radeonsi: implement SAMPLEPOS fragment shader input

The sample positions are read from a constant buffer.

radeonsi: implement SAMPLEID fragment shader input

radeonsi: implement set_min_samples

This is how per-sample shading is enabled.

radeon: add basic register setup for per-sample shading

Only for Cayman, SI, CIK.

radeon: split cayman_emit_msaa_state into 2 functions

The other function will be split up from the framebuffer state.

Revert "glx: load dri driver with RTLD_LOCAL so dlclose never fails to unload"

This reverts commit e3cc0d90e14e62a0a787b6c07a6df0f5c84039be.

It breaks too many apps and completely breaks my desktop too.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=79469

We'll probably need to re-release all stable versions after this is committed.

Cc: "10.1 10.2" <mesa-stable@lists.freedesktop.org>

r600g: use TGSI_PROPERTY to disable viewport and clipping

v2 get rid of magic value, use DEFINES
v3 update clip_disable together with vs_position_window_space

Big thanks to Marek Olšák!

Signed-off-by: David Heidelberger <david.heidelberger@ixit.cz>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>

gallium: create TGSI_PROPERTY to disable viewport and clipping

Marek v2: add a cap

Signed-off-by: Marek Olšák <marek.olsak@amd.com>

r600g: remove assert on draw with count == 0

Signed-off-by: Marek Olšák <marek.olsak@amd.com>

r600g: HW bug workaround for TGSI_OPCODE_BREAKC

Signed-off-by: Marek Olšák <marek.olsak@amd.com>

r600g: implement TGSI_OPCODE_BREAKC

Signed-off-by: Marek Olšák <marek.olsak@amd.com>

r600g: support all channels of TGSI_FILE_ADDRESS

It's allowed in SM3.

v2: fix multi-component tgsi_r600_arl (FLT_TO_INT is trans-only)

Signed-off-by: Marek Olšák <marek.olsak@amd.com>

r600g: check for PIPE_BIND_BLENDABLE in is_format_supported

v2: added !util_format_is_depth_or_stencil(format)

Signed-off-by: Marek Olšák <marek.olsak@amd.com>

r600g: handle PIPE_QUERY_GPU_FINISHED

Signed-off-by: Marek Olšák <marek.olsak@amd.com>

i965/fs: Add fs_inst constructor that takes a list of sources.

Also add an emit() function that calls it.

Reviewed-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/fs: Add a function to resize fs_inst's sources array.

Reviewed-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/fs: Clean up fs_inst constructors.

In a fashion suggested by Ken.

Reviewed-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/fs: Loop from 0 to inst->sources, not 0 to 3.

Reviewed-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/fs: Store the number of sources an fs_inst has.

Reviewed-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/fs: ralloc fs_inst's fs_reg sources.

Reviewed-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/fs: Disable fs_inst assignment operator.

The fs_reg src array is going to turn into a pointer and we'd rather not
consider the implications of shallow copying fs_insts.

Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Reviewed-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/fs: Add and use an fs_inst copy constructor.

Will get more complicated when fs_reg src becomes a pointer.

Reviewed-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965: Skip IR annotations with INTEL_DEBUG=noann.

Running shader-db with INTEL_DEBUG=noann reduces the runtime
from ~90 to ~80 seconds on my machine. It also reduces the disk space
consumed by the .out files from 660 MB (676 on disk) to 343 MB (358 on
disk).

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965/fs: Debug the optimization passes by dumping instr to file.

With INTEL_DEBUG=optimizer, write the output of dump_instructions() to a
file each time an optimization pass makes progress. This lets you easily
diff successive files to see what an optimization pass did.

Example filenames written when running glxgears:
   fs8-0000-00-start
   fs8-0000-01-04-opt_copy_propagate
   fs8-0000-01-06-dead_code_eliminate
   fs8-0000-01-12-compute_to_mrf
   fs8-0000-02-06-dead_code_eliminate
        |   |  |   |
        |   |  |   `-- optimization pass name
        |   |  |
        |   |  `-- optimization pass number in the loop
        |   |
        |   `-- optimization loop interation
        |
        `-- shader program number

Note that with INTEL_DEBUG=optimizer, we disable compact_virtual_grfs,
so that we can diff instruction lists across loop interations without
the register numbers being changes.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965: Give dump_instructions() a filename argument.

