The layer coming from GS needs to be clamped (not sure if that's actually
the correct error behavior but we need something) as the number can be higher
than the amount of layers in the fb. However, this code was using the layer
calculation from the scene, and this was actually calculated in
lp_scene_begin_rasterization() hence too late (so setup was using the value
from the _previous_ scene or just zero if it was the first scene).
Since the value is used in both rasterization and setup, move calculation up
to lp_scene_begin_binning() though it's a bit more inconvenient to calculate
there. (Theoretically could move _all_ code which was in
lp_scene_begin_rasterization() to there, because ever since we got rid of
swizzled render/depth buffers our "map" functions preparing the fb data for
render don't actually change the data in there at all, but it feels like
it would be a hack.)
v2: improve comments
Reviewed-by: Brian Paul <brianp@vmware.com>
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
This prevents unnecessary (and wrong) register allocation in the
scheduler for preloaded values in fixed registers.
Fixes interpolation-mixed.shader_test on rv770
(and probably on all other pre-evergreen chips).
Signed-off-by: Vadim Girlin <vadimgirlin@gmail.com>
Tested-by: Andreas Boll <andreas.boll.dev@gmail.com>
This CAP will determine whether ARB_framebuffer_object can be enabled.
The nv30 driver does not allow mixing swizzled and linear zsbuf/cbuf
textures.
Signed-off-by: Ilia Mirkin <imirkin@alum.mit.edu>
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
The encoding of constant, relative, and relative-const src registers is
a bit more complex than originally thought, which gives an extra bit to
encode const reg # at expense of taking a bit from relative offset.
In most cases a3xx seems to actually use a scheme whereby it can encode
an extra bit for const register. You have three possible encodings in
thirteen bits:
register: (11 bits for N.c)
00........... rN.c
relative: (10 bits for N)
010.......... r<a0.x + N>
011.......... c<a0.x + N>
const: (12 bits for N.c)
1............ cN.c
Which means we can deal w/ more consts than previously thought.
Signed-off-by: Rob Clark <robclark@freedesktop.org>
The gallium vbuf module, which we've been using for some time now, takes
care of uploading user-space vertex/index data into real buffers. The
upload code in the svga driver was unused.
Reviewed-by: José Fonseca <jfonseca@vmware.com>
The new function replaces four old functions: set_fragment/vertex/
geometry/compute_sampler_views().
Note: at this time, it's expected that the 'start' parameter will
always be zero.
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Tested-by: Emil Velikov <emil.l.velikov@gmail.com>
rc_find_free_temporary_list() returns signed integer
(in case of lack of free temporary registers returns -1),
so new_index in radeon_rename_regs() should be signed.
https://bugs.freedesktop.org/show_bug.cgi?id=54867
Signed-off-by: Marek Olšák <marek.olsak@amd.com>
We can't perform DCE using the liveness pass between GVN and GCM
because it relies on the correct schedule, but GVN doesn't care about
preserving correctness - it's rescheduled later by GCM.
This patch makes dce_cleanup pass perform simple DCE
between GVN and GCM instead of relying on liveness pass.
Fixes https://bugs.freedesktop.org/show_bug.cgi?id=70088
Signed-off-by: Vadim Girlin <vadimgirlin@gmail.com>
This reverts commit 94d05bf87a as it has a
few problems:
- it breaks windows builds becuase env[LLVM_CXXFLAGS] is never set there
- it is merging not only rtti, but the whole cxxflags (defines etc)
which has proven to be a source of troubles (breaks debugging etc.)
LLVM 3.3 does not know about CIK processors, and the codes paths for SI
and CIK are the same.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Cc: "9.2" <mesa-stable@lists.freedesktop.org>
* The rtti fix actually dug up a bug in the scons build scripts.
* Autotools took the LLVM cpp and cxx flags, while scons only took
the cpp flags.
* This grabs the cxx flags and applies them where needed. We may
want to make the same change for the llvm cpp flags in scons.
* The only linux platform I can find with LLVM no-rtti is Ubuntu.
