From: Dave Airlie
The API can hit this, as the ogl confirm suite showed, I've
sent a piglit test to also demonstrate hitting it.
Signed-off-by: Dave Airlie
---
src/mesa/state_tracker/st_cb_xformfb.c | 1 -
1 file changed, 1 deletion(-)
diff --git a/src/mesa/state_tracker/st_cb_xformfb
I ran GL41-CTS.gtf32.* from ogl conform, and fell over a lot
in the unreachable in glformats.c
Looks like a bunch of stuff is missing, now I'm not sure
how much is required to add it all, the first patch is probably
okay, the second is just a hint on if I need to flesh out some
types for the _UIN
From: Dave Airlie
these also hit the problem case but I don't think this patch
is the answer,
do I need to flesh out the .8.8.8.8_UNORM ones into .8.8.8.8_UINT
types?
Signed-off-by: Dave Airlie
---
src/mesa/main/glformats.c | 8
1 file changed, 8 insertions(+)
diff --git a/src
From: Dave Airlie
GL3.3 added GL_ARB_texture_rgb10_a2ui, which specifies
a lot more things than just rgb10/a2ui.
While playing with ogl conform one of the tests must
attempted all valid formats for GL3.3 and hits the
unreachable here.
This adds the first chunk of formats that hit the
assert
if we get a request to take the count from feedback, but there
is no buffer to take it from, just draw nothing instead of asserting.
This fixes this assert killing the ogl conform, and a piglit
test I've sent.
Signed-off-by: Dave Airlie
---
src/mesa/state_tracker/st_cb_xformfb.
On 31 July 2015 at 00:26, Marek Olšák wrote:
> From: Marek Olšák
once you rebase onto my last patch, these look fine.
Reviewed-by: Dave Airlie
>
> ---
> src/mesa/state_tracker/st_cb_xformfb.c | 58
> ++
> src/mesa/state_tracker/st_cb_xformf
seems like one I wrote already..
http://lists.freedesktop.org/archives/mesa-dev/2015-July/088404.html
Dave.
On 31 July 2015 at 12:42, Edward O'Callaghan
wrote:
> Don't trigger assert on some stream emit instructions.
>
> Signed-off-by: Edward O'Callaghan
> ---
> src/gallium/drivers/r600/sb/sb_
, I started on that but I have a lot of regressions left
to iron out.
Dave.
>
> Cheers,
> Edward.
>
> --
> Edward O'Callaghan
> edward.ocallag...@koparo.com
>
> On Fri, Jul 31, 2015, at 12:47 PM, Dave Airlie wrote:
> > seems like one I wrote already.
For the series,
Reviewed-by: Dave Airlie
On 2 August 2015 at 23:36, Marek Olšák wrote:
> From: Marek Olšák
>
> No effect, but this is what we should be doing.
> ---
> src/gallium/drivers/r600/evergreen_state.c | 8 ++--
> 1 file changed, 6 insertions(+), 2 deletions(-
On 14 August 2015 at 20:27, Lofstedt, Marta wrote:
>> -Original Message-
>> From: mesa-dev [mailto:mesa-dev-boun...@lists.freedesktop.org] On
>> Behalf Of Tapani Pälli
>> Sent: Friday, August 14, 2015 7:31 AM
>> To: Marta Lofstedt; mesa-dev@lists.freedesktop.org
>> Subject: Re: [Mesa-dev]
This brings in the new ARB extensions.
Signed-off-by: Dave Airlie
---
include/GL/glext.h | 294 +++--
1 file changed, 285 insertions(+), 9 deletions(-)
diff --git a/include/GL/glext.h b/include/GL/glext.h
index e5f1d89..907a582 100644
--- a
From: Dave Airlie
It appears CTS does a lot of
glGetFrameBufferAttachmentParameteriv with GL_FRONT_LEFT,
but gallium doesn't create a GL_FRONT_LEFT attachment,
so these all fail badly, now I'd like to argue the test suite
should probably only hit GL_BACK_LEFT but I'm guessin
e same. This would be a mesa level change, and a bit
of a layering
violation I think.
