mesa.git
5 years agoiris: Drop dead state_size hash table
Kenneth Graunke [Thu, 24 Jan 2019 01:03:54 +0000 (17:03 -0800)]
iris: Drop dead state_size hash table

I inherited this from i965.  It would be nice to track the state size
so INTEL_DEBUG=color,bat decoding can print the right number of e.g.
binding table entries or blend states, but...without a single point
of entry for state, it's a little tricky to get right.  Punt for now,
and drop the dead code in the meantime.

5 years agoiris: Drop comment about ISP_DIS
Kenneth Graunke [Thu, 24 Jan 2019 00:58:30 +0000 (16:58 -0800)]
iris: Drop comment about ISP_DIS

i965 re-emits 3DSTATE_CONSTANT_* on every batch, so there's no point in
restoring the constants from the context.  Iris actually re-pins the
constant buffers properly across the batch, and avoids re-emitting the
constant packets unless it's necessary.  So, we don't want ISP_DIS.

5 years agoiris: Enable PIPE_CAP_COMPACT_ARRAYS
Kenneth Graunke [Wed, 23 Jan 2019 10:58:59 +0000 (02:58 -0800)]
iris: Enable PIPE_CAP_COMPACT_ARRAYS

5 years agoiris: Remap stream output indexes back to VARYING_SLOT_*.
Kenneth Graunke [Wed, 23 Jan 2019 07:28:39 +0000 (23:28 -0800)]
iris: Remap stream output indexes back to VARYING_SLOT_*.

Previously I had a hack in st/mesa to make it stop remapping
VARYING_SLOT_* into the naively compacted slots, which aren't
what we want.  But that wasn't very feasible, as we'd have to
update all drivers, or add capability bits, and it gets messy fast.

It turns out that I can map back to VARYING_SLOT_* in about 5 LOC,
so let's just do that.  It removes the need for hacks, and is easy.

This also fixes KHR-GL46.enhanced_layouts.xfb_capture_struct, which
apparently with my hack was still getting the wrong slot info.

5 years agoiris: Zero the compute predicate when changing the render condition
Kenneth Graunke [Tue, 22 Jan 2019 22:22:55 +0000 (14:22 -0800)]
iris: Zero the compute predicate when changing the render condition

1. Set a render condition.  We emit it immediately on the render
   engine, and stash q->bo as ice->state.compute_predicate in case
   the compute engine needs it.

2. Clear the render condition.  We were incorrectly leaving a stale
   compute_predicate kicking around...

3. Dispatch compute.  We would then read the stale compute predicate,
   and try to load it into MI_PREDICATE_DATA.  But q->bo may have been
   freed altogether, causing us to try and use garbage memory as a BO,
   adding it to the validation list, failing asserts, and tripping
   EINVALs in execbuf.

Huge thanks to Mark Janes for narrowing this sporadic GL CTS failure
down to a list of 48 tests I could easily run to reproduce it.  Huge
thanks to the Valgrind authors for the memcheck tool that immediately
pinpointed the problem.

5 years agoiris: always include an extra constbuf0 if using UBOs
Caio Marcelo de Oliveira Filho [Sat, 19 Jan 2019 19:32:37 +0000 (11:32 -0800)]
iris: always include an extra constbuf0 if using UBOs

In st_nir_lower_uniforms_to_ubo() all UBO access in the shader have
its index incremented to open room for uniforms in constbuf0.  So if
we use UBOs, we always need to include the extra binding entry in the
table.

To avoid doing this checks both when compiling the shader and when
assigning binding tables, store the num_cbufs in iris_compiled_shader.

Fixes a bunch of tests from Piglit and CTS that use UBOs but don't use
uniforms or system values.  Note that some tests fitting this criteria
were passing because the UBOs were moved to be push
constants (avoiding the problem).

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
5 years agoiris: Do binder address allocations per-context, not globally.
Kenneth Graunke [Fri, 18 Jan 2019 20:26:41 +0000 (12:26 -0800)]
iris: Do binder address allocations per-context, not globally.

iris_bufmgr allocates addresses across the entire screen, since buffers
may be shared between multiple contexts.  There used to be a single
special address, IRIS_BINDER_ADDRESS, that was per-context - and all
contexts used the same address.  When I moved to the multi-binder
system, I made a separate memory zone for them.  I wanted there to be
2-3 binders per context, so we could cycle them to avoid the stalls
inherent in pinning two buffers to the same address in back-to-back
batches.  But I figured I'd allow 100 binders just to be wildly
excessive/cautious.

