is not available.
Fixes bug #22604.
José Fonseca [Mon, 6 Jul 2009 17:23:37 +0000 (18:23 +0100)]
wgl: Make the stw_framebuffer destructions threadsafe.
Ensure no other thread is accessing a framebuffer when it is being destroyed by
acquiring both the global and per-framebuffer mutexes. Normal access only
needs the global lock to walk the linked list and acquire the per-framebuffer
mutex.
José Fonseca [Mon, 6 Jul 2009 16:40:08 +0000 (17:40 +0100)]
wgl: Check for multiple threads before GET_CURRENT_CONTEXT.
Fixes wglthreads -- the 2nd thread MakeCurrent call was trying to flush
the first thread context while still in use.
José Fonseca [Mon, 6 Jul 2009 16:08:37 +0000 (17:08 +0100)]
wgl: Listen to WM_WINDOWPOSCHANGED instead of WM_SIZE messages.
According to
http://blogs.msdn.com/oldnewthing/archive/2008/01/15/
7113860.aspx
WM_SIZE is generated from WM_WINDOWPOSCHANGED by DefWindowProc so it
can be masked out by the application.
Also there were some weird bogus WM_SIZE 0x0 messages when starting
sharedtex_mt which we don't get like this.
Jakob Bornecrantz [Sun, 5 Jul 2009 02:14:41 +0000 (04:14 +0200)]
Merge branch 'mesa_7_5_branch'
Brian Paul [Wed, 8 Jul 2009 19:59:03 +0000 (13:59 -0600)]
demos: indentation fix
Brian Paul [Wed, 8 Jul 2009 19:58:30 +0000 (13:58 -0600)]
demos: use glEnable/DisableClientState() for vertex arrays
RALOVICH, Kristóf [Sun, 5 Jul 2009 14:29:44 +0000 (16:29 +0200)]
glx: death to RCS tag
Corbin Simpson [Wed, 8 Jul 2009 18:59:56 +0000 (11:59 -0700)]
r300g: Remove VAP_CNTL_STATUS from invariant state.
Seriously.
Corbin Simpson [Wed, 8 Jul 2009 18:55:27 +0000 (11:55 -0700)]
r300g: Disable MSPOS registers for glisse's CS security checker.
These will come back in someday, when we can properly use them.
Corbin Simpson [Tue, 19 May 2009 00:06:42 +0000 (17:06 -0700)]
r300-gallium: Unify sampler and texture emit.
They have to cross into each other's registers.
Corbin Simpson [Wed, 8 Jul 2009 18:30:59 +0000 (11:30 -0700)]
r300-gallium: Mipmap setup.
(cherry picked from commit
88c01a15da5639dd68a6a0133724994cb66f1316)
Corbin Simpson [Wed, 8 Jul 2009 18:27:26 +0000 (11:27 -0700)]
r300g: Add endian fix to vertex fetcher setup.
As reported and initially tested by MrCooper.
Corbin Simpson [Wed, 8 Jul 2009 18:26:52 +0000 (11:26 -0700)]
gallium: Add endian detection to p_config.
Maciej Cencora [Sat, 4 Jul 2009 09:33:47 +0000 (11:33 +0200)]
r300: fix regression introduced by
ca13937ef97c7779f639dcfc95b3798a11de01bd
Stride == 0 means that we value for first vertex should be copied to every other vertices (e.g. constant color).
This fixes glean/vertProg1 and sauerbraten with enabled shaders.
Maciej Cencora [Sun, 5 Jul 2009 14:17:12 +0000 (16:17 +0200)]
radeon: fix copy and paste typo
Maciej Cencora [Sun, 5 Jul 2009 13:00:42 +0000 (15:00 +0200)]
mesa/vbo: always recalculate min_index and max_index when splitting vertices
Jakob Bornecrantz [Sat, 4 Jul 2009 14:01:45 +0000 (16:01 +0200)]
progs/tests: Use compressed texture in mipmap_comp_tests
Eric Anholt [Tue, 7 Jul 2009 22:10:21 +0000 (15:10 -0700)]
intel: Fix flipped Y for glDrawPixels(GL_STENCIL_INDEX) to window system.
Even after fixing bugs in this code, it doesn't make me feel any cleaner.
Fixes piglit stencil-drawpixels.
Eric Anholt [Tue, 7 Jul 2009 20:39:29 +0000 (13:39 -0700)]
intel: Fall back on glCopyPixels(GL_DEPTH) or GL_STENCIL.
