Without this stall, the test group ES3-CTS.functional.fbo.msaa.\* hangs
about 1 out of every 2 or 3 times on my Sky Lake GT3 laptop. With the
flush and stall, I can run it 6 times in a row without a hang.
Cc: "17.1"
---
src/mesa/drivers/dri/i965/brw_blorp.c |
From: Dave Airlie
This just takes this out to a separate function as it will
get more complex with images.
---
src/gallium/drivers/r600/r600_state_common.c | 52 +++-
1 file changed, 28 insertions(+), 24 deletions(-)
diff --git
From: Dave Airlie
The atomic counters on evergreen are implemented via append/consume
UAV counters. This just adds the register info for them. The EOS
packets are used to get the atomic totals extracted post shader
execution for storing into a buffer.
---
From: Dave Airlie
This just documents in the headers the RAT operation list,
and the RAT encoding for exports.
The immediate registers are used to point to buffers for the
RAT return values (_RTN instructions).
---
src/gallium/drivers/r600/eg_sq.h | 12
From: Dave Airlie
These are used for fragment shader thread calculations.
---
src/gallium/drivers/r600/r600_sq.h | 4
1 file changed, 4 insertions(+)
diff --git a/src/gallium/drivers/r600/r600_sq.h
b/src/gallium/drivers/r600/r600_sq.h
index 3526668..aa38381 100644
---
On 01/06/17 02:37 AM, Jan Vesely wrote:
> On Wed, 2017-05-31 at 17:44 +0200, Marek Olšák wrote:
>>
>> Clover and its driver backends are unmaintained and there is no
>> production quality Mesa OpenCL driver. If you are interested in
>> working on it, feel free, but if you upgrade to a ROCm-capable
On June 4, 2017 5:36:57 PM Kenneth Graunke wrote:
On Sunday, June 4, 2017 3:27:04 PM PDT Jason Ekstrand wrote:
How does the texture even have a format of MESA_FORMAT_NONE? That seems
like the first question to ask.
It's the window system buffer, and it actually has a
From: Dave Airlie
This just collapses a few per-stage things into a loop,
shouldn't affect anything.
Signed-off-by: Dave Airlie
---
src/amd/vulkan/radv_cmd_buffer.c | 58 +---
1 file changed, 12 insertions(+), 46
From: Dave Airlie
This just splits out some non-gfx9 bits in advance to avoid
regressions.
Signed-off-by: Dave Airlie
---
src/amd/vulkan/radv_image.c | 26 +-
1 file changed, 17 insertions(+), 9 deletions(-)
diff --git
From: Dave Airlie
This just moves the code around in preparation for gfx9 support.
Signed-off-by: Dave Airlie
---
src/amd/vulkan/radv_device.c | 60 ++--
1 file changed, 35 insertions(+), 25 deletions(-)
diff
From: Dave Airlie
In advance of GFX9 to reduce chances for regression, refactor
this code out so adding the GFX9 changes will be more obvious.
Signed-off-by: Dave Airlie
---
src/amd/vulkan/radv_device.c | 119 ++-
From: Dave Airlie
This is just ported from radeonsi.
Signed-off-by: Dave Airlie
---
src/amd/vulkan/radv_image.c | 28 ++
src/amd/vulkan/radv_radeon_winsys.h | 29 +++---
On 2 June 2017 at 17:34, Jan Vesely wrote:
> On Fri, 2017-06-02 at 17:22 +0100, Emil Velikov wrote:
>> On 2 June 2017 at 16:34, Jan Vesely wrote:
>> > On Fri, 2017-06-02 at 12:19 +0100, Emil Velikov wrote:
>> > > On 1 June 2017 at 21:28, Jan Vesely
On Sunday, June 4, 2017 3:27:04 PM PDT Jason Ekstrand wrote:
> How does the texture even have a format of MESA_FORMAT_NONE? That seems
> like the first question to ask.
It's the window system buffer, and it actually has a format of
B8G8R8A8_UNORM_SRGB...my guess is just that _Format isn't set
The CL CTS queries the max allocation size, and then attempts to
allocate buffers of that size. If any of the VRAM is in use, this
causes errors in the radeon kernel module.
It's a bit of a hack, but experimentally on my system, I can use 3/4
of the card's VRAM for a single global/constant buffer
clinfo no longer reports my discrete GCN card as unified memory
Signed-off-by: Aaron Watry
---
src/gallium/state_trackers/clover/api/device.cpp | 2 +-
src/gallium/state_trackers/clover/core/device.cpp | 5 +
src/gallium/state_trackers/clover/core/device.hpp | 1 +
3
From: Dave Airlie
This ports: 55445ff1891724c78e6573d2f8c721e14c0449fc from radeonsi
radeonsi: tell LLVM not to remove s_barrier instructions
LLVM 5.0 removes s_barrier instructions if the max-work-group-size
attribute is not set. What a surprise.