What I didn't realize was that we need 2-3 binders per *context*,
and what I did was allocate 100 binders per *screen*.  Web browsers,
for example, might have 1-2 contexts per tab, leading to hundreds of
contexts, and thus binders.

To fix this, we stop allocating VMA for binders in bufmgr, and let
the binder handle it itself.  Binders are per-context, and they can
assign context-local addresses for the buffers by simply doing a
ringbuffer style approach.  We only hold on to one binder BO at a
time, so we won't ever have a conflicting address.

This fixes dEQP-EGL.functional.multicontext.non_shared_clear.

Huge thanks to Tapani Pälli for debugging this whole mess and
figuring out what was going wrong.

Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
5 years agoiris: Fix memzone_for_address for the surface and binder zones
Kenneth Graunke [Fri, 18 Jan 2019 20:20:43 +0000 (12:20 -0800)]
iris: Fix memzone_for_address for the surface and binder zones

We use > for IRIS_MEMZONE_DYNAMIC because IRIS_BORDER_COLOR_POOL_ADDRESS
lives at the very start of that zone.  However, IRIS_MEMZONE_SURFACE and
IRIS_MEMZONE_BINDER are normal zones.  They used to be a single zone
(surface) with a single binder BO at the beginning, similar to the
border color pool.  But when I moved us to multiple binders, I made them
have a real zone (if a small one).  So both zones should use >=.

Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
5 years agoiris: Don't whack SO dirty bits when finishing a BLORP op
Kenneth Graunke [Fri, 18 Jan 2019 08:01:05 +0000 (00:01 -0800)]
iris: Don't whack SO dirty bits when finishing a BLORP op

Re-emitting 3DSTATE_SO_BUFFERS can be hazardous, as it could zero
offsets.  Plus, it's just not necessary - BLORP doesn't change these.

5 years agoiris: Fix SO issue with INTEL_DEBUG=reemit, set fewer bits
Kenneth Graunke [Wed, 16 Jan 2019 09:19:44 +0000 (01:19 -0800)]
iris: Fix SO issue with INTEL_DEBUG=reemit, set fewer bits

INTEL_DEBUG=reemit was breaking streamout tests, by re-emitting
3DSTATE_SO_BUFFER commands that tell the HW to zero the SO write
offsets.  We would need to alter them to use 0xFFFFFFFF for the offset.

Also, have each upload function only flag bits relevant to its own
pipeline.

5 years agoiris: CS stall on VF cache invalidate workarounds
Kenneth Graunke [Fri, 18 Jan 2019 07:44:09 +0000 (23:44 -0800)]
iris: CS stall on VF cache invalidate workarounds

See commit 31e4c9ce400341df9b0136419b3b3c73b8c9eb7e in i965.

5 years agoiris: Pay attention to blit masks
Kenneth Graunke [Wed, 16 Jan 2019 10:02:19 +0000 (02:02 -0800)]
iris: Pay attention to blit masks

For combined depth/stencil formats, we may want to only blit one half.
If PIPE_BLIT_Z is set, blit depth; if PIPE_BLIT_S is set, blit stencil.

5 years agoiris: Assert about blits with color masking
Kenneth Graunke [Wed, 16 Jan 2019 09:53:00 +0000 (01:53 -0800)]
iris: Assert about blits with color masking

st/mesa never asks for this today, but in theory someone might, and we
don't support it.

5 years agoiris: Don't enable smooth points when point sprites are enabled
Kenneth Graunke [Wed, 16 Jan 2019 07:41:34 +0000 (23:41 -0800)]
iris: Don't enable smooth points when point sprites are enabled

dEQP-GLES3.functional.rasterization.fbo.rbo_multisample_*.primitives.points

5 years agoiris: Allow sample mask of 0
Kenneth Graunke [Wed, 16 Jan 2019 07:22:48 +0000 (23:22 -0800)]
iris: Allow sample mask of 0

I think this was an attempt to work around various sample mask bugs I
had early on.  It's not correct.  A sample mask of 0 is legal and means
to disable all samples.

Fixes dEQP-GLES31.functional.texture.multisample.*.*sample_mask*

5 years agoiris: fail to create screen for older unsupported HW
Kenneth Graunke [Mon, 14 Jan 2019 08:25:23 +0000 (00:25 -0800)]
iris: fail to create screen for older unsupported HW

loader shouldn't try, but let's be paranoid

5 years agoiris: Switch to the new PIPELINE_STATISTICS_QUERY_SINGLE capability
Kenneth Graunke [Fri, 11 Jan 2019 21:39:04 +0000 (13:39 -0800)]
iris: Switch to the new PIPELINE_STATISTICS_QUERY_SINGLE capability

I had a hack in place earlier to pass the query type as q->index
for the regular statistics query, but we ended up adjusting the
interface and adding a new query type.  Use that instead, fixing
pipeline statistics queries since the rebase.

