Hi,
the patch no longer applies to mesa master. I've pushed the rebased
patch set to my github mirror (see below), but should I also send the
corrected patches here?
Best,
Gert
Am Montag, den 17.07.2017, 17:47 +0200 schrieb Gert Wollny:
> Hello Nicolai,
>
> many thanks for t
> The original code has a loop over num_renames, so if two renames
> were in the list in order like
>
> 8 renames to 4 and 4 renames to 2, then 8 would rename to 2.
> however if 4 renamed to 2 then 8 renamed to 4 (not sure if this could
> happen), it was different.
>
> My initial patch fails
Just a few comments while updating the patch:
> One more comment about the debug_log: as written, it will getenv
> every time. It would be better to use the approach taken by e.g.
> should_clone_nir() in compiler/nir/nir.h. There's no reason to have
> a whole class for this.
Actually,
ction *op)
-{
- return op->info->is_tex || is_resource_instruction(op->op) ?
- op->info->num_src - 1 : op->info->num_src;
-}
glsl_to_tgsi_instruction *
glsl_to_tgsi_visitor::emit_asm(ir_instruction *ir, unsigned op,
diff --git a/src/mesa/state_tracker/st_glsl_to_tgsi_private.cp
This patch replaces the old register lifetime estiamtion and
rename mapping evaluation with the new one.
Performance to compare between the current and the new implementation
were measured by running the shader-db in one thread.
---
The remapping evaluator first sorts the temporary registers ascending
based on their first life time instruction, and then uses a binary search
to find merge canidates.
For the initial sorting it uses std::sort because qsort is quite slow in
comparison. By removing the define USE_STL_SORT in
/mesa/state_tracker/st_glsl_to_tgsi_temprename.cpp
new file mode 100644
index 00..46a707ec74
--- /dev/null
+++ b/src/mesa/state_tracker/st_glsl_to_tgsi_temprename.cpp
@@ -0,0 +1,881 @@
+/*
+ * Copyright © 2017 Gert Wollny
+ *
+ * Permission is hereby granted, free of charge, to any person
for the evaluation.
Best,
Gert
Gert Wollny (6):
mesa/st: glsl_to_tgsi move some helper classes to extra files
mesa/st: glsl_to_tgsi: implement new temporary register lifetime tracker
mesa/st: glsl_to_tgsi: add tests for the new temporary lifetime tracker
mesa/st: glsl_to_tgsi: add
The patch adds tests for the register rename mapping evaluation and
combined life time estimation and renaming.
---
.../tests/test_glsl_to_tgsi_lifetime.cpp | 192 +
1 file changed, 192 insertions(+)
diff --git
/state_tracker/tests/test_glsl_to_tgsi_lifetime.cpp
b/src/mesa/state_tracker/tests/test_glsl_to_tgsi_lifetime.cpp
new file mode 100644
index 00..b9dadcc03d
--- /dev/null
+++ b/src/mesa/state_tracker/tests/test_glsl_to_tgsi_lifetime.cpp
@@ -0,0 +1,1405 @@
+/*
+ * Copyright © 2017 Gert Wollny
Gert Wollny:
> Dear Nicolai,
>
> this new version of the patch set that should address all the
> comments you gave
> for v5.
>
> Changes are:
>
> - the components are now tracked individually and the life time of a
> temporary
> is evaluated by merging t
Hello Marek,
this patch makes most of the tesselation shaders in piglit crash
because now you are passing MESA_SHADER_TESS_* to st_get_basic_variant,
but this function tests for PIPE_SHADER_TESS_* when allocating new
variants. Hence, the function doesn't fulfill the assert in debug
mode, or it
This patch replaces the old register lifetime estiamtion and
rename mapping evaluation with the new one.
Performance to compare between the current and the new implementation
were measured by running the shader-db in one thread.
---
b/src/mesa/state_tracker/st_glsl_to_tgsi_temprename.cpp
new file mode 100644
index 00..f85a6fa7c9
--- /dev/null
+++ b/src/mesa/state_tracker/st_glsl_to_tgsi_temprename.cpp
@@ -0,0 +1,949 @@
+/*
+ * Copyright © 2017 Gert Wollny
+ *
+ * Permission is hereby granted, free of charge, to any
The remapping evaluator first sorts the temporary registers ascending
based on their first life time instruction, and then uses a binary search
to find merge canidates.