Eric Anholt [Mon, 6 Jul 2009 23:47:57 +0000 (16:47 -0700)]
i965: Remove BRW_NEW_INPUT_VARYING
This state flag has been unused since the ffvertex_prog move to core.
Eric Anholt [Wed, 25 Feb 2009 19:57:44 +0000 (11:57 -0800)]
Cap array elements at 0 when passed an invalid pointer for an array object.
Otherwise, a pointer greater than the size would underflow and give a large
maximum element.
Reviewed-by: Brian Paul <brianp@vmware.com> (previous version)
Roland Scheidegger [Fri, 26 Jun 2009 18:38:07 +0000 (20:38 +0200)]
i965: fix fetching constants from constant buffer in glsl path
the driver used to overwrite grf0 then use implicit move by send instruction
to move contents of grf0 to mrf1. However, we must not overwrite grf0 since
it's still used later for fb write.
Instead, do the move directly do mrf1 (we could use implicit move from another
grf reg to mrf1 but since we need a mov to encode the data anyway it doesn't
seem to make sense).
I think the dp_READ/WRITE_16 functions may suffer from the same issue.
While here also remove unnecessary msg_reg_nr parameter from the dataport
functions since always message register 1 is used.
Eric Anholt [Thu, 7 May 2009 04:44:13 +0000 (21:44 -0700)]
i965: Remove bad constant buffer constant-reg-already-loaded optimization.
Thanks to branching, the state of c->current_const[i].index at the point
of emitting constant loads for this instruction may not match the actual
constant currently loaded in the reg at runtime. Fixes a regression in my
GLSL program for idr's class since
b58b3a786aa38dcc9d72144c2cc691151e46e3d5.
Jakob Bornecrantz [Sat, 4 Jul 2009 05:43:01 +0000 (07:43 +0200)]
Merge branch 'mesa_7_5_branch'
Jakob Bornecrantz [Sat, 4 Jul 2009 05:40:34 +0000 (07:40 +0200)]
progs/tests: Tests more views in mipmap_comp_tests
Zack Rusin [Tue, 7 Jul 2009 01:43:26 +0000 (21:43 -0400)]
Merge commit 'origin/openvg-1.0'
Zack Rusin [Tue, 23 Jun 2009 23:12:46 +0000 (19:12 -0400)]
gallium: rearrange some members to avoid memory holes/padding
plus it saves us a cacheline in the cso
Zack Rusin [Sun, 21 Jun 2009 01:36:40 +0000 (21:36 -0400)]
pipebuffer: handle possible null pointer dereference
reported by clang static analyzer
Zack Rusin [Sun, 21 Jun 2009 01:25:47 +0000 (21:25 -0400)]
cso: possible null pointer dereference
reported by clang static analyzer
Zack Rusin [Sun, 21 Jun 2009 01:19:57 +0000 (21:19 -0400)]
util: fix possible null pointer usage
found by the clang static analyzer
Zack Rusin [Mon, 8 Jun 2009 04:07:04 +0000 (00:07 -0400)]
exa: some infrastucture work for accelerating composite
Dave Airlie [Mon, 6 Jul 2009 10:35:15 +0000 (20:35 +1000)]
radeon: ensure cmdbuf space for state + AOS is available
The problem is if we find out later we don't have any cmdbuf space but
we've already written the arrays to the DMA buffer object, we end up
emitting the current cmdbuf which has references to the current DMA object
we then send that to the hw and we can't reference the arrays we just emitted
to the old DMA buffer. things go bad, crash boom.
This can probably be tuned further + swtcl probably needs some fixes
Dave Airlie [Mon, 6 Jul 2009 07:46:36 +0000 (17:46 +1000)]
r200: fix makefile
Dave Airlie [Mon, 6 Jul 2009 05:50:24 +0000 (15:50 +1000)]
radeon: fixup FBO depth 24 allocations to avoid assert
Dave Airlie [Mon, 6 Jul 2009 04:15:00 +0000 (14:15 +1000)]
radeon/r200/r300: port to new space checking code in libdrm
This moves a big chunk of the space checking code into libdrm so
it can be shared by the DDX.
Corbin Simpson [Sun, 5 Jul 2009 18:56:40 +0000 (11:56 -0700)]
radeon-gallium: Compile warning fix.
Fixes flush during validation, in case a buffer is double-validated.
