radeonsi: use postponed KILL only when derivatives are used

Reviewed-by: Nicolai Hähnle <nicolai.haehnle@amd.com>

diff --git a/src/gallium/drivers/radeonsi/si_shader.c b/src/gallium/drivers/radeonsi/si_shader.c
index 8abacac8c8d3e1f6edaeb5a0b3144a38a914f823..c34304873075e00a7433144c70b79f15357b4c3a 100644 (file)
--- a/src/gallium/drivers/radeonsi/si_shader.c
+++ b/src/gallium/drivers/radeonsi/si_shader.c
@@ -5846,8 +5846,7 @@ static bool si_compile_tgsi_main(struct si_shader_context *ctx,
                }
        }
 
-       if (ctx->type == PIPE_SHADER_FRAGMENT && sel->info.uses_kill &&
-           ctx->screen->b.debug_flags & DBG(FS_CORRECT_DERIVS_AFTER_KILL)) {
+       if (sel->force_correct_derivs_after_kill) {
                ctx->postponed_kill = lp_build_alloca_undef(&ctx->gallivm, ctx->i1, "");
                /* true = don't kill. */
                LLVMBuildStore(ctx->ac.builder, LLVMConstInt(ctx->i1, 1, 0),

diff --git a/src/gallium/drivers/radeonsi/si_shader.h b/src/gallium/drivers/radeonsi/si_shader.h
index ebe956e709e7eb6f95e08d601ba665ffa4b90180..78a2ee14c13a895f74a2e531a3f737cf278cae03 100644 (file)
--- a/src/gallium/drivers/radeonsi/si_shader.h
+++ b/src/gallium/drivers/radeonsi/si_shader.h
@@ -344,6 +344,7 @@ struct si_shader_selector {
        /* PIPE_SHADER_[VERTEX|FRAGMENT|...] */
        unsigned        type;
        bool            vs_needs_prolog;
+       bool            force_correct_derivs_after_kill;
        unsigned        pa_cl_vs_out_cntl;
        ubyte           clipdist_mask;
        ubyte           culldist_mask;

diff --git a/src/gallium/drivers/radeonsi/si_shader_tgsi_alu.c b/src/gallium/drivers/radeonsi/si_shader_tgsi_alu.c
index ad7a42f7b84bb2641adba831db5952b10ffa33d0..283a889d84739fa7eec3993bd5a957bf155fbd29 100644 (file)
--- a/src/gallium/drivers/radeonsi/si_shader_tgsi_alu.c
+++ b/src/gallium/drivers/radeonsi/si_shader_tgsi_alu.c
@@ -68,7 +68,7 @@ static void kil_emit(const struct lp_build_tgsi_action *action,
                visible = LLVMConstInt(ctx->i1, false, 0);
        }
 
-       if (ctx->postponed_kill) {
+       if (ctx->shader->selector->force_correct_derivs_after_kill) {
                LLVMValueRef mask = LLVMBuildLoad(builder, ctx->postponed_kill, "");
                mask = LLVMBuildAnd(builder, mask, visible, "");
                LLVMBuildStore(builder, mask, ctx->postponed_kill);

diff --git a/src/gallium/drivers/radeonsi/si_state_shaders.c b/src/gallium/drivers/radeonsi/si_state_shaders.c
index 9340328a72ab3bcf764e484bc6e4dc7b35384ce2..af7f6ecfd60f3c2e53f6d3667e61964ba68f0680 100644 (file)
--- a/src/gallium/drivers/radeonsi/si_state_shaders.c
+++ b/src/gallium/drivers/radeonsi/si_state_shaders.c
@@ -2052,6 +2052,12 @@ static void *si_create_shader_selector(struct pipe_context *ctx,
                               sel->info.num_inputs &&
                               !sel->info.properties[TGSI_PROPERTY_VS_BLIT_SGPRS];
 
+       sel->force_correct_derivs_after_kill =
+               sel->type == PIPE_SHADER_FRAGMENT &&
+               sel->info.uses_derivatives &&
+               sel->info.uses_kill &&
+               sctx->screen->b.debug_flags & DBG(FS_CORRECT_DERIVS_AFTER_KILL);
+
        /* Set which opcode uses which (i,j) pair. */
        if (sel->info.uses_persp_opcode_interp_centroid)
                sel->info.uses_persp_centroid = true;