This will allow debugging code to dump the IR after an optimization pass
makes progress (the next patch). Only let it open and write to a file if
the effective user isn't root.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965: Give dump_instruction() a FILE* argument.

Use function overloading rather than default arguments, since gdb
doesn't know about default arguments.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965: Add envvar to debug the optimization passes.

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

llvmpipe: (trivial) drop "unswizzled" from some function names

This made sense when swizzled storage layout was used for rendering to tiles.
But nowadays the name just adds confusion (and makes for long lines).

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>

llvmpipe: fix crash when not all attachments are populated in a fb

Framebuffers can have NULL attachments since a while. llvmpipe handled
that properly for lp_rast_shade_quads_mask but it seems the change didn't
make it to lp_rast_shade_tile.
This fixes piglit fbo-drawbuffers-none test (though I need to increase
the FB_SIZE from 32 to 256 so the tris cover some tiles fully).
https://bugs.freedesktop.org/show_bug.cgi?id=79421

Cc: "10.1 10.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>

softpipe: honor the render_condition_enable bit in blits.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>

llvmpipe: honor the render_condition_enable bit in blits.

This fixes piglit nv_conditional_render-blitframebuffer.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>

gallium/docs: improve documentation of render condition wrt blits.

Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>

svga: use svga_shader_too_large() in compile_vs()

And rework the dummy shader code to match the fragment shader case.

Reviewed-by: José Fonseca <jfonseca@vmware.com>

svga: use svga_shader_too_large() in compile_fs()

Reviewed-by: José Fonseca <jfonseca@vmware.com>

svga: added svga_shader_too_large() helper

To check if a shader bytcode exceeds the device limit.  There's no
limit when using GBS.

Reviewed-by: José Fonseca <jfonseca@vmware.com>

darwin: Remove extra kCGLPFAColorSize attribute when requesting an offscreen context

https://xquartz.macosforge.org/trac/ticket/650

Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>

util: Do not use __builtin_clrsb with Intel C++ Compiler.

This patch fixes this build error with icc 14.0.2.

In file included from state_tracker/st_glsl_to_tgsi.cpp(63):
../../src/gallium/auxiliary/util/u_math.h(583): error: identifier "__builtin_clrsb" is undefined
     return 31 - __builtin_clrsb(i);
                 ^

Signed-off-by: Vinson Lee <vlee@freedesktop.org>

i915: add a missing NULL pointer check

mesaVisual can be NULL with configless context since this commit:

    commit 551d459af421a2eb937e9e16301bb64da4624f89
    Author: Neil Roberts <neil@linux.intel.com>
    Date:   Fri Mar 7 18:05:47 2014 +0000

    Add the EGL_MESA_configless_context extension
...
    Previously the i965 and i915 drivers were explicitly creating a zeroed visual
    whenever 0 is passed for the EGLConfig.

We attempt to dereference the visual in i915 and now we don't create a
zeroed-out one one it crashes, breaking at least weston in an i915. There's
no point in doing so as it would be zero anyway.

v2: Fixed a typo in commit message.  Added some tags.

Signed-off-by: Lubomir Rintel <lkundrak@v3.sk>
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1100967
Cc: "10.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

glapi: Duplicate GLES1 prototypes in glapi_dispatch.c

These prototypes are necessary because GLES1 library builds will create
dispatch functions for them.  We can't directly include GLES/gl.h
because it would conflict the previously-included GL/gl.h.  Since GLES1
ABI is not expected to every add more functions, the path of least
resistance is to just duplicate the prototypes for the functions that
aren't already in desktop OpenGL.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=79294
Acked-by: Matt Turner <mattst88@gmail.com>
Tested-by: Andreas Boll <andreas.boll.dev@gmail.com>
Cc: "10.2" <mesa-stable@lists.freedesktop.org>

i965/vec4: Allow writemasking on math instructions on Gen7+.

The math instruction was Align1-only on Gen6 and we never updated this
to let it use Align16 features like writemasking on newer platforms.

total instructions in shared programs: 1686120 -> 1685507 (-0.04%)
instructions in affected programs:     48593 -> 47980 (-1.26%)

Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i965: Fix Line Stipple enable bit in 3DSTATE_SF for Haswell.

