i965: Avoid dependency hints on math opcodes

  Putting NoDDClr and NoDDChk dependency control on instruction
sequences that include math opcodes can cause corruption of channels.
Treat math opcodes like send opcodes and suppress dependency hinting.

Signed-off-by: Mike Stroyan <mike@LunarG.com>
Tested-by: Tony Bertapelli <anthony.p.bertapelli@intel.com>
Cc: mesa-stable@lists.freedesktop.org
Reviewed-by: Matt Turner <mattst88@gmail.com>

i965: Expand INTEL_DEBUG to uint64_t.

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

dri: Expand driParseDebugString return value to uint64_t.

Users will downcast if they don't have >32 debug flags.

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

i965/fs: Remove dead_code_eliminate_local().

Subsumed by the new dead_code_eliminate() function. No shader-db
changes.

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

i965/fs: Clear variable from live-set if it's completely overwritten.

One program affected:

instructions in affected programs:     246 -> 244 (-0.81%)

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

i965/fs: Reimplement dead_code_elimination().

total instructions in shared programs: 1653399 -> 1651790 (-0.10%)
instructions in affected programs:     92157 -> 90548 (-1.75%)
GAINED:                                2
LOST:                                  2

Also significantly reduces the number of optimization loop iterations:

total loop iterations in shared programs: 39724 -> 31651 (-20.32%)
loop iterations in affected programs:     21617 -> 13544 (-37.35%)

Including some great pathological cases, like 29 -> 3 in Strike Suit
Zero and 24 -> 3 in Dota2.

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

i965/vec4: Let DCE eliminate dead writes in other basic blocks.

We previously stopped searching for unread writes after encountering
control flow, but we can instead just search backwards until we hit
control flow.

instructions in affected programs:     22854 -> 22194 (-2.89%)

i965/gs: Add dummy source to prepare_channel_masks instruction.

The generator uses its destination as a source implicitly, which breaks
some assumptions in dead code elimination. Giving the instruction a
source allows us to reason about it better.

glsl: Use M_PI_* macros.

Notice our multiple values for M_PI_2, which rounded ...32 up to
...4 and ...5.

i965: Disable Z16 in all APIs.

We originally thought that GL 3.0 required GL_DEPTH_COMPONENT16 to map
exactly to Z16.  However, we misread the specification, thanks in part
to LaTeX reordering the tables in the PDF.

Page 180 of the GL 3.0 specification (glspec30.20080923.pdf) says:
"[...] memory allocation per texture component is assigned by the GL to
match the allocations listed in tables 3.16-3.18 as closely as possible.
[...]

Required Texture Formats
[...]
In addition, implementations are required to support the following sized
internal formats.  Requesting one of these internal formats for any
texture type will allocate exactly the internal component sizes and
types shown for that format in tables 3.16-3.17:"

Notably, however, GL_DEPTH_COMPONENT16 does /not/ appear in table 3.16
or table 3.17.  It appears in table 3.18, where the "exact" rule doesn't
apply, and it falls back to the "closely as possible" rule.

The confusing part is that the ordering of the tables in the PDF is:

Table 3.16 (pages 182-184)
Table 3.18 (bottom of page 184 to top of 185)
Table 3.17 (page 185)

Presumably, people saw table 3.16, then saw the table immediately
following with DEPTH_COMPONENT* formats, and assumed it was 3.17.

Based on a patch by Chia-I Wu, but without the driconf option to force
Z16 to be used.  It's not required, and there's apparently no benefit
to actually using it.

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

i965: Update comments about Z16 being slow.

We've learned a few things since we originally disabled Z16; this attempts
to summarize the issue.  I am no expert on this subject, though, so the
comment may not be totally accurate.

I did some benchmarking on GM45 and Ironlake, and discovered that for
GLBenchmark 2.7 EgyptHD, using Z16 was 3% slower on GM45 (n=15), and
4.5% slower on Ironlake (n=95).  So, we can drop the "on Ivybridge"
aspect of the comment - it's always slower.

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

r600g/radeonsi: Use caching buffer manager for textures as well

Significantly reduces BO allocation / destruction overhead for transfers,
e.g. measurable via x11perf -shm{ge,pu}t* with glamor.

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

i965/gen8: add debug code to show FS disasm with jump locations

Copied from similar code in gen8_vec4_generator.cpp.