From: Dave Airlie
This is needed to add the max workgroup size attribute.
Signed-off-by: Dave Airlie
---
src/amd/common/ac_llvm_util.c | 10 ++
src/amd/common/ac_llvm_util.h | 5 +
2 files changed, 15 insertions(+)
diff --git
On 5 June 2017 at 01:29, Emil Velikov wrote:
> On 4 June 2017 at 22:47, Mauro Rossi wrote:
>> 2017-05-29 14:30 GMT+02:00 Emil Velikov :
>>> On 26 May 2017 at 16:15, Mauro Rossi wrote:
Fixes
On 4 June 2017 at 22:47, Mauro Rossi wrote:
> 2017-05-29 14:30 GMT+02:00 Emil Velikov :
>> On 26 May 2017 at 16:15, Mauro Rossi wrote:
>>> Fixes the following building error:
>>>
>>>
2017-06-05 1:46 GMT+02:00 Mauro Rossi :
> 2017-06-04 23:21 GMT+02:00 Emil Velikov :
>> On 3 June 2017 at 00:54, Jason Ekstrand wrote:
>>> Acked-by: Jason Ekstrand
>>>
>>> Maybe one of the android people
From: Emil Velikov
There's nothing platform specific in the extension - all the required
core parts are already implemented.
Signed-off-by: Emil Velikov
---
This and the following are an RFC, since I'm not 100% sure if there
isn't
From: Emil Velikov
Analogous to previous commit - only one platform (of 6?) implements
this, and there's no implementations seen in the wild that use this.
Hmm the extension isn't even available anywhere :-\
Signed-off-by: Emil Velikov
From: Emil Velikov
Only the X11+DRI2 platform supports it and AFAICT only Chrome uses it.
Other EGL implementations, barring the Nvidia one, do not implement the
extension.
Which is understandable since it is superseded by EGL_EXT_buffer_age,
2017-06-04 23:21 GMT+02:00 Emil Velikov :
> On 3 June 2017 at 00:54, Jason Ekstrand wrote:
>> Acked-by: Jason Ekstrand
>>
>> Maybe one of the android people would like to test it?
>>
> Mauro and RobH (CCed) are our Android
From: Emil Velikov
As mentioned in last commit - pthread manual suggests using -pthread.
Furthermore, to the best of my knowledge anything built with GCC and
Clang should just work (tm) with the said flag.
AFAICT using the Sun or Intel compiler may need special
From: Emil Velikov
Unneeded, since the PKG_CHECK_MODULES macro already does the
substitution of the package Cflags/Libs.
Signed-off-by: Emil Velikov
---
configure.ac | 2 --
1 file changed, 2 deletions(-)
diff --git a/configure.ac
From: Emil Velikov
The HAVE_ANDROID conditional has been unused as of commit 51accecce77
("mesa/dri: always link against shared glapi") and with that one gone we
no longer need the host detection.
Cc: Chad Versace
Cc: Nicolas Boichat
As described inline - follow what's written in the manual and what works
for all platforms that Mesa supports.
We want to untangle things leaving only -pthread, yet that has a
potential of causing regressions. Thus we'll do it as a follow-up patch.
As a nice side-effect this resolves issues,
How does the texture even have a format of MESA_FORMAT_NONE? That seems
like the first question to ask.
On Sun, Jun 4, 2017 at 12:26 PM, Kenneth Graunke
wrote:
> Since commit e9f5004d5ea38a21019e0affcd63d353b98d499a, KDE's Plasma
> Shell started crashing when
On 4 June 2017 at 21:27, Benedikt Schemmer wrote:
> Hi all,
>
> my reasoning wasn't based on how c handles partial initializations,
> but on how the generated files look like: like something is missing.