For the initial sorting it uses std::sort because qsort is quite slow in
comparison. By removing the define USE_STL_SORT in
/state_tracker/tests/test_glsl_to_tgsi_lifetime.cpp
b/src/mesa/state_tracker/tests/test_glsl_to_tgsi_lifetime.cpp
new file mode 100644
index 00..d99176e86c
--- /dev/null
+++ b/src/mesa/state_tracker/tests/test_glsl_to_tgsi_lifetime.cpp
@@ -0,0 +1,1364 @@
+/*
+ * Copyright © 2017 Gert Wollny
The patch adds tests for the register rename mapping evaluation and
combined life time estimation and renaming.
---
.../tests/test_glsl_to_tgsi_lifetime.cpp | 192 +
1 file changed, 192 insertions(+)
diff --git
ction *op)
-{
- return op->info->is_tex || is_resource_instruction(op->op) ?
- op->info->num_src - 1 : op->info->num_src;
-}
glsl_to_tgsi_instruction *
glsl_to_tgsi_visitor::emit_asm(ir_instruction *ir, unsigned op,
diff --git a/src/mesa/state_tracker/st_glsl_to_tgsi_private.cp
writing a shader
that really needs it and I think I will submit this to piglit.
best regards,
Gert
Gert Wollny (6):
mesa/st: glsl_to_tgsi move some helper classes to extra files
mesa/st: glsl_to_tgsi: implement new temporary register lifetime
tracker
mesa/st: glsl_to_tgsi: add tests
+0200 schrieb Nicolai Hähnle:
> On 27.06.2017 11:32, Gert Wollny wrote:
>
> Yes, I also mean below.
Okay, done.
>
>
> [snip]
> > > + case TGSI_OPCODE_CONT: {
> > > > + cur_scope->set_continue_line(line);
> > >
> > > I
Am Dienstag, den 01.08.2017, 09:32 +1000 schrieb Dave Airlie:
> From: Dave Airlie
>
> This reverts commit 3008161d28e38336ba39aba4769a2deaf9732f55,
> which caused a regression for VMWare.
>
> The initial code had some recursion in it, that I removed by accident
> trying to
/st_glsl_to_tgsi_temprename.cpp
new file mode 100644
index 00..46a707ec74
--- /dev/null
+++ b/src/mesa/state_tracker/st_glsl_to_tgsi_temprename.cpp
@@ -0,0 +1,881 @@
+/*
+ * Copyright © 2017 Gert Wollny
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software
The remapping evaluator first sorts the temporary registers ascending
based on their first life time instruction, and then uses a binary search
to find merge canidates.
For the initial sorting it uses std::sort because qsort is quite slow in
comparison. By removing the define USE_STL_SORT in
The patch adds tests for the register rename mapping evaluation and
combined life time estimation and renaming.
---
.../tests/test_glsl_to_tgsi_lifetime.cpp | 192 +
1 file changed, 192 insertions(+)
diff --git
From: Dave Airlie
Instead of having to search the whole array, just use the whole
thing and store a valid bit in there with the rename.
Removes this from the profile on some of the fp64 tests
Reviewed-by: Timothy Arceri
Signed-off-by: Dave Airlie
instruction *op)
-{
- return op->info->is_tex || is_resource_instruction(op->op) ?
- op->info->num_src - 1 : op->info->num_src;
-}
glsl_to_tgsi_instruction *
glsl_to_tgsi_visitor::emit_asm(ir_instruction *ir, unsigned op,
diff --git a/src/mesa/state_tracker/st_gl
© 2017 Gert Wollny
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction,including without limitation
+ * the rights to use, copy, mod
ert
Dave Airlie (1):
st_glsl_to_tgsi: rewrite rename registers to use array fully.
Gert Wollny (6):
mesa/st: glsl_to_tgsi move some helper classes to extra files
mesa/st: glsl_to_tgsi: implement new temporary register lifetime tracker
mesa/st: glsl_to_tgsi: add tests for the new tempor
This patch replaces the old register lifetime estiamtion and
rename mapping evaluation with the new one.
Performance to compare between the current and the new implementation
were measured by running the shader-db in one thread.