Corbin Simpson [Sun, 5 Jul 2009 18:55:47 +0000 (11:55 -0700)]
radeon-gallium: Use FLINK to do proper global buffers.
Corbin Simpson [Sun, 5 Jul 2009 18:29:13 +0000 (11:29 -0700)]
radeon-gallium: Wait for BOs before mapping them.
Maciej Cencora [Mon, 29 Jun 2009 17:50:39 +0000 (19:50 +0200)]
r300: fix vertex limits
- don't limit vertex count if we are using indices
- max indices count is 65535 not 65536
- remove some comments that don't apply anymore
- remove unreachable code
Jakob Bornecrantz [Fri, 3 Jul 2009 16:53:58 +0000 (18:53 +0200)]
Merge branch 'mesa_7_5_branch'
Conflicts:
src/mesa/main/dlist.c
src/mesa/vbo/vbo_save_api.c
Michel Dänzer [Fri, 3 Jul 2009 15:50:15 +0000 (17:50 +0200)]
intel: Also update stencil bits in intel_update_wrapper().
Fixes assertion failure when binding depth/stencil texture to FBO stencil
attachment.
Keith Whitwell [Fri, 3 Jul 2009 15:40:03 +0000 (16:40 +0100)]
progs: revert damage to progs/SConscript from recent compressed texture commit
Keith Whitwell [Fri, 3 Jul 2009 13:26:26 +0000 (14:26 +0100)]
mesa/shaders: fix gl_NormalMatrix state parameters
gl_NormalMatrix is the inverse transpose of the modelview matrix, but
as every matrix here needs to be transposed, we end up with
{MODELVIEW_MATRIX, INVERSE}.
Keith Whitwell [Fri, 3 Jul 2009 09:37:14 +0000 (10:37 +0100)]
st/wgl: don't advertise WGL_EXT_swap_interval string
This is a tweak to a previous fix -- it's not necessary to actually
advertise this extension to prevent these games from crashing -- they
ignore the extension string anyway. It's sufficient to just have
GetProcAddress return some dummy function addresses for SwapInterval.
Given we don't really implement this funcitonality, this is a better
fix.
José Fonseca [Fri, 3 Jul 2009 13:21:55 +0000 (14:21 +0100)]
wgl: Lookup framebuffers by HWND whenever possible.
Some applications create several HDCs for the same window, so spite the WGL
API is geared towards HDCs it is not reliable searching by HDC.
Michel Dänzer [Fri, 3 Jul 2009 10:44:19 +0000 (12:44 +0200)]
radeon: Wait for BO idle if necessary before mapping it.
Fixes fighting between GPU and software rendering with TTM.
Michel Dänzer [Fri, 3 Jul 2009 10:44:02 +0000 (12:44 +0200)]
r300: Guard debugging output.
Eric Anholt [Fri, 3 Jul 2009 02:21:22 +0000 (19:21 -0700)]
intel: Fall back on glBitmap with fog enabled.
We would have to build the program with the appropriate fog mode, and
also supply the fog coordinate if appropriate.
Bug #19413.
Eric Anholt [Fri, 3 Jul 2009 01:02:44 +0000 (18:02 -0700)]
intel: Flush when mapping buffer objects so writes don't get reordered.
While GEM covers this for execution it knows about, it doesn't know about
the batchbuffer we're preparing. Fixes piglit vbo-map-remap.c testcase.
Corbin Simpson [Thu, 2 Jul 2009 23:31:24 +0000 (16:31 -0700)]
vbo: If MapBufferRange fails, try MapBuffer instead.
Fixes segfaults with radeon winsys.
(Probably libdrm_radeon doing something that it shouldn't.)
Corbin Simpson [Thu, 2 Jul 2009 16:21:56 +0000 (09:21 -0700)]
Nuke old trace-drm integration.
AFAICT nobody will miss it.
Eric Anholt [Thu, 2 Jul 2009 20:30:20 +0000 (13:30 -0700)]
intel: Fix leak of DRI option info due to using the wrong free routine.
Eric Anholt [Thu, 2 Jul 2009 20:19:11 +0000 (13:19 -0700)]
intel: Clean up leak of driver context structure on context destroy.
Eric Anholt [Thu, 2 Jul 2009 00:08:16 +0000 (17:08 -0700)]
intel: Init num_fences to clean up valgrind warning.
Valgrind doesn't know that a successful getparam sets the target of the
pointer, so just set the value beforehand.
