On 25/01/17 07:34 PM, Yu, Qiang wrote:
>> From: Michel Dänzer
>> On 24/01/17 12:36 PM, Qiang Yu wrote:
>>> Third-party can put their backend to a directory configured with
>>> '--with-gbm-backenddir' and create a /etc/gbm.conf.d/*.conf file
>>> which contains the backend so
This extension was not correctly supported, and it conflicts with the
VK_KHR_MAINTENANCE1 spec.
---
src/amd/vulkan/radv_device.c | 4
1 file changed, 4 deletions(-)
diff --git a/src/amd/vulkan/radv_device.c b/src/amd/vulkan/radv_device.c
index 7f68cdc..9f05dd6 100644
---
There is a new error code in Maintenance1 that is more specific to the
situation: VK_ERROR_OUT_OF_POOL_MEMORY_KHR
Fixes CTS test case:
dEQP-VK.api.descriptor_pool.out_of_pool_memory
---
src/amd/vulkan/radv_descriptor_set.c | 2 +-
src/amd/vulkan/radv_util.c | 1 +
2 files changed, 2
---
src/amd/vulkan/radv_cmd_buffer.c | 7 +++
1 file changed, 7 insertions(+)
diff --git a/src/amd/vulkan/radv_cmd_buffer.c b/src/amd/vulkan/radv_cmd_buffer.c
index c62d275..0b090b7 100644
--- a/src/amd/vulkan/radv_cmd_buffer.c
+++ b/src/amd/vulkan/radv_cmd_buffer.c
@@ -1895,6 +1895,13 @@
Previously the z offset of the destination image was being ignored. It
should be taken into account when copying into a 3d target.
Also, img_extent_el.depth was being incorrectly clamped to 1 due to the
source image being VK_IMAGE_TYPE_2D. This would result in the blit
failing to iterate over all
From: Bas Nieuwenhuizen
We really use the depth block for the blits.
Signed-off-by: Bas Nieuwenhuizen
---
src/amd/vulkan/radv_formats.c | 9 +
1 file changed, 5 insertions(+), 4 deletions(-)
diff --git a/src/amd/vulkan/radv_formats.c
---
src/amd/vulkan/radv_device.c | 4
1 file changed, 4 insertions(+)
diff --git a/src/amd/vulkan/radv_device.c b/src/amd/vulkan/radv_device.c
index 4aa6af2..7f68cdc 100644
--- a/src/amd/vulkan/radv_device.c
+++ b/src/amd/vulkan/radv_device.c
@@ -104,6 +104,10 @@ static const
This series implements the VK_KHR_maintenance1 extension. It is loosely
based on jekstrand's series for anv.
This series soft depends on one of Bas's patches that are not yet in master.
I'm not certain of protocol for this kind of situation. I've left them included
in this series, but if I should
From: Bas Nieuwenhuizen
Signed-off-by: Bas Nieuwenhuizen
---
src/amd/vulkan/radv_formats.c | 11 +++
1 file changed, 11 insertions(+)
diff --git a/src/amd/vulkan/radv_formats.c b/src/amd/vulkan/radv_formats.c
index f56f67c..c968cef 100644
This is part of the spec and fixes CTS tests:
dEQP-VK.api.object_management.alloc_callback_fail_multiple.command_buffer_*
---
src/amd/vulkan/radv_cmd_buffer.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/src/amd/vulkan/radv_cmd_buffer.c b/src/amd/vulkan/radv_cmd_buffer.c
index
Question more about the first one
[Mesa-dev] [PATCH 1/2] gallium/radeon: query the CPU accessible size of
VRAM
Is vram_vis_size > vram_size valid?
After some time I get this on r600/Turks XT/6670/2 GB:
/opt/mesa> setenv R600_DEBUG info
/opt/mesa> glxgears
pci_id = 0x6758
family = 45 (AMD
From: Dave Airlie
this fixes deferred shadows with geom shaders enabled.
but I think this fix is fine by itself.
Signed-off-by: Dave Airlie
---
src/amd/vulkan/radv_device.c | 6 --
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git
From: Timothy Arceri
This will be used to remove cache items created with old versions
of Mesa or other invalid cache items from the cache.
V2: rename stub function (cache_* funtions were renamed disk_cache_*)
in master.
