Jason Ekstrand [Sun, 14 Feb 2016 01:14:27 +0000 (17:14 -0800)]
nir/builder: Add a helper for inserting jump instructions
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Jason Ekstrand [Thu, 24 Dec 2015 02:10:08 +0000 (18:10 -0800)]
nir/cf: Make extracting or re-inserting nothing a no-op
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Sat, 26 Dec 2015 18:32:10 +0000 (10:32 -0800)]
nir: Add a function for comparing cursors
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Jason Ekstrand [Fri, 18 Dec 2015 19:27:00 +0000 (11:27 -0800)]
nir/cf: Handle relinking top-level blocks
This can happen if a function ends in a return instruction and you remove
the return.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Sat, 13 Feb 2016 05:52:46 +0000 (21:52 -0800)]
nir: Add a pass to repair SSA form
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Jason Ekstrand [Sat, 13 Feb 2016 05:48:26 +0000 (21:48 -0800)]
nir/vars_to_ssa: Use the new nir_phi_builder helper
The efficiency should be approximately the same. We do a little more work
per phi node because we have to sort the predecessors. However, we no
longer have to walk the blocks a second time to pop things off the stack.
The bigger advantage, however, is that we can now re-use the phi placement
and per-block SSA value tracking in other passes.
As a side-benifit, the phi builder actually handles unreachable blocks
correctly. The original vars_to_ssa code, because of the way it iterated
the blocks and added phi sources, didn't add sources corresponding to
predecessors of unreachable blocks. The new strategy employed by the phi
builder creates a phi source for each predecessor and should correctly
handle unreachable blocks by setting those sources to SSA undefs.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Jason Ekstrand [Tue, 29 Dec 2015 23:25:43 +0000 (15:25 -0800)]
nir/dominance: Handle unreachable blocks
Previously, nir_dominance.c didn't properly handle unreachable blocks.
This can happen if, for instance, you have something like this:
loop {
if (...) {
break;
} else {
break;
}
}
In this case, the block right after the if statement will be unreachable.
This commit makes two changes to handle this. First, it removes an assert
and allows block->imm_dom to be null if the block is unreachable. Second,
it properly skips unreachable blocks in calc_dom_frontier_cb.
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Connor Abbott <cwabbott0@gmail.com>
Jason Ekstrand [Sat, 13 Feb 2016 05:41:42 +0000 (21:41 -0800)]
nir: Add a phi node placement helper
Right now, we have phi placement code in two places and there are other
places where it would be nice to be able to do this analysis. Instead of
repeating it all over the place, this commit adds a helper for placing all
of the needed phi nodes for a value.
v2: Add better documentation
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Jason Ekstrand [Sun, 17 Jan 2016 00:42:06 +0000 (16:42 -0800)]
util/bitset: Allow iterating over const bitsets
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Rob Clark [Thu, 24 Mar 2016 19:44:35 +0000 (15:44 -0400)]
ttn: remove stray global from header
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Samuel Pitoiset [Wed, 23 Mar 2016 22:29:20 +0000 (23:29 +0100)]
nv50/ir: silence unhandled TGSI_PROPERTY_NEXT_SHADER info
radeonsi uses this property to make the best decision about which
shader to compile, but this is not currently used by our codegen.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Kenneth Graunke [Thu, 24 Mar 2016 06:46:12 +0000 (23:46 -0700)]
mesa: Handle negative length in glPushDebugGroup().
The KHR_debug spec doesn't actually say we should handle this, but that
is most likely an oversight - it says to check against strlen and
generate errors if length is negative. It appears they just forgot to
explicitly spell out that we should then proceed to actually handle it.
Fixes crashes from uncaught std::string exceptions in many
dEQP-GLES31.functional.debug.error_filters.* tests.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>
Kenneth Graunke [Thu, 24 Mar 2016 06:35:40 +0000 (23:35 -0700)]
mesa: Make glDebugMessageInsert deal with negative length for all types.
From the KHR_debug spec, section 5.5.5 (Externally Generated Messages):
"If <length> is negative, it is implied that <buf> contains a null
terminated string. The error INVALID_VALUE will be generated if the
number of characters in <buf>, excluding the null terminator when
<length> is negative, is not less than the value of
MAX_DEBUG_MESSAGE_LENGTH."
This indicates that length should be set to strlen for all types, not
just GL_DEBUG_TYPE_MARKER. We want it to be after validate_length()
so we still generate appropriate errors.
Fixes crashes from uncaught std::string exceptions in many
dEQP-GLES31.functional.debug.error_filters.* tests.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>
Kenneth Graunke [Thu, 24 Mar 2016 04:38:42 +0000 (21:38 -0700)]
mesa: Include null terminator in GL_DEBUG_NEXT_LOGGED_MESSAGE_LENGTH.
From the KHR_debug spec:
"Applications can query the number of messages currently in the log by
obtaining the value of DEBUG_LOGGED_MESSAGES, and the string length
(including its null terminator) of the oldest message in the log
through the value of DEBUG_NEXT_LOGGED_MESSAGE_LENGTH."