Dave.
>
> Marek
>
> On Mon, Aug 17, 2015 at 5:23 AM, Dave Airlie wrote:
>> From: Dave Airlie
>>
>> It appears CTS does a lot of
>>
>> glGetFrameBufferAttachmentPara
Hey,
while running CTS under valgrind I got to see a lot of
==32256== Invalid read of size 2
==32256==at 0x5B53F07: convert_ushort (format_utils.c:1155)
==32256==by 0x5B8523A: _mesa_swizzle_and_convert (format_utils.c:1453)
==32256==by 0x5B11151: _mesa_format_convert (format_utils.c:3
On 19 August 2015 at 18:58, Edward O'Callaghan
wrote:
> From: Edward O'Callaghan
>
> This struct was getting a bit crowded, following the lead of
> radeonsi, mirror the idea of having sub-structures for each
> shader type. Turning 'r600_shader_key' into an union saves
> some trivial memory and CP
> Shader-db results for vec4 on i965:
reads like a Johnny Cash cover of NIN.
sorry couldn't resist :-P
Dave.
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
> Can anyone take a look at this simple one-liner ? Afaics Dave added
> the pass earlier in 10.7-devel.
Looks good, pushed it, thanks.
Dave.
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
http://lists.freedesktop.org/mailman/listinfo/mesa-dev
shader scan has already happened so just the results,
also nobody uses gs_input_prim so drop it.
Signed-off-by: Dave Airlie
---
src/gallium/drivers/r600/r600_shader.c | 38 +-
src/gallium/drivers/r600/r600_shader.h | 1 -
2 files changed, 10 insertions(+), 29
I sent this a while ago,
mesa: enable texture stencil8 for multisample
and I forgot to push it, pushed now with stable tags.
Dave.
On 25 August 2015 at 09:59, Marek Olšák wrote:
> From: Marek Olšák
>
> Cc: 10.6 11.0
> ---
> src/mesa/main/teximage.c | 4 +++-
> 1 file changed, 3 insertions(+
On 25 August 2015 at 09:59, Marek Olšák wrote:
> From: Marek Olšák
>
> This works if drivers upsample on upload (like all radeon ones do).
> The alternative is an unexpected GL error from anything calling
> _mesa_update_state and possibly other issues.
>
Reviewed-by: Dave A
From: Dave Airlie
GetTexImage can read to stencil8 but only from
a stencil or depthstencil textures.
This fixes a bunch of failures in CTS
GL33-CTS.gtf32.GL3Tests.packed_pixels
Signed-off-by: Dave Airlie
---
src/mesa/main/texgetimage.c | 7 +++
1 file changed, 7 insertions(+)
diff --git
This adds multiple stream support for ARB_gpu_shader5, and one
other patch.
It doesn't expose ARB_gpu_shader5 yet, as I think we'd like
to try and get SB support for it into some sort of shape first.
Dave.
___
mesa-dev mailing list
mesa-dev@lists.freede
From: Dave Airlie
This adds a peephole and removes an assert that isn't
actually valid with some of the stream emit instructions.
Signed-off-by: Dave Airlie
---
src/gallium/drivers/r600/sb/sb_bc_finalize.cpp | 2 --
src/gallium/drivers/r600/sb/sb_bc_parser.cpp | 18 +++-
From: Dave Airlie
This just aids debugging.
Signed-off-by: Dave Airlie
---
src/gallium/drivers/r600/sb/sb_bc_dump.cpp | 4
1 file changed, 4 insertions(+)
diff --git a/src/gallium/drivers/r600/sb/sb_bc_dump.cpp
b/src/gallium/drivers/r600/sb/sb_bc_dump.cpp
index 631fac2..0fc73c4 100644
From: Dave Airlie
This just adds support to the assembler dumper and allows
stream instructions to be generated. Also fix up the stream
debugging to add stream info.
