| Age | Commit message (Collapse) | Author |
|---|
|
random output is bad
|
|
|
|
It is possible to bind texture images of an incomplete mipmapped texture.
Software fallbacks in this case incorrectly tried to mmap the entire texture.
Additionally, add span functions for 1555 and 4444 formats.
This fixes crashes in piglit's fbo-readpixels test; unfortunately, the test
itself still fails - this needs to be investigated.
Signed-off-by: Nicolai Hähnle <nhaehnle@gmail.com>
|
|
This updates some of the clear code from Intel gives a 5x clearspd perf
for me here. played openarena also, not sure if the viewport changes
broke anything,
|
|
it might still be misrendering not sure
|
|
fgl_glxgears -fbo runs, though the gears don't look right yet.
|
|
|
|
make sure to turn off when no texture is used in hw
|
|
|
|
|
|
When a function parameter is const-qualified we can avoid making a copy
of the actual parameter (we basically do a search/replace when inlining).
This is now done for array element params too, resulting in better code
(fewer MOV instructions).
We should allow some other types of function arguments here but let's be
conservative for the moment.
|
|
The two indexes were mixed up when accessing a row of a matrix in an array
of matrices.
|
|
two)
The semantics are a little different for shaders vs. fixed-function when
trying to use an incomplete texture. The fallback texture returning
(0,0,0,1) should only be used with shaders. For fixed function, the texture
unit is truly disabled/ignored.
Fixes glean fbo test regression.
(cherry picked from commit 01e16befd4809e3e93ee7855a4a5d7df9fe010d9)
(cherry picked from commit 51325f44d1e7e62b47795a79f8038d10dc5bc30b)
[anholt: squashed these two together from master, skipping the mess in between]
|
|
|
|
Also fix one case where a 32 bit depth value was incorrectly converted to a
combined depth/stencil value.
|
|
Such visuals are subject to automatic compositing in the X server, so DRI1
can't render to them properly.
|
|
Fixes potential crash when SwapBuffers is called but there's no back buffer.
|
|
fix the cases when fog coord/front face/point coord are used in the same
shader.
|
|
Ensure no other thread is accessing a framebuffer when it is being destroyed by
acquiring both the global and per-framebuffer mutexes. Normal access only
needs the global lock to walk the linked list and acquire the per-framebuffer
mutex.
|
|
Fixes wglthreads -- the 2nd thread MakeCurrent call was trying to flush
the first thread context while still in use.
|
|
According to
http://blogs.msdn.com/oldnewthing/archive/2008/01/15/7113860.aspx
WM_SIZE is generated from WM_WINDOWPOSCHANGED by DefWindowProc so it
can be masked out by the application.
Also there were some weird bogus WM_SIZE 0x0 messages when starting
sharedtex_mt which we don't get like this.
|
|
|
|
|
|
Seriously.
|
|
These will come back in someday, when we can properly use them.
|
|
They have to cross into each other's registers.
|
|
(cherry picked from commit 88c01a15da5639dd68a6a0133724994cb66f1316)
|
|
As reported and initially tested by MrCooper.
|
|
|
|
Stride == 0 means that we value for first vertex should be copied to every other vertices (e.g. constant color).
This fixes glean/vertProg1 and sauerbraten with enabled shaders.
|
|
|
|
|
|
Even after fixing bugs in this code, it doesn't make me feel any cleaner.
Fixes piglit stencil-drawpixels.
|
|
|
|
This state flag has been unused since the ffvertex_prog move to core.
|
|
Otherwise, a pointer greater than the size would underflow and give a large
maximum element.
Reviewed-by: Brian Paul <brianp@vmware.com> (previous version)
|
|
the driver used to overwrite grf0 then use implicit move by send instruction
to move contents of grf0 to mrf1. However, we must not overwrite grf0 since
it's still used later for fb write.
Instead, do the move directly do mrf1 (we could use implicit move from another
grf reg to mrf1 but since we need a mov to encode the data anyway it doesn't
seem to make sense).
I think the dp_READ/WRITE_16 functions may suffer from the same issue.
While here also remove unnecessary msg_reg_nr parameter from the dataport
functions since always message register 1 is used.
|
|
Thanks to branching, the state of c->current_const[i].index at the point
of emitting constant loads for this instruction may not match the actual
constant currently loaded in the reg at runtime. Fixes a regression in my
GLSL program for idr's class since b58b3a786aa38dcc9d72144c2cc691151e46e3d5.
|
|
|
|
|
|
plus it saves us a cacheline in the cso
|
|
reported by clang static analyzer
|
|
reported by clang static analyzer
|
|
found by the clang static analyzer
|
|
|
|
The problem is if we find out later we don't have any cmdbuf space but
we've already written the arrays to the DMA buffer object, we end up
emitting the current cmdbuf which has references to the current DMA object
we then send that to the hw and we can't reference the arrays we just emitted
to the old DMA buffer. things go bad, crash boom.
This can probably be tuned further + swtcl probably needs some fixes
|
|
|
|
|
|
This moves a big chunk of the space checking code into libdrm so
it can be shared by the DDX.
|
|
Fixes flush during validation, in case a buffer is double-validated.
|
