freedreno/a3xx/compiler: cat4 cannot use const reg as src

Category 4 instructions (rsq, rcp, sqrt, etc) seem to be unable to take
a const register as src.  In these cases we need to move the src to a
temporary gpr first.

This is the second case of such a restriction, where the instruction
encoding appears to support a const src, but in fact the hw appears to
ignore that bit.  So split things out into a helper that can be re-used
for any instructions which have this limitation.

Signed-off-by: Rob Clark <robclark@freedesktop.org>

diff --git a/src/gallium/drivers/freedreno/a3xx/fd3_compiler.c b/src/gallium/drivers/freedreno/a3xx/fd3_compiler.c
index dc5c87395f6f988722620a743d54c3bd164d06bc..772c7d2a2e35bd3e3c81b81b9f34961273a3b2d9 100644 (file)
--- a/src/gallium/drivers/freedreno/a3xx/fd3_compiler.c
+++ b/src/gallium/drivers/freedreno/a3xx/fd3_compiler.c
@@ -135,6 +135,8 @@ struct fd3_compile_context {
 static void vectorize(struct fd3_compile_context *ctx,
                struct ir3_instruction *instr, struct tgsi_dst_register *dst,
                int nsrcs, ...);
+static void create_mov(struct fd3_compile_context *ctx,
+               struct tgsi_dst_register *dst, struct tgsi_src_register *src);
 
 static unsigned
 compile_init(struct fd3_compile_context *ctx, struct fd3_shader_stateobj *so,
@@ -374,6 +376,23 @@ get_immediate(struct fd3_compile_context *ctx,
        reg->SwizzleW  = swiz2tgsi[swiz];
 }
 
+/* for instructions that cannot take a const register as src, if needed
+ * generate a move to temporary gpr:
+ */
+static struct tgsi_src_register *
+get_unconst(struct fd3_compile_context *ctx, struct tgsi_src_register *src,
+               struct tgsi_src_register *tmp_src)
+{
+       static struct tgsi_dst_register tmp_dst;
+       if ((src->File == TGSI_FILE_CONSTANT) ||
+                       (src->File == TGSI_FILE_IMMEDIATE)) {
+               get_internal_temp(ctx, &tmp_dst, tmp_src);
+               create_mov(ctx, &tmp_dst, src);
+               src = tmp_src;
+       }
+       return src;
+}
+
 static type_t
 get_type(struct fd3_compile_context *ctx)
 {
@@ -1027,8 +1046,7 @@ instr_cat3(const struct instr_translater *t,
                struct tgsi_full_instruction *inst)
 {
        struct tgsi_dst_register *dst = get_dst(ctx, inst);
-       struct tgsi_src_register *src1 = &inst->Src[1].Register;
-       struct tgsi_dst_register tmp_dst;
+       struct tgsi_src_register *src1;
        struct tgsi_src_register tmp_src;
        struct ir3_instruction *instr;
 
@@ -1038,12 +1056,7 @@ instr_cat3(const struct instr_translater *t,
         * const.  Not sure if this is a hw bug, or simply that the
         * disassembler lies.
         */
-       if ((src1->File == TGSI_FILE_CONSTANT) ||
-                       (src1->File == TGSI_FILE_IMMEDIATE)) {
-               get_internal_temp(ctx, &tmp_dst, &tmp_src);
-               create_mov(ctx, &tmp_dst, src1);
-               src1 = &tmp_src;
-       }
+       src1 = get_unconst(ctx, &inst->Src[1].Register, &tmp_src);
 
        instr = ir3_instr_create(ctx->ir, 3,
                        ctx->so->half_precision ? t->hopc : t->opc);
@@ -1060,13 +1073,17 @@ instr_cat4(const struct instr_translater *t,
                struct tgsi_full_instruction *inst)
 {
        struct tgsi_dst_register *dst = get_dst(ctx, inst);
+       struct tgsi_src_register *src;
+       struct tgsi_src_register tmp_src;
        struct ir3_instruction *instr;
 
+       /* seems like blob compiler avoids const as src.. */
+       src = get_unconst(ctx, &inst->Src[0].Register, &tmp_src);
+
        ir3_instr_create(ctx->ir, 0, OPC_NOP)->repeat = 5;
        instr = ir3_instr_create(ctx->ir, 4, t->opc);
 
-       vectorize(ctx, instr, dst, 1,
-                       &inst->Src[0].Register, 0);
+       vectorize(ctx, instr, dst, 1, src, 0);
 
        regmask_set(ctx->needs_ss, instr->regs[0]);