Signed-off-by: Dave Airlie
---
src/gallium/drivers/r600/eg_asm.c | 1 +
src/gallium/drivers/r600/r600_asm.c| 2
From: Glenn Kennard
This patch is taken from work by Glenn and myself,
and I've spent some time making it all work here.
This adds support for the multiple streams part of
ARB_gpu_shader5 to r600g.
It doesn't enable ARB_gpu_shader5 yet.
Signed-off-by: Dave Airlie
---
src/galli
> case MESA_FORMAT_B8G8R8X8_SRGB:
> case MESA_FORMAT_X8R8G8B8_SRGB:
>return GL_FALSE;
> + default:
> + assert(_mesa_is_format_compressed(format));
This assert is clearly wrong, _mesa_is_format_compressed should take a
mesa_format not a format.
Dave.
From: Dave Airlie
commit 26c549e69d12e44e2e36c09764ce2cceab262a1b
Author: Nanley Chery
Date: Fri Jul 31 10:26:36 2015 -0700
mesa/formats: remove compressed formats from matching function
caused a regression in my CTS testing, this looks like a clear
thinko.
sSigned-off-by: Dave Airlie
From: Dave Airlie
Same as for other paths, need to take into a/c rowlength
vs width differences for 7/RED/UNSIGNED_BYTE cases.
Signed-off-by: Dave Airlie
---
src/mesa/main/readpix.c | 16
1 file changed, 12 insertions(+), 4 deletions(-)
diff --git a/src/mesa/main/readpix.c b
This code wasn't handling the 7-width/GL_RED/GL_BYTE case where
GL_PACK_ALIGNMENT was set to 4. This fixes it to iterate
on a row by row basis over the image instead, and avoids
accessing uninitialised memory.
Signed-off-by: Dave Airlie
---
src/mesa/main/texgetimage.c | 15 +++--
From: Dave Airlie
This code doesn't handle the the user setting
GL_UNPACK_ALIGNMENT at all well since we have
the cases where 7 byte wide (GL_RED/GL_UNSIGNED_BYTE)
still has it's rows aligned to 8, and the old
code failed in that case.
Just iterate the swaps over rows and images
to h
The glcts has two tests
GL33-CTS.gtf32.GL3Tests.packed_pixels.packed_pixels
and
GL33-CTS.gtf32.GL3Tests.packed_pixels.packed_pixels_pixelstore
that try all the valid combinations of things at TexImage, GetTexImage
and ReadPixels.
This series of patches is a first round of fixes:
a) SwapBytes ha
From: Dave Airlie
GL3.3 added GL_ARB_texture_rgb10_a2ui, which specifies
a lot more things than just rgb10/a2ui.
While playing with ogl conform one of the tests must
attempted all valid formats for GL3.3 and hits the
unreachable here.
This adds the first chunk of formats that hit the
assert
From: Dave Airlie
Add enough 8-bit channel formats to handle all the
different things CTS throws at us.
Signed-off-by: Dave Airlie
---
src/mesa/main/formats.c | 43 +++
src/mesa/main/formats.csv| 4
src/mesa/main/formats.h | 5
This case just wasn't handled, so add support for row by
row swapping when we've decompressed the texture.
Signed-off-by: Dave Airlie
---
src/mesa/main/texgetimage.c | 19 +++
1 file changed, 19 insertions(+)
diff --git a/src/mesa/main/texgetimage.c b/src
From: Dave Airlie
GL33-CTS.shaders.preprocessor.definitions.*
has 4 tests the undefine these,
I can't find anything in the spec saying that isn't correct.
Signed-off-by: Dave Airlie
---
src/glsl/glcpp/glcpp-parse.y | 4 +---
1 file changed, 1 insertion(+), 3 deletions(-)
diff -
From: Dave Airlie
GL33-CTS.shaders.preprocessor.definitions.*
has 4 tests the undefine these,
It appears GLES doesn't allow this, but desktop GL
does, yay specs.