---
Am Freitag, den 18.08.2017, 18:25 +0200 schrieb Marek Olšák:
>
> Sure. The changes are here:
> https://cgit.freedesktop.org/~mareko/mesa/log/?h=blitter-optz
>
> If somebody wants to test it on r600g, that would be appreciated too.
>
>
I did this and unfortunately it crashes with 744c3e2254
Am Freitag, den 18.08.2017, 09:43 +1000 schrieb Timothy Arceri:
> This fixes an assert during IR validation in LLVMpipe.
>
> Fixes: e2e2c5abd279 (glsl: calculate number of operands in an
> expression once)
> Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=102274
> Cc: Brian Paul
Am Samstag, den 19.08.2017, 12:13 +0200 schrieb Gert Wollny:
>
> I did this and unfortunately it crashes with 744c3e2254 applied.
> Specifically, in
>
> src/gallium/drivers/r600/evergreen_state.c2302:
> evergreen_emit_vertex_fetch_shader
>
> de-referenc
Dear all,
(and specifically to Dave Arlie),
when looking at the TGSI code of shaders that fail on r600g because of
too many temporaries used, I notes that often their lifetime is very
short, like just assigned in one line and read the only time in the
next line.
Looking on the
Hello Nicolai,
Am Montag, den 12.06.2017, 12:17 +0200 schrieb Nicolai Hähnle:
> > spec/arb_shader_bit_encoding/execution/and-clamp: fail fail pass
> > fail
> >
> >
> >
> It's disconcerting that you have tests here whose pass status is
> unstable. Those tests really should be deterministic.
Am Montag, den 12.06.2017, 12:28 +0200 schrieb Nicolai Hähnle:
> On 10.06.2017 01:15, Gert Wollny wrote:
> >
> Plenty of style issues aside, can you explain where and why you get
> tighter lifetimes?
In the original code if a temporary is used within a loop it gets the
This patch replaces the old register livetime estimation with the
new approach.
---
src/mesa/state_tracker/st_glsl_to_tgsi.cpp | 17 +++--
1 file changed, 15 insertions(+), 2 deletions(-)
diff --git a/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
/src/mesa/state_tracker/st_glsl_to_tgsi_temprename.cpp
new file mode 100644
index 00..389a4b6b5f
--- /dev/null
+++ b/src/mesa/state_tracker/st_glsl_to_tgsi_temprename.cpp
@@ -0,0 +1,551 @@
+/*
+ * Copyright © 2017 Gert Wollny
+ *
+ * Permission is hereby granted, free of charge, to any person
->op) ?
- op->info->num_src - 1 : op->info->num_src;
-}
glsl_to_tgsi_instruction *
glsl_to_tgsi_visitor::emit_asm(ir_instruction *ir, unsigned op,
diff --git a/src/mesa/state_tracker/st_glsl_to_tgsi_private.cpp
b/src/mesa/state_tracker/st_glsl_to_tgsi_private.cpp
new file mode
no commit rights.
many thanks,
Gert
[1] http://www.geeks3d.com/gputest/
Gert Wollny (3):
mesa/st: glsl_to_tgsi move some helper classes to extra files
mesa/st: glsl_to_tgsi Implement a new lifetime tracker for temporaries
mesa/st: glsl_to_tgsi: tie in the new register renaming approach
co
@@
+/*
+ * Copyright © 2017 Gert Wollny
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, co
glsl_to_tgsi_instruction *op)
-{
- return op->info->is_tex || is_resource_instruction(op->op) ?
- op->info->num_src - 1 : op->info->num_src;
-}
glsl_to_tgsi_instruction *
glsl_to_tgsi_visitor::emit_asm(ir_instruction *ir, unsigned op,
diff --git a/src/mesa/state
This patch ties in the new temporary register lifetime estiamtion and
rename mapping evaluation. In order to enable it, the evironment
variable MESA_GLSL_TO_TGSI_NEW_MERGE must be set.
Performance to compare between the current and the new implementation
were measured by running the shader-db in
The remapping evaluator first sorts the temporary registers ascending
based on their first life time instruction, and then uses a binary search
to find merge canidates.
For the initial sorting it uses std::sort because qsort is quite slow in
comparison. By removing the define USE_STL_SORT in
/test_glsl_to_tgsi_lifetime.cpp
new file mode 100644
index 00..5f3378637a
--- /dev/null
+++ b/src/mesa/state_tracker/tests/test_glsl_to_tgsi_lifetime.cpp
@@ -0,0 +1,976 @@
+/*
+ * Copyright © 2017 Gert Wollny
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining
ndications that registers would be megrged
wrongly.
