Samuel Pitoiset [Thu, 13 Sep 2018 13:39:43 +0000 (15:39 +0200)]
radv: gather which GS stream is used for every outputs
To only emit outputs for the given stream.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Samuel Pitoiset [Tue, 11 Sep 2018 09:08:49 +0000 (11:08 +0200)]
radv: gather the number of output components per stream
This will be also used for splitting the GS->VS ring buffer.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Samuel Pitoiset [Tue, 11 Sep 2018 09:08:23 +0000 (11:08 +0200)]
radv: gather the number of streams used by geometry shaders
This will be used for splitting the GS->VS ring buffer. The
stream ID is always 0 for now.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Jason Ekstrand [Fri, 5 Oct 2018 14:13:25 +0000 (09:13 -0500)]
nir: Add a pass for gathering transform feedback info
This is different from the GL_ARB_spirv pass because it generates a much
simpler data structure that isn't tied to OpenGL and mtypes.h.
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Jason Ekstrand [Mon, 29 Oct 2018 14:42:21 +0000 (09:42 -0500)]
vulkan: Update the XML and headers to 1.1.90
This doesn't include any new features but it does include an XML and
header typo fix for modifiers.
Acked-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Samuel Pitoiset [Mon, 29 Oct 2018 11:13:13 +0000 (12:13 +0100)]
radv: remove wrong comment in calculate_gs_ring_sizes() about streams
The computation seems correct compared to RadeonSI.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Rob Clark [Sun, 28 Oct 2018 14:45:06 +0000 (10:45 -0400)]
freedreno: don't flush when new and old pfb is identical
In the 'inorder' case (ie. FD_MESA_DEBUG=inorder, or old kernel), if the
u_blitter clear path is used (a3xx, a4xx, and some fallback cases on
newer gens), util_blitter_restore_fb_state() will set_framebuffer_state()
to something that is identical to the current fb state, which triggers
an unnecessary flush, and then eventually an assert:
(gdb) bt
#0 0x0000007fbf24a078 in kill () from /lib64/libc.so.6
#1 0x0000007fbe061278 in _debug_assert_fail (expr=0x7fbe93a820 "!batch->flushed", file=0x7fbe93a628 "../src/gallium/drivers/freedreno/freedreno_batch.c", line=491, function=0x7fbe93a990 <__func__.17380> "fd_batch_check_size") at ../src/gallium/auxiliary/util/u_debug.c:322
#2 0x0000007fbe1ccb8c in fd_batch_check_size (batch=0x55556d5a70) at ../src/gallium/drivers/freedreno/freedreno_batch.c:491
#3 0x0000007fbe1d0e08 in fd_clear (pctx=0x55555c61e0, buffers=5, color=0x55556e388c, depth=1, stencil=0) at ../src/gallium/drivers/freedreno/freedreno_draw.c:463
#4 0x0000007fbe57afa4 in st_Clear (ctx=0x55556e17b0, mask=18) at ../src/mesa/state_tracker/st_cb_clear.c:452
The assert was introduced in
4b847b38ae3, so from a functionality
standpoint this patch fixes that commit. But it should also avoid an
unnecessary flush in the 'inorder' case, fixing a performance bug.
Fixes: 4b847b38ae3 freedreno: make fd_batch a one-shot thing
Signed-off-by: Rob Clark <robdclark@gmail.com>
Rob Clark [Sat, 27 Oct 2018 18:20:22 +0000 (14:20 -0400)]
freedreno: dependency tracking for z/s depends on ZSA state
ZSA state can change whether depth or stencil is enabled
This plus previous patch fix stk, and various things w/
FD_MESA_DEBUG=inorder
Fixes: ec717fc629 freedreno: reduce resource dependency tracking overhead
Signed-off-by: Rob Clark <robdclark@gmail.com>
Rob Clark [Sat, 27 Oct 2018 18:07:09 +0000 (14:07 -0400)]
freedreno: mark all state dirty after switching batch
The problem isn't directly with
ec717fc629 but rather that commit
exposes the problem. When we switch batch we cannot assume previous
state is clean so we should mark all state dirty.
Fixes: ec717fc629 freedreno: reduce resource dependency tracking overhead
Signed-off-by: Rob Clark <robdclark@gmail.com>
Jason Ekstrand [Tue, 16 Oct 2018 21:59:37 +0000 (16:59 -0500)]
anv: Use absolute timeouts in wait_for_bo_fences
We were previously using relative timeouts and decrementing the
user-provided timeout as we waited. Instead, this commit refactors
things to use absolute timeouts throughout. This should fix a subtle
bug in the waitAll case where we aren't decrementing the timeout after a
successful GPU wait. Since pthread_cond_timedwait already takes an
absolute timeout, it's also significantly simpler.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Jason Ekstrand [Fri, 26 Oct 2018 18:36:01 +0000 (13:36 -0500)]
anv: Flag semaphore BOs as external
It probably doesn't actually break anything but it does cause some
assertions in debug builds.