* Fixes bug #70471
Tested-by: Vinson Lee <vlee@freedesktop.org>
Textures that likely reside in VRAM, are mapped for reading and
don't require direct mapping should be staged into GTT, to avoid bad
performance. This fixes readback performance of VDPAU surfaces.
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
This new bind flag forces linear storage, but does not have other
side effects like R600_RESOURCE_FLAG_TRANSFER.
Reviewed-by: Christian König <christian.koenig@amd.com>
Currently it's hardcoded in the shader, so every change requires
compilation of the shader variant, killing the performance
in Serious Sam 3 and probably other apps.
This patch passes alpha_ref in the user sgpr and removes it from
the shader key.
Signed-off-by: Vadim Girlin <vadimgirlin@gmail.com>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
This fixes the issue when dst and src is the same reg and operation on one
channel overwrites the source for other channels, e.g.:
UMUL TEMP[2].xyz, TEMP[0].xyzz, TEMP[2].xxxx
In this example the result of the operation on channel x is written in
TEMP[2].x and then used as a second source operand for channels y and z
instead of original value in TEMP[2].x.
This patch stores the results in temp reg and moves them to
dst after performing operation on all channels.
Fixes https://bugs.freedesktop.org/show_bug.cgi?id=70327
Signed-off-by: Vadim Girlin <vadimgirlin@gmail.com>
With code dump enabled LLVM may generate disassembly during compilation.
Show this disassembly when available and prefer it to SI bytecode dump.
Reviewed-by: Tom Stellard <thomas.stellard@amd.com>
Signed-off-by: Jay Cornwall <jay@jcornwall.me>
Fix coord wrapping (and face selection too) in case of edges.
Unfortunately, the coord wrapping is way more complicated than what
the code did, as it depends on the face and the direction where the
texel falls off the face (the logic needed to get this right in fact
seems utterly ridiculous).
Also fix a bug in (y direction under/overflow) face selection.
And get rid of complicated cube corner handling. Just like edge case,
the coord wrapping was wrong and it seems very difficult to fix.
I'm near certain it can't always work anyway (though ordinary seamless
filtering on edge has actually a similar problem but not as severe)
because we don't have per-pixel face, hence could have multiple corner
texels which would make it very difficult to average the remaining texels
correctly. Hence simply pick a texel which would only have fallen off one
edge but not both instead, which is not quite accurate but actually I think
should be enough to meet OpenGL (but not d3d10) requirements.
v2: small fixes suggested by Brian, add some comments.
Reviewed-by: Brian Paul <brianp@vmware.com>
The previous limit of of 128*1024 was reported to cause frequent recompiles
in some apps due to shader variant thrashing on IRC in some apps leading
to noticeable lags.
Note that the LP_MAX_SHADER_VARIANTS limit (1024) was more or less impossible
to reach, since even simple fragment shaders without texturing (glxgears) used
more than twice than 128 instructions, hence the instruction limit would have
always been reached first (excluding things like trivial shaders not writing
color). Even with the new limit it is VERY likely the instruction limit is hit
first.
Should help with such lags due to recompiles (though other shader types have
their own limits, LP_MAX_SETUP_VARIANTS and DRAW_MAX_SHADER_VARIANTS, in
particular the latter seems a bit small (128)).
Reviewed-by: Brian Paul <brianp@vmware.com>
We should be able to safely set the framebuffer state without a
fragment shader bound. bind_ps_state will take care of updating the
necessary state bits later.
v2: check in update_db_shader_control
Unless the polygon fill mode is different from PIPE_POLYGON_MODE_FILL,
so checking the the polygon mode is sufficient.
Testing done: no regression in polygon-mode-offset
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
As we're moving towards expanding the number of subpixel
bits and the width of the variables used in the computations
we need to make this code a bit more centralized.
Signed-off-by: Zack Rusin <zackr@vmware.com>
Reviewed-by: José Fonseca <jfonseca@vmware.com>
Reviewed-by: Brian Paul <brianp@vmware.com>
It doesn't work (decodes to garbage) with most videos on UVD 3.0. Worse
yet, it often results in random memory corruption or GPU hangs. Rumor
has it only the newest UVD hardware could do it anyway.
Reviewed-by: Christian König <christian.koenig@amd.com>