---
src/util/disk_cache.c | 22
On 17-01-23 15:32:32, Chad Versace wrote:
On Fri 20 Jan 2017, Rafael Antognolli wrote:
I have tested this series with the branches that you mentioned, and with
piglit with the patches from my own branch:
https://github.com/rantogno/piglit/tree/review/fences-v02
Everything seems to work fine.
From: Timothy Arceri
This uses disk_cache.c to write out a serialization of various
state that's required in order to successfully load and use a
binary written out by a drivers backend, this state is referred to as
"metadata" throughout the implementation.
This
On 26/01/17 08:07 PM, Christian König wrote:
> Am 26.01.2017 um 12:01 schrieb Samuel Pitoiset:
>> On 01/26/2017 03:45 AM, Michel Dänzer wrote:
>>> On 25/01/17 11:19 PM, Samuel Pitoiset wrote:
>>>
I would like to approach the problem by reducing the amount of vram
needed by the userspace
Hi Christian,
2017-01-26 12:00 GMT+01:00 Christian König :
> Hi Peter,
>
> Am 25.01.2017 um 19:45 schrieb Peter Frühberger:
>
>>
>>
>> Peter, Rainer any idea what I'm missing here? Do you guys use some
>> modified ffmpeg for Kodi or how does that work for you?
>>
2017-01-26 16:36 GMT+01:00 Christian König :
> Am 26.01.2017 um 12:16 schrieb Peter Frühberger:
>
> Hi Christian,
>
> 2017-01-26 12:00 GMT+01:00 Christian König :
>
>> Hi Peter,
>>
>> Am 25.01.2017 um 19:45 schrieb Peter Frühberger:
>>
>>>
>>>
>>>
Hi Christian,
2017-01-26 19:34 GMT+01:00 Christian König :
> Am 26.01.2017 um 16:59 schrieb Peter Frühberger:
>
>
>
> 2017-01-26 16:36 GMT+01:00 Christian König :
>
>> Am 26.01.2017 um 12:16 schrieb Peter Frühberger:
>>
>> Hi Christian,
>>
>>
Andy Furniss wrote:
Andy Furniss wrote:
Christian König wrote:
Hi guys,
ok this is completely work in progress and untested except for a
compile run.
Most of the stuff necessary should be there for VDPAU, but I'm
honestly not sure how to approach VAAPI.
These regress R9 285 8bit h264 vaapi
On Thu 26 Jan 2017, Chad Versace wrote:
> On Thu 26 Jan 2017, Chris Wilson wrote:
> > Since the workaround bo is used strictly as a write-only buffer, we need
> > only allocate one per screen and use the same one from all contexts.
> >
> > (The caveat here is during extension initialisation,
From: Marek Olšák
The time spent in the function dropped by 37% for torcs.
---
src/gallium/winsys/amdgpu/drm/amdgpu_cs.c | 16
src/gallium/winsys/amdgpu/drm/amdgpu_cs.h | 5 +
2 files changed, 21 insertions(+)
diff --git
From: Marek Olšák
For lower memory usage and more efficient updates of the buffer residency
list. (e.g. if drivers keep seeing the same buffer for many consecutive
"add" calls, the calls can be turned into no-ops trivially)
---
src/gallium/include/pipe/p_context.h | 7
On Thu 26 Jan 2017, Chris Wilson wrote:
> On Thu, Jan 26, 2017 at 09:39:51AM -0800, Chad Versace wrote:
> > On Thu 26 Jan 2017, Chris Wilson wrote:
> > > Since the workaround bo is used strictly as a write-only buffer, we need
> > > only allocate one per screen and use the same one from all
Sorry, ignore this. I sent to the wrong list.
On 17-01-26 14:16:23, Ben Widawsky wrote:
Originally based off of a patch by Kristian.
This new ioctl extends DRM_IOCTL_MODE_GETPLANE, by returning information
about the modifiers that will work with each format.
It's modified from Kristian's
See "glsl: Rewrite atan2 implementation to fix accuracy and handling
of zero/infinity." for the rationale, but note that the instruction
count benefit discussed there is somewhat less important for the SPIRV
implementation, because the current code already emitted no control
flow instructions --
This addresses several issues of the current atan2 implementation:
- Negative zero (and negative denorms which end up getting flushed to
zero) isn't handled correctly by the current implementation. The
reason is that it does 'y >= 0' and 'x < 0' comparisons to decide
on which side of
Hey Matt,
yeah sorry about this, it is a known problem:
https://bugs.freedesktop.org/show_bug.cgi?id=99010
A fix was commited but broke scons and Jose reverted it and doesn't accept the
way it was fixed in general.