Fixes: 7a89a0d9edae6 "anv: Use separate MOCS settings for external BOs"
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Jason Ekstrand [Tue, 2 Oct 2018 22:29:33 +0000 (17:29 -0500)]
anv: Improve the asserts in anv_buffer_get_range
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Rob Clark [Fri, 26 Oct 2018 18:34:04 +0000 (14:34 -0400)]
freedreno/a6xx: inline draw_impl()
Now that it is just called once per draw (instead of once for binning
and once for draw), let's just inline it. If nothing else, it makes
perf-annotate easier to look at.
Signed-off-by: Rob Clark <robdclark@gmail.com>
Rob Clark [Fri, 26 Oct 2018 17:50:58 +0000 (13:50 -0400)]
freedreno/a6xx: small cleanup
Signed-off-by: Rob Clark <robdclark@gmail.com>
Rob Clark [Fri, 26 Oct 2018 17:48:38 +0000 (13:48 -0400)]
freedreno/a6xx: move where we handle dirty vbo state
Historically this wasn't in fdN_emit_state(), because prior to addition
of blitter in a5xx, fdN_emit_state() was also used in the clear path.
These days that is only true for a2xx (a3xx and a4xx use u_blitter). So
the reason for it not to be in fd6_emit_state() no longer exists.
Signed-off-by: Rob Clark <robdclark@gmail.com>
Rob Clark [Thu, 4 Oct 2018 12:05:49 +0000 (08:05 -0400)]
freedreno: avoid no-op flushes by re-using last-fence
Noticed that with webgl (in chromium, at least) we end up generating a
lot of no-op submits just to get a fence. Tracking the last fence and
returning that if there is no rendering since last flush avoids this.
Signed-off-by: Rob Clark <robdclark@gmail.com>
Kristian H. Kristensen [Wed, 24 Oct 2018 19:02:00 +0000 (12:02 -0700)]
freedreno/a6xx: Move stencil/depth/alpha state to IB
Signed-off-by: Kristian H. Kristensen <hoegsberg@chromium.org>
Kristian H. Kristensen [Thu, 25 Oct 2018 20:46:24 +0000 (13:46 -0700)]
freedreno/a6xx: Move stencil mask emit to FD_DIRTY_ZSA group
Signed-off-by: Kristian H. Kristensen <hoegsberg@chromium.org>
Kristian H. Kristensen [Thu, 25 Oct 2018 20:35:15 +0000 (13:35 -0700)]
freedreno/a6xx: Rename FD6_GROUP_ZSA ro FD6_GROUP_LRZ
Signed-off-by: Kristian H. Kristensen <hoegsberg@chromium.org>
Kristian H. Kristensen [Mon, 22 Oct 2018 16:35:39 +0000 (09:35 -0700)]
freedreno/a6xx: Move rasterizer state to state object
Signed-off-by: Kristian H. Kristensen <hoegsberg@chromium.org>
Kristian H. Kristensen [Wed, 17 Oct 2018 21:18:56 +0000 (14:18 -0700)]
freedreno/a6xx: Fix set_blit_scissor helper
The scissor maxx/maxy are non-inclusive, so don't subtract one from
framebuffer width and height.
Signed-off-by: Kristian H. Kristensen <hoegsberg@chromium.org>
Kristian H. Kristensen [Tue, 16 Oct 2018 21:50:58 +0000 (14:50 -0700)]
freedreno/a2xx: Squash a compiler warning
We get a warning here for assigning a const char * pointer to
char *swizzle in struct ir2_src_register. The constructor strdups a 4
byte string here, so just memcpy to that instead.
Signed-off-by: Kristian H. Kristensen <hoegsberg@chromium.org>
Kristian H. Kristensen [Tue, 16 Oct 2018 21:28:57 +0000 (14:28 -0700)]
freedreno/a6xx: Use fd6_emit_ib from a6xx
Move it to a header and use it where possible to avoid vfunc call.
Signed-off-by: Kristian H. Kristensen <hoegsberg@chromium.org>
Rob Clark [Sun, 21 Oct 2018 14:22:11 +0000 (10:22 -0400)]
freedreno: import libdrm_freedreno + redesign submit
In the pursuit of lowering driver overhead, it became clear that some
amount of redesign of how libdrm_freedreno constructs the submit ioctl
would be needed. In particular, as the gallium driver is starting to
make heavier use of CP_SET_DRAW_STATE state groups/objects, the over-
head of tracking cmd buffers and relocs becomes too much. And for
"streaming" state, which isn't ever reused (like uniform uploads) the
overhead of allocating/freeing ringbuffer[1] objects is too high.
