fde6aeb886
mesa/main: wire up glapi bits for EXT_multi_draw_indirect
...
Turns out we were missing the glapi bits, making it impossible to use get
the function pointers for this extension. Whoops?!
[daniels: Squashed in a618 SkQP fails, presumably caused by these not
being skipped anymore.]
Fixes: 9f5af68995antomani103@gmail.com >
Tested-by: Chris Healy <healych@amazon.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33546 >
2025-02-21 09:22:03 +00:00
244c9cc45e
mapi/glx: Remove FASTCALL/PURE
...
This isn't worth the complexity.
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io >
Reviewed-by: Eric Engestrom <eric@igalia.com >
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33623 >
2025-02-20 15:47:23 +00:00
32a10ccbdd
glx: Remove (almost) all usage of _X_HIDDEN / _X_INTERNAL
...
It's redundant at this point. The one exception is for GLX_PUBLIC when
building for glvnd, because then we really do want the GLX API to be
hidden.
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io >
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33623 >
2025-02-20 15:47:23 +00:00
43fb26f8ea
mapi/glx: Remove xserver code generation
...
This hasn't been hooked up to the build since we deleted autotools back
in 2019. It's effectively dead code anyway, as GLX is not a moving
target, and at this point is it easier to modify the generated code
directly than to modify the generator. xserver is encouraged to copy
the generators from 2019 into its own build if it wants, or -
preferably, in this GLX greybeard's opinion - find a prettier codegen
solution in the process of finishing GL 3.0 support.
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io >
Reviewed-by: Eric Engestrom <eric@igalia.com >
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Acked-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33623 >
2025-02-20 15:47:23 +00:00
be106bd6c6
mesa: Handle getting GL_MAX_VIEWS_OVR
...
Add support for GL_OVR_multiview's GL_MAX_VIEWS_OVR which can be
accessed with glGetIntegerv().
MaxViews is accessed via the hash table set up by get_hash_params.py as
a constant (MAX_VIEWS_OVR) using GL_MAX_VIEWS_OVR.
v2: Add this patch (thanks to Mike's guidance)
v3: Drop unnecessary enum size element in OVR_multiview.XML
v4: Switch to CONST(MAX_VIEWS_OVR) instead of gl_constants::MaxViews
(Marek's suggestion)
Fixes: 328c29d600michael.blumenkrantz@gmail.com >
Signed-off-by: James Hogan <james@albanarts.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32992 >
2025-02-19 19:12:33 +00:00
da6971f865
mapi: Fix typos
...
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33556 >
2025-02-15 17:43:44 +00:00
57d47f717a
mesa: Add GL_EXT_protected_textures support
...
Add support for GL_EXT_protected_textures to create protected
texture in OpenGL ES 3.2. This enables allocating standard
GL textures as protected surfaces. This allows use-cases such
as depth, stencil, or mipmapped textures to be supported as
destinations for rendering within a protected context.
Signed-off-by: Saroj Kumar <saroj.kumar@amd.com >
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33358 >
2025-02-11 08:14:03 +00:00
1aa21c27d4
treewide: remove unneeded executable bit in non-scripting files
...
Noticed one by chance and searched for any others with find that were
clearly not meant to be executable.
For the curious:
33aa039acfed176e2c71d0e5203855328c29d600ac912b3754random342@airmail.cc >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/33436 >
2025-02-06 16:51:15 -06:00
e7214b9446
glapi: rename exported symbols so as not to conflict with old libglapi
...
libwaffle 1.7.0 has a hack that dlopen's libglapi with RTLD_GLOBAL, which
was meant to preload libglapi, but with this MR it overwrites libgallium's
own symbols, which breaks libgallium.
Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com >
Reviewed-by: Eric Engestrom <None>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32789 >
2025-01-23 00:49:05 +00:00
464dde302c
glapi: remove the remap table
...
it's unused now
Reviewed-by: Adam Jackson <ajax@redhat.com >
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Reviewed-by: Eric Engestrom <None>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32789 >
2025-01-23 00:49:05 +00:00
b22f682a31
glapi: stop using the remap table
...