See discussion here:
Ian Romanick writes:
> On 01/26/2017 12:20 PM, Francisco Jerez wrote:
>> Ian Romanick writes:
>>
>>> On 01/25/2017 10:53 AM, Francisco Jerez wrote:
Hi Ian, and thank you for your comments,
Ian Romanick writes:
Originally based off of a patch by Kristian.
This new ioctl extends DRM_IOCTL_MODE_GETPLANE, by returning information
about the modifiers that will work with each format.
It's modified from Kristian's patch in that the modifiers and formats
are setup by the driver, and then a callback is used to
On 01/26/2017 12:20 PM, Francisco Jerez wrote:
> Ian Romanick writes:
>
>> On 01/25/2017 10:53 AM, Francisco Jerez wrote:
>>> Hi Ian, and thank you for your comments,
>>>
>>> Ian Romanick writes:
>>>
On 01/24/2017 03:26 PM, Francisco Jerez wrote:
On Thursday, January 26, 2017 12:00:14 PM PST Eric Anholt wrote:
> Kenneth Graunke writes:
>
> > Applications may delete a shader program, create a new one, and bind it
> > before the next draw. With terrible luck, malloc may randomly return a
> > chunk of memory for the
- Original Message -
> From: "Marek Olšák"
> To: "Nicolai Hähnle"
> Cc: "Ilia Mirkin" , mesa-dev@lists.freedesktop.org,
> "Dave Airlie"
> Sent: Friday, 27 January, 2017 6:18:32 AM
> Subject: Re:
https://bugs.freedesktop.org/show_bug.cgi?id=97879
matt...@familycampground.org changed:
What|Removed |Added
CC|
Reported against 17.0.0-rc2 [1], but occurs on master as well
Reproduce with
./configure --with-dri-drivers= --with-gallium-drivers=swrast
--disable-gallium-llvm --enable-gallium-osmesa
Fails linking Gallium OSMesa with undefined references to draw_llvm_destroy.
https://bugs.freedesktop.org/show_bug.cgi?id=95460
--- Comment #4 from Shmerl ---
FYI: Freedreno is visible in the MesaMatrix: https://mesamatrix.net
--
You are receiving this mail because:
You are the assignee for the bug.
You are the QA Contact for the
My preference would be not to use static variables in pipe_screen
functions, because all functions needs to be thread-safe, but anyway:
Reviewed-by: Marek Olšák
Marek
On Thu, Jan 26, 2017 at 7:27 PM, Emil Velikov wrote:
> From: Emil Velikov
For the series:
Reviewed-by: Marek Olšák
Marek
On Thu, Jan 26, 2017 at 8:09 PM, Nicolai Hähnle wrote:
> From: Nicolai Hähnle
>
> All of these have had support for the TGSI opcodes since before most of
> the glsl compiler work
For the series:
Tested-by: Edmondo Tommasina
Tested with:
* The Witcher 2
* Talos Principle
* Shadow Tactics
* Wasteland 2
* X3: AP
* Pillars of Eternity
* Uningine Heaven & Valley
Thanks
edmondo
On Thu, Jan 26, 2017 at 5:04 PM, Marek Olšák
Ian Romanick writes:
> On 01/25/2017 10:53 AM, Francisco Jerez wrote:
>> Hi Ian, and thank you for your comments,
>>
>> Ian Romanick writes:
>>
>>> On 01/24/2017 03:26 PM, Francisco Jerez wrote:
Will avoid a regression in a future commit that
On Thu, Jan 26, 2017 at 9:04 PM, Nicolai Hähnle wrote:
> On 26.01.2017 20:23, Ilia Mirkin wrote:
>>
>> I have no serious preference, but for doubles, we use a shader cap.
>
>
> I think this:
>
>> On Thu, Jan 26, 2017 at 2:09 PM, Nicolai Hähnle
>> wrote:
Am 26.01.2017 um 19:27 schrieb Emil Velikov:
> From: Emil Velikov
>
> Currently we can build draw without LLVM thus honouring SOFTPIPE_USE_LLVM
> is misleading even if most of the code nicely falls-back to no-op in the
> lack of LLVM.