This redesign makes two main changes:
1) Introduces a fd_submit object for tracking bos and cmds table
for the submit ioctl, making ringbuffer objects more light-
weight. This was previously done in the ringbuffer. But we
have many ringbuffer instances involved in a submit (gmem +
draw + potentially 1000's of state-group rbs), and only need
a single bos and cmds table. (Reloc table is still per-rb)
The submit is also a convenient place for a slab allocator for
ringbuffer objects. Other options would have required locking
because, while we can guarantee allocations will only happen on
a single thread, free's could happen either on the application
thread or the flush_queue thread. With the slab allocator in
the submit object, any frees that happen on the flush_queue
thread happen after we know that the application thread is done
with the submit.
2) Introduce a new "softpin" msm_ringbuffer_sp implementation that
does not use relocs and only has cmds table entries for IB1 (ie.
the cmdstream buffers that kernel needs to CP_INDIRECT_BUFFER
to from the RB). To do this properly will require some updates
on the kernel side, so whether you get the softpin or legacy
submit/ringbuffer implementation at runtime depends on your
kernel version.
To make all these changes in libdrm would basically require adding a
libdrm_freedreno2, so this is a good point to just pull the libdrm code
into mesa. Plus it allows for using mesa's hashtable, slab allocator,
etc. And it lets us have asserts enabled for debug mesa buids but
omitted for release builds. And it makes life easier if further API
changes become necessary.
At this point I haven't tried to pull in the kgsl backend. Although
I left the level of vfunc indirection which would make it possible
to have other backends. (And this was convenient to keep to allow
for the "softpin" ringbuffer to coexist.)
NOTE: if bisecting a build error takes you here, try a clean build.
There are a bunch of ways things can go wrong if you still have
libdrm_freedreno cflags.
[1] "ringbuffer" is probably a bad name, the only level of cmdstream
buffer that is actually a ring is RB managed by kernel. User-
space cmdstream is all IB1/IB2 and state-groups.
Reviewed-by: Kristian H. Kristensen <hoegsberg@chromium.org>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
Signed-off-by: Rob Clark <robdclark@gmail.com>
Jason Ekstrand [Tue, 16 Oct 2018 19:58:41 +0000 (14:58 -0500)]
Revert "anv/skylake: disable ForceThreadDispatchEnable"
This reverts commit
0fa9e6d7b304f6a8064ed78a4b9c557e1026e7e5. The real
issue appears to have been that HiZ ops don't like having WM thread
dispatch force-enabled. The previous commit fixes that problem so we
can go back to using the ForceThreadDispatchEnable bit even on SKL+.
Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Jason Ekstrand [Tue, 16 Oct 2018 19:58:18 +0000 (14:58 -0500)]
blorp: Emit a dummy 3DSTATE_WM prior to 3DSTATE_WM_HZ_OP
Cc: mesa-stable@lists.freedesktop.org
Suggested-by: Francisco Jerez <currojerez@riseup.net>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Axel Davy [Sun, 16 Sep 2018 15:43:56 +0000 (17:43 +0200)]
st/nine: Handle window resize when a presentation buffer is used
Usually when a window is resized, the app calls d3d to resize the back
buffer to the window size. In some cases, it is not done,
and it expects the output resizes to the window size, even if
the back buffer size is unchanged.
This patch introduces the behaviour when a presentation buffer
is used.
ID3DPresent_GetWindowInfo is a function available with
D3DPresent v1.0, and thus we don't need to check if the
function is available.
The function had been introduced to implement this very
feature.
Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Axel Davy [Sun, 16 Sep 2018 15:24:45 +0000 (17:24 +0200)]
d3dadapter: Fix wrong naming in header file
GetWindowInfo used to be GetWindowSize before gallium
nine was merged. A left-over remained...
Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Axel Davy [Sun, 14 Oct 2018 21:31:07 +0000 (23:31 +0200)]
st/nine: Reduce MaxSimultaneousTextures to 8
Windows drivers don't set this flag (which affects ff) to more than 8.
Do the same in case some games check for 8.
v2: Remove any dependence on MaxSimultaneousTextures. For non-ff
the number of textures is 16 when the device is able of vs/ps3.
Add this requirement of 16 textures to the driver requirements.
Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Axel Davy [Sun, 14 Oct 2018 20:02:06 +0000 (22:02 +0200)]
st/nine: Enable shadow mapping for ps 1.X
We didn't implement shadow textures for ps 1.X,
assuming the case couldn't happen...
Well it does.
Fixes: https://github.com/iXit/Mesa-3D/issues/261
Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Axel Davy [Sat, 13 Oct 2018 21:33:47 +0000 (23:33 +0200)]
st/nine: Do not set unused states for stateblocks
A lot of these states are used only for the context,
and are unused for stateblocks (which just uses the
changed.* fields instead for a lot of them).
Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Axel Davy [Sat, 13 Oct 2018 21:21:36 +0000 (23:21 +0200)]
st/nine: Fix aliasing states for stateblocks
If NINE_STATE_FF_MATERIAL is set, the stateblock will upload
its recorded materials matrix.