Cc: "10.1 10.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Signed-off-by: Pavel Popov <pavel.e.popov@intel.com>

st/wgl: use _debug_printf() instead of fprintf()

This should print output both for debug and release builds.
Suggested by Jose.

Reviewed-by: José Fonseca <jfonseca@vmware.com>
Reviewed-by: Charmaine Lee <charmainel@vmware.com>

st/wgl: formatting fixes in stw_framebuffer.c

And remove some unneeded #includes and INLINE qualifiers.

Reviewed-by: Charmaine Lee <charmainel@vmware.com>

st/wgl: make stw_lookup_context_locked() an inline function

Reviewed-by: Charmaine Lee <charmainel@vmware.com>

st/wgl: fix implementation of wglCreateContextAttribsARB()

wglCreateContextAttribsARB() didn't work previously since it returned
a context ID that wasn't allocated by OPENGL32.DLL.  So if that context
ID was later passed to wglMakeCurrent(), etc. it was rejected.

Now when wglCreateContextAttribsARB() is called we actually call
wglCreateContext() in order to get a valid context ID.  Then we
replace the context data which was created with new context data
which reflects the arguments passed to wglCreateContextAttribsARB().

If there were a DrvCreateContextAttribs() function in the ICD this
work-around wouldn't be necessary.

Reviewed-by: Charmaine Lee <charmainel@vmware.com>
Conflicts:
src/gallium/state_trackers/wgl/stw_ext_extensionsstring.c
src/gallium/state_trackers/wgl/stw_getprocaddress.c

st/wgl: add debug code to check that pixel format initialization worked

If the assertion fails, it means something is really broken.  Before,
if this happened we reverted to the GDI renderer without any warning.

Reviewed-by: Matthew McClure <mcclurem@vmware.com>

st/wgl: change PFD_SWAP_COPY to PFD_SWAP_EXCHANGE.

To reflect our actual SwapBuffers implementation.  See
stw_st_swap_framebuffer_locked().  This fixes various rendering issues
with SolidEdge.

Reviewed-by: Jose Fonseca <jfonseca@vmware.com>

docs: Document how to replace Windows built-in OpenGL software rasterizer with llvmpipe.

Just happened to stumble across this registry key while debugging
something else.

This technique is much neater than trying to override opengl32.dll.

Also a few minors cleanups.

scons: add common.c as part of glcpp build

to have _mesa_error_no_memory function available

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=79440
Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>

mesa: Add missing null checks into prog_hash_table.c

Check calloc return values in hash_table_insert() and
hash_table_replace()

Signed-off-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

glcpp: link with tests/common.c

So that prog_hash_table can use _mesa_error_no_memory function.

Signed-off-by: Tapani Pälli <tapani.palli@intel.com>
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>

mesa/main: Add missing null check in _mesa_CreatePerfQueryINTEL()

Signed-off-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Reviewed-by: Petri Latvala <petri.latvala@intel.com>

mesa/drivers: Add extra null check in blitframebuffer_texture()

If texObj == NULL here it mean there is already GL_INVALID_VALUE
or GL_OUT_OF_MEMORY error set to context.

Signed-off-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

glsl: Add null check in loop_analysis.cpp

Check return value from hash_table_find before using it as a pointer

Signed-off-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

mesa: add missing null check in _mesa_NewHashTable()

Signed-off-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

loader: add optional /sys filesystem method for PCI identification.

Introduce a simple PCI identification method of looking up the answer
the /sys filesystem (available on Linux).  Attempted after libudev, but
before DRM.

Disabled by default (available only when the --enable-sysfs configure
option is specified).

Signed-off-by: Gary Wong <gtw@gnu.org>
Acked-by: Emil Velikov <emil.l.velikov@gmail.com>

loader: allow attempting more than one method of PCI identification.

loader_get_pci_id_for_fd() and loader_get_device_name_for_fd() now attempt
all available strategies to identify the hardware, instead of conditionally
compiling in a single test.  The existing libudev and DRM approaches have
been retained, attempting first libudev (if available) and then DRM (if
necessary).