>
> That does not seem to be the
Am 26.01.2017 um 21:11 schrieb Ilia Mirkin:
> On Thu, Jan 26, 2017 at 3:04 PM, Nicolai Hähnle wrote:
>> On 26.01.2017 20:23, Ilia Mirkin wrote:
>>>
>>> I have no serious preference, but for doubles, we use a shader cap.
>>
>>
>> I think this:
>>
>>> On Thu, Jan 26, 2017 at
On Thu, Jan 26, 2017 at 3:04 PM, Nicolai Hähnle wrote:
> On 26.01.2017 20:23, Ilia Mirkin wrote:
>>
>> I have no serious preference, but for doubles, we use a shader cap.
>
>
> I think this:
>
>> On Thu, Jan 26, 2017 at 2:09 PM, Nicolai Hähnle
>> wrote:
On 26.01.2017 20:26, Emil Velikov wrote:
From: Emil Velikov
Signed-off-by: Emil Velikov
---
I forget this a bit too often :-\
:D
Reviewed-by: Nicolai Hähnle
---
docs/releasing.html | 2 ++
1 file changed,
On 26.01.2017 20:20, Ilia Mirkin wrote:
On Thu, Jan 26, 2017 at 2:09 PM, Nicolai Hähnle wrote:
+ case ir_unop_i642b:
+ emit_asm(ir, TGSI_OPCODE_U64SNE, result_dst, op[0],
st_src_reg_for_int(0));
+ break;
Does this work reliably? I would have imagined you'd
On 26.01.2017 20:23, Ilia Mirkin wrote:
I have no serious preference, but for doubles, we use a shader cap.
I think this:
On Thu, Jan 26, 2017 at 2:09 PM, Nicolai Hähnle wrote:
v1.1: move to using a normal CAP. (Marek)
... suggests that this has already gone back and
Kenneth Graunke writes:
> Applications may delete a shader program, create a new one, and bind it
> before the next draw. With terrible luck, malloc may randomly return a
> chunk of memory for the new gl_program that happened to be the exact
> same pointer as our
On Thu, Jan 26, 2017 at 8:43 PM, Ernst Sjöstrand wrote:
> Should this code be able to handle drm 4.0.0?
DRM 4.0.0 doesn't exist and hopefully won't exist.
DRM 1.x.0 = intel
DRM 2.x.0 = radeon
DRM 3.x.0 = amdgpu
That's just a coincidence.
Marek
We also want to monitor other MMIO counters like SRBM_STATUS2 in
order to know if SDMA is busy.
Signed-off-by: Samuel Pitoiset
---
src/gallium/drivers/radeon/r600_gpu_load.c| 30 +++
src/gallium/drivers/radeon/r600_pipe_common.h | 35
There are even more counters in the CP_STAT register but I think
these ones are enough for now.
Signed-off-by: Samuel Pitoiset
---
src/gallium/drivers/radeon/r600_gpu_load.c| 34 +++
src/gallium/drivers/radeon/r600_pipe_common.h | 9
Signed-off-by: Samuel Pitoiset
---
src/gallium/drivers/radeon/r600_gpu_load.c| 11 +++
src/gallium/drivers/radeon/r600_pipe_common.h | 3 +++
src/gallium/drivers/radeon/r600_query.c | 5 -
src/gallium/drivers/radeon/r600_query.h | 1 +
4
Yeah, I'd much prefer this to be
if (nouveau_screen(pscreen)->class_3d < NV50_3D_CLASS)
llvm = " LLVM";
or something along those lines. [Using dev->chipset will be annoying
since nv50 = 0x50, and nv67 (nv4x) = 0x67.]
On Thu, Jan 26, 2017 at 1:27 PM, Emil Velikov
https://bugs.freedesktop.org/show_bug.cgi?id=97250
Vedran Miletić changed:
What|Removed |Added
Blocks||99553
Referenced
Should this code be able to handle drm 4.0.0?