If NINE_STATE_FF_LIGHTING is set, the lighting set is uploaded.
These flags could be set by a NineDevice9_SetTransform call
or by setting some states related to ff, but that shouldn't trigger
these stateblock behaviours.
We don't need to follow the context states dirtied by render states.
NINE_STATE_FF_VSTRANSF is exactly the state controlling stateblock
updates of transformation matrices, NINE_STATE_FF is too broad.
These two changes avoid setting the two mentionned states when we
shouldn't.
Fixes: https://github.com/iXit/Mesa-3D/issues/320
Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Axel Davy [Sat, 13 Oct 2018 20:45:16 +0000 (22:45 +0200)]
st/nine: Never update device changed.* fields
The device state changed.* field are never used.
These fields are used only for stateblocks.
Avoid setting them at all for clarity.
Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Axel Davy [Sun, 23 Sep 2018 14:22:01 +0000 (16:22 +0200)]
st/nine: Capture also default matrices for D3DSBT_ALL
We avoid allocating space for never unused matrices.
However we must do as if we had captured them.
Thus when a D3DSBT_ALL stateblock apply has fewer matrices
than device state, allocate the default matrices for the stateblock
before applying.
Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Axel Davy [Sun, 23 Sep 2018 14:45:30 +0000 (16:45 +0200)]
st/nine: Mark transform matrices dirty for D3DSBT_ALL
D3DSBT_ALL stateblocks capture the transform matrices.
Fixes some d3d test programs not displaying properly.
Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Axel Davy [Sun, 23 Sep 2018 20:28:07 +0000 (22:28 +0200)]
st/nine: Don't update unused world matrices
While to the application we have to track
accurately all 256 world matrices (including
in stateblocks), hw vertex processing enables
to set a limit to the number of world matrices
the hardware can access to in the advertised caps,
which is 8 for nine.
Thus don't bother in the stateblock code to send
the updated values for the unreachable matrices.
Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Axel Davy [Sat, 13 Oct 2018 20:28:34 +0000 (22:28 +0200)]
st/nine: Remove two unused states.
NINE_STATE_MATERIAL was used incorrectly at one location.
Replace it with the correct state.
Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Axel Davy [Sat, 13 Oct 2018 20:35:22 +0000 (22:35 +0200)]
st/nine: Remove commented nine_context_apply_stateblock
At some point the project was to adapt the
commented version to csmt.
The csmt rework enabled to fix some state aliasing
issues between stateblocks and internal state updates.
The commented version needs a lot of work to work with that.
Just drop it.
Signed-off-by: Axel Davy <davyaxel0@gmail.com>
Brian Paul [Fri, 26 Oct 2018 18:34:09 +0000 (12:34 -0600)]
nir: Fix array initializer
Empty initializer is not standard C. This fixes MSVC build.
Trivial.
Jason Ekstrand [Fri, 26 Oct 2018 13:32:39 +0000 (08:32 -0500)]
anv: Return VK_ERROR_DEVICE_LOST from anv_device_set_lost
This lets us get rid of a bunch of duplicated error messages.
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
Jason Ekstrand [Fri, 26 Oct 2018 13:24:49 +0000 (08:24 -0500)]
anv/util: Split a vk_errorv helper out of vk_errorf
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
Brian Paul [Fri, 26 Oct 2018 16:23:39 +0000 (10:23 -0600)]
scons/svga: remove opt from the list of valid build types
This reverts commit
a5fd54f8bf6713312fa5efd7ef5cd125557a0ffe.
The whole point was to add a way to pass -DVMX86_STATS to the build,
but we can do that with a command line argument when we invoke scons.
Reviewed-by: José Fonseca <jfonseca@vmware.com>
Nanley Chery [Thu, 25 Oct 2018 21:08:52 +0000 (14:08 -0700)]
intel/blorp: Define the clear value bounds for HiZ clears
Follow the restriction of making sure the clear value is between the min
and max values defined in CC_VIEWPORT. Avoids a simulator warning for
some piglit tests, one of them being:
./bin/depthstencil-render-miplevels 146 d=z32f_s8
Jason found this to fix incorrect clearing on SKL.
Fixes: 09948151ab1d5184b4dd9052bb1f710fa1e00a7b
("intel/blorp: Add the BDW+ optimized HZ_OP sequence to BLORP")
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Tested-by: Jason Ekstrand <jason@jlekstrand.net>
Eric Engestrom [Thu, 25 Oct 2018 16:36:30 +0000 (17:36 +0100)]
radv: remove duplicate brackets in version string
MESA_GIT_SHA1 resolves to either an empty "" string if not build from git,
or " (git-
DEADBEEF)" if it is. No need to wrap it in additional "()".
Fixes: 9d40ec2cf6ec6d3d9d78 "radv: Add support for VK_KHR_driver_properties."