Signed-off-by: Jordan Justen <jordan.l.justen@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

ilo: remove GPE state size estimation

Use size defines from genhw.

ilo: remove GPE command size estimation

Use size defines from genhw.

ilo: remove unused headers

Remove intel_*.h.  brw_*.h is still needed by the state dumper and
disassembler.

ilo: use only defines from genhw headers

Stop including classic driver headers in genhw.h, with some formatting fixes.

ilo: scripted conversion to genhw headers

Hopefully my four hundred line sed script is correct.

ilo: add genhw headers

All except genhw.h are generated by https://github.com/olvaffe/envytools/.
intel_chipset.h is deprecated.

ilo: avoid brw_wm_barycentric_interp_mode in compiler

In preparation for genhw.

ilo: add TOY_OPCODE_DO

We used to give BRW_OPCODE_DO a special meaning, while we should have used
TOY_OPCODE_DO.

gtest: Update to 1.7.0.

This patch fixes gtest build errors on Mac OS X 10.9.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=73106
Signed-off-by: Vinson Lee <vlee@freedesktop.org>
Tested-by: Ian Romanick <ian.d.romanick@intel.com>
Acked-by: Ian Romanick <ian.d.romanick@intel.com>

mesa: Consider gl_VertexID and gl_InstanceID active attribs

Fixes piglit's spec/gl-3.2/get-active-attrib-returns-all-inputs.

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

mesa: Extract is_active_attrib() in shaderapi

The rules are about to get a bit more complex to account for
gl_InstanceID and gl_VertexID, which are system values.

Extracting this first avoids introducing duplication.

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

glsl: Fix typo in interface block comment

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>

egl-static: fix build after recent radeon winsys changes

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

mesa: Fix typo in error message

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>

i965: glClearBuffer() should only clear a single buffer.

glClearBuffer() is currently clearing all active draw color buffers (all
buffers that have not been set to GL_NONE when calling glDrawBuffers) instead
of only clearing the one it receives as parameter. Altough brw_clear()
receives a bit mask indicating the color buffers that should be cleared,
this mask is ignored when calling brw_blorp_clear_color().

This was breaking the 'fbo-drawbuffers-none glClearBuffer' piglit test.

The patch provides the bit mask to brw_blorp_clear_color() so it can limit
clearing to the color buffers present in the mask.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76832
Reviewed-by: Eric Anholt <eric@anholt.net>

i965: Add comment to explain the weird-looking shadow compares.

This always looks crazy when I stumble across it, until I remember
what the hardware is doing. Describing it ought to short-circuit
that process next time :)

V2: Fix indents to 6 spaces, not 7.

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

glsl: Ignore loop-too-large heuristic if there's bad variable indexing.

Many shaders use a pattern such as:

for (int i = 0; i < NUM_LIGHTS; i++) {
   ...access a uniform array, or shader input/output array...
}

where NUM_LIGHTS is a small constant (such as 2, 4, or 8).

The expectation is that the compiler will unroll those loops, turning
the array access into constant indexing, which is more efficient, and
which may enable array splitting and other optimizations.

In many cases, our heuristic fails - either there's another tiny nested
loop inside, or the estimated number of instructions is just barely
beyond the threshold.  So, we fail to unroll the loop, leaving the
variable indexing in place.

Drivers which don't support the particular flavor of variable indexing
will call lower_variable_index_to_cond_assign(), which generates piles
and piles of immensely inefficient code.  We'd like to avoid generating
that.

This patch detects unsupported forms of variable-indexing in loops, where
the array index is a loop induction variable.  In that case, it bypasses
the loop-too-large heuristic and forces unrolling.

Improves performance in various microbenchmarks: Gl32PSBump8 by 47%,
Gl32ShMapVsm by 80%, and Gl32ShMapPcf by 27%.  No changes in shader-db.

v2: Check ir->array for being an array or matrix, rather than the
    ir_dereference_array itself.
v3: Fix and expand statistics in commit message.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>

glsl: Rename loop_unroll_count::fail to "nested_loop."

The "fail" flag is set if loop_unroll_count encounters a nested loop;
calling the flag "nested_loop" is a bit clearer.

The original reasoning was that count is inaccurate (too small) if there
are nested loops, as we don't do any sort of analysis on the inner loop.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>

glsl: Pass gl_shader_compiler_optimizations to unroll_loops().

Loop unrolling will need to know a few more options in the future.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>

glsl: Drop do_common_optimization's max_unroll_iterations parameter.

Now that we pass in gl_shader_compiler_options, it makes sense to just
use options->MaxUnrollIterations, rather than passing a separate
parameter.

Half of the invocations already passed options->MaxUnrollIterations,
while the other half passed in a hardcoded value of 32.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>

i965: Use EmitNoIndirect flags in lower_variable_index_to_cond_assign.

This will prevent the two from getting out of sync again.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>

i965: Correct EmitNoIndirect shader compiler option flags.