Because we weren't including the null terminator, many dEQP tests
called glGetDebugMessageLog with a bufSize parameter that was 1 too
small, and unable to contain the message, so we skipped returning it,
failing many cases.
Fixes 298 dEQP-GLES31.functional.debug.* tests.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Stephane Marchesin <stephane.marchesin@gmail.com>
Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>
Nicolai Hähnle [Wed, 23 Mar 2016 20:22:16 +0000 (15:22 -0500)]
st/mesa: use RGBA instead of BGRA for SRGB_ALPHA
This fixes a regression introduced by commit
a8eea696 "st/mesa: honour sized
internal formats in st_choose_format (v2)".
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94657
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94671
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Wed, 23 Mar 2016 16:58:28 +0000 (11:58 -0500)]
radeonsi: silence a coverity warning
The following Coverity warning
5378 tmpl.fetch_args = atomic_fetch_args;
5379 tmpl.emit = atomic_emit;
>>> CID
1357115: Uninitialized variables (UNINIT)
>>> Using uninitialized value "tmpl". Field "tmpl.intr_name" is uninitialized.
5380 bld_base->op_actions[TGSI_OPCODE_ATOMUADD] = tmpl;
5381 bld_base->op_actions[TGSI_OPCODE_ATOMUADD].intr_name = "add";
... is a false positive, but what the hell. This change should "fix" it.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Bas Nieuwenhuizen [Thu, 24 Mar 2016 14:30:09 +0000 (08:30 -0600)]
mesa: replace gl_context->Multisample._Enabled with _mesa_is_multisample_enabled.
This removes any dependency on driver validation of the number of
framebuffer samples.
Signed-off-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Brian Paul <brianp@vmware.com>
Rob Clark [Tue, 22 Mar 2016 19:02:42 +0000 (15:02 -0400)]
nir: fix dangling ssadef->name ptrs
In many places, the convention is to pass an existing ssadef name ptr
when construction/initializing a new nir_ssa_def. But that goes badly
(as noticed by garbage in nir_print output) when the original string
gets freed.
Just use ralloc_strdup() instead, and add ralloc_free() in the two
places that would care (not that the strings wouldn't eventually get
freed anyways).
Also fixup the nir_search code which was directly setting ssadef->name
to use the parent instruction as memctx.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Jason Ekstrand [Thu, 24 Mar 2016 04:04:18 +0000 (21:04 -0700)]
glsl: Add propagate_invariance to the other makefile
This fixes the scons build
Jason Ekstrand [Thu, 17 Mar 2016 22:20:20 +0000 (15:20 -0700)]
nir/glsl: Propagate invariant into NIR alu ops
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Jason Ekstrand [Thu, 17 Mar 2016 21:44:57 +0000 (14:44 -0700)]
glsl/rebalance_tree: Don't handle invariant or precise trees
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Jason Ekstrand [Thu, 17 Mar 2016 21:41:14 +0000 (14:41 -0700)]
glsl/opt_algebraic: Don't handle invariant or precise trees
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Jason Ekstrand [Thu, 17 Mar 2016 20:58:40 +0000 (13:58 -0700)]
glsl: Add a pass to propagate the "invariant" and "precise" qualifiers
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Jason Ekstrand [Thu, 17 Mar 2016 23:13:40 +0000 (16:13 -0700)]
nir/alu_to_scalar: Propagate the "exact" bit
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Jason Ekstrand [Thu, 17 Mar 2016 20:39:07 +0000 (13:39 -0700)]
i965/peephole_ffma: Don't fuse exact adds
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Jason Ekstrand [Thu, 17 Mar 2016 18:38:54 +0000 (11:38 -0700)]
nir/cse: Properly handle nir_ssa_def.exact
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Jason Ekstrand [Thu, 17 Mar 2016 18:31:48 +0000 (11:31 -0700)]
nir/algebraic: Flag inexact optimizations
Many of our optimizations, while great for cutting shaders down to size,
aren't really precision-safe. This commit tries to flag all of the
inexact floating-point optimizations so they don't get run on values that
are flagged "exact". It's a bit conservative and maybe flags some safe
optimizations as unsafe but that's better than missing one.
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Jason Ekstrand [Wed, 23 Mar 2016 21:30:29 +0000 (14:30 -0700)]
nir/algebraic: Fix fmin detection to match the spec
The previous transformation got the arguments to fmin backwards. When NaNs
are involved, the GLSL min/max aren't commutative so it matters.
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Jason Ekstrand [Wed, 23 Mar 2016 21:25:56 +0000 (14:25 -0700)]
nir/algebraic: Get rid of an invlid fxor optimization
The fxor opcode is required to return 1.0f or 0.0f but the input variable
may not be 1.0f or 0.0f.