5 years agoiris: Use new PIPE_STAT_QUERY enums rather than hardcoded numbers.
Kenneth Graunke [Fri, 11 Jan 2019 08:21:06 +0000 (00:21 -0800)]
iris: Use new PIPE_STAT_QUERY enums rather than hardcoded numbers.

5 years agoiris: Fix Broadwell WaDividePSInvocationCountBy4
Kenneth Graunke [Fri, 11 Jan 2019 08:28:07 +0000 (00:28 -0800)]
iris: Fix Broadwell WaDividePSInvocationCountBy4

We were dividing by 4 in calculate_result_on_gpu(), and also in
iris_get_query_result().  We should stop doing the latter, and instead
divide by 4 in calculate_result_on_cpu() as well.

Otherwise, if snapshots were available, and you hit the
calculate_result_on_cpu() path, but requested it be written to a QBO,
you'd fail to get a divide.

5 years agoiris: Delete genx->bound_vertex_buffers
Kenneth Graunke [Sun, 6 Jan 2019 23:56:26 +0000 (15:56 -0800)]
iris: Delete genx->bound_vertex_buffers

This is actually stored in ice->state, as it isn't gen-specific

5 years agoiris: Drop a dead comment
Kenneth Graunke [Fri, 4 Jan 2019 06:34:49 +0000 (22:34 -0800)]
iris: Drop a dead comment

5 years agoiris: Don't check other batches for our batch BO
Kenneth Graunke [Wed, 2 Jan 2019 10:45:00 +0000 (02:45 -0800)]
iris: Don't check other batches for our batch BO

This is an awkward corner case.  We create batches in order, each of
which creates and pins a BO.  The other batches may not be set up yet,
so it may not be safe to ask whether they reference a BO.

Just avoid this for now.  We could avoid it for other context-local BOs
too, but we currently don't have a flag for that (and I'm not certain
whether it's worth it).

5 years agoiris: Handle PIPE_TRANSFER_DISCARD_WHOLE_RESOURCE somewhat
Kenneth Graunke [Tue, 1 Jan 2019 06:03:35 +0000 (22:03 -0800)]
iris: Handle PIPE_TRANSFER_DISCARD_WHOLE_RESOURCE somewhat

Various places in the transfer code need to know whether they must
read the existing resource's values.  Rather than checking both flags
everywhere, just make PIPE_TRANSFER_DISCARD_WHOLE_RESOURCE also flag
PIPE_TRANSFER_DISCARD_RANGE - if we can discard everything, we can
discard a subrange, too.

Obviously, we can do better for PIPE_TRANSFER_DISCARD_WHOLE_RESOURCE,
but eventually u_threaded_context should handle swapping out buffers
for new idle buffers, anyway.  In the meantime, this is at least better.

5 years agoiris: Flush the render cache in flush_and_dirty_for_history
Kenneth Graunke [Sun, 23 Dec 2018 05:24:02 +0000 (21:24 -0800)]
iris: Flush the render cache in flush_and_dirty_for_history

BLORP uses the render engine to write to buffers, and we need to flush
that data out to the actual surface (finishing the write).  Then, the
rest of this function invalidates any caches that might have stale data
which needs to be refetched.

5 years agoiris: Implement multi-slice copy_region
Kenneth Graunke [Mon, 24 Dec 2018 07:04:37 +0000 (23:04 -0800)]
iris: Implement multi-slice copy_region

I don't know if this is required - surprisingly, I haven't seen it
matter - but I'd like to use it for multi-slice transfer maps.  We may
as well do the right thing.

5 years agoiris: Leave a comment about why Broadwell images are broken
Kenneth Graunke [Mon, 31 Dec 2018 17:19:07 +0000 (09:19 -0800)]
iris: Leave a comment about why Broadwell images are broken

There are a variety of ways to fix this, many of which are simple, but
I could use some advice on which ones other people prefer, and so we'll
punt until after the holidays.