These were out of sync with the flags used to control
lower_variable_index_to_cond_assign in brw_shader.cpp.

Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>

i965/fs: Reset reg_from when we can't coalesce.

Not setting this would prevented coalescing after a failed attempt if
the sources for both MOVs were the same.

total instructions in shared programs: 1654531 -> 1650224 (-0.26%)
instructions in affected programs:     423167 -> 418860 (-1.02%)
GAINED:                                2
LOST:                                  0

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

i965: Fill in a bunch of gen7/hsw data cache-related disasm.

This gets us disasm of atomic ops.

v2: Fix fallthrough on pre-gen7.  (bug caught by Ilia Mirkin).

Reviewed-by: Matt Turner <mattst88@gmail.com>

i965: Stop setting up a 1:1 "attrib" member in our vertex inputs.

It's just the array index, so we can just go look at the array and see
which element we are.

No significant performance difference (n=140)

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

i965: Skip a bunch of IB BO refcount twiddling.

Improves cairo performance on glamor by 1.64828% +/- 1.04742% (n=65).

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

i965/gen7: Skip repeated NULL depth/stencil state emits.

Improves cairo performance on glamor by 2.87752% +/- 0.966977 (n=57).

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

docs: Fix ubo indexing description

Ian points out that this being unrestricted was an oversight in the
spec, and is corrected in GLSL4.40.

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>

draw: remove unused 'start' variable in draw_stats_clipper_primitives()

It was computed, but never actually used.

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

glsl: Try vectorizing when seeing a repeated assignment to a channel.

When considering assignment expressions like:

    v.x += u.x;
    v.x += u.x;

the vectorizer would incorrectly keep going, attempting to find more
instructions to vectorize.  It would overwrite the saved assignment
to point at the second one, and increment channels a second time,
resulting in try_vectorize thinking the expression was a vec2 instead of
a float.

Instead, if we see a repeated assignment to a channel, just try to
vectorize everything we've found so far.  This clears the saved state
so it will start over.

Fixes Piglit's repeated-channel-assignments.vert.

Cc: "10.1" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Matt Turner <mattst88@gmail.com>

glsl: Propagate explicit binding information from the AST all the way to the linker

Information about the binding was not being properly communicated from
the front-end compiler to the linker.  As a result, the linker never
knew that any UBOs had explicit bindings!

Fixes the piglit test arb_shading_language_420pack-binding-layout.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76323
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Tested-by: github@socker.lepus.uberspace.de [v0]
Cc: "10.1" <mesa-stable@lists.freedesktop.org>
Cc: github@socker.lepus.uberspace.de

linker: Set binding for all elements of UBO array

Previously, a UBO like

    layout(binding=2) uniform U {
        ...
    } my_constants[4];

wouldn't get any bindings set.  The code would try to set the binding of
U, but that would fail.  It should instead set the bindings for U[0],
U[1], ...

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76323
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: "10.1" <mesa-stable@lists.freedesktop.org>
Cc: github@socker.lepus.uberspace.de

linker: Set block bindings based on UniformBlocks rather than UniformStorage

For blocks, gl_shader_program::UniformStorage isn't very useful.  The
names stored there are the names of the elements of the block, so
finding blocks with an instance name is hard.  There is also only one
entry in ::UniformStorage for each element of a block array, and that is
a deal breaker.

Using ::UniformBlocks is what _mesa_GetUniformBlockIndex does.  I
contemplated sharing code between set_block_binding and
_mesa_GetUniformBlockIndex, but building the stand-alone compiler and
the unit tests make this hard.  I plan to return to this effort shortly.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76323
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: "10.1" <mesa-stable@lists.freedesktop.org>
Cc: github@socker.lepus.uberspace.de

linker: Clean up "unused parameter" warnings

../../src/glsl/link_uniform_initializers.cpp:87:1: warning: unused parameter 'mem_ctx' [-Wunused-parameter]
../../src/glsl/link_uniform_initializers.cpp:87:1: warning: unused parameter 'type' [-Wunused-parameter]
../../src/glsl/link_uniform_initializers.cpp:127:1: warning: unused parameter 'mem_ctx' [-Wunused-parameter]
../../src/glsl/link_uniform_initializers.cpp:127:1: warning: unused parameter 'type' [-Wunused-parameter]

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76323
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: "10.1" <mesa-stable@lists.freedesktop.org>
Cc: github@socker.lepus.uberspace.de

linker: Fold set_uniform_binding into call site

In the next patch, we'll see that using
gl_shader_program::UniformStorage is not correct for uniform blocks.
That means we can't use ::UniformStorage to select between the sampler
path and the block path.  Instead we want to just use the type of the
variable.  That's never passed to set_uniform_binding, and it's easier
to just remove the function (especially for later patches in the series)
than to add another parameter.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76323
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: "10.1" <mesa-stable@lists.freedesktop.org>
Cc: github@socker.lepus.uberspace.de

linker: Various trivial clean-ups in set_sampler_binding

- Remove the spurious block left from the previous commit and re-indent.