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Jason Ekstrand [Thu, 17 Mar 2016 18:04:49 +0000 (11:04 -0700)]
nir/algebraic: Allow for flagging operations as being inexact
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Jason Ekstrand [Thu, 17 Mar 2016 22:20:34 +0000 (15:20 -0700)]
nir/search: Propagate exactness into newly created expressions
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Jason Ekstrand [Thu, 17 Mar 2016 22:54:26 +0000 (15:54 -0700)]
nir/builder: Add a flag for setting exact
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Jason Ekstrand [Thu, 17 Mar 2016 17:50:27 +0000 (10:50 -0700)]
nir: Add an "exact" bit to nir_alu_instr
Reviewed-by: Francisco Jerez <currojerez@riseup.net>
Jason Ekstrand [Wed, 23 Mar 2016 22:05:55 +0000 (15:05 -0700)]
nir/clone: Export nir_variable_clone
Reviewed-by: Rob Clark <robclark@gmail.com>
Jason Ekstrand [Thu, 31 Dec 2015 02:44:19 +0000 (18:44 -0800)]
nir/clone: Expose nir_constant_clone
Reviewed-by: Rob Clark <robclark@gmail.com>
Jason Ekstrand [Wed, 23 Mar 2016 21:57:57 +0000 (14:57 -0700)]
nir: Fix whitespace
Reviewed-by: Rob Clark <robclark@gmail.com>
Brian Paul [Wed, 23 Mar 2016 18:55:45 +0000 (12:55 -0600)]
docs: use latest libDRM version
Signed-off-by: Brian Paul <brianp@vmware.com>
Lars Hamre [Wed, 23 Mar 2016 14:14:23 +0000 (10:14 -0400)]
compiler/glsl: allow sequence op as a const expr in gles 1.0
Allow the sequence operator to be a constant expression in GLSL ES
versions prior to GLSL ES 3.0
Fixes the following piglit test:
/all/spec/glsl-es-1.0/compiler/array-sized-by-sequence-in-parenthesis.vert
This is similar to the logic from process_initializer() which performs
the same check for constant variable initialization with sequence
operators.
v2: Fixed regression pointed out by Eduardo Lima Mitev
Signed-off-by: Lars Hamre <chemecse@gmail.com>
Reviewed-by: Eduardo Lima Mitev <elima@igalia.com>
Nicolai Hähnle [Mon, 21 Mar 2016 19:50:50 +0000 (14:50 -0500)]
radeonsi: fix out-of-bounds indexing of shader images
Results are undefined but may not crash. Without this change, out-of-bounds
indexing can lead to VM faults and GPU hangs.
Constant buffers, samplers, and possibly others will eventually need similar
treatment to support GL_ARB_robust_buffer_access_behavior.
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-and-Tested-by: Michel Dänzer <michel.daenzer@amd.com>
Nicolai Hähnle [Thu, 17 Mar 2016 01:47:47 +0000 (20:47 -0500)]
radeonsi: cache flush/invalidation for missing PIPE_BARRIER_*_BUFFER bits (v2)
This fixes arb_shader_image_load_store-host-mem-barrier.
v2: flush TC L2 for index buffers on <= CIK (Marek)
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Fri, 18 Mar 2016 00:49:26 +0000 (19:49 -0500)]
st/mesa: add missing MemoryBarrier bits and some explanations
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Fri, 18 Mar 2016 00:49:03 +0000 (19:49 -0500)]
gallium: add PIPE_BARRIER_STREAMOUT_BUFFER
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Marek Olšák [Tue, 22 Mar 2016 17:26:53 +0000 (18:26 +0100)]
radeonsi: fix 2D array MSAA failures since image support landed
Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Reviewed-and-Tested-by: Michel Dänzer <michel.daenzer@amd.com>
Jason Ekstrand [Thu, 17 Mar 2016 02:31:02 +0000 (19:31 -0700)]
i965/fs: Don't constant-fold RCP
No shader-db changes on Broadwell
Reviewed-by: Matt Turner <mattst88@gmail.com>
Jason Ekstrand [Wed, 16 Mar 2016 23:06:10 +0000 (16:06 -0700)]
i965: Remove the RCP+RSQ algebraic optimizations
NIR already has this optimization and it can do much better than the little
peephole in the backend.
No shader-db change on Haswell or Broadwell.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Ian Romanick [Wed, 2 Mar 2016 21:47:56 +0000 (13:47 -0800)]
nir: Don't abs slt and friends
No shader-db changes, but this is symmetric with the previous commit.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Ian Romanick [Wed, 2 Mar 2016 21:46:50 +0000 (13:46 -0800)]
nir: Don't abs the result of b2f or b2i
In the results below, 2 SIMD16 shaders in Trine are lost.