Signed-off-by: Dave Airlie
---
src/glsl/glcpp/glcpp-parse.y | 9 +
1 file changed, 5 insertions(+), 4 deletions(-)
On 26 August 2015 at 16:48, Ilia Mirkin wrote:
> On Wed, Aug 26, 2015 at 2:38 AM, Dave Airlie wrote:
>> From: Dave Airlie
>>
>> GL33-CTS.shaders.preprocessor.definitions.*
>> has 4 tests the undefine these,
>>
>> I can't find anything in the spec sa
On 26 August 2015 at 16:55, Dave Airlie wrote:
> On 26 August 2015 at 16:48, Ilia Mirkin wrote:
>> On Wed, Aug 26, 2015 at 2:38 AM, Dave Airlie wrote:
>>> From: Dave Airlie
>>>
>>> GL33-CTS.shaders.preprocessor.definitions.*
>>> has 4 tests the und
From: Dave Airlie
Instead of looking this up lots, lets just cache it in the instruction
translation up front. I just noticed this function what high in a profile
of shader-db on radeonsi.
Signed-off-by: Dave Airlie
---
src/mesa/state_tracker/st_glsl_to_tgsi.cpp | 39
I profiled shader-db running on radeonsi, and the temp register
handling passes in the glsl->tgsi convertor were quite heavy
users of CPU.
These two passes are my attempt at cleaning them up to avoid
looping over the instruction set so much. I'm in the process
of piglitting then,
On my CPU, a sha
From: Dave Airlie
This can be done with a single pass for the instruction base,
and takes renumber_registers out of its spot on the profile.
Signed-off-by: Dave Airlie
---
src/mesa/state_tracker/st_glsl_to_tgsi.cpp | 57 --
1 file changed, 38 insertions(+), 19
On 27 August 2015 at 11:29, Dave Airlie wrote:
> I profiled shader-db running on radeonsi, and the temp register
> handling passes in the glsl->tgsi convertor were quite heavy
> users of CPU.
>
> These two passes are my attempt at cleaning them up to avoid
> looping over t
From: Dave Airlie
This can be done with a single pass for the instruction base,
and takes renumber_registers out of its spot on the profile.
Signed-off-by: Dave Airlie
---
src/mesa/state_tracker/st_glsl_to_tgsi.cpp | 57 --
1 file changed, 38 insertions(+), 19
Oh I missed one of these patches on the initial send, at least
shader-db shows no output changes with these on radeonsi,
and also it removes the merge/renumber passes from the profiles
for a shader-db run.
Dave.
___
mesa-dev mailing list
mesa-dev@lists.
From: Dave Airlie
The glsl->tgsi convertor does some temporary register reduction
however in profiling shader-db this shows up quite highly,
so optimise things to reduce the number of loops through
all the instructions we do. This drops merge_registers
from 4-5% on the profile to 1%. I th
cares can do it sooner).
it doesn't change however what this patch does, it looks useful for r600
as is.
Dave.
>
> Marek
>
> On Thu, Aug 27, 2015 at 5:30 AM, Dave Airlie wrote:
>> From: Dave Airlie
>>
>> The glsl->tgsi convertor does some temporary register
From: Dave Airlie
On 32-bit we need to use PRIu64 flags for printfs,
otherwise this segfaults in R600_DEBUG=help otherwise.
Cc: "11.0"
Signed-off-by: Dave Airlie
---
src/gallium/auxiliary/util/u_debug.c | 7 ---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/s
From: Dave Airlie
Otherwise this will crash on 32-bit, and it gets rid of
warnings building on 32-bit.
Signed-off-by: Dave Airlie
---
src/gallium/drivers/r600/compute_memory_pool.c | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/src/gallium/drivers/r600
From: Dave Airlie
This code we broken by the tess merge, and I totally missed it
until now. I'm not sure this fixes anything but it stops the assert.