- Constify elements.

- Make the spec reference in the code look like other spec references in
  the compiler.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76323
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: "10.1" <mesa-stable@lists.freedesktop.org>
Cc: github@socker.lepus.uberspace.de

linker: Split set_uniform_binding into separate functions for blocks and samplers

The two code paths are quite different, and there are some problems in
the handling of uniform blocks.  Future changes will cause these paths
to diverge further.  Ultimately, selecting between the two functions
will happen at the set_uniform_binding call site, and
set_uniform_binding will be deleted.

NOTE: This patch just moves code around.

Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=76323
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Cc: "10.1" <mesa-stable@lists.freedesktop.org>
Cc: github@socker.lepus.uberspace.de

softpipe: fix shadow sampling

And remove nonsensical approximation of linear interpolation behavior
for shadow samplers.

Reviewed-by: Brian Paul <brianp@vmware.com>
Tested-by: Brian Paul <brianp@vmware.com>

softpipe: add PIPE_CAP_MIN/MAX_TEXTURE_GATHER_OFFSET query cases

To silence compiler warnings.

Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>

mesa: use _mesa_get_srgb_format_linear() in sRGB texstore functions

Instead of switch statements.

Reviewed-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>

swrast: use macros to initialize texfetch_funcs[] table

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

swrast: fix more fetch_texel function names

These were missed/typo'd in the previous patch series:
s/R8G8B8A/R8G8B8A8/
s/rgba_16/RGBA_UNORM16/
s/rgba_uint/RGBA_UINT/
s/rgba_int/RGBA_SINT/

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

egl-static: Fix missing radeon_surface.h includes.

Fixes

  fatal error: radeon_surface.h: No such file or directory

when libdrm is not present, or non-Linux OSes.

Trivial.

gallium/radeon: fix missing winsys include in pipe-loader.

The commit 3b0b44f7def0acb4f7a7aef086c0bece321418a6 introduced a build
error:

error: dereferencing pointer to incomplete type

This patch fixes this issue in all the affected files.

Reviewed-by: Tom Stellard <thomas.stellard@amd.com>

st/omx/enc: separate input buffer private and task structure

Keep tasks as linked list, this way we can associate
more than one encoding task with each buffer.

Signed-off-by: Christian König <christian.koenig@amd.com>

radeon/vce: implement B-frame support

Signed-off-by: Slava Grigorev <slava.grigorev@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>

radeon/vce: add proper CPB backtrack

Remember what frames we encoded at which position.

Signed-off-by: Christian König <christian.koenig@amd.com>

vl: add interface for H264 B-frame encoding

Signed-off-by: Christian König <christian.koenig@amd.com>

radeon/vce: remove RVCE_NUM_CPB_EXTRA_FRAMES

Doesn't seems to be needed any more.

Signed-off-by: Christian König <christian.koenig@amd.com>

docs/relnotes: Fix consistency, add i965 to ARB_buffer_storage.

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>

i965: Fix missing _NEW_SCISSOR in Broadwell SF_CLIP_VIEWPORT state.

The _Xmin/_Xmax/_Ymin/_Ymax values need to be guarded by _NEW_SCISSOR.

Fixes Piglit's scissor-many, and rendering in GNOME Shell.
Hopefully fixes similar issues with Unity and ChromeOS.

Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=75879
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Eric Anholt <eric@anholt.net>
Tested-by: James Ausmus <james.ausmus@intel.com>
Tested-by: Timo Aaltonen <tjaalton@ubuntu.com>

mesa/st: set min/max texture gather offset to driver-reported value

It was always getting set to -8/7 unconditionally.  Use the
driver-reported value instead.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

gallium: add a way to query min/max texture gather offsets

Defaults to providing the same offsets as MIN/MAX_TEXEL_OFFSET. For
nvc0, the offset can be -32/31.

Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>

configure.ac: require libdrm_radeon 2.4.53

We need latest radeon_drm.h.

winsys/radeon: fix a race condition in initialization of radeon_winsys::screen

Create the screen in the winsys while the mutex is locked.
This also results in a nice code cleanup!