G4X
total instructions in shared programs:
4012279 ->
4011108 (-0.03%)
instructions in affected programs: 116776 -> 115605 (-1.00%)
helped: 339
HURT: 0
total cycles in shared programs:
84315862 ->
84313584 (-0.00%)
cycles in affected programs:
1767232 ->
1764954 (-0.13%)
helped: 274
HURT: 81
Ironlake
total instructions in shared programs:
6399073 ->
6396998 (-0.03%)
instructions in affected programs: 218050 -> 215975 (-0.95%)
helped: 600
HURT: 0
total cycles in shared programs:
128892088 ->
128888810 (-0.00%)
cycles in affected programs:
2867452 ->
2864174 (-0.11%)
helped: 422
HURT: 137
Sandy Bridge
total instructions in shared programs:
8462174 ->
8460759 (-0.02%)
instructions in affected programs: 178529 -> 177114 (-0.79%)
helped: 596
HURT: 0
total cycles in shared programs:
117542276 ->
117534098 (-0.01%)
cycles in affected programs:
1239166 ->
1230988 (-0.66%)
helped: 369
HURT: 150
Ivy Bridge
total instructions in shared programs:
7775131 ->
7773410 (-0.02%)
instructions in affected programs: 162903 -> 161182 (-1.06%)
helped: 590
HURT: 0
total cycles in shared programs:
65759882 ->
65747268 (-0.02%)
cycles in affected programs:
1004354 -> 991740 (-1.26%)
helped: 467
HURT: 141
Haswell
total instructions in shared programs:
7107786 ->
7106327 (-0.02%)
instructions in affected programs: 140954 -> 139495 (-1.04%)
helped: 590
HURT: 0
total cycles in shared programs:
64668028 ->
64655322 (-0.02%)
cycles in affected programs: 967080 -> 954374 (-1.31%)
helped: 452
HURT: 149
LOST: 2
GAINED: 0
Broadwell
total instructions in shared programs:
8980029 ->
8978287 (-0.02%)
instructions in affected programs: 197232 -> 195490 (-0.88%)
helped: 715
HURT: 0
total cycles in shared programs:
70070448 ->
70055970 (-0.02%)
cycles in affected programs: 975724 -> 961246 (-1.48%)
helped: 471
HURT: 111
LOST: 2
GAINED: 0
Skylake
total instructions in shared programs:
9115178 ->
9113436 (-0.02%)
instructions in affected programs: 203012 -> 201270 (-0.86%)
helped: 715
HURT: 0
total cycles in shared programs:
68848660 ->
68834004 (-0.02%)
cycles in affected programs: 993888 -> 979232 (-1.47%)
helped: 473
HURT: 116
LOST: 2
GAINED: 0
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Ian Romanick [Wed, 2 Mar 2016 23:36:14 +0000 (15:36 -0800)]
nir: Simplify 0 < fabs(a)
Sandy Bridge / Ivy Bridge / Haswell
total instructions in shared programs:
8462180 ->
8462174 (-0.00%)
instructions in affected programs: 564 -> 558 (-1.06%)
helped: 6
HURT: 0
total cycles in shared programs:
117542462 ->
117542276 (-0.00%)
cycles in affected programs: 9768 -> 9582 (-1.90%)
helped: 12
HURT: 0
Broadwell / Skylake
total instructions in shared programs:
8980833 ->
8980826 (-0.00%)
instructions in affected programs: 626 -> 619 (-1.12%)
helped: 7
HURT: 0
total cycles in shared programs:
70077900 ->
70077714 (-0.00%)
cycles in affected programs: 9378 -> 9192 (-1.98%)
helped: 12
HURT: 0
G45 and Ironlake showed no change.
v2: Modify the comments to look more like a proof.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Ian Romanick [Wed, 2 Mar 2016 23:18:34 +0000 (15:18 -0800)]
nir: Simplify 0 >= b2f(a)
This also prevented some regressions with other patches in my local
tree.
Broadwell / Skylake
total instructions in shared programs:
8980835 ->
8980833 (-0.00%)
instructions in affected programs: 45 -> 43 (-4.44%)
helped: 1
HURT: 0
total cycles in shared programs:
70077904 ->
70077900 (-0.00%)
cycles in affected programs: 122 -> 118 (-3.28%)
helped: 1
HURT: 0
No changes on earlier platforms.
v2: Modify the comments to look more like a proof.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Ian Romanick [Wed, 2 Mar 2016 02:59:57 +0000 (18:59 -0800)]
nir: Simplify i2b with negated or abs operand
This enables removing ssa_201 and ssa_202 in sequences like:
vec1 ssa_200 = flt ssa_199, ssa_194
vec1 ssa_201 = b2i ssa_200
vec1 ssa_202 = i2b -ssa_201
shader-db results:
Sandy Bridge
total instructions in shared programs:
8462257 ->
8462180 (-0.00%)
instructions in affected programs: 3846 -> 3769 (-2.00%)
helped: 35
HURT: 0
total cycles in shared programs:
117542934 ->
117542462 (-0.00%)
cycles in affected programs: 20072 -> 19600 (-2.35%)
helped: 20
HURT: 1
Ivy Bridge
total instructions in shared programs:
7775252 ->
7775137 (-0.00%)
instructions in affected programs: 3645 -> 3530 (-3.16%)
helped: 35
HURT: 0
total cycles in shared programs:
65760522 ->
65760068 (-0.00%)
cycles in affected programs: 21082 -> 20628 (-2.15%)
helped: 25
HURT: 2
Haswell
total instructions in shared programs:
7108666 ->
7108589 (-0.00%)
instructions in affected programs: 3253 -> 3176 (-2.37%)
helped: 35
HURT: 0
total cycles in shared programs:
64675726 ->
64675272 (-0.00%)
cycles in affected programs: 21034 -> 20580 (-2.16%)
helped: 26
HURT: 1
Broadwell / Skylake
total instructions in shared programs:
8980912 ->
8980835 (-0.00%)
instructions in affected programs: 3223 -> 3146 (-2.39%)
helped: 35
HURT: 0
total cycles in shared programs:
70077926 ->
70077904 (-0.00%)
cycles in affected programs: 21886 -> 21864 (-0.10%)
helped: 21
HURT: 6
G45 and Ironlake showed no change.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Suggested-by: Jason Ekstrand <jason.ekstrand@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Ian Romanick [Mon, 7 Mar 2016 21:09:30 +0000 (13:09 -0800)]
nir: Lower flrp with Boolean interpolator to bcsel
On Intel platforms that don't set lower_flrp, using bcsel instead of
flrp seems to be a small amount worse. On those platforms, the use of
flrp, bcsel, and multiply of b2f is still an active area of research.