Signed-off-by: Eric Engestrom <eric.engestrom@intel.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Eric Engestrom [Thu, 25 Oct 2018 10:15:38 +0000 (11:15 +0100)]
vulkan: drop always-true param
Signed-off-by: Eric Engestrom <eric.engestrom@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Boyuan Zhang [Tue, 23 Oct 2018 15:22:13 +0000 (11:22 -0400)]
radeon/vcn: use util function to get h264 profile idc
Use utility function for converting h264 pipe video profile to profile idc,
instead of using array.
Signed-off-by: Boyuan Zhang <boyuan.zhang@amd.com>
Acked-by: Christian König <christian.koenig at amd.com>
Boyuan Zhang [Tue, 23 Oct 2018 15:20:33 +0000 (11:20 -0400)]
radeon/vce: use util function to get h264 profile idc
Use utility function for converting h264 pipe video profile to profile idc,
instead of using array.
Signed-off-by: Boyuan Zhang <boyuan.zhang@amd.com>
Acked-by: Christian König <christian.koenig at amd.com>
Boyuan Zhang [Tue, 23 Oct 2018 15:15:52 +0000 (11:15 -0400)]
vl: get h264 profile idc
Adding a function for converting h264 pipe video profile to profile idc
Signed-off-by: Boyuan Zhang <boyuan.zhang@amd.com>
Acked-by: Christian König <christian.koenig at amd.com>
Jason Ekstrand [Fri, 19 Oct 2018 17:06:36 +0000 (12:06 -0500)]
intel/nir: Use the OPT macro for more passes
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Jason Ekstrand [Fri, 19 Oct 2018 21:32:15 +0000 (16:32 -0500)]
spirv: Initialize subgroup destinations with the destination type
Instead of initializing them manually, just use the type that we already
have sitting there.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Jason Ekstrand [Sat, 20 Oct 2018 00:08:58 +0000 (19:08 -0500)]
spirv: Use the right bit-size for spec constant ops
Previously, we would always pull the bit size from the destination which
is wrong for opcodes like nir_ilt where the sources are variable-sized
but the destination is a fixed size. We were getting lucky before
because nir_op_ilt returns a 32-bit value and basically everyone who
uses spec constants uses 32-bit ones.
Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Jason Ekstrand [Fri, 19 Oct 2018 17:12:28 +0000 (12:12 -0500)]
nir/prog: Use nir_bany in kill handling
We have a helper that does exactly what the bany_inequal was doing. It
emits the same code but is a bit higher level and is designed to operate
on a bvec4.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Jason Ekstrand [Thu, 18 Oct 2018 22:55:49 +0000 (17:55 -0500)]
glsl/nir: Use i2b instead of ine for fixing UBO/SSBO Booleans
They do the same thing in the end but i2b is a bit simpler. Also, let's
clean up the mess of code for SSBO handling with one line of builder.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Jason Ekstrand [Fri, 19 Oct 2018 15:51:47 +0000 (10:51 -0500)]
nir/system_values: Use the bit size from the load_deref
This isn't a great solution for bit-sizes but we don't have a
particularly convenient way to get a bit size from the system value enum
and this keeps the lowering pass from changing it.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Jason Ekstrand [Sat, 20 Oct 2018 02:42:22 +0000 (21:42 -0500)]
nir/opt_if: Rework condition propagation
Instead of doing our own constant folding, we just emit instructions and
let constant folding happen. This is substantially simpler and lets us
use the nir_imm_bool helper instead of dealing with the const_value's
ourselves.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Jason Ekstrand [Mon, 22 Oct 2018 19:08:13 +0000 (14:08 -0500)]
nir/search: Use the nir_imm_* helpers from nir_builder
This requires that we rework the interface a bit to use nir_builder but
that's a nice little modernization anyway.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Jason Ekstrand [Mon, 22 Oct 2018 19:08:44 +0000 (14:08 -0500)]
nir/builder: Handle 16-bit floats in nir_imm_floatN_t
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Jason Ekstrand [Fri, 19 Oct 2018 14:35:49 +0000 (09:35 -0500)]
nir/builder: Add a nir_imm_true/false helpers
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Jason Ekstrand [Mon, 22 Oct 2018 20:53:14 +0000 (15:53 -0500)]
nir/constant_folding: Use nir_src_as_bool for discard_if
Missed one while converting to the nir_src_as_* helpers.