5 years agoiris: Fix surface states for Gen8 lowered-to-untype images
Kenneth Graunke [Wed, 26 Dec 2018 10:06:13 +0000 (02:06 -0800)]
iris: Fix surface states for Gen8 lowered-to-untype images

We have to use SURFTYPE_BUFFER and ISL_FORMAT_RAW for these.

5 years agoiris: Fill out brw_image_params for storage images on Broadwell
Kenneth Graunke [Fri, 30 Nov 2018 10:27:07 +0000 (02:27 -0800)]
iris: Fill out brw_image_params for storage images on Broadwell

5 years agoiris: Don't make duplicate system values
Kenneth Graunke [Thu, 27 Dec 2018 09:27:44 +0000 (01:27 -0800)]
iris: Don't make duplicate system values

We were relying on CSE/GVN/etc to coalesce all intrinsics that load the
same value, but that's a bad idea.  We might have a couple intrinsics
that reload the same value.  If so, we only want to set up the uniform
on the first one we see.

5 years agoiris: Don't enable push constants just because there are system values
Kenneth Graunke [Thu, 27 Dec 2018 08:49:56 +0000 (00:49 -0800)]
iris: Don't enable push constants just because there are system values

System values are built-in uniforms.  We set them up as UBO values, and
might pull or push them.  UBO push analysis will take care of that.  We
only want to enable push constants if there's an actual range being
pushed.  Otherwise, we might get into a scenario where 3DSTATE_PS
enables push constants but 3DSTATE_CONSTANT_PS isn't pushing anything.

This fixes GPU hangs in Broadwell image load store tests which have
unused image param system values but no other uniforms.  (We shouldn't
be making those anyway, but that's a separate fix...)

5 years agoiris: Fix framebuffer layer count
Kenneth Graunke [Mon, 24 Dec 2018 02:22:44 +0000 (18:22 -0800)]
iris: Fix framebuffer layer count

cso_fb->layers is only valid for no-attachment framebuffers.  Use the
helper function to get the real value, then stash it so we don't have
to call the helper function on the old value for comparison, or at draw
time for Force Zero RTA Index setting.

This fixes Force Zero RTA Index being set even when attempting layered
rendering.

5 years agoiris: handle qbo fragment shader invocation workaround
Dave Airlie [Fri, 9 Nov 2018 02:13:30 +0000 (12:13 +1000)]
iris: handle qbo fragment shader invocation workaround

5 years agoiris: add fs invocations query workaround for broadwell
Dave Airlie [Thu, 8 Nov 2018 07:46:22 +0000 (17:46 +1000)]
iris: add fs invocations query workaround for broadwell

5 years agoiris: setup gen8 caps
Dave Airlie [Thu, 8 Nov 2018 00:38:48 +0000 (10:38 +1000)]
iris: setup gen8 caps

5 years agoiris: limit gen8 to 8 samples
Dave Airlie [Thu, 8 Nov 2018 00:20:54 +0000 (10:20 +1000)]
iris: limit gen8 to 8 samples

5 years agoiris/WIP: add broadwell support
Dave Airlie [Wed, 7 Nov 2018 04:23:27 +0000 (14:23 +1000)]
iris/WIP: add broadwell support

This adds all the state changes, MOCS changes,

5 years agoiris: Delete bogus comment about cube array counting.
Kenneth Graunke [Sun, 23 Dec 2018 07:09:16 +0000 (23:09 -0800)]
iris: Delete bogus comment about cube array counting.

Both 'z' and 'depth' are counted in slices, according to the Gallium
docs (context.rst).  In our temporary memory, we allocate `box.depth`
slices, so we need to rebase the starting slice (box.z) down to 0,
and back again when writing on unmap.

There's nothing strange about cubes here.

5 years agoiris: Fix compute scratch pinning
Kenneth Graunke [Fri, 21 Dec 2018 09:00:34 +0000 (01:00 -0800)]
iris: Fix compute scratch pinning

Thanks to Eero Tamminen for helping catch this.

5 years agoiris: Add a more long term TODO about timebase scaling
Kenneth Graunke [Fri, 14 Dec 2018 08:51:00 +0000 (00:51 -0800)]
iris: Add a more long term TODO about timebase scaling

5 years agoiris: Only resolve inputs for actual shader stages
Kenneth Graunke [Thu, 20 Dec 2018 04:34:55 +0000 (20:34 -0800)]
iris: Only resolve inputs for actual shader stages

We don't need to consider compute at render time, and don't need to
consider disabled stages.  4% on drawoverhead.