2017-01-26 17:04 GMT+01:00 Marek Olšák :
> From: Marek Olšák
>
> ---
> src/gallium/drivers/radeon/r600_buffer_common.c | 4
> 1 file changed, 4 insertions(+)
>
> diff --git
On Thu, Jan 26, 2017 at 7:58 PM, Emil Velikov wrote:
> On 26 January 2017 at 18:54, Marek Olšák wrote:
>> They still have to ship LLVM to have GCN support. Or do they simply not ship
>> radeonsi?
>>
> The latter - they omit anything that requires LLVM.
From: Emil Velikov
Signed-off-by: Emil Velikov
---
I forget this a bit too often :-\
---
docs/releasing.html | 2 ++
1 file changed, 2 insertions(+)
diff --git a/docs/releasing.html b/docs/releasing.html
index 2ed66a13f4..020f3dec70
I have no serious preference, but for doubles, we use a shader cap.
On Thu, Jan 26, 2017 at 2:09 PM, Nicolai Hähnle wrote:
> From: Dave Airlie
>
> v1.1: move to using a normal CAP. (Marek)
>
> v2: fill in the cap everywhere
>
> Signed-off-by: Dave Airlie
On Thu, Jan 26, 2017 at 2:09 PM, Nicolai Hähnle wrote:
> + case ir_unop_i642b:
> + emit_asm(ir, TGSI_OPCODE_U64SNE, result_dst, op[0],
> st_src_reg_for_int(0));
> + break;
Does this work reliably? I would have imagined you'd need a
st_srg_reg_for_int64()
From: Dave Airlie
v1.1: move to using a normal CAP. (Marek)
v2: fill in the cap everywhere
Signed-off-by: Dave Airlie
Reviewed-by: Marek Olšák (v1)
---
src/gallium/drivers/etnaviv/etnaviv_screen.c | 1 +
From: Nicolai Hähnle
All of these have had support for the TGSI opcodes since before most of
the glsl compiler work landed.
Also update the docs accordingly, including the missing note about i965.
---
docs/features.txt| 2 +-
Hi all,
now that the GLSL compiler work has landed recently, and TGSI has actually
had the opcodes for several months, it's time to finally flip the switch!
The bulk of the work here is actually due to Dave, with some cleanups from
Ian, so kudos to them. I just filled in some minor bits and
From: Dave Airlie
v2: add conversion opcodes.
v3 (idr): Rebase on replacemtn of TGSI_OPCODE_I2U64 with
TGSI_OPCODE_I2I64.
v4 (idr): "cut them down later" => Remove ir_unop_b2u64 and
ir_unop_u642b. Handle these with extra i2u or u2i casts just like
uint(bool) and bool(uint)
From: Dave Airlie
Signed-off-by: Dave Airlie
Reviewed-by: Nicolai Hähnle
---
src/mesa/state_tracker/st_extensions.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/src/mesa/state_tracker/st_extensions.c
On 01/25/2017 10:53 AM, Francisco Jerez wrote:
> Hi Ian, and thank you for your comments,
>
> Ian Romanick writes:
>
>> On 01/24/2017 03:26 PM, Francisco Jerez wrote:
>>> Will avoid a regression in a future commit that introduces some
>>> additional rcp operations.
>>
>>
On 26 January 2017 at 18:48, Matt Turner wrote:
> I ran into this with mesa-17.0.0_rc2. This patch should go into the 17.0
> branch.
Seem like I forgot to run ./bin/get-extra-pick-list.sh which flags it up.
Thanks Matt, will ensure it's in for rc3.
Emil
On 26 January 2017 at 18:54, Marek Olšák wrote:
> They still have to ship LLVM to have GCN support. Or do they simply not ship
> radeonsi?
>
The latter - they omit anything that requires LLVM.
I dare not discuss how good/bad/etc of a choice that it, but some
people have their
On 25 January 2017 at 22:10, Chad Versace wrote:
> On Tue 24 Jan 2017, Jason Ekstrand wrote:
>> On Tue, Jan 24, 2017 at 11:25 AM, Emil Velikov
>> > I'd rather not. That would make sense if we all lived in the
>> open-source
>> >
They still have to ship LLVM to have GCN support. Or do they simply not
ship radeonsi?
Marek
On Jan 26, 2017 7:30 PM, "Emil Velikov" wrote:
> Hi all,
>
> Here's a few small fixes/functionality improvements when dealing with
> LLVM.
>
> Most notably the series adds
I ran into this with mesa-17.0.0_rc2. This patch should go into the 17.0 branch.