Cc: "11.0"
Signed-off-by: Dave Airlie
---
src/gallium/drivers/r600/r600_pipe.h | 31 ---
1 f
From: Dave Airlie
The geom shader rings require a VGT FLUSH, but up until now
it was just hacked into the function, add proper support for it.
Signed-off-by: Dave Airlie
---
src/gallium/drivers/r600/evergreen_state.c | 9 +
src/gallium/drivers/r600/r600_hw_context.c | 5 +
src
From: Dave Airlie
This should avoid C++ fail including this header.
Signed-off-by: Dave Airlie
---
src/gallium/drivers/r600/r600_pipe.h | 26 +-
src/gallium/drivers/r600/r600_state_common.c | 25 +
2 files changed, 26 insertions(+), 25
From: Dave Airlie
I've no idea why this is restricted to r600 only, but we should
use vertex fetches anywhere the input shader would use them from
what I can see.
Signed-off-by: Dave Airlie
---
src/gallium/drivers/r600/sb/sb_shader.cpp | 2 +-
1 file changed, 1 insertion(+), 1 del
From: Dave Airlie
As Glenn did for finalize_loop we need to update_cf when we
add a POP at the end of a shader.
I think this fixes one of the earlier shader going off end
of memory problems we've stopped.
Signed-off-by: Dave Airlie
---
src/gallium/drivers/r600/sb/sb_bc_finalize.cp
On 1 September 2015 at 09:34, Brian Paul wrote:
> On 08/25/2015 07:14 PM, Dave Airlie wrote:
>>
>> From: Dave Airlie
>>
>> GL3.3 added GL_ARB_texture_rgb10_a2ui, which specifies
>> a lot more things than just rgb10/a2ui.
>>
>> While playing with ogl c
From: Dave Airlie
I've been chasing a geom shader hang on rv635 since I wrote
r600 geom code, and finally I hacked some values from fglrx
in and I could run texelfetch without failures.
This is totally my fault as well, maths fail 101.
This makes geom shaders on r600 not fail heavily
From: Dave Airlie
In a number of places the SwapBytes handling
didn't handle cases with GL_(UN)PACK_ALIGNMENT
set and 7 byte width cases aligned to 8 bytes.
This adds a common routine to swap bytes a 2D
image and uses this code in the:
texture store
texture getting
readpixels
and s
This case just wasn't handled, so add support for it.
Signed-off-by: Dave Airlie
---
src/mesa/main/texgetimage.c | 7 +++
1 file changed, 7 insertions(+)
diff --git a/src/mesa/main/texgetimage.c b/src/mesa/main/texgetimage.c
index 0c23687..52ed1ef 100644
--- a/src/mesa/main/texgetim
On 28 August 2015 at 09:16, Brian Paul wrote:
> Some of the commit messages, like this one, seem to line-wrapped shorter
> than necessary.
>
> For patches 1-4, the changes look OK, AFAICT. Though, it seems there's some
> very similar byte-swap code duplicated in several places. Would it be
> pos
From: Dave Airlie
The fastpath currently checks the stride != width, but
if you have a RowLength of 7, and Alignment of 4, then
that shuoldn't match.
align the rowlength to the pack alignment before comparing.
This fixes compressed cases in CTS packed_pixels_pixelstore
test when SKIP_PIXE
From: Dave Airlie
We don't need to use the 3d image address here as that will
include SKIP_IMAGES, and we are only blitting a single
2D anyways, so just use the 2D path.
This fixes some memory overruns under CTS
packed_pixels.packed_pixels_pixelstore when PACK_SKIP_IMAGES
is used.
Signe
From: Dave Airlie
The CTS packed_pixels test checks that readpixels doesn't write
into the space between rows, however we fail that here unless
we check the format and stride match.
This fixes all the core mesa problems with CTS packed_pixels
tests.