The remap table adds an array lookup into 75% of CALL_* macros, which are
used to call GL functions through the dispatch table. Removing the array
lookup reduces overhead of dispatch table calls.
Since libglapi is now required to be from the same build as libgallium,
the remap table is no longer needed.
This change doesn't remove the remapping table. It only disables it.
Compare asm:
Before:
0000000000000000 <_mesa_unmarshal_Uniform1f>:
0: f3 0f 1e fa endbr64
4: 48 83 ec 08 sub $0x8,%rsp
8: 48 8b 05 00 00 00 00 mov 0x0(%rip),%rax # f <_mesa_unmarshal_Uniform1f+0xf>
f: 8b 4e 04 mov 0x4(%rsi),%ecx
12: 31 d2 xor %edx,%edx
14: f3 0f 10 46 08 movss 0x8(%rsi),%xmm0
19: 48 63 80 a8 01 00 00 movslq 0x1a8(%rax),%rax
20: 85 c0 test %eax,%eax
22: 78 08 js 2c <_mesa_unmarshal_Uniform1f+0x2c>
24: 48 8b 57 40 mov 0x40(%rdi),%rdx
28: 48 8b 14 c2 mov (%rdx,%rax,8),%rdx
2c: 89 cf mov %ecx,%edi
2e: ff d2 call *%rdx
30: b8 02 00 00 00 mov $0x2,%eax
35: 48 83 c4 08 add $0x8,%rsp
39: c3 ret
After:
0000000000000000 <_mesa_unmarshal_Uniform1f>:
0: f3 0f 1e fa endbr64
4: 48 89 f8 mov %rdi,%rax
7: 48 83 ec 08 sub $0x8,%rsp
b: f3 0f 10 46 08 movss 0x8(%rsi),%xmm0
10: 8b 7e 04 mov 0x4(%rsi),%edi
13: 48 8b 40 40 mov 0x40(%rax),%rax
17: ff 90 10 10 00 00 call *0x1010(%rax)
1d: b8 02 00 00 00 mov $0x2,%eax
22: 48 83 c4 08 add $0x8,%rsp
26: c3 ret
Reviewed-by: Adam Jackson <ajax@redhat.com >
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Reviewed-by: Eric Engestrom <None>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32789 >
2025-01-23 00:49:05 +00:00
44bda7c258
dri: put shared-glapi into libgallium.*.so
...
so that we don't have to maintain a stable ABI for it.
This will allow removal of the remapping table to reduce CALL_* overhead
for GL dispatch tables.
Also we can now clean it up.
Reviewed-by: Adam Jackson <ajax@redhat.com >
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Reviewed-by: Eric Engestrom <None>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/32789 >
2025-01-23 00:49:05 +00:00
894c4f0c78
meson: remove selinux option
...
Reviewed-by: Michel Dänzer <mdaenzer@redhat.com >
Reviewed-by: Eric Engestrom <eric@igalia.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31656 >
2024-10-21 01:14:35 +00:00
ac912b3754
mesa: OVR_multiview_multisampled_render_to_texture
...
this is automatically supported for anyone that supports OVR_multiview
and EXT_multisampled_render_to_texture
Acked-by: Marek Olšák <marek.olsak@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31235 >
2024-09-20 18:54:26 +00:00
328c29d600
mesa,glsl,gallium: add GL_OVR_multiview
...
Co-authored-by: Mike Blumenkrantz <michael.blumenkrantz@gmail.com >
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/31128 >
2024-09-18 13:07:41 +00:00
34f4b5e0ea
mesa: add KHR_shader_subgroup extension
...
Signed-off-by: Qiang Yu <yuq825@gmail.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/30610 >
2024-08-26 10:46:16 +08:00
91e1ea52c9
mesa_interface: Move out of GL/internal/
...
Move it into src/gallium/include/ to make it absolutely clear this is a Mesa detail.