>
> Also I don't want to suppress these warnings, because I don't have
> the
On 01/06/17 23:05, Samuel Pitoiset wrote:
Signed-off-by: Samuel Pitoiset
---
src/mesa/main/teximage.c | 18 ++
1 file changed, 18 insertions(+)
diff --git a/src/mesa/main/teximage.c b/src/mesa/main/teximage.c
index 9285a96b68..ca8090f653 100644
---
On 01/06/17 23:05, Samuel Pitoiset wrote:
Signed-off-by: Samuel Pitoiset
---
src/mesa/main/copyimage.c | 78 ---
1 file changed, 47 insertions(+), 31 deletions(-)
diff --git a/src/mesa/main/copyimage.c
On 01/06/17 23:05, Samuel Pitoiset wrote:
Signed-off-by: Samuel Pitoiset
---
src/mapi/glapi/gen/ARB_direct_state_access.xml | 2 +-
src/mesa/main/blit.c | 30 ++
src/mesa/main/blit.h | 9
Reviewed-by: Timothy Arceri
On 01/06/17 23:05, Samuel Pitoiset wrote:
Signed-off-by: Samuel Pitoiset
---
src/mapi/glapi/gen/ARB_framebuffer_object.xml | 2 +-
src/mesa/main/blit.c | 15 +++
5-7:
Reviewed-by: Timothy Arceri
On 01/06/17 23:04, Samuel Pitoiset wrote:
Signed-off-by: Samuel Pitoiset
---
src/mapi/glapi/gen/gl_API.xml | 2 +-
src/mesa/main/bufferobj.c | 10 ++
src/mesa/main/bufferobj.h | 3 +++
3
2017-05-29 14:30 GMT+02:00 Emil Velikov :
> On 26 May 2017 at 16:15, Mauro Rossi wrote:
>> Fixes the following building error:
>>
>> external/mesa/src/gallium/drivers/svga/svga_screen.c:26:10:
>> fatal error: 'git_sha1.h' file not found
>>
On 01/06/17 23:04, Samuel Pitoiset wrote:
Signed-off-by: Samuel Pitoiset
---
src/mesa/main/bufferobj.c | 15 +++
1 file changed, 11 insertions(+), 4 deletions(-)
diff --git a/src/mesa/main/bufferobj.c b/src/mesa/main/bufferobj.c
index
You should probably at least mention you are adding no_error support to
vertex_array_vertex_buffers() which will be used in a following patch in
the commit message.
With that:
Reviewed-by: Timothy Arceri
On 01/06/17 23:04, Samuel Pitoiset wrote:
Signed-off-by: Samuel
Hi Lucas,
On Sun, Jun 4, 2017 at 4:06 PM, Lucas Stach wrote:
> Hi all,
>
> I decided to take a look at our current piglit status and the following are
> some stright foward fixes for some of the issues.
>
> The first 3 patches fix some resource copy issues, that may send the
On 3 June 2017 at 00:54, Jason Ekstrand wrote:
> Acked-by: Jason Ekstrand
>
> Maybe one of the android people would like to test it?
>
Mauro and RobH (CCed) are our Android people.
Gents can you give the following a try
From: Dave Airlie
This just aligns with how anv does it.
Signed-off-by: Dave Airlie
---
src/amd/vulkan/radv_formats.c | 76 ++-
1 file changed, 46 insertions(+), 30 deletions(-)
diff --git
Signed-off-by: Bas Nieuwenhuizen
---
src/amd/vulkan/radv_device.c | 24 ++--
src/amd/vulkan/radv_private.h | 1 +
2 files changed, 23 insertions(+), 2 deletions(-)
diff --git a/src/amd/vulkan/radv_device.c b/src/amd/vulkan/radv_device.c
index
Signed-off-by: Bas Nieuwenhuizen
---
src/amd/vulkan/radv_image.c | 7 ++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/src/amd/vulkan/radv_image.c b/src/amd/vulkan/radv_image.c
index 9a76d285242..22bc6b41da8 100644
--- a/src/amd/vulkan/radv_image.c
+++
From: Dave Airlie
This adds support for exporting 2D images, to an
opaque fd.
This implements the:
VK_KHX_external_memory_capabilities
VK_KHX_external_memory
VK_KHX_external_memory_fd
extensions.
These are used by SteamVR, we should work with anv
to decide if we should
On Sun, Jun 4, 2017 at 3:06 PM, Lucas Stach wrote:
> + /* If the complete render target is written, set full_overwrite:
> +* - The color mask is
> +* - No blending is used
> +*/
> + bool full_overwrite = (rt0->colormask == 0xf) && !blend->enable;
> +
For patches 25 - 31:
Reviewed-by: Marek Olšák
Marek
On Tue, May 30, 2017 at 10:35 PM, Samuel Pitoiset
wrote:
> Hi,
>
> This is the v2 of ARB_bindless_texture for RadeonSI.
>
> Major changes since v1:
> - fix use of 64-bit handles on 32-bit
Hi all,
my reasoning wasn't based on how c handles partial initializations,
but on how the generated files look like: like something is missing.
Also I don't want to suppress these warnings, because I don't have
the experience you have with the codebase and would kind of like to
see if my
Hi Marek,
Do you have any other reasons besides it not improving correctness?
I'd like to pick at least the radv one, as the code doesn't get less
clear, and using 5 zeros for struct with 6 members is just plain
silly.
- Bas
On Sun, Jun 4, 2017 at 9:57 PM, Marek Olšák wrote:
Hi Jason,
How about just using vblank_mode=0 for mailbox and vblank_mode=3 for
fifo? While the semantics probably don't line up perfectly I think it
is close enough, and means people don't need to use different flags
depending on whether the app uses vulkan or GL.
- Bas
On Sun, Jun 4, 2017 at
NAK.