In review, Matt suggested this is because bcsel turns into CMP+SEL, and
because of the flag register we can't schedule instructions well.
shader-db results:
G4X / Ironlake
total instructions in shared programs:
4016538 ->
4012279 (-0.11%)
instructions in affected programs: 161556 -> 157297 (-2.64%)
helped: 1077
HURT: 1
total cycles in shared programs:
84328296 ->
84315862 (-0.01%)
cycles in affected programs:
4174570 ->
4162136 (-0.30%)
helped: 926
HURT: 53
Unsurprisingly, no changes on later platforms.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Ian Romanick [Mon, 7 Mar 2016 18:55:21 +0000 (10:55 -0800)]
i965: Have NIR lower flrp on pre-GEN6 vec4 backend
Previously we were doing the lowering by hand in vec4_visitor::emit_lrp.
By doing it in NIR, we have the opportunity for NIR to do additional
optimization of the expanded code.
This also enables optimizations added by the next commit.
shader-db results:
G4X / Ironlake
total instructions in shared programs:
4024401 ->
4016538 (-0.20%)
instructions in affected programs: 447686 -> 439823 (-1.76%)
helped: 2623
HURT: 0
total cycles in shared programs:
84375846 ->
84328296 (-0.06%)
cycles in affected programs:
16964960 ->
16917410 (-0.28%)
helped: 2556
HURT: 41
Unsurprisingly, no changes on later platforms.
v2: Formatting and comment changes suggested by Matt.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Brian Paul [Tue, 22 Mar 2016 14:35:25 +0000 (08:35 -0600)]
swrast: fix discarded const warning in s_texture.c
Signed-off-by: Brian Paul <brianp@vmware.com>
Marc-André Lureau [Fri, 18 Mar 2016 19:01:07 +0000 (20:01 +0100)]
i965: fix invalid memory write
I noticed some heap corruption running virgl tests, and valgrind
helped me to track it down to the following error:
==29272== Invalid write of size 4
==29272== at 0x90283D4: push_loop_stack (brw_eu_emit.c:1307)
==29272== by 0x9029A7D: brw_DO (brw_eu_emit.c:1750)
==29272== by 0x90554B0: fs_generator::generate_code(cfg_t const*, int) (brw_fs_generator.cpp:1999)
==29272== by 0x904491F: brw_compile_fs (brw_fs.cpp:5685)
==29272== by 0x8FC5DC5: brw_codegen_wm_prog (brw_wm.c:137)
==29272== by 0x8FC7663: brw_fs_precompile (brw_wm.c:638)
==29272== by 0x8FA4040: brw_shader_precompile(gl_context*, gl_shader_program*) (brw_link.cpp:51)
==29272== by 0x8FA4A9A: brw_link_shader (brw_link.cpp:260)
==29272== by 0x8DEF751: _mesa_glsl_link_shader (ir_to_mesa.cpp:3006)
==29272== by 0x8C84325: _mesa_link_program (shaderapi.c:1042)
==29272== by 0x8C851D7: _mesa_LinkProgram (shaderapi.c:1515)
==29272== by 0x4E4B8E8: add_shader_program (vrend_renderer.c:880)
==29272== Address 0xf2f3cb0 is 0 bytes after a block of size 112 alloc'd
==29272== at 0x4C2AA98: calloc (vg_replace_malloc.c:711)
==29272== by 0x8ED11F7: ralloc_size (ralloc.c:113)
==29272== by 0x8ED1282: rzalloc_size (ralloc.c:134)
==29272== by 0x8ED14C0: rzalloc_array_size (ralloc.c:196)
==29272== by 0x9019C7B: brw_init_codegen (brw_eu.c:291)
==29272== by 0x904F565: fs_generator::fs_generator(brw_compiler const*, void*, void*, void const*, brw_stage_prog_data*, unsigned int, bool, gl_shader_stage) (brw_fs_generator.cpp:124)
==29272== by 0x9044883: brw_compile_fs (brw_fs.cpp:5675)
==29272== by 0x8FC5DC5: brw_codegen_wm_prog (brw_wm.c:137)
==29272== by 0x8FC7663: brw_fs_precompile (brw_wm.c:638)
==29272== by 0x8FA4040: brw_shader_precompile(gl_context*, gl_shader_program*) (brw_link.cpp:51)
==29272== by 0x8FA4A9A: brw_link_shader (brw_link.cpp:260)
==29272== by 0x8DEF751: _mesa_glsl_link_shader (ir_to_mesa.cpp:3006)
if_depth_in_loop is an array of size p->loop_stack_array_size, and
push_loop_stack() will access if_depth_in_loop[p->loop_stack_depth+1],
thus the condition to grow the array should be
p->loop_stack_array_size <= (p->loop_stack_depth + 1) (it's currently
off by 2...)