Signed-off-by: Dave Airlie
---
src
On 1 September 2015 at 20:34, Iago Toral wrote:
> Hey,
>
> On Tue, 2015-08-18 at 12:52 +1000, Dave Airlie wrote:
>> Hey,
>>
>> while running CTS under valgrind I got to see a lot of
>>
>> ==32256== Invalid read of size 2
>> ==32256==at 0
it should.
For both:
Reviewed-by: Dave Airlie
> ---
> src/compiler/spirv/vtn_glsl450.c | 54
> +++-
> 1 file changed, 53 insertions(+), 1 deletion(-)
>
> diff --git a/src/compiler/spirv/vtn_glsl450.c
> b/src/compiler/spirv/vtn_glsl450.c
On 15 September 2016 at 13:03, Ilia Mirkin wrote:
> On Wed, Sep 14, 2016 at 10:15 PM, Michel Dänzer wrote:
>>> No, the current impl is pretty radeon-specific (note - it doesn't work
>>> on nouveau, and no other drivers support the interfaces, so ... it's
>>> radeon-specific).
>>
>> We're getting
From: Dave Airlie
This reverts commit d180de35320eafa3df3d76f0e82b332656530126.
This is a radeon specific hack that causes problems on nouveau
when combined with the SHARED flag later. If radeonsi needs a fix
for this, please fix it in the driver.
Signed-off-by: Dave Airlie
---
src/gallium
From: Dave Airlie
SPIR-V/Vulkan have a special image type for input attachments
called the subpass type. It has different characteristics than
other images types.
The main one being it can only be an input image to fragment
shaders and loads from it are relative to the frag coord.
This adds
From: Dave Airlie
This adds support for the input attachments subpass type
to the SPIRV->NIR pass.
---
src/compiler/spirv/spirv_to_nir.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/src/compiler/spirv/spirv_to_nir.c
b/src/compiler/spirv/spirv_to_nir.c
index 7e7a
On 15 September 2016 at 17:43, Christian König wrote:
> Am 15.09.2016 um 06:00 schrieb Ilia Mirkin:
>>
>> On Wed, Sep 14, 2016 at 11:58 PM, Dave Airlie wrote:
>>>
>>> From: Dave Airlie
>>>
>>> This reverts commit d180de35320eafa3df3d76f0e82b3
but this makes sense after
I looked through it.
It also fixes vkQuake on radv.
Tested-by: Dave Airlie
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=97233
Dave.
> ---
> src/compiler/spirv/vtn_cfg.c | 6 --
> src/compiler/spirv/vtn_private.h | 4 ++--
> 2 files changed
On 16 September 2016 at 14:47, Jason Ekstrand wrote:
> On Thu, Sep 15, 2016 at 1:23 AM, Dave Airlie wrote:
>>
>> From: Dave Airlie
>>
>> SPIR-V/Vulkan have a special image type for input attachments
>> called the subpass type. It has different character
On 14 September 2016 at 04:00, wrote:
> From: Marc-André Lureau
>
> When updating the sampler views, virgl might need to flush. After
> flushing, the resources are reattached, however, the sampler
> enabled_mask isn't yet updated, and some views could be in the process
> of being removed, which l
On 30 September 2016 at 01:07, Jason Ekstrand wrote:
> On Sep 29, 2016 7:56 AM, "Emil Velikov" wrote:
>>
>> On 28 September 2016 at 19:53, Marek Olšák wrote:
>> > Hi,
>> >
>> > It's been almost 4 months since the 12.0 branch was created, and soon
>> > it will have been 3 months since Mesa 12.0 w
On 1 October 2016 at 06:11, Marek Olšák wrote:
> On Fri, Sep 30, 2016 at 4:42 PM, Ilia Mirkin wrote:
>> So is this getting pushed, or should I push my "don't build vdpau for
>> nouveau" patch?
>
> Please wait.