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Jason Ekstrand [Fri, 19 Oct 2018 03:26:03 +0000 (22:26 -0500)]
nir/constant_folding: Add an unreachable to a switch
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Jason Ekstrand [Thu, 18 Oct 2018 20:18:30 +0000 (15:18 -0500)]
nir/validate: Print when the validation failed
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Jason Ekstrand [Tue, 13 Mar 2018 19:26:20 +0000 (12:26 -0700)]
anv: Handle the device loss abort in anv_device_set_lost
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
Jason Ekstrand [Tue, 13 Mar 2018 18:50:33 +0000 (11:50 -0700)]
anv: Add helpers for setting/checking device lost
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
Jason Ekstrand [Thu, 25 Oct 2018 15:13:12 +0000 (10:13 -0500)]
anv: Provide a error message with a DEVICE_LOST
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Reviewed-by: Eric Engestrom <eric.engestrom@intel.com>
Alex Smith [Thu, 25 Oct 2018 09:50:52 +0000 (10:50 +0100)]
anv: Fix sanitization of stencil state when the depth test is disabled
When depth testing is disabled, we shouldn't pay attention to the
specified depthCompareOp, and just treat it as always passing. Before,
if the depth test is disabled, but depthCompareOp is VK_COMPARE_OP_NEVER
(e.g. from the app having zero-initialized the structure), then
sanitize_stencil_face() would have incorrectly changed passOp to
VK_STENCIL_OP_KEEP.
v2: Roll the depthTestEnable check into the ds_aspect check below since
they now both do the same thing.
Fixes: 028e1137e6 "anv/pipeline: Be smarter about depth/stencil state"
Signed-off-by: Alex Smith <asmith@feralinteractive.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Samuel Pitoiset [Wed, 24 Oct 2018 06:50:26 +0000 (08:50 +0200)]
radv: implement image to image operations for R32G32B32
This should address the remaining failures in Batman Arkhman City.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=107765
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Samuel Pitoiset [Wed, 24 Oct 2018 06:50:25 +0000 (08:50 +0200)]
radv: fix a comment in radv_meta_buffer_to_image_cs_r32g32b32()
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Samuel Pitoiset [Wed, 24 Oct 2018 06:50:24 +0000 (08:50 +0200)]
radv: add get_image_stride_for_r32g32b32() helper
For the special R32G32B32 paths.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Samuel Pitoiset [Wed, 24 Oct 2018 06:50:23 +0000 (08:50 +0200)]
radv: add create_bview_for_r32g32b32() helper
For the special R32G32B32 paths.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Samuel Pitoiset [Wed, 24 Oct 2018 06:50:22 +0000 (08:50 +0200)]
radv: add create_buffer_from_image() helper
For the special R32G32B32 paths.
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Bas Nieuwenhuizen <bas@basnieuwenhuizen.nl>
Sagar Ghuge [Wed, 24 Oct 2018 23:25:53 +0000 (16:25 -0700)]
intel/compiler: Print message descriptor as immediate source
While disassembling send(c) instruction print message descriptor as
immediate source operand along with message descriptor. This allows
assembler to read immediate source operand and set bits accordingly.
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Sagar Ghuge [Wed, 24 Oct 2018 20:27:27 +0000 (13:27 -0700)]
intel/compiler: Print hex representation along with floating point value
While encoding the immediate floating point values in instruction we use
values upto precision 9, but while disassembling, we print precision to
6 places, which round up the value and gives wrong interpretation for
encoded immediate constant.
To avoid misinterpretation of encoded immediate values in instruction
and disassembled output, print hex representation along with floating
point value which can be used by assembler in future.
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
David McFarland [Wed, 24 Oct 2018 00:51:09 +0000 (21:51 -0300)]
util: Change remaining uint32 cache ids to sha1
After discussion with Timothy Arceri. disk_cache_get_function_identifier
was using only the first byte of the sha1 build-id. Replace
disk_cache_get_function_identifier with implementation from
radv_get_build_id. Instead of writing a uint32_t it now writes to a
mesa_sha1. All drivers using disk_cache_get_function_identifier are
updated accordingly.
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Fixes: 83ea8dd99bb1 ("util: add disk_cache_get_function_identifier()")
Hyunjun Ko [Wed, 24 Oct 2018 01:57:15 +0000 (10:57 +0900)]
freedreno: use fd_bc_alloc_batch instead of fd_batch_create.
Following the commit
2385d7b066 and
8e798e28f7, for resource dependancy
tracking.
Fixes: dEQP-GLES31.functional.image_load_store.early_fragment_tests.no_early_fragment_tests_depth_fbo
with FD_MESA_DEBUG=inorder
Signed-off-by: Rob Clark <robdclark@gmail.com>
Hyunjun Ko [Thu, 25 Oct 2018 08:26:19 +0000 (17:26 +0900)]
freedreno/ir3: take reg->num out of union in ir3_register
To avoid wrong result when identifying the type of register.
Ie. If the reg is an array, it might be identified as address or
predicate register.
Fixes: dEQP-GLES31.functional.ssbo.layout.random.arrays_of_arrays.6
Signed-off-by: Rob Clark <robdclark@gmail.com>
Rob Clark [Thu, 25 Oct 2018 19:27:10 +0000 (15:27 -0400)]
freedreno/a6xx: disable unused groups
Don't leave vsconst/fsconst group enabled if we switch to shader with no
uniforms.