Many thanks for any commenst,
Gert
** BLURB HERE ***
Gert Wollny (6):
mesa/st: glsl_to_tgsi move some helper classes to extra files
mesa/st: glsl_to_tgsi: implement new temporary register lifetime
tracker
mesa/st: glsl_to_tgsi:
The patch adds tests for the register rename mapping evaluation.
---
.../tests/test_glsl_to_tgsi_lifetime.cpp | 94 ++
1 file changed, 94 insertions(+)
diff --git a/src/mesa/state_tracker/tests/test_glsl_to_tgsi_lifetime.cpp
Thanks for the comments,
I've fixed these little issues locally, but I think in order to not to
spam the list, I'll send the changes later. I kind of suspect that
Nicolai might have one or the other additional comment :)
best,
Gert
___
mesa-dev
The patch adds tests for the register rename mapping evaluation.
---
.../tests/test_glsl_to_tgsi_lifetime.cpp | 94 ++
1 file changed, 94 insertions(+)
diff --git a/src/mesa/state_tracker/tests/test_glsl_to_tgsi_lifetime.cpp
/state_tracker/tests/test_glsl_to_tgsi_lifetime.cpp
b/src/mesa/state_tracker/tests/test_glsl_to_tgsi_lifetime.cpp
new file mode 100644
index 00..5f3378637a
--- /dev/null
+++ b/src/mesa/state_tracker/tests/test_glsl_to_tgsi_lifetime.cpp
@@ -0,0 +1,976 @@
+/*
+ * Copyright © 2017 Gert Wollny
The remapping evaluator first sorts the temporary registers ascending
based on their first life time instruction, and then uses a binary search
to find merge canidates.
For the initial sorting it uses std::sort because qsort is quite slow in
comparison. By removing the define USE_STL_SORT in
glsl_to_tgsi_instruction *op)
-{
- return op->info->is_tex || is_resource_instruction(op->op) ?
- op->info->num_src - 1 : op->info->num_src;
-}
glsl_to_tgsi_instruction *
glsl_to_tgsi_visitor::emit_asm(ir_instruction *ir, unsigned op,
diff --git a/src/mesa/state
This patch ties in the new temporary register lifetime estiamtion and
rename mapping evaluation. In order to enable it, the evironment
variable MESA_GLSL_TO_TGSI_NEW_MERGE must be set.
Performance to compare between the current and the new implementation
were measured by running the shader-db in
@@
+/*
+ * Copyright © 2017 Gert Wollny
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, co
without noting any regressions.
As noted before, I don't have write access to mesa-git, so I'll need someone
who sponsors this patch.
Many thanks for any additional comments,
Gert
Gert Wollny (6):
mesa/st: glsl_to_tgsi move some helper classes to extra files
mesa/st: glsl_to_tgsi
Thanks for your comments
Am Montag, den 26.06.2017, 14:52 +0200 schrieb Nicolai Hähnle:
> Thanks for the update. First off, you're still not tracking
> individual
> components, but that's absolute necessary. Think:
>
> BGNLOOP
> MOV TEMP[1].x, ...
>
> UIF ...
> MOV TEMP[1].y, ...
Hello all,
Hardware; Radeon 6850HD,
Mesa: mesa 17.0.1 and git (sha 531887),
llvm:
4.0.0
Playing a bit around with the Unreal Editor I was confronted with the
same error message reported in #99349, i.e. "Failed to build shader
(translation from TGSI).
After some digging though the code I
mments,
Gert Wollny (1):
r600g: work around shaders allocating too many superflous temporaries
src/gallium/drivers/r600/r600_shader.c | 52 +++---
1 file changed, 42 insertions(+), 10 deletions(-)
--
2.13.0
___
mesa-dev m
Related bugs:
https://bugs.freedesktop.org/show_bug.cgi?id=99349
https://bugs.freedesktop.org/show_bug.cgi?id=50338
1. Allocate ctx.temp_reg and a limited number of registers
(R600_TEMP_REG_RESERVED=10) that are given out via r600_get_temp() before
the temporaries of the TGSI are allocated. That
Hello,
I should probably add that I don't have write access to the mesa-git.