5 years agoiris: Fix assertion in iris_resource_from_handle() tiling usage
Rhys Kidd [Mon, 17 Dec 2018 06:25:11 +0000 (01:25 -0500)]
iris: Fix assertion in iris_resource_from_handle() tiling usage

Assertion error:

  iris_resource_from_handle: Assertion `res->bo->tiling_mode ==
      isl_tiling_to_i915_tiling(res->surf.tiling)' failed.

This patch fixes 16 piglit tests on KBL:
glx/glx-multithread-texture
glx/glx-query-drawable-glx_fbconfig_id-glxpbuffer
glx/glx-query-drawable-glx_fbconfig_id-glxpixmap
glx/glx-query-drawable-glx_preserved_contents
glx/glx-query-drawable-glxpbuffer-glx_height
glx/glx-query-drawable-glxpbuffer-glx_width
glx/glx-query-drawable-glxpixmap-glx_height
glx/glx-query-drawable-glxpixmap-glx_width
glx/glx-swap-pixmap
glx/glx-swap-pixmap-bad
glx/glx-tfp
glx/glx-visuals-depth -pixmap
glx/glx-visuals-stencil -pixmap
spec/egl 1.4/eglcreatepbuffersurface and then glclear
spec/egl 1.4/largest possible eglcreatepbuffersurface and then glclear
spec/egl_nok_texture_from_pixmap/basic

Cc: Kenneth Graunke <kenneth@whitecape.org>
Cc: Jason Ekstrand <jason.ekstrand@intel.com>
Signed-off-by: Rhys Kidd <rhyskidd@gmail.com>
5 years agoiris: Fix scratch space allocation on Icelake.
Kenneth Graunke [Fri, 14 Dec 2018 23:48:07 +0000 (15:48 -0800)]
iris: Fix scratch space allocation on Icelake.

Gen9-10 have fewer than 4 subslices per slice, so they need this to be
rounded up.  Gen11 isn't documented as needing this hack, and it can
also have more than 4 subslices, so the hack actually can break things.

Fixes tests/spec/arb_enhanced_layouts/execution/component-layout/
sso-vs-gs-fs-array-interleave

5 years agoiris: better MOCS
Kenneth Graunke [Wed, 12 Dec 2018 08:02:25 +0000 (00:02 -0800)]
iris: better MOCS

5 years agoiris: fix gpu calcs for timestamp queries
Dave Airlie [Fri, 14 Dec 2018 04:05:27 +0000 (14:05 +1000)]
iris: fix gpu calcs for timestamp queries

5 years agoiris: only mark depth/stencil as writable if writes are actually enabled
Kenneth Graunke [Wed, 12 Dec 2018 10:03:57 +0000 (02:03 -0800)]
iris: only mark depth/stencil as writable if writes are actually enabled

5 years agoiris: more dead comments
Kenneth Graunke [Wed, 12 Dec 2018 09:52:23 +0000 (01:52 -0800)]
iris: more dead comments

5 years agoiris: pin and re-pin the scratch BO
Kenneth Graunke [Wed, 12 Dec 2018 09:41:39 +0000 (01:41 -0800)]
iris: pin and re-pin the scratch BO

5 years agoiris: delete finished comments
Kenneth Graunke [Wed, 12 Dec 2018 06:51:15 +0000 (22:51 -0800)]
iris: delete finished comments

5 years agoiris: always pin the binder...in the compute context, too.
Kenneth Graunke [Wed, 12 Dec 2018 06:46:40 +0000 (22:46 -0800)]
iris: always pin the binder...in the compute context, too.

not sure why this hasn't tripped things up

5 years agoiris: Track blend enables, save outbound for resolve code
Kenneth Graunke [Tue, 11 Dec 2018 07:22:54 +0000 (23:22 -0800)]
iris: Track blend enables, save outbound for resolve code

5 years agoiris: whitespace fixes
Kenneth Graunke [Mon, 10 Dec 2018 08:49:35 +0000 (00:49 -0800)]
iris: whitespace fixes

5 years agoiris: Make a alloc_surface_state helper
Kenneth Graunke [Fri, 7 Dec 2018 19:20:39 +0000 (11:20 -0800)]
iris: Make a alloc_surface_state helper

This does the gtt_offset addition for us

5 years agoiris: Use a surface state fill helper
Kenneth Graunke [Fri, 7 Dec 2018 07:29:20 +0000 (23:29 -0800)]
iris: Use a surface state fill helper

This will check aux_usage eventually

5 years agoiris: don't print the pointer in INTEL_DEBUG=submit
Kenneth Graunke [Sat, 8 Dec 2018 19:21:48 +0000 (11:21 -0800)]
iris: don't print the pointer in INTEL_DEBUG=submit

lots of noise in diff, hope was it would be useful for gdb, but the
the GEM handle is good enough

5 years agoiris: Fix the prototype for iris_bo_alloc_tiled
Kenneth Graunke [Fri, 7 Dec 2018 07:05:14 +0000 (23:05 -0800)]
iris: Fix the prototype for iris_bo_alloc_tiled

This now matches the actual function in iris_bufmgr.c, as well as the
equivalent brw_bufmgr.c function...