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

winsys/radeon: fix a race condition between winsys_create and winsys_destroy

This also hides the reference count from drivers.

v2: update the reference count while the mutex is locked in winsys_create

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

winsys/radeon: fix a race condition between 2 calls to radeon_winsys_create

This fixes random crashes of: piglit/glx-multithread-shader-compile.

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

winsys/radeon: remove unused radeon_info variables, move backend_map

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

winsys/radeon: unify radeon_bo::flink and radeon_bo::name

Both contained the GEM flink name.

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

winsys/radeon: remove definitions already present in radeon_drm.h

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

winsys/radeon: handle squared micro tiling from GEM_GET_TILING

Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>

gallium/u_gen_mipmap: rewrite using pipe->blit (v2)

This replaces u_gen_mipmap with an extremely simple implementation based
on pipe->blit. st/mesa is also cleaned up.

Pros:
- less code
- correct mipmap generation for NPOT 3D textures (u_blitter uses a better
  formula)
- queries are not affected by mipmap generation if drivers disable them

v2: add "first_layer", "last_layer" parameters, drop "face"
v2.1: add format
v2.2: document the format parameter

st/mesa: properly implement MapTextureImage with multiple mapped slices (v2)

This is needed by _mesa_generate_mipmap.

This adds an array of pipe_transfers to st_texture_image. Each transfer is
for mapping a single layer.

v2: allocate the array of transfers on demand

mesa: remove the MALLOC, CALLOC and FREE macros

No longer used anywhere.  These also caused trouble in the Gallium
state tracker code where we include both core Mesa and Gallium util
headers (and the macros were defined differently in each world.)
Removing these macros should help avoid macro mix-ups in the future.

Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

xlib: s/FREE/free/

Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

mesa: s/FREE/free/ in vdpau code

Reviewed-by: Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Christian König <christian.koenig@amd.com>

mesa: s/FREE/free/ in _mesa_free_errors_data()

Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

mesa: use malloc/free instead of MALLOC/FREE in attrib stack code

We moved away from MALLOC/FREE in the rest of core Mesa a while ago.

Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>

st/mesa: fix sampler_view REALLOC/FREE macro mix-up

We were using REALLOC() from u_memory.h but FREE() from imports.h.
This mismatch caused us to trash the heap on Windows after we
deleted a texture object.

This fixes a regression from commit 6c59be7776e4d.

Reviewed-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Jakob Bornecrantz <jakob@vmware.com>

docs: Expand ARB_gpu_shader5 to describe status of individual features

This extension is a huge grab-bag of "stuff that's in DX11". Break it
apart to make it clear what still needs to be done.

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Matt Turner <mattst88@gmail.com>

docs: Mark off ARB_texture_view and add to release notes for 10.2.

V4: Don't claim Gen8 yet.

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>

i965: Enable ARB_texture_view on Gen7

V4: Don't enable this for Gen8 yet -- that still needs wired up.

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>

i965: Account for view parameters in blit CTSI path

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>

i965: Account for MinLayer/MinLevel in blorp CTSI path

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>

i965: Account for view parameters in fast depth clears

V2: - No need for layer_multiplier; multisampled depth surfaces are IMS.
    - Remove unused num_layers.

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>

i965/blorp: Account for nonzero MinLayer in layered clears.

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>

i965/blorp: Use irb->layer_count in clear

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>

i965: Add layer_count to intel_renderbuffer

This is the effective layer count, for clears etc. This differs from the
depth of the miptree level when views are involved.

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>

i965: Pull out layer_multiplier in intel_update_renderbuffer_wrapper

We're about to need this in another place.

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>

i965: Add `layered` parameter to intel_update_renderbuffer_wrapper

We're about to need this so we can determine the layer count of the
wrapper.

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>

i965: Adjust renderbuffer wrapper to account for MinLevel/MinLayer

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>

i965: Enable texture upload fast path with MinLevel

We'll still avoid MinLayer here since the fast path doesn't understand
arrays at all, but it's straightforward to do levels.

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>

i965: Account for MinLevel in texture upload fast path

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>

i965: Adjust map/unmap code for MinLevel/MinLayer

This allows core mesa's TexSubImage paths etc to work correctly
with views which have nonzero MinLevel or MinLayer.

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>

i965: Don't try to use fast upload path for nontrivial views

This will eventually be relaxed, but we'll get the fallback path
working first.

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>

i965: Adjust surface_state emission to account for view parameters

V4: Comment style, remove magic shift.

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>

i965: Add _Format to intel_texobj.

This is the actual mesa_format to use. In non-view cases this is always
the same as the mt's format.

V4: Comment style

Signed-off-by: Chris Forbes <chrisf@ijw.co.nz>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Acked-by: Eric Anholt <eric@anholt.net>