This can be reproduced by running the following test with virgl test
server:
LIBGL_ALWAYS_SOFTWARE=y GALLIUM_DRIVER=virpipe bin/shader_runner
./tests/shaders/glsl-fs-unroll-explosion.shader_test -auto
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Dave Airlie [Tue, 22 Mar 2016 00:28:44 +0000 (10:28 +1000)]
tgsi: drop unused set_exec/kill_mask interfaces.
These don't get used and haven't been in git history from what I can
see, so drop them.
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Dave Airlie [Mon, 21 Mar 2016 23:53:47 +0000 (09:53 +1000)]
docs/relnotes: update ARB_internalformat_query2 status.
Signed-off-by: Dave Airlie <Airlied@redhat.com>
Dave Airlie [Fri, 4 Mar 2016 02:33:46 +0000 (12:33 +1000)]
st/mesa: add support for internalformat query2.
Add code to handle GL_INTERNALFORMAT_PREFERRED.
Add code to deal with GL_RENDERBUFFER being passes into ChooseTextureFormat.
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Anuj Phogat [Fri, 11 Mar 2016 23:24:36 +0000 (15:24 -0800)]
i965: Fix assert conditions for src/dst x/y offsets
Cc: mesa-stable@lists.freedesktop.org
Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Anuj Phogat [Mon, 14 Mar 2016 17:25:50 +0000 (10:25 -0700)]
swrast: Move assert for 'slice' in to check_map_teximage
Signed-off-by: Anuj Phogat <anuj.phogat@gmail.com>
Reviewed-by: Alejandro Piñeiro <apinheiro@igalia.com>
xavier [Wed, 9 Mar 2016 08:58:48 +0000 (09:58 +0100)]
r600/sb: Do not distribute neg in expr_handler::fold_assoc() when folding multiplications.
Previously it was doing this transformation for a Trine 3 shader:
MUL R6.x.12, R13.x.23, 0.5|
3f000000
- MULADD R4.x.12, -R6.x.12, 2|
40000000, 1|
3f800000
+ MULADD R4.x.12, -R13.x.23, -1|
bf800000, 1|
3f800000
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=94412
Signed-off-by: Xavier Bouchoux <xavierb@gmail.com>
Cc: "11.0 11.1 11.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Glenn Kennard <glenn.kennard@gmail.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Samuel Pitoiset [Mon, 21 Mar 2016 12:15:44 +0000 (13:15 +0100)]
nvc0: make sure to delete samplers used by compute shaders
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "11.1 11.2" <mesa-stable@lists.freedesktop.org>
Kenneth Graunke [Thu, 17 Mar 2016 03:19:50 +0000 (20:19 -0700)]
i965/blorp: Make BlitFramebuffer() do sRGB encoding in ES 3.x.
According to the ES 3.0 and GL 4.4 specifications, glBlitFramebuffer
is supposed to perform sRGB decoding and encoding whenever sRGB formats
are in use. The ES 3.0 specification is completely clear, and has
always stated this.
However, the GL specification has changed behavior in 4.1, 4.2, and
4.4. The original behavior stated that no sRGB encoding should occur.
The 4.4 behavior matches ES 3.0's wording. However, implementing the
new behavior appears to break applications such as Left 4 Dead 2.
This patch changes Meta to apply the ES 3.x rules in ES 3.x, but
leaves OpenGL alone for now, to avoid breaking applications.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Kenneth Graunke [Thu, 17 Mar 2016 03:15:52 +0000 (20:15 -0700)]
i965/blorp: Refactor sRGB encoding/decoding.
Because the rules for sRGB are so insane, we change brw_blorp_miptrees
to take decode_srgb and encode_srgb flags, which control linearization
of the source and destination separately.
This should make it easy to implement whatever crazy combination of
rules people throw at us. For now, it should be equivalent.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Kenneth Graunke [Tue, 8 Mar 2016 08:34:14 +0000 (00:34 -0800)]
meta: Make BlitFramebuffer() do sRGB encoding in ES 3.x.
