e13ebc57e1
renderer: move away from pipe bind flag definitions.
10 years ago
dfd7749ddb
renderer: refactor iov bounds checking
...
take the user supplied stride and layer stride into a/c
when calculating the amount of memory required to read/write
from the user supplied iovec.
10 years ago
20e86f1771
renderer: validate layer stride isn't set for some texture types
...
Some texture types have no layers, so a non-zero layer stride makes
no sense.
10 years ago
1fe0822ff4
renderer: fix bug transfer inline write decoding
...
cleanup a little, but also read layer stride from correct place
found writing validation tests.
10 years ago
6495f870b4
renderer: block resource creation for mipmaps
...
Some scenarios for mipmap creation are illegal, so block them.
10 years ago
5e975c6a2e
renderer: switch hw contexts outside the transfer iov
...
this fixes a regression since the transfer inline rewrite
10 years ago
b6971e08b8
shader: ignore empty color mask in emit_color_select()
...
Fixes piglit interpolation-flat-other-flat-none.shader_test
10 years ago
dd954de6e0
shader: fix uniform buffer object size
...
Size of array is Range.Last+1, like other declarations
Fixes piglit spec/glsl-1.40/uniform_buffer/fs-basic
10 years ago
dbd90fb682
Rename Elements ARRAY_SIZE
...
Similar to upstream mesa change
10 years ago
cd69deebad
Remove INLINE, use inline instead
...
Similar to MESA recent change
10 years ago
e4880f6bd3
shader: fix UCMP translation
10 years ago
ee78525252
shader: promote type for comparison with same type
...
Fixes some piglit test.
10 years ago
aea7785887
renderer: handle resource inline writes using common code
...
inline resource writes should use common code for transfers
10 years ago
cccbc3b5e4
renderer: return values from submit_cmd/decode block
...
makes easier to write unit tests.
10 years ago
bc975ef808
virgl/transfer: take transfer offset into a/c
...
This checks the offset can fit inside the iovec
10 years ago
e7afcd62cf
build: fix make distcheck
10 years ago
5783d84b30
egl: overhaul the fourcc retrieval function
...
We really need a new mesa interface that works properly for all this.
10 years ago
58ef21b0b8
vrend: accept empty vertex elements
...
Vertex elements may have a count of 0, this is an acceptable value.
Fix piglit spec/glsl-1.50/execution/geometry/primitive-types
gl_triangle_fan test.
10 years ago
e094ca79f9
vrend: return EINVAL for unknown object type
10 years ago
6948e60c74
renderer: fix check for streamout target object insert
10 years ago
135dc05084
gallium: drop unused header
10 years ago
e440d4b48c
gallium: drop some debug code
10 years ago
51d84f851c
gallium: drop large fast math tables
...
no need for these in virgl
10 years ago
55d226385b
gallium: drop tgsi execute code.
...
don't need this in virgl
10 years ago
d01f462c53
gallium: drop unneeded format code
...
we don't need the accessor methods in virgl so far, only
the description tables, so drop all the extra stuff for now.
this introduces the python from mesa to generate our cut down
table
10 years ago
3a04ee4a16
renderer: drop pointless emit depth range
10 years ago
12bf136ae7
renderer: drop hw_emit_blend_color
...
no longer required since ctx
10 years ago
1b3960a150
renderer: some minor cleanups
10 years ago
85bfbb4443
query: fix glean/occluQry test
...
It is bad to sum up all the subsequent Begin/End values of a query,
since glBeginQuery should always reset the counter.
[airlied: I'm not 100% sure we'll never need this, but I think
sub contexts work around the problem I was initially seeing.]
10 years ago
2258a66f61
renderer: drop ctx query handling code
...
also drop active_hw flag
10 years ago
6fc3a2c9b5
renderer: link shaders and programs so things can be freed.
...
We need to link the shaders to the programs to ensure things
can be freed properly. This fixes a problem with glean
texCombine also.
10 years ago
5ddb7c3d0a
renderer: split destroy program out from free programs
...
Split out from Marc-André's larger patchset.
10 years ago
e2b6de9a89
renderer: drop reference to any shaders in sub context destroy.
...
We were holding onto a reference for these, causing a leak.
10 years ago
9537d5a6dc
renderer: only allocate ubo_locs when required
...
This fixes a valgrind warning.
10 years ago
a0074a2f1b
renderer: free timer query properly
...
Reported-by: Marc-André Lureau <marcandre.lureau@gmail.com>
10 years ago
040df0b86c
renderer: use names instead of numbers for shader indexes
10 years ago
4a6383f4ce
build: make distcheck work
...
Mostly taken from Marc's branch, just didn't change
as much stuff around
10 years ago
639f0a6002
gallium/imports: drop reference and bmp debugging
...
this makes building with -DDEBUG work.
10 years ago
fa71f3761d
renderer: free vertex element data
...
this is duplicated by the renderer so free the decoder data.
found with rendering test
10 years ago
abff45cff0
transfer: check iov size on all transfers
10 years ago
8954f84727
tgsi: try and handle overflowing shaders.
10 years ago
d404b2279b
tgsi: add option to dump floats as hex values
...
This adds support to the parser to accept hex values as floats,
and then adds support to the dumper to allow the user to select
to dump float as 32-bit hex numbers.
[pulled from virgl driver repo]
Signed-off-by: Dave Airlie <airlied@redhat.com>
10 years ago
9fcf1e0ec0
build-sys: add git.mk
10 years ago
d9953a56c9
build-sys: add coverage
10 years ago
368b62fd43
build-sys: add missing util/u_debug_describe.c
10 years ago
b5be63c8dc
Fix cursor resource creation
10 years ago
83d7fbb0d7
renderer: overhaul transfer code a bit
...
This merges the error/bounds checking on the transfer
code, but keeps the same API, it also uses a struct
to pass through the transfer info.
this also passes a return value out to make testing easier.
10 years ago
ca38be7a2d
renderer: check transfer read/write args better.
...
This confirms the transfer read/write args are non-NULL
10 years ago
6a2d49f37a
renderer: check detach_iov interface
...
check the return values aren't NULL
10 years ago
3d91ff730e
renderer: cleanup context create errors and context destroy
10 years ago
Dave Airlie
Marc-André Lureau
Dave Airlie