Fixes: abcdf5627a2 freedreno/a6xx: move const emit to state group
Signed-off-by: Rob Clark <robdclark@gmail.com>
Rob Clark [Thu, 18 Oct 2018 13:05:52 +0000 (09:05 -0400)]
freedreno: add useful assert
Would have been useful to catch the problem fixed in
8e798e28f736e22e9e1e4534ab42a36cde14b142
Signed-off-by: Rob Clark <robdclark@gmail.com>
Alok Hota [Tue, 16 Oct 2018 23:15:29 +0000 (18:15 -0500)]
swr/rast: ignore CreateElementUnorderedAtomicMemCpy
This function's API changed between LLVM 5 and 6. Compile errors occur
when building with LLVM 6+ if LLVM 5 was used for a dist tarball
CC: <mesa-stable@lists.freedesktop.org>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=107865
Reviewed-by: Emil Velikov <emil.velikov@collabora.com>
Alok Hota [Wed, 19 Sep 2018 17:42:57 +0000 (12:42 -0500)]
swr/rast: fix intrinsic/function for LLVM 7 compatibility
Converted from x86 VFMADDPS intrinsic to generic LLVM intrinsic, and
removed createInstructionSimplifierPass, which were both removed in LLVM
7.0.0
These changes combine patches we received from the community and our own
internal patches
Reviewed-by: Bruce Cherniak <bruce.cherniak@intel.com>
Tested-by: Chuck Atkins <chuck.atkins@kitware.com>
Rhys Perry [Thu, 4 Oct 2018 12:40:43 +0000 (13:40 +0100)]
nvc0: increase NOUVEAU_TRANSFER_PUSHBUF_THRESHOLD to 1024 on Kepler+
Gives a +3.89% to +5.27% FPS improvement with Hitman and +2.73% to +2.82%
FPS improvement with Dirt Rally on my GTX 1060.
Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Bas Nieuwenhuizen [Tue, 23 Oct 2018 08:54:24 +0000 (10:54 +0200)]
radv: Emit enqueued pipeline barriers on event write.
Since the CPU can read them we need to execute any GPU->CPU
flushes before the event is written.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=108524
Fixes: f4e499ec791 "radv: add initial non-conformant radv vulkan driver"
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Bas Nieuwenhuizen [Sun, 30 Sep 2018 18:02:04 +0000 (20:02 +0200)]
radv: Add support for VK_KHR_driver_properties.
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Eric Engestrom [Sat, 20 Oct 2018 17:00:09 +0000 (18:00 +0100)]
util: use C99 declaration in the for-loop set_foreach() macro
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Eric Engestrom [Sat, 20 Oct 2018 17:00:08 +0000 (18:00 +0100)]
util: use C99 declaration in the for-loop hash_table_foreach() macro
Signed-off-by: Eric Engestrom <eric@engestrom.ch>
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Dylan Baker [Tue, 23 Oct 2018 17:02:05 +0000 (10:02 -0700)]
gen: Add AMD_gpu_shader_int64.xml to tarball
CC: Ian Romanick <ian.d.romanick@intel.com>
CC: Marek Olšák <marek.olsak@amd.com>
Fixes: b3c17330e631695b5e5dc209ba9ea1a528618c97
("mesa: expose AMD_gpu_shader_int64")
Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Dylan Baker [Tue, 23 Oct 2018 17:00:01 +0000 (10:00 -0700)]
gen: Add EXT_vertex_attrib_64bit.xml to dependency lists
Which is also required to put it in the tarball, a requirement for
building with meson from the tarball.
CC: Ian Romanick <ian.d.romanick@intel.com>
CC: Marek Olšák <marek.olsak@amd.com>
Fixes: 263c962cfdee6b43578ee5f28601309ea77d1434
("mesa: expose EXT_vertex_attrib_64bit")
Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Eric Engestrom [Tue, 23 Oct 2018 14:37:21 +0000 (15:37 +0100)]
anv: move variable to proper scope and mark as MAYBE_UNUSED
Signed-off-by: Eric Engestrom <eric.engestrom@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Eric Engestrom [Tue, 23 Oct 2018 14:27:51 +0000 (15:27 +0100)]
anv: use snprintf() instead of memset()+strcpy()
snprintf() guarantees that it will not write more chars than allowed,
and that the string will be null-terminated, without the need to fill
the whole thing with zeroes to begin with.
Signed-off-by: Eric Engestrom <eric.engestrom@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Eric Engestrom [Tue, 23 Oct 2018 14:25:45 +0000 (15:25 +0100)]
anv: drop unused includes
Signed-off-by: Eric Engestrom <eric.engestrom@intel.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Dylan Baker [Tue, 23 Oct 2018 18:01:12 +0000 (11:01 -0700)]
autotools: include intel_tiled_memcopy.c
There are two problems with the fixed patch. First, it fails to create a
dependency on the sourced .c file, so changes to intel_tiled_memcpy.c
won't trigger a rebuild. It also doesn't get included in the dist
tarball.