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
On Thu, Jan 26, 2017 at 09:39:51AM -0800, Chad Versace wrote:
> On Thu 26 Jan 2017, Chris Wilson wrote:
> > Since the workaround bo is used strictly as a write-only buffer, we need
> > only allocate one per screen and use the same one from all contexts.
> >
> > (The caveat here is during
On Thu, Jan 26, 2017 at 10:23 AM, Ian Romanick wrote:
> I keep seeing patches like this... is it time to move BITFIELD64_* from
> mtypes.h to somewhere in util for more general use?
>
It may be time
> On 01/26/2017 05:18 AM, Emil Velikov wrote:
> > From: Emil Velikov
It should be OK.
Marek
On Jan 26, 2017 7:17 PM, "Alex Deucher" wrote:
> On Thu, Jan 26, 2017 at 12:39 PM, Gustaw Smolarczyk
> wrote:
> > 2017-01-26 17:04 GMT+01:00 Marek Olšák :
> >> From: Marek Olšák
> >>
>
Am 26.01.2017 um 16:59 schrieb Peter Frühberger:
2017-01-26 16:36 GMT+01:00 Christian König >:
Am 26.01.2017 um 12:16 schrieb Peter Frühberger:
Hi Christian,
2017-01-26 12:00 GMT+01:00 Christian König
On 26 January 2017 at 11:13, Andreas Boll wrote:
> b3119a3 introduced a strict LLVM requirement for r300 on all
> architectures and thus configure fails on architectures where LLVM is
> not available or buggy.
>
> r300 doesn't strictly require LLVM, but for performance
On Thu, Jan 26, 2017 at 10:11 AM, Eric Engestrom
wrote:
> This is git, not svn :P
>
> Signed-off-by: Eric Engestrom
Reviewed-by: Matt Turner
> ---
> I also noticed some difference between this file and the one on
>
For the series:
Reviewed-by: Marek Olšák
On Jan 25, 2017 5:50 PM, "Nicolai Hähnle" wrote:
Reviewed-by: Nicolai Hähnle
On 25.01.2017 16:56, Samuel Pitoiset wrote:
> This new query returns the current visible usage of VRAM
On Thu, Jan 26, 2017 at 10:22 AM, Jason Ekstrand wrote:
> Ugh... windows defines long to be 32-bit on 32-bit platforms Yeah.
long is 32-bit on 32-bit x86/Linux too...
I think you mean they define long to be 32-bits on 64-bit platforms.
From: Emil Velikov
Provides quick and direct feedback to the user/developer.
Cc: Stéphane Marchesin
Signed-off-by: Emil Velikov
---
src/gallium/drivers/i915/i915_screen.c | 7 ++-
1 file changed, 6
From: Emil Velikov
Signed-off-by: Emil Velikov
---
src/gallium/drivers/virgl/virgl_screen.c | 1 -
1 file changed, 1 deletion(-)
diff --git a/src/gallium/drivers/virgl/virgl_screen.c
b/src/gallium/drivers/virgl/virgl_screen.c
index
From: Emil Velikov
Cc: Roland Scheidegger
Cc: Jose Fonseca
Signed-off-by: Emil Velikov
---
src/gallium/drivers/llvmpipe/lp_screen.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff
From: Emil Velikov
Currently one can build i915g without LLVM thus the current handling is
wrong. Whether using i915g w/o LLVM is a good idea or not is a question
for another time.
Cc: Stéphane Marchesin
Signed-off-by: Emil Velikov
From: Emil Velikov
Analogous to previous two commits. Afaict only nv30 uses draw, so if
people prefer we can restrict this print only to those devices.
Signed-off-by: Emil Velikov
---
Unrelated:
Wasn't there a kernel/libdrm helper which
From: Emil Velikov
Provides quick and direct feedback to the user/developer.
Cc: Marek Olšák
Signed-off-by: Emil Velikov
---
src/gallium/drivers/r300/r300_screen.c | 10 +-
1 file changed, 9 insertions(+), 1
From: Emil Velikov
Provides quick and direct feedback to the user/developer.
Cc: Roland Scheidegger
Cc: Jose Fonseca
Signed-off-by: Emil Velikov
---
We can move all the ifdef to a helper, if
From: Emil Velikov
Currently we can build draw without LLVM thus honouring SOFTPIPE_USE_LLVM
is misleading even if most of the code nicely falls-back to no-op in the
lack of LLVM.