Best,
Gert
Am Dienstag, den 30.05.2017, 14:14 +0200 schrieb Gert Wollny:
> Dear all,
>
> like I pointed out yesterday I've got a workaround for shaders that
> allocate
> too many temporaries afte
Am Montag, den 12.06.2017, 21:00 +0200 schrieb Nicolai Hähnle:
Thanks for you comments, although I do not agree with most of them.
>
> > Okay. I think you should seriously re-think your algorithm in a way
> > that makes it a more natural evolution from the algorithm that's
> > already there.
into this.
Best,
Gert
> >
> > Marek
> >
> > On Sat, Jun 10, 2017 at 1:15 AM, Gert Wollny <gw.foss...@gmail.com>
> > wrote:
> > > Dear all,
> > >
> > > as I wrote before, I was looking into the temporary register
> > > ren
Am Sonntag, den 11.06.2017, 19:21 +0200 schrieb Gert Wollny:
>
> > >
> > > Have you measured the CPU overhead of the new code?
>
> So far no, I guess one would do that with the shader-db to get
> reasonable complex shaders, but I only have a r600 based card so I'
Am Montag, den 12.06.2017, 15:44 +0900 schrieb Michel Dänzer:
> On 10/06/17 08:15 AM, Gert Wollny wrote:
> >
> > Piglit shows 7 fixes and 6 regressions compared to git 8fac894f,
> > but they don't
> > seem to be related to shaders.
>
> Which tests regressed (m
Am Dienstag, den 13.06.2017, 11:07 +0200 schrieb Nicolai Hähnle:
>
>
> I'm curious what you'd suggest for getting rid of allocations anyway.
As the refactoring goes I think I will end up with a hybrid approach:
In the temporaries I will not keep the full time line, but the
important read/write
@@ -0,0 +1,202 @@
+/*
+ * Copyright © 2010 Intel Corporation
+ * Copyright © 2011 Bryan Cain
+ * Copyright © 2017 Gert Wollny
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+
This patch replaces the old register livetime estimation with the
new approach.
---
src/mesa/state_tracker/st_glsl_to_tgsi.cpp | 16 ++--
1 file changed, 14 insertions(+), 2 deletions(-)
diff --git a/src/mesa/state_tracker/st_glsl_to_tgsi.cpp
To prepare the implementation of a temp register lifetime tracker
some of the classes and functions are moved into seperate header/
implementation files to make them accessible from other files.
Specifically these are:
class st_src_reg;
class st_dst_reg;
class glsl_to_tgsi_instruction;
any thanks for any comments,
Gert
Gert Wollny (3):
mesa/st: glsl_to_tgsi move some helper classes to extra files
mesa/st: glsl_to_tgsi Implement a new lifetime tracker for temporaries
mesa/st: glsl_to_tgsi: tie in the new register renaming approach
co
Hello Emil,
Am Freitag, den 16.06.2017, 15:21 +0100 schrieb Emil Velikov:
>
> Please don't use STL within core mesa code.
May I ask why? I always try to not re-implement already available
functionality and since mesa already uses C++ it seems kind of natural
to use the STL because it provides
+0200 schrieb Gert Wollny:
> Dear all,
>
> following the comments of Emil and Nicolai I've updated the patch
> set.
>
> Changes with respect to the old version are:
>
> - split the changes into more patches
> - correct formatting errors
> - remove the use of the
/src/mesa/state_tracker/tests/test_glsl_to_tgsi_lifetime.cpp
new file mode 100644
index 00..7e07f8868f
--- /dev/null
+++ b/src/mesa/state_tracker/tests/test_glsl_to_tgsi_lifetime.cpp
@@ -0,0 +1,969 @@
+/*
+ * Copyright © 2017 Gert Wollny
+ *
+ * Permission is hereby granted, free of charge
glsl_to_tgsi_instruction *op)
-{
- return op->info->is_tex || is_resource_instruction(op->op) ?
- op->info->num_src - 1 : op->info->num_src;
-}
glsl_to_tgsi_instruction *
glsl_to_tgsi_visitor::emit_asm(ir_instruction *ir, unsigned op,
diff --git a/src/mesa/state
The patch adds tests for the register rename mapping evaluation.