While we're at it, clean up its include sites, including some places
where we can just include kopper_interface.h instead since it includes
mesa_interface.h as its first act.
Acked-by: Daniel Stone <daniels@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28378 >
2024-07-17 23:47:05 +00:00
68215332a8
build: pass licensing information in SPDX form
...
Acked-by: Alyssa Rosenzweig <alyssa@rosenzweig.io >
Acked-by: Dylan Baker <dylan.c.baker@intel.com >
Acked-by: Eric Engestrom <eric@igalia.com >
Acked-by: Daniel Stone <daniels@collabora.com >
Signed-off-by: David Heidelberg <david@ixit.cz >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29972 >
2024-06-29 12:42:49 -07:00
21d3eacd23
gles1: fix glBufferSubData()
...
InternalBufferSubDataCopyMESA is required for
PIPE_CAP_ALLOW_GLTHREAD_BUFFER_SUBDATA_OPT.
Signed-off-by: Zack Middleton <zack@cloemail.com >
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29795 >
2024-06-24 17:08:12 +00:00
31841c6b11
gles1: fix GL_OES_vertex_array_object
...
Export functions for GL_OES_vertex_array_object through GetProcAddress
on gles1.
Signed-off-by: Zack Middleton <zack@cloemail.com >
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/29794 >
2024-06-24 16:21:47 +00:00
bc134ade0b
mapi: add EXT_texture_storage_compression extension
...
This extension enables applications to opt-in to fixed-rate
compression for immutable textures.
Signed-off-by: Louis-Francis Ratté-Boulianne <lfrb@collabora.com >
Reviewed-by: Erik Faye-Lund <erik.faye-lund@collabora.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27109 >
2024-06-12 21:20:06 +00:00
72ed71877c
glapi: fix param type in TexGenxOES
...
The spec [1] and _mesa_TexGenxOES both agree that it's GLfixed
instead of GLint.
[1]: https://registry.khronos.org/OpenGL/extensions/OES/OES_texture_cube_map.txt
Signed-off-by: Simon Ser <contact@emersion.fr >
Reviewed-by: Marek Olšák <marek.olsak@amd.com >
Reviewed-by: Ian Romanick <ian.d.romanick@intel.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28825 >
2024-05-01 00:47:59 +02:00
1ac1c0843f
treewide: Replace usage of macro DEBUG with MESA_DEBUG when possible
...
This is achieved by the following steps:
#ifndef DEBUG => #if !MESA_DEBUG
defined(DEBUG) => MESA_DEBUG
#ifdef DEBUG => #if MESA_DEBUG
This is done by replace in vscode
excludes
docs,*.rs,addrlib,src/imgui,*.sh,src/intel/vulkan/grl/gpu
These are safe because those files should keep DEBUG macro is already excluded;
and not directly replace DEBUG, as we have some symbols around it.
Use debug or NDEBUG instead of DEBUG in comments when proper
This for reduce the usage of DEBUG,
so it's easier migrating to MESA_DEBUG
These are found when migrating DEBUG to MESA_DEBUG,
these are all comment update, so it's safe
Replace comment /* DEBUG */ and /* !DEBUG */ with proper /* MESA_DEBUG */ or /* !MESA_DEBUG */ manually
DEBUG || !NDEBUG -> MESA_DEBUG || !NDEBUG
!DEBUG && NDEBUG -> !(MESA_DEBUG || !NDEBUG)
Replace the DEBUG present in comment with proper new MESA_DEBUG manually
Signed-off-by: Yonggang Luo <luoyonggang@gmail.com >
Acked-by: David Heidelberg <david.heidelberg@collabora.com >
Reviewed-by: Eric Engestrom <eric@igalia.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28092 >
2024-03-22 18:22:34 +00:00
97b6851815
apple: Extended Apple feature support using GLX_USE_APPLE.
...
On MacOS/Apple/Dawin you can only get MESA to forward the GL funtions to
the systems OpenGL.framework or run SWRast directly. There is no way to use a gallium driver, even if they have been compiled.