That does not seem to be the case in softpipe_draw_vbo() where extra
prepare
Hi all,
Here's a few small fixes/functionality improvements when dealing with
LLVM.
Most notably the series adds "LLVM" string [when applicable] to the .get_name()
callback for drivers that use draw.
Thus developer can respond accordingly - be that "rebuild with LLVM or
enjoy the bad
Tested with ffmpeg and gst-vaapi. Without this bits per
frame is set way too low.
Signed-off-by: Andy Furniss
---
src/gallium/state_trackers/va/picture.c | 32
1 file changed, 24 insertions(+), 8 deletions(-)
diff --git
I keep seeing patches like this... is it time to move BITFIELD64_* from
mtypes.h to somewhere in util for more general use?
On 01/26/2017 05:18 AM, Emil Velikov wrote:
> From: Emil Velikov
>
> MSVC warns about implicit conversion as below. Annotate the literal
>
Ugh... windows defines long to be 32-bit on 32-bit platforms Yeah.
Reviewed-by: Jason Ekstrand
On Thu, Jan 26, 2017 at 6:05 AM, Lionel Landwerlin <
lionel.g.landwer...@intel.com> wrote:
> Reviewed-by: Lionel Landwerlin
>
> On 26/01/17
LGTM
On Thu, Jan 26, 2017 at 6:05 AM, Lionel Landwerlin <
lionel.g.landwer...@intel.com> wrote:
> Reviewed-by: Lionel Landwerlin
>
> On 26/01/17 13:18, Emil Velikov wrote:
>
>> From: Emil Velikov
>>
>> MSVC warns about different const
On 01/25/2017 12:55 PM, Francisco Jerez wrote:
> Ian Romanick writes:
>
>> It's a real bummer that we have two implementations of this function
>> that are basically written in assembly... I'm not sure what else you'd
>> call generating IR by hand. The code review and
On Thu, Jan 26, 2017 at 12:39 PM, Gustaw Smolarczyk
wrote:
> 2017-01-26 17:04 GMT+01:00 Marek Olšák :
>> From: Marek Olšák
>>
>> this cleanup is based on the vulkan driver, which seems to do the same thing
>
> Is this also ok for
This is git, not svn :P
Signed-off-by: Eric Engestrom
---
I also noticed some difference between this file and the one on
mesa3d.org; it might be worth making sure everything is sync'ed between
the two (most likely just push the version on git to the webserver).
---
Michel Dänzer writes:
> From: Michel Dänzer
>
> Signed-off-by: Michel Dänzer
Reviewed-by: Francisco Jerez
> ---
>
> Not sure if PropagateAttrs should be set to true or false. Setting it to
> true
On Thu, Jan 26, 2017 at 09:39:51AM -0800, Chad Versace wrote:
> On Thu 26 Jan 2017, Chris Wilson wrote:
> > Since the workaround bo is used strictly as a write-only buffer, we need
> > only allocate one per screen and use the same one from all contexts.
> >
> > (The caveat here is during
On 01/25/2017 01:42 PM, Francisco Jerez wrote:
> Ian Romanick writes:
>
>> On 01/24/2017 03:26 PM, Francisco Jerez wrote:
>>> This does point at the front-end emitting silly code that could have
>>> been optimized out, but the current fsign implementation would emit
>>>
Reviewed-by: Marek Olšák
On Jan 26, 2017 2:22 PM, "Emil Velikov" wrote:
> From: Emil Velikov
>
> Follow the approach set in the file and handle all the ops, as otherwise
> the compiler throws a bunch of lovely
Reviewed-by: Marek Olšák
On Jan 26, 2017 2:21 PM, "Emil Velikov" wrote:
> From: Emil Velikov
>
> Analogous to previous commit.
>
> Signed-off-by: Emil Velikov
> ---
>
Reviewed-by: Marek Olšák
On Jan 26, 2017 12:13 PM, "Andreas Boll" wrote:
> b3119a3 introduced a strict LLVM requirement for r300 on all
> architectures and thus configure fails on architectures where LLVM is
> not available or buggy.
>
> r300
1 - 100 of 190 matches
Mail list logo