According to the ES 3.0 and GL 4.4 specifications, glBlitFramebuffer
is supposed to perform sRGB decoding and encoding whenever sRGB formats
are in use. The ES 3.0 specification is completely clear, and has
always stated this.
However, the GL specification has changed behavior in 4.1, 4.2, and
4.4. The original behavior stated that no sRGB encoding should occur.
The 4.4 behavior matches ES 3.0's wording. However, implementing the
new behavior appears to break applications such as Left 4 Dead 2.
This patch changes Meta to apply the ES 3.x rules in ES 3.x, but
leaves OpenGL alone for now, to avoid breaking applications.
Meta implements several other functions in terms of BlitFramebuffer,
and many of those explicitly do not perform sRGB encoding. So, this
patch explicitly disables sRGB encoding in those other functions,
preserving the existing (correct) behavior.
If you're from the future and are reading this, hi! Welcome to
the "fun" of debugging sRGB problems! Best of luck!
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Jordan Justen <jordan.l.justen@intel.com>
Nicolai Hähnle [Tue, 15 Mar 2016 18:08:21 +0000 (13:08 -0500)]
docs: mark GL_ARB_shader_image_load_store/_size as done for radeonsi
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Edward O'Callaghan [Sun, 10 Jan 2016 13:50:32 +0000 (00:50 +1100)]
radeonsi: Set PIPE_SHADER_CAP_MAX_SHADER_IMAGES
This enables ARB_shader_image_load_store and ARB_shader_image_size.
Signed-off-by: Edward O'Callaghan <eocallaghan@alterapraxis.com>
[allow the same number of images for all shader stages and require LLVM 3.9]
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Wed, 16 Mar 2016 01:58:12 +0000 (20:58 -0500)]
radeonsi: disable early Z if the fragment shader writes to memory
Empirically, both the EXEC_ON_* flags and LATE_Z are necessary.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Wed, 16 Mar 2016 01:54:30 +0000 (20:54 -0500)]
tgsi/scan: add writes_memory to flag presence of stores or atomics
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Sun, 13 Mar 2016 19:44:46 +0000 (14:44 -0500)]
radeonsi: force the DCC enable bit off in image descriptors for writing (v2)
This avoids a lockup at least on Tonga.
v2: only force DCC off on VI+ (Marek)
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Sun, 13 Mar 2016 16:37:27 +0000 (11:37 -0500)]
radeonsi: implement MemoryBarrier (v2)
v2: invalidate both constant and VMEM/TC L1 for constant buffers (Marek)
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Mon, 14 Mar 2016 15:22:21 +0000 (10:22 -0500)]
radeonsi: implement volatile memory access
Prevent loads from being re-ordered or coalesced.
Atomics don't need special handling by definition, and stores don't need
special handling because LLVM is unable to detect dead image or buffer
stores.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Sun, 13 Mar 2016 02:32:34 +0000 (21:32 -0500)]
radeonsi: implement coherent memory access (v2)
v2: set glc=1 for volatile also on buffers
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Thu, 10 Mar 2016 23:12:44 +0000 (18:12 -0500)]
radeonsi: Lower TGSI_OPCODE_MEMBAR down to LLVM op
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Fri, 12 Feb 2016 01:54:25 +0000 (20:54 -0500)]
radeonsi: Lower TGSI_OPCODE_ATOM* down to LLVM op
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Tue, 9 Feb 2016 16:51:31 +0000 (11:51 -0500)]
radeonsi: Lower TGSI_OPCODE_STORE down to LLVM op
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Tue, 9 Feb 2016 15:59:14 +0000 (10:59 -0500)]
radeonsi: Lower TGSI_OPCODE_LOAD down to LLVM op (v3)
v2: new signature style for buffer intrinsics (offsets)
v3: new signature style for llvm.amdgcn.buffer.load.format (overloaded return)
Reviewed-by: Marek Olšák <marek.olsak@amd.com> (v2)
Nicolai Hähnle [Tue, 9 Feb 2016 20:01:35 +0000 (15:01 -0500)]
radeonsi: extract the LLVM type name construction into its own function
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Sun, 7 Feb 2016 23:41:09 +0000 (18:41 -0500)]
radeonsi: Lower TGSI_OPCODE_RESQ down to LLVM op
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Sun, 7 Feb 2016 22:34:57 +0000 (17:34 -0500)]
radeonsi: extract TXQ buffer size computation into its own function
This will allow it to be reused for RESQ.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Mon, 8 Feb 2016 03:30:46 +0000 (22:30 -0500)]
radeonsi: decompress shader images
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Wed, 16 Mar 2016 03:01:39 +0000 (22:01 -0500)]
radeonsi: update shader image descriptor for invalidated buffer
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Sat, 6 Feb 2016 23:32:13 +0000 (18:32 -0500)]
radeonsi: implement set_shader_images (v2)
Whether DCC is disabled depends on the access flags with which the image
is bound: image_load supports DCC, but store and atomic don't.
v2: remove an unnecessary masking of images->desc.enabled_mask
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Sat, 12 Mar 2016 00:39:18 +0000 (19:39 -0500)]
gallium/radeon: make r600_texture_disable_dcc externally accessible
We will need it in radeonsi for shader images.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Sun, 13 Mar 2016 20:06:15 +0000 (15:06 -0500)]
tgsi/scan: track which shader images are really buffers
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Sun, 13 Mar 2016 20:00:40 +0000 (15:00 -0500)]
tgsi/scan: add images_writemask
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Sun, 13 Mar 2016 16:37:10 +0000 (11:37 -0500)]
st/mesa: translate additional flags in MemoryBarrier
Re-order flags in the order in which they appear in the OpenGL spec in the
description of MemoryBarrier().