The two gallium drivers of interest are SWRast and Zink, as the rest are hardware drivers and not relavent on MacOS.
The code changes add a new define GLX_USE_APPLE. This is used in combination with the existing GLX_USE_APPLEGL.
GLX_USE_APPLEGL calls the systems OpenGL.framework, Apple's OpenGL.
GLX_USE_APPLE calls the non-system OpenGL code, i.e. Gallium, hence the subtle naming difference. Apple systems are still used, just not the GL ones.
When GLX_USE_APPLE is defined the code will use the DRI/gallium driver sub-system so SWRast and Zink can selected at runtime on MacOS.
This also allows Zink to be run on MacOS, once it is fixed up.
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/28205 >
2024-03-18 18:06:45 +00:00
74c28b93c8
glthread: invert _mesa_glthread_has_no_{un}pack_buffer by removing the negation
...
transform _mesa_glthread_has_no_unpack_buffer into
_mesa_glthread_has_unpack_buffer, etc.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27490 >
2024-03-11 22:44:42 +00:00
9939f20a49
glthread: execute small glDrawPixels asynchronously
...
Compute the image size and copy the image into the batch.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27490 >
2024-03-11 22:44:41 +00:00
b43b111b19
glthread: execute small glBitmap asynchronously
...
Compute the bitmap size and copy the bitmap into the batch.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27490 >
2024-03-11 22:44:41 +00:00
ada351f2ea
glthread: track glPixelStore(GL_UNPACK_*)
...
so that glthread can compute the size of images passed to GL functions.
Reviewed-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27490 >
2024-03-11 22:44:41 +00:00
c9abb7ff6e
glthread: generate packed versions of gl*Pointer/Offset calls
...
The pointer/offset parameter is often NULL or a small number with VBOs.
The idea is:
- If the pointer/offset parameter is NULL/0, use a different cmd structure
and unmarshal function that doesn't contain the pointer/offset parameter
to save 8 bytes per call.
- If the cmd structure has a hole and the pointer/offset parameter is
a small number that fits into the hole, use a different cmd structure
and unmarshal function that stores the value within the hole using
a smaller type to save 8 bytes per call.
This implements those ideas. It will continue generating the most optimal
code even if we change the packing of other parameters.
This decreases the size of 1 frame in glthread batches by 21%
in Viewperf2020/Catia1.
Example of generated code for glVertexPointer with and without the pointer
parameter if it's NULL. See the arrows for comments.
/* VertexPointer: marshalled asynchronously */
struct marshal_cmd_VertexPointer
{
struct marshal_cmd_base cmd_base;
GLpacked16i size;
GLenum16 type;
GLclamped16i stride;
const GLvoid * pointer;
};
struct marshal_cmd_VertexPointer_packed
{
struct marshal_cmd_base cmd_base;
GLpacked16i size;
GLenum16 type;
GLclamped16i stride; // <------- no "pointer"
};
uint32_t _mesa_unmarshal_VertexPointer(struct gl_context *ctx, const struct marshal_cmd_VertexPointer *restrict cmd)
{
GLpacked16i size = cmd->size;
GLenum16 type = cmd->type;
GLclamped16i stride = cmd->stride;
const GLvoid * pointer = cmd->pointer;
CALL_VertexPointer(ctx->Dispatch.Current, (size, type, stride, pointer));
return align(sizeof(struct marshal_cmd_VertexPointer), 8) / 8;
}
uint32_t _mesa_unmarshal_VertexPointer_packed(struct gl_context *ctx, const struct marshal_cmd_VertexPointer_packed *restrict cmd)
{
GLpacked16i size = cmd->size;
GLenum16 type = cmd->type;
GLclamped16i stride = cmd->stride;
const GLvoid * pointer = (const GLvoid *)(uintptr_t)0; // <------- using NULL