In C/C++, the initializer is used to clear the memory to 0s, thus,
adding 0s to the initializer is redundant and unnecessary. Empty
initializer {} is also commonly used instead of memset.
You need to suppress this warning if you don't want to see it.
Marek
On Sun, Jun 4, 2017 at 9:45 AM,
Nothing like that exists at the moment. I've considered adding it. It
would probably be more of a VK_FORCE_SWAPCHAIN_MODE=MAILBOX.
On June 4, 2017 8:24:29 AM Jan Ziak <0xe2.0x9a.0...@gmail.com> wrote:
Hi
Exists there an equivalent to vblank_mode in the radv implementation? Such as:
$
Since commit e9f5004d5ea38a21019e0affcd63d353b98d499a, KDE's Plasma
Shell started crashing when intel_disable_rb_aux_buffer tried to
dereference irb->mt, and there was no miptree. This happened during
intel_texture_view_requires_resolve, coming from a glClear().
Apparently intel_tex->_Format is
Now that we support RB swapped targets by using a shader variant, we
must derive the color mask from both the blend state and the bound
framebuffer.
Fixes piglit: fbo-colormask-formats
Signed-off-by: Lucas Stach
---
src/gallium/drivers/etnaviv/etnaviv_blend.c | 48
The maximum LOD bias supported is the same as the max texture level
supported.
Fixes piglit: ext_texture_lod_bias
Signed-off-by: Lucas Stach
---
src/gallium/drivers/etnaviv/etnaviv_screen.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git
This replaces the open coded etnaviv version of the color pack with the
common util_pack_color.
Fixes piglits:
arb_color_buffer_float-clear
fcc-front-buffer-distraction
fbo-clearmipmap
Signed-off-by: Lucas Stach
---
src/gallium/drivers/etnaviv/etnaviv_clear_blit.c | 13 ++-
When copying a resource fully we can just blit the whole level. This allows
to use the RS even for level sizes not aligned to the RS min alignment. This
is especially useful, as etna_copy_resource is part of the software fallback
paths (used in etna_transfer), that are used for doing unaligned
etna_resource_copy_region handles resources with multiple samples
by falling back to the software path. There is no need to kill the
application there.
Signed-off-by: Lucas Stach
---
src/gallium/drivers/etnaviv/etnaviv_clear_blit.c | 2 --
1 file changed, 2 deletions(-)
diff
If the blit region is not aligned to the RS min alignment don't try
to execute the blit, but fall back to the software path.
Signed-off-by: Lucas Stach
---
src/gallium/drivers/etnaviv/etnaviv_clear_blit.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git
The rules to generate egd_tables.h are added in Android makefile
Fixes: f42fb00 "r600/eg: add support for tracing IBs after a hang."
---
src/gallium/drivers/r600/Android.mk | 10 ++
1 file changed, 10 insertions(+)
diff --git a/src/gallium/drivers/r600/Android.mk
Hi
Exists there an equivalent to vblank_mode in the radv implementation? Such as:
$ vblank_mode=0 vulkan-demos/bin/computenbody
-Jan
___
mesa-dev mailing list
mesa-dev@lists.freedesktop.org
https://lists.freedesktop.org/mailman/listinfo/mesa-dev
On Jun 4, 2017 4:04 AM, "Dieter Nützel" wrote:
It improves Uningine Heaven performance here (RX 580, 8 GB),too.
without Tessellation
Before:
FPS: 79.0
Score:1991
Min FPS: 20.8
Max FPS: 189.8
After:
FPS: 79.1
Score:1993
Min FPS: 19.6
Max FPS: 185.9
https://bugs.freedesktop.org/show_bug.cgi?id=95346
--- Comment #24 from Médéric Boquien ---
I made a new test with Stellaris 1.6.1. With Mesa 17.1.0 on an RX480, the
results are unchanged, with broken planets both in the intro screen and
in-game. Seconding the previous comment,
Fix missing initializer warning in vk_format_table.h by adding appropriate
default fields in vk_format_table.py
and correct the autogenerated from message
---
src/amd/vulkan/vk_format_table.py | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git
Fix missing initializer warning in sid_tables.h by adding appropriate default
fields in sid_tables.py
---
src/amd/common/sid_tables.py | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/amd/common/sid_tables.py b/src/amd/common/sid_tables.py
index fd88d3c9d5..691d766b08
Fix missing initializer warning in egd_tables.h by adding appropriate default
fields in egd_tables.py
---
src/gallium/drivers/r600/egd_tables.py | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/src/gallium/drivers/r600/egd_tables.py
I send these patches as a series because they share the same problem:
all have missing fields in the default initialization path for field/register
definitions.
This causes several hundreds (thousands?) of "missing initializer" warnings
from the compiler making it hard to spot less frequent
67 matches
Mail list logo