fdo-mirrors/mesa
1
0
Fork 0
mirror of https://gitlab.freedesktop.org/mesa/mesa.git synced 2026-02-20 23:10:30 +01:00
Code Issues Projects Releases Packages Wiki Activity Actions 4

aco: only load streamout buffers if streamout is enabled

Browse source 
The streamout_config SGPR is used to determine if streamout is enabled.

This fixes a GPU hang with various transform feedback tests:
 - dEQP-GLES3.functional.transform_feedback.*
 - KHR-GL46.transform_feedback.api_errors_test
 - KHR-GL46.draw_indirect.basic-draw*-xfbPaused
 - KHR-GL46.geometry_shader.api.draw_calls_while_tf_is_paused

Cc: 21.3 mesa-stable
Signed-off-by: Samuel Pitoiset <samuel.pitoiset@gmail.com>
Reviewed-by: Rhys Perry <pendingchaos02@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/13514>
This commit is contained in:
Samuel Pitoiset 2021-10-25 13:58:20 +02:00 • committed by Marge Bot
parent db82d90451
commit dc74285d32
1 changed files with 5 additions and 11 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

16
src/amd/compiler/aco_instruction_selection.cpp
View file

@ -11196,17 +11196,6 @@ emit_streamout(isel_context* ctx, unsigned stream)
{
Builder bld(ctx->program, ctx->block);
Temp so_buffers[4];
Temp buf_ptr = convert_pointer_to_64_bit(ctx, get_arg(ctx, ctx->args->streamout_buffers));
for (unsigned i = 0; i < 4; i++) {
unsigned stride = ctx->program->info->so.strides[i];
if (!stride)
continue;
Operand off = bld.copy(bld.def(s1), Operand::c32(i * 16u));
so_buffers[i] = bld.smem(aco_opcode::s_load_dwordx4, bld.def(s4), buf_ptr, off);
}
Temp so_vtx_count =
bld.sop2(aco_opcode::s_bfe_u32, bld.def(s1), bld.def(s1, scc),
get_arg(ctx, ctx->args->ac.streamout_config), Operand::c32(0x70010u));
@ -11223,13 +11212,18 @@ emit_streamout(isel_context* ctx, unsigned stream)
Temp so_write_index =
bld.vadd32(bld.def(v1), get_arg(ctx, ctx->args->ac.streamout_write_index), tid);
Temp so_buffers[4];
Temp so_write_offset[4];
Temp buf_ptr = convert_pointer_to_64_bit(ctx, get_arg(ctx, ctx->args->streamout_buffers));
for (unsigned i = 0; i < 4; i++) {
unsigned stride = ctx->program->info->so.strides[i];
if (!stride)
continue;
so_buffers[i] = bld.smem(aco_opcode::s_load_dwordx4, bld.def(s4), buf_ptr,
bld.copy(bld.def(s1), Operand::c32(i * 16u)));
if (stride == 1) {
Temp offset = bld.sop2(aco_opcode::s_add_i32, bld.def(s1), bld.def(s1, scc),
get_arg(ctx, ctx->args->ac.streamout_write_index),