CALL_VertexPointer(ctx->Dispatch.Current, (size, type, stride, pointer));
return align(sizeof(struct marshal_cmd_VertexPointer_packed), 8) / 8;
}
static void GLAPIENTRY
_mesa_marshal_VertexPointer(GLint size, GLenum type, GLsizei stride, const GLvoid *pointer)
{
GET_CURRENT_CONTEXT(ctx);
if (!pointer) { // <------- the condition
int cmd_size = sizeof(struct marshal_cmd_VertexPointer_packed);
struct marshal_cmd_VertexPointer_packed *cmd = _mesa_glthread_allocate_command(ctx, DISPATCH_CMD_VertexPointer_packed, cmd_size);
cmd->size = size < 0 ? UINT16_MAX : MIN2(size, UINT16_MAX);
cmd->type = MIN2(type, 0xffff); /* clamped to 0xffff (invalid enum) */
cmd->stride = CLAMP(stride, INT16_MIN, INT16_MAX);
} else {
int cmd_size = sizeof(struct marshal_cmd_VertexPointer);
struct marshal_cmd_VertexPointer *cmd = _mesa_glthread_allocate_command(ctx, DISPATCH_CMD_VertexPointer, cmd_size);
cmd->size = size < 0 ? UINT16_MAX : MIN2(size, UINT16_MAX);
cmd->type = MIN2(type, 0xffff); /* clamped to 0xffff (invalid enum) */
cmd->stride = CLAMP(stride, INT16_MIN, INT16_MAX);
cmd->pointer = pointer;
}
_mesa_glthread_AttribPointer(ctx, VERT_ATTRIB_POS, MESA_PACK_VFORMAT(type, size, 0, 0, 0), stride, pointer);
}
Example of generated code for glNormalPointer using a smaller type:
/* NormalPointer: marshalled asynchronously */
struct marshal_cmd_NormalPointer
{
struct marshal_cmd_base cmd_base;
GLenum16 type;
GLclamped16i stride;
const GLvoid * pointer;
};
struct marshal_cmd_NormalPointer_packed
{
struct marshal_cmd_base cmd_base;
GLenum16 type;
GLclamped16i stride;
GLushort pointer; // <-------- truncated "pointer"
};
uint32_t _mesa_unmarshal_NormalPointer(struct gl_context *ctx, const struct marshal_cmd_NormalPointer *restrict cmd)
{
GLenum16 type = cmd->type;
GLclamped16i stride = cmd->stride;
const GLvoid * pointer = cmd->pointer;
CALL_NormalPointer(ctx->Dispatch.Current, (type, stride, pointer));
return align(sizeof(struct marshal_cmd_NormalPointer), 8) / 8;
}
uint32_t _mesa_unmarshal_NormalPointer_packed(struct gl_context *ctx, const struct marshal_cmd_NormalPointer_packed *restrict cmd)
{
GLenum16 type = cmd->type;
GLclamped16i stride = cmd->stride;
const GLvoid * pointer = (const GLvoid *)(uintptr_t)cmd->pointer; // <-------- upcasting
CALL_NormalPointer(ctx->Dispatch.Current, (type, stride, pointer));
return align(sizeof(struct marshal_cmd_NormalPointer_packed), 8) / 8;
}
static void GLAPIENTRY
_mesa_marshal_NormalPointer(GLenum type, GLsizei stride, const GLvoid *pointer)
{
GET_CURRENT_CONTEXT(ctx);
if (((uintptr_t)pointer & 0xffff) == (uintptr_t)pointer) { // <-------- the condition
int cmd_size = sizeof(struct marshal_cmd_NormalPointer_packed);
struct marshal_cmd_NormalPointer_packed *cmd = _mesa_glthread_allocate_command(ctx, DISPATCH_CMD_NormalPointer_packed, cmd_size);
cmd->type = MIN2(type, 0xffff); /* clamped to 0xffff (invalid enum) */
cmd->stride = CLAMP(stride, INT16_MIN, INT16_MAX);
cmd->pointer = (uintptr_t)pointer; /* truncated */ // <-------- the truncation
} else {
int cmd_size = sizeof(struct marshal_cmd_NormalPointer);
struct marshal_cmd_NormalPointer *cmd = _mesa_glthread_allocate_command(ctx, DISPATCH_CMD_NormalPointer, cmd_size);
cmd->type = MIN2(type, 0xffff); /* clamped to 0xffff (invalid enum) */
cmd->stride = CLAMP(stride, INT16_MIN, INT16_MAX);
cmd->pointer = pointer;
}
_mesa_glthread_AttribPointer(ctx, VERT_ATTRIB_NORMAL, MESA_PACK_VFORMAT(type, 3, 1, 0, 0), stride, pointer);
}
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:59 +00:00
24f14f8daa
glthread: add a packed version of DrawElementsUserBuf
...