Signed-off-by: Gary Wong <gtw@gnu.org>
Reviewed-by: Emil Velikov <emil.l.velikov@gmail.com>

st/egl: do not link against libloader

Move the link to the final targets, like any other place in
mesa/gallium. This allows better visibilty and will prevent
us from including the library archive twice.

Resolves multiple definition of `loader_get_pci_id_for_fd'

multiple definition of `loader_get_pci_id_for_fd'

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=79263
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=79382
Cc: Chia-I Wu <olv@lunarg.com>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>
Reviewed-by: Chia-I Wu <olv@lunarg.com>
Tested-by: Michel Dänzer <michel.daenzer@amd.com>

egl_dri2: fix wayland_platform when drm_platform is not set

The build fails with implicit delaration of drmGetCap (xf86drm.h)
Were we're including the header only when building the DRM_PLATFORM.

Wayland backend can operate without DRM_PLATFORM so replace the
guard, and fold in drmGetCap() usage to silence compiler warnings.

Cc: Chad Versace <chad.versace@linux.intel.com>
Cc: Kristian Høgsberg <krh@bitplanet.net>
Signed-off-by: Emil Velikov <emil.l.velikov@gmail.com>

i965/fs: Set correct number of regs_written for MCS fetches.

regs_written is in units of virtual GRFs.

Reviewed-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

glx: load dri driver with RTLD_LOCAL so dlclose never fails to unload

There is no reason anymore to load with RTLD_GLOBAL and for some driver
this even result in dlclose failing to unload leading to catastrophic
failure with swrast fallback.

Cc: "10.1 10.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Signed-off-by: Jérôme Glisse <jglisse@redhat.com>

i915g: Support B5G5R5A1 render targets and textures

i915g: Support R4G4B4A4 render targets and textures

i915g: Fix copy region code

This fixes a few issues with it, also cleans up the code.

glsl/tests: remove generated tests from the repo

They were made unneccesary by the last commit.

Signed-off-by: Connor Abbott <cwabbott0@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>

glsl/tests: call create_test_cases.py in optimization-test

This way, when someone modifies create_test_cases.py and forgets to
commit their changes again, people will notice.

v2: make sure we parse the right directories and check for existance the
right way.

v3 (Ken): Use $PYTHON2 instead of calling python directly.

Signed-off-by: Connor Abbott <cwabbott0@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>

glsl/tests/lower_jumps: fix generated sexpr's for loops

In 088494aa (as well as other commits in the series) Paul Berry modified
the tests for lower_jumps to account for the fact that the s-expression
for the loop IR instruction changed from
(loop () () () () (statements...)) to (loop (statements...)), but he
forgot to update create_test_cases.py which he used to create the tests.
Fix that, so that now create_test_cases.py is synced with the generated
tests.

Signed-off-by: Connor Abbott <cwabbott0@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>

glsl: be more consistent about printing constants

Make sure that we print the same number of digits when printing 0.0 as
any other floating-point number. This will make generating expected
output files for tests easier. To avoid breaking "make check," update
the generated tests for lower_jumps before the next commit which will
bring create_test_cases.py in line with them.

Signed-off-by: Connor Abbott <cwabbott0@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>

glsl: replace strncmp("gl_") calls with new is_gl_identifier() helper

Makes things a little easier to read.

Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

glsl: fix use-after free bug/crash in ast_declarator_list::hir()

The call to get_variable_being_redeclared() may delete 'var' so we
can't reference var->name afterward.  We fix that by examining the
var's name before making that call.

Fixes valgrind warnings and possible crash when running the piglit
tests/spec/glsl-1.30/execution/clipping/vs-clip-distance-in-param.shader_test
test (and probably others).

Cc: "10.1 10.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

i965: Fix repeated usage of rectangle texture coordinate scaling.

Previously, we set up new entries in the params[] array on every access
of a rectangle texture.  Unfortunately, we only reserve space for
(2 * MaxTextureImageUnits) extra entries, so programs which accessed
rectangle textures more times than that would write off the end of the
array and likely crash.