---
.../tests/test_glsl_to_tgsi_lifetime.cpp | 71 --
1 file changed, 66 insertions(+), 5 deletions(-)
diff --git a/src/mesa/state_tracker/tests/test_glsl_to_tgsi_lifetime.cpp
@@
+/*
+ * Copyright © 2017 Gert Wollny
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, co
The remapping evaluator first sorts the temporary registers ascending
based on their first life time instruction, and then uses a binary search
to find merge canidates.
For the initial sorting it uses std::sort because qsort is quite slow in
comparison. By removing the define USE_STL_SORT in
For the new register renaming approach C++11 is wanted.
---
src/mesa/Makefile.am | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/src/mesa/Makefile.am b/src/mesa/Makefile.am
index 53f311d2a9..3339926d93 100644
--- a/src/mesa/Makefile.am
+++ b/src/mesa/Makefile.am
@@ -101,7
This patch ties in the new temporary register lifetime estiamtion and
rename mapping evaluation. In order to enable it, the evironment
variable MESA_GLSL_TO_TGSI_NEW_MERGE must be set.
Performance to compare between the current and the new implementation
were measured by running the shader-db in
Hello Dieter,
> W'll start with your attachment
> https://bugs.freedesktop.org/attachment.cgi?id=131683
This one is a dirty hack that works around the use of too many
registers mostly by allowing to allocate more than the allowed limit
and let the bytecode optimizer sort it out (there is also
r register-renaming.
The performance numbers estimated by running the shader-db are given in the
commit message of the last patch, the trend is the same like reported before.
Many thanks for any commenst,
Gert
Gert Wollny (7):
mesa/st: glsl_to_tgsi move some helper classes to extra files
mesa:
Hi,
> > >+typedef int scope_idx;
> > >
> > > Please remove this, it's an unnecessary and distracting
> > > abstraction that doesn't gain you anything.
> >
> > Actually, with the refactoring it did the last two days it helped a
> > lot.
>
> How? Perhaps your variable names stand to be
Hello Nicolai,
Am Sonntag, den 18.06.2017, 12:05 +0200 schrieb Nicolai Hähnle:
>
> > if HAVE_XLIB_GLX
> > SUBDIRS += drivers/x11
> > @@ -101,7 +101,7 @@ AM_CFLAGS = \
> > $(VISIBILITY_CFLAGS) \
> > $(MSVC2013_COMPAT_CFLAGS)
> > AM_CXXFLAGS = \
> > - $(LLVM_CFLAGS) \
> > +
> > +TEST_F(LifetimeEvaluatorExactTest, SimpleMoveAddMoveTexoffset)
> > +{
> > + const vector code = {
> > + { TGSI_OPCODE_MOV, {1}, {in0}, {}},
> > + { TGSI_OPCODE_MOV, {2}, {in1}, {}},
> > + { TGSI_OPCODE_UADD, {out0}, {}, {1,2}},
>
> UADD doesn't have texoffsets.
The test
Am Montag, den 26.06.2017, 15:13 +0200 schrieb Nicolai Hähnle:
> Thanks for the update. Do you have the series on an accessible git
> repository somewhere? E.g. on GitHub or Gitlab or wherever? That
> would be helpful.
I've put the code on
https://github.com/gerddie/mesa
and I am already
Sorry for the lapse and thanks for fixing it.
Reviewed-By: Gert Wollny <gw.foss...@gmail.com>
Am Mittwoch, den 13.09.2017, 19:05 +0200 schrieb Nicolai Hähnle:
> From: Nicolai Hähnle <nicolai.haeh...@amd.com>
>
> It can't *really* happen since we don't use subrout
libunwind is a optional dependency used by the gallium aux module (libgallium)
and consequently the final binaries must be linked against it. To test whether
the library is properly specified in the link pass add it to the travis-ci
build environment and force its use.
---
.travis.yml | 11
In Ubuntu Trusty the default version of llvm is 3.4 and the build
was actually randomly picking 3.5 or 3.9. Adding libunwind would then
result is build success or failure depending of what version was picked.
Install the llvm-3.3-dev package and force its use: On one hand it is the
minimum
The second version of the two patches now forces llvm-3.3 for
"make Gallium ST Other" and no longer requires changing any Makefile.am.
In addition, like suggested by Emil, --enable/disable-libunwind is set
appropriately.