The reduces the call size by 24 bytes.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:59 +00:00
c566df8b39
glthread: add a packed variant of glDrawElements with 16-bit count and indices
...
This is just to decrease the size of glDrawElements by 8 more bytes.
This packed glDrawElements call occupies only 1 slot in glthread_batch.
This decreases the size of 1 frame in glthread batches by 13%
in Viewperf2020/Catia1.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:59 +00:00
1f9b554839
glthread: use marshal_count instead of count for more functions
...
Same as the previous commit, just applied to more functions.
This removes safe_mul and checking whether cmd_size is too large
because the size is always small with these functions.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:59 +00:00
e8721fefcd
glthread: don't check cmd_size for small variable-sized calls
...
This removes the size checking, syncing, and direct execution if
the variable-sized call is always small. Don't use safe_mul in that case
either. Only calls already using marshal_count are affected. Example:
Before:
static void GLAPIENTRY
_mesa_marshal_PointParameterfv(GLenum pname, const GLfloat *params)
{
GET_CURRENT_CONTEXT(ctx);
int params_size = safe_mul(_mesa_point_param_enum_to_count(pname), 1 * sizeof(GLfloat));
int cmd_size = sizeof(struct marshal_cmd_PointParameterfv) + params_size;
if (unlikely(params_size < 0 || (params_size > 0 && !params) || (unsigned)cmd_size > MARSHAL_MAX_CMD_SIZE)) {
_mesa_glthread_finish_before(ctx, "PointParameterfv");
CALL_PointParameterfv(ctx->Dispatch.Current, (pname, params));
return;
}
struct marshal_cmd_PointParameterfv *cmd = _mesa_glthread_allocate_command(ctx, DISPATCH_CMD_PointParameterfv, cmd_size);
cmd->num_slots = align(cmd_size, 8) / 8;
cmd->pname = MIN2(pname, 0xffff); /* clamped to 0xffff (invalid enum) */
char *variable_data = (char *) (cmd + 1);
memcpy(variable_data, params, params_size);
}
After:
static void GLAPIENTRY
_mesa_marshal_PointParameterfv(GLenum pname, const GLfloat *params)
{
GET_CURRENT_CONTEXT(ctx);
int params_size = _mesa_point_param_enum_to_count(pname) * 1 * sizeof(GLfloat);
int cmd_size = sizeof(struct marshal_cmd_PointParameterfv) + params_size;
assert(cmd_size >= 0 && cmd_size <= MARSHAL_MAX_CMD_SIZE);
struct marshal_cmd_PointParameterfv *cmd = _mesa_glthread_allocate_command(ctx, DISPATCH_CMD_PointParameterfv, cmd_size);
cmd->num_slots = align(cmd_size, 8) / 8;
cmd->pname = MIN2(pname, 0xffff); /* clamped to 0xffff (invalid enum) */
char *variable_data = (char *) (cmd + 1);
memcpy(variable_data, params, params_size);
}
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:59 +00:00
eda0b9f8d4
glthread: pack glVertexAttribPointer calls better
...
These parameters can use 8 bits.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:59 +00:00
ad34c932cd
glapi: pass pointer size to python for glthread from meson
...
glthread (the python generator) needs to know the pointer size at compile
time to sort structure fields of calls for optimal structure packing based
on the CPU.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:59 +00:00
610bc4d115
glthread: remove "if True" from print_marshal_async_code
...