José Fonseca [Thu, 2 Jul 2009 16:24:25 +0000 (17:24 +0100)]
mesa: Assume depth textures have a single level unless told otherwise.
Keith Whitwell [Thu, 2 Jul 2009 13:43:16 +0000 (14:43 +0100)]
mesa: s/TRUE/GL_TRUE
Fix compile breakage on Linux.
Jakob Bornecrantz [Thu, 2 Jul 2009 13:05:51 +0000 (15:05 +0200)]
r300g: Readd trace driver support
Jakob Bornecrantz [Thu, 2 Jul 2009 13:05:17 +0000 (15:05 +0200)]
i915g: Readd trace driver support
Jakob Bornecrantz [Thu, 2 Jul 2009 12:57:55 +0000 (14:57 +0200)]
trace: Add drm api integration
This is okay since drm_api.h doesn't have any external
dependancies, one could make it only compile on platforms
that support drm.
Keith Whitwell [Thu, 2 Jul 2009 12:28:20 +0000 (13:28 +0100)]
glapi: ensure _mesa_lookup_prim_by_nr() is not clobbered on regeneration
Propogate changes to enums.c back up to the python source.
Keith Whitwell [Thu, 2 Jul 2009 12:29:46 +0000 (13:29 +0100)]
Merge commit 'origin/dlist-statechange-shortcircuit' into mesa_7_5_branch
Conflicts:
progs/trivial/Makefile
Pull in a minimal version of statechange shortcircuiting in display
list compilation. This affects only glMaterial and glShadeModel state,
and includes quite a few tests to exercise various tricky cases.
If this goes well, will consider extending to all state in the future.
Dave Airlie [Thu, 2 Jul 2009 10:57:45 +0000 (20:57 +1000)]
radeon/r200/r300: drop radeon renderbuffer private width/height
half stealing the code without taking the intel regions
Dave Airlie [Thu, 2 Jul 2009 10:44:30 +0000 (20:44 +1000)]
radeon/r300: use base width/height.
I suspect this might break TFP in some way but it makes firecube run here
Keith Whitwell [Thu, 2 Jul 2009 10:28:56 +0000 (11:28 +0100)]
mesa: ensure UsesFogFragCoord value is set for non-glsl shaders
With recent changes to support frontfacing in glsl, it is necessary
to ensure that the UsesFogFragCoord value is accurate in all shaders.
We were previously not setting it for fixed-function and ARB_fs shaders.
Xiang, Haihao [Thu, 2 Jul 2009 08:32:19 +0000 (16:32 +0800)]
i965: fixes for JMPI
1. the data type of <src1> (JMPI offset) must be D
2. execution size must be 1
3. NoMask
4. instruction compression isn't allowed.
Corbin Simpson [Thu, 2 Jul 2009 06:25:47 +0000 (23:25 -0700)]
r300g: Fix recursive Draw flush.
Also just noticed that demos/copypix walks around the overlapping blit rules.
Bad, bad Mesa. :3
Corbin Simpson [Thu, 2 Jul 2009 06:09:18 +0000 (23:09 -0700)]
r300g: Use floats for surface_copy texcoords, use correct src and dest.
This makes demos/copypix better-looking. Horizontal dimensions are right now.
Corbin Simpson [Thu, 2 Jul 2009 05:33:17 +0000 (22:33 -0700)]
radeon-gallium: Adapt to drm_api changes.
Note that trace debugging is temporarily gone. I'll rework it later.
Eric Anholt [Wed, 1 Jul 2009 05:57:56 +0000 (22:57 -0700)]
intel: Avoid pointer arithmetic on void *.
Bug #22000.
Eric Anholt [Tue, 30 Jun 2009 21:26:06 +0000 (14:26 -0700)]
i965: Increase G4X default VS URB allocation to actually allow 32 threads.
This improves the performance of my GLSL demo by 30%. It also fixes the
VS deadlock that ut2004 had, for reasons I can't explain. Bug #21330.
Ian Romanick [Wed, 1 Jul 2009 00:29:14 +0000 (17:29 -0700)]
demos: Set the depth mask correctly to get the desired blending
Ben Skeggs [Tue, 30 Jun 2009 23:15:51 +0000 (09:15 +1000)]
nouveau: return some supported zeta formats
Ben Skeggs [Tue, 30 Jun 2009 22:56:46 +0000 (08:56 +1000)]
nouveau: deal with PIPE_CAP_TGSI_CONT_SUPPORTED
Ben Skeggs [Tue, 30 Jun 2009 22:50:24 +0000 (08:50 +1000)]
nouveau: fix build...
