r600g: LIT: swap MUL_LIT operands to fix 0^0

For 0^0 case result of "LOG_CLAMPED ...,0" is -MAX_FLOAT, and then result of
"MUL_LIT ...,0,-MAX_FLOAT,..." is -MAX_FLOAT instead of 0 because of special
src1 checks for -MAX_FLOAT. So swap src0/1:
"MUL_LIT ...,-MAX_FLOAT,0,..." to get expected 0, then result of
"EXP_IEEE ...,0" is 1 as expected for LIT.

Signed-off-by: Vadim Girlin <vadimgirlin@gmail.com>

diff --git a/src/gallium/drivers/r600/r600_shader.c b/src/gallium/drivers/r600/r600_shader.c
index f83d707..f3cbf98 100644 (file)
--- a/src/gallium/drivers/r600/r600_shader.c
+++ b/src/gallium/drivers/r600/r600_shader.c
@@ -1388,7 +1388,7 @@ static int tgsi_lit(struct r600_shader_ctx *ctx)
                                        return r;
                        }
                } else {
-                       /* dst.z = log(src.y) */
+                       /* tmp.z = log(src.y) */
                        memset(&alu, 0, sizeof(struct r600_bc_alu));
                        alu.inst = CTX_INST(V_SQ_ALU_WORD1_OP2_SQ_OP2_INST_LOG_CLAMPED);
                        r600_bc_src(&alu.src[0], &ctx->src[0], 1);
@@ -1404,13 +1404,12 @@ static int tgsi_lit(struct r600_shader_ctx *ctx)
                chan = alu.dst.chan;
                sel = alu.dst.sel;
 
-               /* tmp.x = amd MUL_LIT(src.w, dst.z, src.x ) */
+               /* tmp.x = amd MUL_LIT(tmp.z, src.w, src.x ) */
                memset(&alu, 0, sizeof(struct r600_bc_alu));
                alu.inst = CTX_INST(V_SQ_ALU_WORD1_OP3_SQ_OP3_INST_MUL_LIT);
-               r600_bc_src(&alu.src[0], &ctx->src[0], 3);
-               alu.src[1].sel  = sel;
-               alu.src[1].chan = chan;
-
+               alu.src[0].sel  = sel;
+               alu.src[0].chan = chan;
+               r600_bc_src(&alu.src[1], &ctx->src[0], 3);
                r600_bc_src(&alu.src[2], &ctx->src[0], 0);
                alu.dst.sel = ctx->temp_reg;
                alu.dst.chan = 0;