vc4: Introduce scheduling of QPU instructions.

This doesn't reschedule much currently, just tries to fit things into the
regfile A/B write-versus-read slots (the cause of the improvements in
shader-db), and hide texture fetch latency by scheduling setup early and
results collection late (haven't performance tested it).  This
infrastructure will be important for doing instruction pairing, though.

shader-db2 results:
total instructions in shared programs: 61874 -> 59583 (-3.70%)
instructions in affected programs:     50677 -> 48386 (-4.52%)

1 files changed, 12 insertions, 0 deletions

diff --git a/src/gallium/drivers/vc4/vc4_qpu.c b/src/gallium/drivers/vc4/vc4_qpu.c
index 093ca077e6d..723b3613665 100644
--- a/src/gallium/drivers/vc4/vc4_qpu.c
+++ b/src/gallium/drivers/vc4/vc4_qpu.c
@@ -22,6 +22,7 @@
  */
 
 #include <stdbool.h>
+#include "vc4_qir.h"
 #include "vc4_qpu.h"
 
 static uint64_t
@@ -267,3 +268,14 @@ qpu_inst_is_tlb(uint64_t inst)
                 sig == QPU_SIG_COLOR_LOAD ||
                 sig == QPU_SIG_WAIT_FOR_SCOREBOARD);
 }
+
+void
+qpu_serialize_one_inst(struct vc4_compile *c, uint64_t inst)
+{
+        if (c->qpu_inst_count >= c->qpu_inst_size) {
+                c->qpu_inst_size = MAX2(16, c->qpu_inst_size * 2);
+                c->qpu_insts = realloc(c->qpu_insts,
+                                       c->qpu_inst_size * sizeof(uint64_t));
+        }
+        c->qpu_insts[c->qpu_inst_count++] = inst;
+}