>
> Christian, your commit says it fixes a radeonsi issue, but Ilia says
> it breaks no
On 3 October 2016 at 05:45, Tobias Droste wrote:
> This reuse the same logic gallium uses to determine if LLVM is needed or
> not:
> --enable-vulkan-llvm is set to yes if at least one vulkan driver is
> active and the host is i3*6 or x86_64.
> To build vulkan drivers without LLVM (e.g. intel) one
Hi all,
I fully expect this won't make it to the list in one piece due to size.
I've pushed the same patchset to
https://github.com/airlied/mesa/tree/radv-submit
This is an initial submission of the open source radv vulkan driver
for AMD GPUs supported by the amdgpu driver. It is a project Bas an
From: Dave Airlie
This just brings these files into the tree, it doesn't integrate
them with the build system.
The radv winsys is based on the gallium one with some changes,
due to how command buffers are built and lack of flushing behaviour.
Authors: Bas Nieuwenhuizen and Dave Airlie
S
From: Dave Airlie
This moves some of the llvm checks around to allow them
to be used for non-gallium drivers as well.
radv requires llvm 3.9.0 as vulkan requires compute shaders.
Authors: Bas Nieuwenhuizen and Dave Airlie
Signed-off-by: Dave Airlie
---
configure.ac| 33
> It would have been great if the build/integration bits were properly
> split, but I won't sweat too much over it.
I thought they were, all the bits outside src/amd are in this patch really.
I suppose I could move the Makefile.am and Makefile.sources into this
patch to make it easier to review.
On 4 October 2016 at 20:47, Albert Freeman wrote:
> It might be a good idea to consider moving the nir -> llvm code out of
> the amd folder for reuse by a Vulkan software rasterizer or something.
> I have heard another developers recent interest in developing a Vulkan
> software rasterizer. I at o
On 4 October 2016 at 20:09, Nicolai Hähnle wrote:
> On 04.10.2016 03:48, Dave Airlie wrote:
>>
>> From: Bas Nieuwenhuizen
>>
>> This adds the basic files for the NIR->LLVM translation layer,
>> along with some hopefully generic code to load the binary
&
On 4 October 2016 at 21:05, Emil Velikov wrote:
> Hi Dave,
>
> On 4 October 2016 at 02:48, Dave Airlie wrote:
>> From: Bas Nieuwenhuizen
>>
>> This adds the basic files for the NIR->LLVM translation layer,
>> along with some hopefully generic code to load the
From: Dave Airlie
This moves some of the llvm checks around to allow them
to be used for non-gallium drivers as well.
radv requires llvm 3.9.0 as vulkan requires compute shaders.
v1.1: add all make infrastructure to this patch for easier
review.
Authors: Bas Nieuwenhuizen and Dave Airlie
From: Dave Airlie
This just brings these files into the tree, it doesn't integrate
them with the build system.
The radv winsys is based on the gallium one with some changes,
due to how command buffers are built and lack of flushing behaviour.
v1.1: cleanup whitespace issues, move Makefil
Again I'm sure this will hit limits, and I've asked Michel to
drop the big patch before it gets here.
All of these are in a new branch radv-submit2.
https://github.com/airlied/mesa/commits/radv-submit2
I've incorporated all the feedback except two things:
a) #pragma once - this is used in lots o
On 13 March 2016 at 11:29, Ian Romanick wrote:
> On 03/11/2016 03:46 PM, Eric Anholt wrote:
>> Ian Romanick writes:
>>
>>> On 03/10/2016 05:53 PM, Francisco Jerez wrote:
Iago Toral writes:
> On Wed, 2016-03-09 at 19:04 -0800, Francisco Jerez wrote:
>> Matt Turner writes:
>
Please snip out stuff if you have just one question.
That code comes from the radeonsi driver, no point wondering here
about why the default is there, if you never wondered why it's in the
radeonsi code :)
Dave.
___
mesa-dev mailing list
mesa-dev@lists.