5 years agoiris: Fix for PIPE_CAP_SIGNED_VERTEX_BUFFER_OFFSET
Kenneth Graunke [Wed, 5 Dec 2018 11:30:42 +0000 (03:30 -0800)]
iris: Fix for PIPE_CAP_SIGNED_VERTEX_BUFFER_OFFSET

This fixes ext_transform_feedback-builtin-varyings gl_Position after the
combination of my transform feedback reworks and my vertex buffer
reworks (?)

5 years agoiris: drop unnecessary genx->streamout field
Kenneth Graunke [Wed, 5 Dec 2018 11:16:54 +0000 (03:16 -0800)]
iris: drop unnecessary genx->streamout field

5 years agoiris: Implement DrawTransformFeedback()
Kenneth Graunke [Wed, 5 Dec 2018 06:19:33 +0000 (22:19 -0800)]
iris: Implement DrawTransformFeedback()

We get the count by dividing the offset by the stride.

5 years agoiris: Copy anv's MI_MATH helpers for multiplication and division
Jason Ekstrand [Wed, 5 Dec 2018 06:00:33 +0000 (22:00 -0800)]
iris: Copy anv's MI_MATH helpers for multiplication and division

(import done by Ken but with author set to Jason because it's his
code that's being imported, so he deserves the credit)

5 years agoiris: only get space for one offset in stream output targets
Kenneth Graunke [Wed, 5 Dec 2018 09:16:23 +0000 (01:16 -0800)]
iris: only get space for one offset in stream output targets

Target corresponds to a buffer, buffer only records one offset, not
multiple.

5 years agoiris: Move iris_stream_output_target def to iris_context.h
Kenneth Graunke [Wed, 5 Dec 2018 08:57:07 +0000 (00:57 -0800)]
iris: Move iris_stream_output_target def to iris_context.h

now that it doesn't have genxml

5 years agoiris: Don't bother packing 3DSTATE_SO_BUFFER at create time
Kenneth Graunke [Wed, 5 Dec 2018 08:55:45 +0000 (00:55 -0800)]
iris: Don't bother packing 3DSTATE_SO_BUFFER at create time

We have to do half the packet late anyway, we may as well just do it
all at set time.  This also lets us move the struct def out of genxml

5 years agoiris: Add _MI_ALU helpers that don't paste
Kenneth Graunke [Wed, 5 Dec 2018 06:04:16 +0000 (22:04 -0800)]
iris: Add _MI_ALU helpers that don't paste

This lets you pass arguments as function parameters

5 years agoiris: Reorder LRR parameters to have dst first.
Kenneth Graunke [Wed, 5 Dec 2018 06:02:50 +0000 (22:02 -0800)]
iris: Reorder LRR parameters to have dst first.

LRI and LRM both put dst first, be consistent.

5 years agoiris: rewrite set_vertex_buffer and VB handling
Kenneth Graunke [Wed, 5 Dec 2018 00:38:14 +0000 (16:38 -0800)]
iris: rewrite set_vertex_buffer and VB handling

I was using the Gallium API wrong.  set_* functions with start_slot
and count parameters are supposed to update a subrange of the items.
I had been trashing all bound vertex buffers and starting over.

This should hopefully also make it easier to slot in additional
VERTEX_BUFFER_STATEs at draw time, say, for shader draw parameters.

5 years agoiris: handle PatchVerticesIn as a system value.
Kenneth Graunke [Tue, 4 Dec 2018 22:11:51 +0000 (14:11 -0800)]
iris: handle PatchVerticesIn as a system value.