We don't really have a decent mapping between the index returned by
_mesa_add_state_reference and our index into the params array, so we
have to manually search for it.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=78691
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Cc: mesa-stable@lists.freedesktop.org

egl-static: Fix undefined reference to `loader_*'

Trivial.  Better than a broken build.

meta/blit: Use gl_FragColor also in the msaa blit shader

Fixes framebuffer_blit_functionality_multisampled_to_singlesampled_blit
es3 cts test on bdw. Also fixes this on ivb when ivb is forced to use
the meta path.

No piglit regressions on IVB.

Further input from Ken:

 "Unfortunately, this doesn't fix MRT for integer data.

  In the single-sampled case, since we're directly copying data, we were
  read/copy/write data as "float" values, which actually contained the
  integer bits.  Here, we can't do that since we need to process the
  actual integer data.

  I do wonder if we could use intBitsToFloat/uintBitsToFloat to stuff the
  integer bits in the float gl_FragColor output.  Just a crazy idea.

  In the long term (post 10.2), I think we should draft an extension that
  allows you to do "layout(location = all)" on user-defined fragment
  shader outputs.  (Or some similar syntax.)"

Signed-off-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Cc: "10.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Anuj Phogat <anuj.phogat@gmail.com>

nvc0/ir: use SM35 ISA with GK20A

GK20A is mostly compatible with GK104, but uses the SM35 ISA. Use
the GK110 path when this chip is detected.

Signed-off-by: Alexandre Courbot <acourbot@nvidia.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>

nvc0: add GK20A 3D class

GK20A is mostly compatible with GK104, but features a new 3D
class. Add it to the relevant header and use it when GK20A is
detected.

Signed-off-by: Alexandre Courbot <acourbot@nvidia.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>

i965/sf: Replace push/pop in brw_emit_anyprim_setup.

Each of the subroutine emitters alter the predication state, but
otherwise don't change anything (or put it back when they do).
Resetting predication at the end makes these functions idempotent with
regard to the default instruction state - which is a nice property.

With that in place, push/pop is no longer necessary.

v2: Improve whitespace (requested by Matt).

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>

i965/sf: Drop unnecessary push/pop in copy_z_inv_w.

brw_MOV doesn't alter the default instruction state, so this does
nothing.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>

i965/sf: Drop unnecessary push/pop in flatshading code.

brw_JMPI sets predicate_control to BRW_PREDICATE_NONE, but that's
already the value coming in.  Otherwise, nothing changes state.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>

i965/sf: Move brw_compile::flag_value to brw_sf_compile.

This field is only used to track the current value of the flag register
during the SF compile.  It has no place in the common compiler code.

While we're changing every call, drop the 'brw' prefix from the function
since it's static.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>

i965/sf: Move brw_set_predicate_control_flag_value to brw_sf_emit.c.

Only the Gen4-5 SF program compiler actually uses this function; move
it there.  Soon the fields will be moved out of brw_compile.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>

i965/sf: Drop useless push/pop state from flag register mashing code.

There's no point in pushing and popping the default state; the code
between the two stack operations doesn't alter anything.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>

i965/sf: Drop unnecessary push/pop in do_twoside_color.

None of the assembly emitters called between push and pop actually
change the state.  So, we can drop these.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>

i965: Don't implicitly set predicate default state in brw_CMP.

Previously, brw_CMP with a null destination implicitly set the default
state to make future instructions predicated.  This is messy and
confusing - emitting a CMP that populates the flag register and later
using it to predicate instructions are logically separate.  With the
main compiler, we may even schedule instructions between the CMP and the
user of the flag value.

This patch simplifies brw_CMP to just emit a CMP instruction, and not
mess with predication.  It also updates all necessary callers.  These
mostly fell into two patterns:

1. brw_CMP followed by brw_IF.

   We don't need to do anything special here; brw_IF already sets up
   predication appropriately.

2. brw_CMP followed by a single predicated instruction.

   The old model was to call brw_CMP, emit the next (predicated)
   instruction, then disable predication for any instructions beyond
   that.  Instead, just explicitly set predicate_control on the single
   instruction we want to predicate.  It's no more code, and requires
   less cross-module knowledge.

This drops setting flag_value to 0xff as well, which is a field only
used by the SF compile.  There is only one brw_CMP call in the SF code,
which is in do_twoside_caller, and called at the start of
brw_emit_tri_setup, where flag_value is already 0xff.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>

i965: Drop unnecessary predication default state resets in clip code.