On 6 October 2016 at 07:07, Gustaw Smolarczyk wrote:
> Additional problems when configuring radv without the rest of mesa
> (might be also applicable to anv):
> 1. For radv to advertise VK_KHR_xcb_surface, HAVE_PLATFORM_X11 has to
> be defined. This variable is set from the list of EGL platforms,
If nobody has any remaining major objections I'd rather not spend too
much time rebasing this stuff, so I'd like to propose merging it
approx now + 24hr, if anyone wants to explicitly ack or r-b anything
let me know, but I'm most likely going to squash merge down the
contents of this branch
https:
On 7 October 2016 at 12:05, Dieter Nützel wrote:
> gcc (SUSE Linux) 4.8.3
>
> make[4]: Entering directory '/opt/mesa/src/amd/vulkan'
Should already be fixed, must be compiler/libc mismatches or something.
Dave.
___
mesa-dev mailing list
mesa-dev@lists.
This is a bit of a workaround to how llvm does spilling, and
we should fix llvm, I'm just not good enough yet, and this will
get us further with CTS tests for now.
Details in patch 2.
Dave.
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https
From: Dave Airlie
This adds some comments and adds defines for the user sgprs,
so that we can move them around easier later and not have
to change/revalidate every one of these.
Signed-off-by: Dave Airlie
---
src/amd/common/ac_nir_to_llvm.c | 7 +--
src/amd/common/ac_nir_to_llvm.h | 17
From: Dave Airlie
This is a bit of a hack due to how llvm currently handles
spilling in it's shader ABI. Currently llvm amdgpu backend
uses relocations to patch the shader with the address of
the tmpring. The driver loads the shader and patches the
relocations.
However for vulkan this do
On 10 October 2016 at 13:25, Dave Airlie wrote:
> From: Dave Airlie
>
> This is a bit of a hack due to how llvm currently handles
> spilling in it's shader ABI. Currently llvm amdgpu backend
> uses relocations to patch the shader with the address of
> the tmpring. The driv
On 10 October 2016 at 13:24, Dave Airlie wrote:
> This is a bit of a workaround to how llvm does spilling, and
> we should fix llvm, I'm just not good enough yet, and this will
> get us further with CTS tests for now.
>
> Details in patch 2.
I've implemented LLVM side f
On 10 October 2016 at 21:45, Arsenault, Matthew
wrote:
> I don't like adding explicit IR arguments for ABI arguments, especially this
> one. Adding a special case for the first index feels dirty. The rest of llvm
> also won't be aware of the specialness of the argument. It would be
> problematic b
as hex for flt values.
Reviewed-by: Dave Airlie
>
> diff --git a/src/gallium/auxiliary/tgsi/tgsi_text.c
> b/src/gallium/auxiliary/tgsi/tgsi_text.c
> index be80842..36dc979 100644
> --- a/src/gallium/auxiliary/tgsi/tgsi_text.c
> +++ b/src/gallium/auxiliary/tgsi/tgsi_text.c
> @@ -
On 10 October 2016 at 18:32, Nicolai Hähnle wrote:
> From: Nicolai Hähnle
>
> For specifying an exact location/component.
Any reason you have the args ordering different, that seems like it
might get confusing.
I'd prefer you kept the output_layout and output_masked ordering the
same and stick
On 11 October 2016 at 05:50, Dave Airlie wrote:
> On 10 October 2016 at 21:45, Arsenault, Matthew
> wrote:
>> I don't like adding explicit IR arguments for ABI arguments, especially this
>> one. Adding a special case for the first index feels dirty. The rest of llvm
>&g
On 11 October 2016 at 11:42, Dave Airlie wrote:
> On 11 October 2016 at 05:50, Dave Airlie wrote:
>> On 10 October 2016 at 21:45, Arsenault, Matthew
>> wrote:
>>> I don't like adding explicit IR arguments for ABI arguments, especially this
>>> one. Add
401 - 500 of 4069 matches
Mail list logo