This only changes indentation. No functional change.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:59 +00:00
e299473968
glthread: separate marshal code generation into print_marshal_async_code
...
I added "if True" to make the diff readable.
No functional change.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:59 +00:00
2d796de157
glthread: separate unmarshal function generation into print_unmarshal_func
...
No functional change.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:59 +00:00
dd6b0ea015
gltrhead: merge 3 blocks conditional on marshal_sync in print_async_body
...
There are 3 blocks next to each other that check marshal_sync. Merge them.
No functional change.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:59 +00:00
e9819744b3
glthread: precompute fixed_params and variable_params lists
...
This removes functions get_fixed_params and get_variable_params.
No functional change.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:59 +00:00
6eef0c60f8
glthread: move global marshal_XML.py functions into class marshal_function
...
No functional change.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:58 +00:00
aea16b5ae9
glthread: sort fixed-sized parameters before returning them
...
No functional change.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:58 +00:00
ee0d869d66
glapi: fix type names for glthread and handle all types
...
glthread will compare the whole type string, so the string must not have
trailing spaces.
No functional change.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:58 +00:00
12754aec1a
glthread: rewrite glDrawArrays call packing
...
Since changing 1 field to 8 bits and the removal of cmd_size, call sizes
have decreased, so we have 4 unused bytes in 2 DrawArrays structures
So far we use:
- DrawArrays
- DrawArraysInstancedBaseInstance
- DrawArraysInstancedBaseInstanceDrawID
Change them to these by either removing 4 more bytes or adding 4 bytes,
so that we don't waste space, which drops the number of used calls by 1:
- DrawArraysInstanced
- DrawArraysInstancedBaseInstanceDrawID
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:58 +00:00
5e5d91671e
glthread: rewrite glDrawElements call packing
...
Since changing 2 fields to 8 bits and the removal of cmd_size, call sizes
have decreased by 4 bytes, so we have 4 unused bytes in most DrawElements
structures. So far we have used these calls for all DrawElements variants:
- DrawElementsBaseVertex
- DrawElementsInstanced
- DrawElementsInstancedBaseVertexBaseInstance
- DrawElementsInstancedBaseVertexBaseInstanceDrawID
Change them to these by either removing 4 more bytes or adding 4 bytes,
so that we don't waste space.
- DrawElements
- DrawElementsInstancedBaseVertex
- DrawElementsInstancedBaseInstance
- DrawElementsInstancedBaseVertexBaseInstanceDrawID
This decreases the size of 1 frame in glthread batches by 12%
in Viewperf2020/Catia1.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:58 +00:00
99f8f01dfb
glthread: pack the index type to 8 bits
...
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:58 +00:00
15bc7e1d62
glthread: pack the primitive type to 8 bits
...
The maximum valid enum is only 14.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:58 +00:00
1388be4d39
glthread: pack "size" in Pointer calls as 16 bits
...
The only legal values are {1, 2, 3, 4, GL_BGRA}.
We need GLpacked16i to be unsigned, not signed, because GL_BGRA is
greater than 0x8000.
This decreases the size of 1 frame by 10% in Viewperf2020/Catia1.
It decreases the size of many Pointer calls by 8 bytes.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:58 +00:00
13a8efcb2c
glthread: clean up how vertex stride is packed
...
Use a better type name. Also check the function name more accurately
(no change in behavior).
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:58 +00:00
617cdedd35
glthread: remove cmd_size from constant-sized calls
...
Only variable-sized calls keep cmd_size in their structures, and it's
renamed to num_slots because it's in units of 8-byte elements.
The motivation is to make room for reducing call sizes.
Acked-by: Pierre-Eric Pelloux-Prayer <pierre-eric.pelloux-prayer@amd.com >
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/27350 >
2024-02-23 18:03:58 +00:00
Erik Faye-Lund