Presumably, this was to reset the default state of predication_control
from brw_CMP.  But brw_CMP only sets that if dst is ARF null, which it
isn't here.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>

i965/sf: Reset flag_value to 0xff before emitting SF subroutines.

When compiling any of the SF program variants, flag_value starts off as
0xff and will be modified when generating code.

brw_emit_anyprim_setup emits several subroutines, saving and restoring
flag_value across each of them.  Since it starts out as 0xff, this is
equivalent to simply setting it to 0xff at the start of each subroutine.

Resetting the value makes more logical sense; each subroutine doesn't
know whether one of the others even executed, much less what it did
to the flag register.

This also lets us to drop the brw_set_predicate_control_flag_value call
from brw_init_compile: predicate is already initialized to
BRW_PREDICATE_NONE by the memset, and the value of flag_value is
irrelevant (as it's only used by the SF compiler).

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>

st/omx/enc: implement restricted b frames pattern

Signed-off-by: Leo Liu <leo.liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

radeon/vce: implement non-referenced frames

Signed-off-by: Leo Liu <leo.liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

vl: add interface for non-referenced frames

Signed-off-by: Leo Liu <leo.liu@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

i965/meta: Store stencil texturing mode

Meta path needs to keep the current texture object's state. Fixes
the following gles3 cts tests on bdw:

framebuffer_blit_functionality_negative_width_blit.test: fail
framebuffer_blit_functionality_all_buffer_blit.test: fail
framebuffer_blit_functionality_negative_height_blit.test: fail
framebuffer_blit_functionality_missing_buffers_blit.test: fail
framebuffer_blit_functionality_negative_dimensions_blit.test: fail
framebuffer_blit_functionality_minifying_blit.test: fail
framebuffer_blit_functionality_magnifying_blit.test: fail

Signed-off-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Cc: "10.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

meta/blit: Add stencil texturing mode save and restore

v2 (Ken): Only restore the mode if it has changed.

Signed-off-by: Topi Pohjolainen <topi.pohjolainen@intel.com>
Cc: "10.2" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

i915g: Fix shader disasm code

This broke when I separated declarations/shader.

i915g: Fallback to sw for npot copies

i915g's npot support is incomplete, so let's not use it for copies.
This fixes a bunch of piglit tests.

i915g: handle more formats in copy

We can handle depth, luminance,... copies by simply replacing the
format with a known format of the same bpp.

nvc0: implement clear_buffer

Provide an accelerated path for ARB_clear_buffer_object

Signed-off-by: Tobias Klausmann <tobias.johannes.klausmann@mni.thm.de>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>

i965: Switch types D->UD when possible to allow compaction.

Number of compacted instructions: 827404 -> 833045 (0.68%)

Reviewed-by: Eric Anholt <eric@anholt.net>

Revert "i965: Don't make instructions with a null dest a barrier to scheduling."

This reverts commit 42a26cb5e441a01d5288b299980f23affaad53fe.

Cc: "10.2" <mesa-stable@lists.freedesktop.org>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=78648

Revert "i965/fs: Simplify interference scan in register coalescing."

This reverts commit 5ff1e446d44bb9d50f84883c7058635cb070e069.

Cc: "10.2" <mesa-stable@lists.freedesktop.org>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=77704

Revert "i965/fs: Give up in interference check if we see a WHILE."

This reverts commit 55de1c035cbca2b7087b3aa21a8c3dfc900a4ad9.

Cc: "10.2" <mesa-stable@lists.freedesktop.org>

Revert "i965/fs: Reduce restrictions on interference in register coalescing."

This reverts commit f770123f58b46459e8dbd27525162ee8ba89f30b.

Cc: "10.2" <mesa-stable@lists.freedesktop.org>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=78692

nvc0: revert mistaken logic to collapse color outputs to the beginning

In commit af38ef907, I added a "fix" to color outputs not being assigned
correctly when sample mask was being output. This was totally wrong --
the color indices (i.e. "si" values) were the ones that were wrong. Undo
that hunk.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Acked-by: Emil Velikov <emil.l.velikov@gmail.com>