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Nicolai Hähnle [Sun, 13 Mar 2016 16:36:53 +0000 (11:36 -0500)]
gallium: add additional PIPE_BARRIER_* bits
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Brian Paul [Wed, 24 Feb 2016 01:02:11 +0000 (18:02 -0700)]
svga: add svga_winsys_context::pipe_debug_callback pointer
The svga winsys modules can use this to send debug messages to the
state tracker and Mesa.
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
Charmaine Lee [Tue, 26 Jan 2016 19:12:09 +0000 (11:12 -0800)]
svga: Fix the index buffer rebind regression
The index buffer handle saved in the hw_state structure could
be invalid after the index buffer is destroyed. Instead of
rebinding the index buffer using the saved index buffer handle,
we will reset the index buffer handle in the hw_state structure
to force resending of the index buffer.
Fixes bug
1593320
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Charmaine Lee [Wed, 20 Jan 2016 18:35:56 +0000 (10:35 -0800)]
svga: rebind stream output targets
To ensure stream output target surfaces are available for the draw commands,
we need to rebind the current stream output targets at the first draw in the
command buffer.
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Charmaine Lee [Wed, 20 Jan 2016 04:25:39 +0000 (20:25 -0800)]
svga: rebind index buffer
Similar to other resources, current index buffer needs to be
rebound at the first draw of the current command buffer to make
sure the buffer is available for the draw command.
Fixes bug
1587263.
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Brian Paul [Mon, 21 Mar 2016 19:23:04 +0000 (13:23 -0600)]
svga: minor formatting fix, comment addition
To sync with our internal tree.
Signed-off-by: Brian Paul <brianp@vmware.com>
Charmaine Lee [Fri, 11 Mar 2016 22:54:36 +0000 (14:54 -0800)]
svga: optimize constant buffer uploads
When a constant buffer slot is allocated in the upload buffer,
the allocated slot size is always in multiple of 256. But the actual buffer
size might not be in multiple of 256. This causes a gap between
the ending offset of a slot and the starting offset of the next slot.
The gap will prevent the two slots to be updated in a single update command.
In order to maximize the chance of merging the contiguous dirty ranges,
when a slot is to be allocated in the constant upload buffer,
specify a buffer size in multiple of 256.
There is about 10% performance improvement with Lightsmark2008 and
30% with Cinebench R11.
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Charmaine Lee [Fri, 11 Mar 2016 22:33:39 +0000 (14:33 -0800)]
svga: add a few more resource updates HUD query
This patch adds the following HUD queries:
.num-resource-updates -- number of resource update. Commands include
UPDATE_SUBRESOURCE, UPDATE_GB_IMAGE.
.num-buffer-uploads -- number of buffer uploads.
.num-const-buf-updates -- number of set constant buffer.
.num-const-updates -- number of set shader constant.
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Charmaine Lee [Thu, 10 Mar 2016 18:57:24 +0000 (10:57 -0800)]
svga: add new num-readbacks HUD query
To find out how many image readback command is issued.
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Thomas Hellstrom <thellstrom@vmware.com>
Brian Paul [Fri, 18 Mar 2016 15:55:57 +0000 (09:55 -0600)]
svga: use shader sampler view declarations
Previously, we looked at the bound textures (via the pipe_sampler_views)
to determine texture dimensions (1D/2D/3D/etc) and datatype (float vs.
int). But this could fail in out of memory conditions. If we failed to
allocate a texture and didn't create a pipe_sampler_view, we'd default
to using 0 (PIPE_BUFFER) as the texture type. This led to device errors
because of inconsistent shader code.
This change relies on all TGSI shaders having an SVIEW declaration for
each SAMP declaration. The previous patch series does that.
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Brian Paul [Fri, 18 Mar 2016 20:20:44 +0000 (14:20 -0600)]
gallium/tests: declare sampler views in shaders
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Brian Paul [Fri, 18 Mar 2016 20:20:27 +0000 (14:20 -0600)]
gallium/util: declare sampler view in util_make_fs_blit_msaa_depthstencil()
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Brian Paul [Fri, 18 Mar 2016 20:20:01 +0000 (14:20 -0600)]
postprocess: declare sampler views in shaders
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Brian Paul [Fri, 18 Mar 2016 20:07:47 +0000 (14:07 -0600)]
hud: add sampler view declaration in text fragment shader
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Reviewed-by: Charmaine Lee <charmainel@vmware.com>