On 02/18/2016 11:36 AM, Charmaine Lee wrote:
From: Brian Paul <bri...@vmware.com> Sent: Thursday, February 18, 2016 7:51 AM To: mesa-dev@lists.freedesktop.org Cc: Charmaine Lee Subject: [PATCH] svga: allow non-contiguous VS input declarations This fixes a glDrawPixels regression since b63fe0552b5f. The new quad-drawing utility code uses 3 vertex attributes (xyz, rgba, st). For glDrawPixels path we don't use the rgba attribute so there's a gap in the TGSI VS input declarations (INPUT[0] = pos, INPUT[2] = texcoord). The TGSI->VGPU10 translations code did not handle this correctly. I missed this because my VM was configured for HWv11 while testing. Another way to fix this would be to change the tgsi_scan.c code so that the tgsi_shader_info::num_inputs (and num_outputs) included the unused inputs/outputs. These counts would then actually be "max input register index + 1" rather than "number of used inputs". But that change could impact all drivers so put it off for now. No regressions found with piglit or typical GL apps. --- src/gallium/drivers/svga/svga_tgsi_vgpu10.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/gallium/drivers/svga/svga_tgsi_vgpu10.c b/src/gallium/drivers/svga/svga_tgsi_vgpu10.c index 1223e44..384fb9e 100644 --- a/src/gallium/drivers/svga/svga_tgsi_vgpu10.c +++ b/src/gallium/drivers/svga/svga_tgsi_vgpu10.c @@ -2446,7 +2446,7 @@ emit_input_declarations(struct svga_shader_emitter_v10 *emit) else { assert(emit->unit == PIPE_SHADER_VERTEX);- for (i = 0; i < emit->info.num_inputs; i++) { + for (i = 0; i < emit->info.file_max[TGSI_FILE_INPUT] + 1; i++) { unsigned usage_mask = emit->info.input_usage_mask[i]; unsigned index = i;num_inputs are referenced in other functions as well such as svga_get_generic_inputs_mask, don't we need to fix those as well?
svga_get_generic_inputs_mask() is only concerned with FS inputs so we're OK there. IIRC, the GLSL linker (or maybe the GLSL->TGSI converter) allocates contiguous outputs/inputs between shader stages so there should be no gaps there. I think it's just VS inputs where we can have gaps.
But now I'm looking at alloc_system_value_index() which uses info.num_inputs, potentially for the VS...
-Brian _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