Jakob Bornecrantz [Tue, 30 Jun 2009 23:16:56 +0000 (01:16 +0200)]
drm/st: Remove drm_api struct from drivers
Remove the drm_api from the functions in the softpipe and
i915simple drivers. Create wrapper functions in the
backends instead.
Brian Paul [Tue, 30 Jun 2009 23:12:34 +0000 (17:12 -0600)]
i965: first attempt at handling URB overflow when there's too many vs outputs
If we can't fit all the VS outputs into the MRF, we need to overflow into
temporary GRF registers, then use some MOVs and a second brw_urb_WRITE()
instruction to place the overflow vertex results into the URB.
This is hit when a vertex/fragment shader pair has a large number of varying
variables (12 or more).
There's still something broken here, but it seems close...
Brian Paul [Tue, 30 Jun 2009 23:08:40 +0000 (17:08 -0600)]
i965: use BRW_MAX_MRF
Brian Paul [Tue, 30 Jun 2009 22:36:11 +0000 (16:36 -0600)]
i965: use BRW_MAX_GRF, BRW_MAX_MRF
Brian Paul [Tue, 30 Jun 2009 22:35:23 +0000 (16:35 -0600)]
i965: move BRW_MAX_GRF, define BRW_MAX_MRF
Brian Paul [Tue, 30 Jun 2009 22:33:09 +0000 (16:33 -0600)]
i965: defined BRW_MAX_MRF
Brian Paul [Tue, 30 Jun 2009 21:36:42 +0000 (15:36 -0600)]
i965: comments and a new assertion
Keith Whitwell [Tue, 30 Jun 2009 18:52:44 +0000 (19:52 +0100)]
mesa/vbo: fix compile and replay of nodes ending in a FALLBACK
Where vbo save nodes are terminated with a call to DO_FALLBACK(), as in
the case of a recursive CallList which is itself within a Begin/End pair,
there two problems:
1) The display list node's primitive information was incorrect, stating
the cut-off prim had zero vertices
2) On replay, we would get confused by a primitive that started in a
node, but was terminated by individual opcodes.
This change fixes the first problem by correctly terminating the last
primitive on fallback, and the second by forcing the display list to
use the Loopback path, converting all nodes into immediate-mode rendering.
The loopback fix is a performance hit, but avoiding this would require
a fairly large rework of this code.
Eric Anholt [Tue, 30 Jun 2009 01:48:27 +0000 (18:48 -0700)]
i915: Fix assertion failure on remapping a non-BO-backed VBO.
Failure to set the obj->Pointer back to null tripped up the assertion.
Bug #22428.
Keith Whitwell [Tue, 30 Jun 2009 17:40:20 +0000 (18:40 +0100)]
mesa/dlist: fixes and improvements for material caching
Only short-circuit material call if *all* statechanges from this call
are cached. Some material calls (eg with FRONT_AND_BACK) change more
than one piece of state -- need to check all of them before returning.
Also, Material calls are legal inside begin/end pairs, so don't need
to be as careful about begin/end state as with regular statechanges
(like ShadeModel) when caching. Take advantage of this and do better
caching.
Keith Whitwell [Tue, 30 Jun 2009 17:35:17 +0000 (18:35 +0100)]
progs/trivial: add dlist-mat-tri.c
Keith Whitwell [Tue, 30 Jun 2009 17:17:39 +0000 (18:17 +0100)]
progs/trivial: add test case for short-circuiting material changes
Similar to dlist-tri-flat-tri, but using glMaterial calls, which
have the extra property of being legal within Begin/End calls.
Keith Whitwell [Tue, 30 Jun 2009 16:10:56 +0000 (17:10 +0100)]
mesa/dlist: don't cache state which may not be replayed on CallList
Statechanges which occur before the first End in a display list may
not be replayed when the list is called, in particular if it is called
from within a begin/end pair.
Recognize vulnerable statechanges and do not use them to fill in the
state cache.
Keith Whitwell [Tue, 30 Jun 2009 16:06:50 +0000 (17:06 +0100)]
progs/trivial: add dlist-flat-tri.c
State-change functions which precede the first call to glEnd() in
a compiled list are vulnerable to not being executed when that list
is called.
In particular this can happen if a list is invoked from within a
begin/end pair, as in this example.