5 years agoiris: add Android build
Tapani Pälli [Mon, 3 Dec 2018 12:15:13 +0000 (14:15 +0200)]
iris: add Android build

Note that at least following additional libs/components require changes
since they refer to BOARD_GPU_DRIVERS variable which is used to select
the driver:

  - mixins
  - minigbm
  - libdrm
  - drm_gralloc

v2: (feedback by Gustaw Smolarczyk) Fix trailing \ in a few cases

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
5 years agoiris: override alpha to one src1 blend factors
Kenneth Graunke [Tue, 4 Dec 2018 08:18:41 +0000 (00:18 -0800)]
iris: override alpha to one src1 blend factors

No idea why this used to pass and doesn't after updating...seems like
we should have been handling it all along...

5 years agoiris: Always do rasterizer discard in clipper
Kenneth Graunke [Mon, 3 Dec 2018 10:59:08 +0000 (02:59 -0800)]
iris: Always do rasterizer discard in clipper

but continue doing it in SOL if possible because it's faster

Fixes ./bin/ext_transform_feedback-discard-drawarrays - simpler too

5 years agoiris: Fix primitive generated query active flag
Kenneth Graunke [Mon, 3 Dec 2018 10:59:08 +0000 (02:59 -0800)]
iris: Fix primitive generated query active flag

5 years agoiris: Enable guardband clipping
Kenneth Graunke [Mon, 3 Dec 2018 10:08:23 +0000 (02:08 -0800)]
iris: Enable guardband clipping

5 years agoiris: Clamp viewport extents to the framebuffer dimensions
Kenneth Graunke [Mon, 3 Dec 2018 10:02:49 +0000 (02:02 -0800)]
iris: Clamp viewport extents to the framebuffer dimensions

Fixes arb_framebuffer_no_attachments-query's resize subtest.

5 years agoiris: Fix clear dimensions
Kenneth Graunke [Mon, 3 Dec 2018 09:03:17 +0000 (01:03 -0800)]
iris: Fix clear dimensions

Fixes depthstencil-render-miplevels 1024 s=z24_s8

5 years agoiris: Drop continues in resolve
Kenneth Graunke [Mon, 3 Dec 2018 07:30:09 +0000 (23:30 -0800)]
iris: Drop continues in resolve

Now that we u_bit_scan we know it exists

5 years agoiris: Replace num_textures etc with a bitmask we can scan
Kenneth Graunke [Mon, 3 Dec 2018 07:17:44 +0000 (23:17 -0800)]
iris: Replace num_textures etc with a bitmask we can scan

More accurate bounds, plus can skip dead ones

5 years agoiris: Fix set_sampler_views with start > 0
Kenneth Graunke [Mon, 3 Dec 2018 07:15:41 +0000 (23:15 -0800)]
iris: Fix set_sampler_views with start > 0

5 years agoiris: fix set_sampler_views to not unbind, be better about bounds
Kenneth Graunke [Mon, 3 Dec 2018 07:07:27 +0000 (23:07 -0800)]
iris: fix set_sampler_views to not unbind, be better about bounds

5 years agoiris: fix overhead regression from flushing for storage images
Kenneth Graunke [Mon, 3 Dec 2018 07:03:08 +0000 (23:03 -0800)]
iris: fix overhead regression from flushing for storage images

st calls us with count = 32 but a NULL pointer...we only really care
about the highest non-NULL image...

5 years agoiris: Fix NOS mechanism
Kenneth Graunke [Mon, 3 Dec 2018 06:30:07 +0000 (22:30 -0800)]
iris: Fix NOS mechanism

Set bits, not values

5 years agoiris: re-pin inherited streamout buffers
Kenneth Graunke [Mon, 3 Dec 2018 00:06:01 +0000 (16:06 -0800)]
iris: re-pin inherited streamout buffers

5 years agoiris: reemit SBE when sprite coord origin changes
Kenneth Graunke [Sun, 2 Dec 2018 23:52:46 +0000 (15:52 -0800)]
iris: reemit SBE when sprite coord origin changes

fixes arb_point_sprite-checkerboard

5 years agoiris: omask can kill
Kenneth Graunke [Sun, 2 Dec 2018 23:41:13 +0000 (15:41 -0800)]
iris: omask can kill

5 years agoiris: reject all clipping when we can't use streamout render disabled
Kenneth Graunke [Sun, 2 Dec 2018 22:51:52 +0000 (14:51 -0800)]
iris: reject all clipping when we can't use streamout render disabled

5 years agoiris: make clipper statistics dynamic
Kenneth Graunke [Sun, 2 Dec 2018 22:39:29 +0000 (14:39 -0800)]
iris: make clipper statistics dynamic

5 years agoiris: CS stall for stream out -> VB
Kenneth Graunke [Sun, 2 Dec 2018 22:16:08 +0000 (14:16 -0800)]
iris: CS stall for stream out -> VB

i965 doesn't do this, but I suspect it just stalls a lot and doesn't hit
this.  Fixes ext_transform_feedback-position render among others.