Fixes: 11b1afdc92db98e93f2ca50beeb7fc481a11e708
("i965/tiled_memcpy: inline movntdqa loads in tiled_to_linear")
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Dylan Baker [Tue, 23 Oct 2018 17:40:15 +0000 (10:40 -0700)]
meson: fix formatting and add extra_files to i965
extra_files is just a nice way to to tell certain IDEs (and those
reading the file) that this file is also a dependency. Meson will use
the .d file generated by the compiler to figure out what the target
actually depends on.
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Juan A. Suarez <jasuarez@igalia.com>
Eduardo Lima Mitev [Tue, 23 Oct 2018 19:24:11 +0000 (21:24 +0200)]
ir3_compiler/nir: fix imageSize() for buffer-backed images
GL_EXT_texture_buffer introduced texture buffers, which can be used
in shaders through a new type imageBuffer.
Because how image access is implemented in freedreno, calling
imageSize on an imageBuffer returns the size in bytes instead of texels,
which is incorrect.
This patch adds a division of imageSize result by the bytes-per-pixel
of the image format, when image is buffer-backed.
Fixes all tests under
dEQP-GLES31.functional.image_load_store.buffer.image_size.*
v2: Pre-compute and submit the log2 of the image format's bpp as shader
constant instead of emitting the LOG2 instruction in code. (Rob Clark)
v3: Use ffs (find-first-bit) helper for computing log2 (Ilia Mirkin)
Reviewed-by: Rob Clark <robdclark@gmail.com>
Jose Fonseca [Wed, 24 Oct 2018 10:33:09 +0000 (11:33 +0100)]
nir: Fix array initializer.
Empty initializer is not standard C. This fixes MSVC build.
Trivial.
Liviu Prodea [Wed, 24 Oct 2018 10:08:35 +0000 (11:08 +0100)]
scons: Put to rest zombie texture_float build option.
I found a remnant of texture_float build option that wasn't removed in
commit
66673bef941af344314fe9c91cad8cd330b245eb
This patch removes it.
Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
Alex Smith [Thu, 18 Oct 2018 16:29:37 +0000 (17:29 +0100)]
anv: Allow presenting via a different GPU
anv_GetPhysicalDeviceSurfaceSupportKHR will already return success for
this, but anv_GetPhysicalDevice{Xcb,Xlib}PresentationSupportKHR do not.
Apps which check for presentation support via the latter (all Feral
Vulkan games at least) will therefore fail.
This allows me to render on an Intel GPU and present to a display
connected to an AMD card (tested HD 530 + Vega 64).
v2: Rebase on current master.
Signed-off-by: Alex Smith <asmith@feralinteractive.com>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Juan A. Suarez Romero [Tue, 23 Oct 2018 13:55:11 +0000 (15:55 +0200)]
nir: fix nir_copy_propagation test
Use nir_src_comp_as_uint() to read the proper second component, as
nir_src_as_uint() returns the first one.
v2: Use nir_src_comp_as_uint() [Jason]
Fixes: 16870de8a0a ("nir: Use nir_src_is_const and nir_src_as_* in core
code")
Signed-off-by: Juan A. Suarez Romero <jasuarez@igalia.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=108532
Tested-by: Michel Dänzer <michel.daenzer@amd.com>
Tested-by: Vinson Lee <vlee@freedesktop.org>
Reviewed-by: Jason Ekstrand <jason@jlekstrand.net>
Timothy Arceri [Tue, 23 Oct 2018 10:56:31 +0000 (21:56 +1100)]
radv: call nir_link_xfb_varyings()
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Timothy Arceri [Tue, 23 Oct 2018 10:56:30 +0000 (21:56 +1100)]
radv: move nir_lower_io_to_scalar_early() to radv_link_shaders()
nir_lower_io_to_scalar_early() is really part of the link time
optimisations. Moving it here allows the code to be simplified
and also keeps the code easy to follow in the next patch.
Reviewed-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Samuel Pitoiset [Tue, 23 Oct 2018 10:56:29 +0000 (21:56 +1100)]
nir: add linking helper nir_link_xfb_varyings()
The linking opts shouldn't try removing or compacting XFB varyings
in the consumer. To avoid this we copy the always_active_io flag
from the producer.
Reviewed-by: Timothy Arceri <tarceri@itsqueeze.com>
Sagar Ghuge [Sat, 20 Oct 2018 01:25:23 +0000 (18:25 -0700)]
intel/compiler: Change src1 reg type to unsigned doubleword
To have uniform behavior while disassembling send(c) instruction use
register type of unsigned doubleword for src1 when message descriptor is
immediate value. Bspec does not specifiy anything for src1 immediate
default type.
Reviewed-by: Samuel Iglesias Gonsálvez <siglesias@igalia.com>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Sagar Ghuge <sagar.ghuge@intel.com>