Best,
Gert
Gert Wollny (2):
.travis.yml: force llvm-3.3
Am Freitag, den 15.09.2017, 13:07 +0100 schrieb Emil Velikov:
> On 14 September 2017 at 11:27, Gert Wollny <gw.foss...@gmail.com>
> wrote:
> > The second version of the two patches now forces llvm-3.3 for
> > "make Gallium ST Other" and no longer req
Although gl_SampleMaskIn is declared as an array in GLSL, it is
effectively a 32 bit mask on all hardware supported by mesa, so the
array indexing is ignored (Thanks Glenn Kennard for the explanation).
Add a comment that the assert is not made superfluos by the else branch.
Corrects: piglit
Adding gbm_device_get_format_modifier_plane_count made the
test gbm-symbols-check fail, this patch adds the according
function name to the test.
Fixes: 8824141b8d48d9120ddbf542d6fb661046c41c62
(gbm: Add a gbm_device_get_format_modifier_plane_count function)
Signed-off-by: Gert Wollny <gw.f
Am Freitag, den 15.09.2017, 21:41 -0400 schrieb Jan Vesely:
[...]
> > > > > a/src/gallium/state_trackers/clover/llvm/compat.hpp
> > > > > b/src/gallium/state_trackers/clover/llvm/compat.hpp
> > > > > index b22ecfa6b2..9bca6d82e8 100644
> > > > > ---
The tests in mesa/st/tests require C++11. Check whether the compiler
supports the -std=c++11 flag and if so, enable the tests. Otherwise
skip compiling and running the tests altogether.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=102665
---
- Submitter has no write access to mesa-git.
Compiles fine with this patch,
Tested-by: Gert Wollny <gw.foss...@gmail.com>
Am Samstag, den 16.09.2017, 22:05 -0400 schrieb Jan Vesely:
> Fixes build issues with llvm-3.6
> Fixes: 3115687f9b9830417c408228db2bc679e346bba6 (clover: Fix build
> after
> LLVM r313390)
>
> S
The tests in mesa/st/tests require C++11. Check whether the compiler
supports the -std=c++11 flag and if so, enable the tests. Otherwise
skip compiling and running the tests altogether.
Fixes: 7be6d8fe12 ("mesa/st: glsl_to_tgsi: add tests for the
new temporary lifetime tracker")
Bugzilla:
Am Montag, den 18.09.2017, 16:07 +0100 schrieb Emil Velikov:
> Hi Gert,
>
> On 17 September 2017 at 08:32, Gert Wollny <gw.foss...@gmail.com>
> wrote:
> > The tests in mesa/st/tests require C++11. Check whether the
> > compiler
> > supports the -std=c+
oy Stogner <royst...@ices.utexas.edu>
+# Copyright (c) 2014, 2015 Google Inc.; contributed by Alexey Sokolov
<soko...@google.com>
+# Copyright (c) 2015 Paul Norman <penor...@mac.com>
+# Copyright (c) 2015 Moritz Klammler <mor...@klammler.eu>
+# Copyright (c) 2016 Krzesimir
then adds a test to see whether the mesa/st tests are not
build with a compilet that doesn't offer -std=c++11 support.
Best,
Gert
--
- Submitter has no mesa-git write access
Gert Wollny (2):
configure+mesa/st: unify check for -std=c++11 support and enable
accordingly
travis: Add test
Add a build check for g++-4.4 which is the oldest gcc based compiler
directly supported in ubuntu/trusty, and one of the compilers that don't
have full c++11 support, i.e. it doesn't understand -std=c++11.
Since llvm >= 3.9 propagates -std=c++11 in the LLVM_CXXFLAGS, an
older llvm version is
Am Freitag, den 22.09.2017, 09:22 +0200 schrieb Samuel Pitoiset:
> That's unnecessary to double-check that dcc_offset is not 0
> because all callers already check that.
In that case I would suggest to add an assert(rtex->dcc_offset)
to ensure this.
Best,
Gert
>
> Signed-off-by: Samuel
04.10.2017, 11:45 +0200 schrieb Gert Wollny:
> Dear all,
>
> this set of patches fixes some memory leak in mesa/st/tests, enhances
> the debugging output used in the temporary register lifetime
> estimation, and further improves the lifetime tracking by handling
> if-else pat
1 - 100 of 826 matches
Mail list logo