5 years agoiris: fix dma buf import strides
Kenneth Graunke [Sun, 2 Dec 2018 21:37:55 +0000 (13:37 -0800)]
iris: fix dma buf import strides

5 years agoiris: fix alpha channel for RGB BC1 formats
Kenneth Graunke [Sun, 2 Dec 2018 01:07:09 +0000 (17:07 -0800)]
iris: fix alpha channel for RGB BC1 formats

5 years agoiris: Allocate buffer resources separately
Jason Ekstrand [Thu, 25 Oct 2018 22:02:56 +0000 (17:02 -0500)]
iris: Allocate buffer resources separately

(cleaned up by Ken - make sure a bunch of things were more obviously
not using res->surf, do allow checking res->surf.tiling == LINEAR,
drop format cpp checks that aren't needed, drop memzone handling for
images, assume buffers / non-buffers in a few places...)

5 years agoiris: Don't bother considering if the underlying surface is a cube
Kenneth Graunke [Fri, 30 Nov 2018 07:03:20 +0000 (23:03 -0800)]
iris: Don't bother considering if the underlying surface is a cube

Dave fixed it to consider whether the sampler view is a cube.
With that, there's no point (possibly harm) in looking if the original
resource was a cube...if it's an array view, we don't want to treat it
as a cube anymore...

5 years agoiris: move some non-buffer case code in a bit
Kenneth Graunke [Fri, 30 Nov 2018 06:23:34 +0000 (22:23 -0800)]
iris: move some non-buffer case code in a bit

5 years agoiris: Stop leaking iris_uncompiled_shaders like mad
Kenneth Graunke [Thu, 29 Nov 2018 00:43:21 +0000 (16:43 -0800)]
iris: Stop leaking iris_uncompiled_shaders like mad

Now shader-db actually executes.  We still need a plan for culling
dead iris_compiled_shaders...

5 years agoiris: Destroy the bufmgr
Kenneth Graunke [Wed, 28 Nov 2018 23:23:13 +0000 (15:23 -0800)]
iris: Destroy the bufmgr

Plugs a 12360 byte leak

5 years agoiris: Fix IRIS_MEMZONE_COUNT to exclude the border color pool
Kenneth Graunke [Wed, 28 Nov 2018 23:29:38 +0000 (15:29 -0800)]
iris: Fix IRIS_MEMZONE_COUNT to exclude the border color pool

This is supposed to exclude single address zones.  We were getting
too many VMA allocators but failing to set them up, which worked out
because we also forgot to destroy them...

5 years agoiris: Unref unbound_tex resource
Kenneth Graunke [Wed, 28 Nov 2018 23:22:07 +0000 (15:22 -0800)]
iris: Unref unbound_tex resource

Plugs a 12536 byte leak

5 years agoiris: Destroy the border color pool
Kenneth Graunke [Wed, 28 Nov 2018 23:15:21 +0000 (15:15 -0800)]
iris: Destroy the border color pool

This plugs a 12224 byte leak

5 years agoiris: Destroy transfer helper on screen teardown
Kenneth Graunke [Wed, 28 Nov 2018 23:10:07 +0000 (15:10 -0800)]
iris: Destroy transfer helper on screen teardown

Plugs a 16 byte leak

5 years agoiris: Fix failed to compile TCS message
Kenneth Graunke [Wed, 28 Nov 2018 23:06:00 +0000 (15:06 -0800)]
iris: Fix failed to compile TCS message

5 years agoiris: Rework tiling/modifiers handling
Kenneth Graunke [Wed, 28 Nov 2018 10:30:42 +0000 (02:30 -0800)]
iris: Rework tiling/modifiers handling

We were being very picky about things being Y tiled.  But, not
everything can be - for example, > 16382 surfaces on SKL GT1-3
have to fall back to linear.

Instead, give ISL options and let it pick.

5 years agoiris: fix conditional compute, don't stomp predicate for pipelined queries
Kenneth Graunke [Tue, 27 Nov 2018 23:30:16 +0000 (15:30 -0800)]
iris: fix conditional compute, don't stomp predicate for pipelined queries

5 years agoiris: check query first
Kenneth Graunke [Tue, 27 Nov 2018 22:20:48 +0000 (14:20 -0800)]
iris: check query first

this lets us avoid the predicate bit in more cases, which is nice