From 8ffa3c58d4bfec789872c58a16da4af4352119f3 Mon Sep 17 00:00:00 2001 From: Dave Airlie Date: Tue, 26 Apr 2016 14:28:56 +1000 Subject: [PATCH] tgsi/exec: make inputs/outputs optional for compute shaders. compute shaders don't need input/outputs so don't bother allocating memory for these. Acked-by: Roland Scheidegger Signed-off-by: Dave Airlie --- src/gallium/auxiliary/tgsi/tgsi_exec.c | 43 ++++++++++++++------------ 1 file changed, 24 insertions(+), 19 deletions(-) diff --git a/src/gallium/auxiliary/tgsi/tgsi_exec.c b/src/gallium/auxiliary/tgsi/tgsi_exec.c index 879ce6f0752..9348528c42a 100644 --- a/src/gallium/auxiliary/tgsi/tgsi_exec.c +++ b/src/gallium/auxiliary/tgsi/tgsi_exec.c @@ -1049,10 +1049,12 @@ tgsi_exec_machine_create(enum pipe_shader_type shader_type) mach->MaxGeometryShaderOutputs = TGSI_MAX_TOTAL_VERTICES; mach->Predicates = &mach->Temps[TGSI_EXEC_TEMP_P0]; - mach->Inputs = align_malloc(sizeof(struct tgsi_exec_vector) * PIPE_MAX_SHADER_INPUTS, 16); - mach->Outputs = align_malloc(sizeof(struct tgsi_exec_vector) * PIPE_MAX_SHADER_OUTPUTS, 16); - if (!mach->Inputs || !mach->Outputs) - goto fail; + if (shader_type != PIPE_SHADER_COMPUTE) { + mach->Inputs = align_malloc(sizeof(struct tgsi_exec_vector) * PIPE_MAX_SHADER_INPUTS, 16); + mach->Outputs = align_malloc(sizeof(struct tgsi_exec_vector) * PIPE_MAX_SHADER_OUTPUTS, 16); + if (!mach->Inputs || !mach->Outputs) + goto fail; + } /* Setup constants needed by the SSE2 executor. */ for( i = 0; i < 4; i++ ) { @@ -5850,7 +5852,8 @@ tgsi_exec_machine_run( struct tgsi_exec_machine *mach, int start_pc ) uint inst = 1; memset(mach->Temps, 0, sizeof(temps)); - memset(mach->Outputs, 0, sizeof(outputs)); + if (mach->Outputs) + memset(mach->Outputs, 0, sizeof(outputs)); memset(temps, 0, sizeof(temps)); memset(outputs, 0, sizeof(outputs)); #endif @@ -5886,21 +5889,23 @@ tgsi_exec_machine_run( struct tgsi_exec_machine *mach, int start_pc ) } } } - for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { - if (memcmp(&outputs[i], &mach->Outputs[i], sizeof(outputs[i]))) { - uint j; - - memcpy(&outputs[i], &mach->Outputs[i], sizeof(outputs[i])); - debug_printf("OUT[%2u] = ", i); - for (j = 0; j < 4; j++) { - if (j > 0) { - debug_printf(" "); + if (mach->Outputs) { + for (i = 0; i < PIPE_MAX_ATTRIBS; i++) { + if (memcmp(&outputs[i], &mach->Outputs[i], sizeof(outputs[i]))) { + uint j; + + memcpy(&outputs[i], &mach->Outputs[i], sizeof(outputs[i])); + debug_printf("OUT[%2u] = ", i); + for (j = 0; j < 4; j++) { + if (j > 0) { + debug_printf(" "); + } + debug_printf("(%6f %u, %6f %u, %6f %u, %6f %u)\n", + outputs[i].xyzw[0].f[j], outputs[i].xyzw[0].u[j], + outputs[i].xyzw[1].f[j], outputs[i].xyzw[1].u[j], + outputs[i].xyzw[2].f[j], outputs[i].xyzw[2].u[j], + outputs[i].xyzw[3].f[j], outputs[i].xyzw[3].u[j]); } - debug_printf("(%6f %u, %6f %u, %6f %u, %6f %u)\n", - outputs[i].xyzw[0].f[j], outputs[i].xyzw[0].u[j], - outputs[i].xyzw[1].f[j], outputs[i].xyzw[1].u[j], - outputs[i].xyzw[2].f[j], outputs[i].xyzw[2].u[j], - outputs[i].xyzw[3].f[j], outputs[i].xyzw[3].u[j]); } } } -- 2.30.2