We were returning the most recently freed BO, without checking if it
was idle yet. This meant that we generally stalled immediately on the
previous frame when generating a new one. Instead, allocate new BOs
when the *oldest* BO is still busy, so that the cache scales with how
much is needed to keep some frames outstanding, as originally
intended.
Note that if you don't have some throttling happening, this means that
you can accidentally run the system out of memory. The kernel is now
applying some throttling on all execs, to hopefully avoid this.
AFAICT, there is no real way to make sure a send message with EOT is properly
ignored from compact, nor can I see a way to actually encode EOT while
compacting. Before the single send optimization we'd always bail because we hit
the is_immediate && !is_compactable_immediate case. However, with single send,
is_immediate is not true, and so we end up trying to compact the un-compactible.
Without this, any compacting single send instruction will hang because the EOT
isn't there. I am not sure how I didn't hit this when I originally enabled the
optimization. I didn't check if some surrounding code changed.
I know Neil and Matt were both looking into this. I did a quick search and
didn't see any patches out there to handle this. Please ignore if this has
already been sent by someone. (Direct me to it and I will review it).
Reported-by: Neil Roberts <neil@linux.intel.com>
Reported-by: Mark Janes <mark.a.janes@intel.com>
Tested-by: Mark Janes <mark.a.janes@intel.com>
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
Reviewed-by: Matt Turner <mattst88@gmail.com>
Pure integer formats cannot be sampled with linear tex / mip filters. In GL
such a setup would make the texture incomplete.
We shouldn't rely on the state tracker though to filter that out, just return
all zeros instead of dying in the lerp.
Reviewed-by: Jose Fonseca <jfonseca@vmware.com>
gallivm now depends on it. And depending on particular LLVM version /
configure options, the build can fail without this change due to
undefined reference to `LLVM*Disasm*' symbols.
Trivial.
It doesn't do everything we want. In particular it doesn't allow to
detect jumps or return opcodes. Currently we detect the x86's RET
opcode.
Even though it's worse for LLVM 3.3, it's an improvement for LLVM 3.7,
which was totally busted.
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Roland pointed out my previous attempt was lacking, so I enhanced the
texwrap piglit test, and tested them. This fixes the offset calculations
in a number of areas by adding the offset first, it also fixes the fastpaths,
which I forgot to address in the previous commit.
v2: try and avoid divides in most paths, the repeat mirror path
really was ugly no matter which way I went, so I left it having
the divide.
Also fix the gather lod calculation bug.
Reviewed-by: Roland Scheidegger <sroland@vmware.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Whether or not to use NIR is now equivalent to brw->scalar_vs. We can
simplify the logic and make it far less confusing.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Now that everything is running through NIR, this is all dead.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
Now that everything is running through NIR, this is all dead.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
This is using multiple inheritance in C++. However, ir_visitor is really
just an interface with no data so it shouldn't be so bad.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
The backend_shader class really is a representation of a shader. The fact
that it inherits from ir_visitor is somewhat immaterial.
Reviewed-by: Matt Turner <mattst88@gmail.com>
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
And core profile only.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Fredrik Höglund <fredrik@kde.org>
Cc: "10.6" <mesa-stable@lists.freedesktop.org>
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Suggested-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "10.6" <mesa-stable@lists.freedesktop.org>
The next patch will add a test for compatibility profile dispatch, and
it seems to make more sense to share the lists.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Cc: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "10.6" <mesa-stable@lists.freedesktop.org>
GL_ARB_vertex_attrib_64bit is exclusive to core profile, and none of the
other functions added by the extension are advertised in other profiles.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: "10.6" <mesa-stable@lists.freedesktop.org>
Currently on the functions that are exclusive to core-profile are
implemented. The remainder continue to live in the XML. Additional
functions can be moved later.
The functions for GL_ARB_draw_indirect and GL_ARB_multi_draw_indirect
are put in the dispatch table inside the VBO module, so they do not need
to be moved over.
The diff of src/mesa/main/api_exec.c before and after this patch is as
expected. All of the functions listed in apiexec.py moved out of a 'if
(_mesa_is_desktop(ctx))' block into a new 'if (ctx->API ==
API_OPENGL_CORE)' block.
v2: Remove stray shebang line in apiexec.py. Suggested by Ilia.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Ilia Mirkin <imirkin@alum.mit.edu>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Dylan Baker <baker.dylan.c@gmail.com>
Cc: "10.6" <mesa-stable@lists.freedesktop.org>
The extension on which this depends will always be enabled in core
profile, and the extension bit is about to be removed.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Fredrik Höglund <fredrik@kde.org>
Cc: "10.6" <mesa-stable@lists.freedesktop.org>
The string is only applied when the context is API_OPENGLES2.
The bulk of the change is to prevent overriding the context to
API_OPENGL_CORE based on the requested version. If the context is
API_OPENGL_ES2, don't change it.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
Remove _mesa_get_gl_version_override. We don't need two functions that
do basically the same thing. This change seemed easier (esp. with the
next patch) than going the other way.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
This is a bit of a hack for now. Several of the extensions required for
OpenGL ES 3.1 have no support, at all, in Mesa. However, with this
patch and a patch to allow MESA_GL_VERSION_OVERRIDE to work with ES
contexts, people can begin testing the ES "version" of the functionality
that is supported.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
A couple functions are missing because there are no implementations of
them yet. These are:
glFramebufferParameteri (from GL_ARB_framebuffer_no_attachments)
glGetFramebufferParameteriv (from GL_ARB_framebuffer_no_attachments)
glMemoryBarrierByRegion
v2: Rebase on updated dispatch_sanity.cpp test.
v3: Add support for glDraw{Arrays,Elements}Indirect in vbo_exec_array.c.
The updated dispatch_sanity.cpp test discovered this omission.
v4: Rebase on glapi changes.
Signed-off-by: Ian Romanick <ian.d.romanick@intel.com>
Reviewed-by: Tapani Pälli <tapani.palli@intel.com>
If the multiplication's result is unused, except by a conditional_mod,
the destination will be null. Since the final instruction in the lowered
sequence is a partial-write, we can't put the conditional mod on it and
we have to store the full result to a register and do a MOV with a
conditional mod.
Cc: "10.6" <mesa-stable@lists.freedesktop.org>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=90580
Reviewed-by: Kenneth Graunke <kenneth@whitecape.org>
When we compute the output swizzle we want to consider the number of
components in the add operation. So far we were using the writemask
of the multiplication for this instead, which is not correct.
Reviewed-by: Jason Ekstrand <jason.ekstrand@intel.com>
