v3d: implement simultaneous peripheral access exceptions for V3D 4.1+

Shader-db results:

total instructions in shared programs: 9117550 -> 9102719 (-0.16%)
instructions in affected programs: 1752873 -> 1738042 (-0.85%)
helped: 7076
HURT: 478
helped stats (abs) min: 1 max: 22 x̄: 2.19 x̃: 2
helped stats (rel) min: 0.07% max: 13.89% x̄: 1.70% x̃: 1.07%
HURT stats (abs)   min: 1 max: 7 x̄: 1.41 x̃: 1
HURT stats (rel)   min: 0.09% max: 10.17% x̄: 0.86% x̃: 0.54%
95% mean confidence interval for instructions value: -2.00 -1.92
95% mean confidence interval for instructions %-change: -1.58% -1.50%
Instructions are helped.

total max-temps in shared programs: 1327774 -> 1327728 (<.01%)
max-temps in affected programs: 1025 -> 979 (-4.49%)
helped: 47
HURT: 2
helped stats (abs) min: 1 max: 2 x̄: 1.02 x̃: 1
helped stats (rel) min: 2.63% max: 20.00% x̄: 7.67% x̃: 5.26%
HURT stats (abs)   min: 1 max: 1 x̄: 1.00 x̃: 1
HURT stats (rel)   min: 4.17% max: 4.17% x̄: 4.17% x̃: 4.17%
95% mean confidence interval for max-temps value: -1.06 -0.82
95% mean confidence interval for max-temps %-change: -8.89% -5.49%
Max-temps are helped.

Reviewed-by: Eric Anholt <eric@anholt.net>

src/broadcom/compiler/qpu_schedule.c

@@ -604,6 +604,37 @@ qpu_accesses_peripheral(const struct v3d_qpu_instr *inst)
                 inst->sig.wrtmuc);
 }
 
+static bool
+qpu_compatible_peripheral_access(const struct v3d_device_info *devinfo,
+                                 const struct v3d_qpu_instr *a,
+                                 const struct v3d_qpu_instr *b)
+{
+        const bool a_uses_peripheral = qpu_accesses_peripheral(a);
+        const bool b_uses_peripheral = qpu_accesses_peripheral(b);
+
+        /* We can always do one peripheral access per instruction. */
+        if (!a_uses_peripheral || !b_uses_peripheral)
+                return true;
+
+        if (devinfo->ver < 41)
+                return false;
+
+        /* V3D 4.1 and later allow TMU read along with a VPM read or write, and
+         * WRTMUC with a TMU magic register write (other than tmuc).
+         */
+        if ((a->sig.ldtmu && v3d_qpu_uses_vpm(b)) ||
+            (b->sig.ldtmu && v3d_qpu_uses_vpm(a))) {
+                return true;
+        }
+
+        if ((a->sig.wrtmuc && v3d_qpu_writes_tmu_not_tmuc(b)) ||
+            (b->sig.wrtmuc && v3d_qpu_writes_tmu_not_tmuc(a))) {
+                return true;
+        }
+
+        return false;
+}
+
 static bool
 qpu_merge_inst(const struct v3d_device_info *devinfo,
                struct v3d_qpu_instr *result,
@@ -615,12 +646,7 @@ qpu_merge_inst(const struct v3d_device_info *devinfo,
                 return false;
         }
 
-        /* Can't do more than one peripheral access in an instruction.
-         *
-         * XXX: V3D 4.1 allows TMU read along with a VPM read or write, and
-         * WRTMUC with a TMU magic register write (other than tmuc).
-         */
-        if (qpu_accesses_peripheral(a) && qpu_accesses_peripheral(b))
+        if (!qpu_compatible_peripheral_access(devinfo, a, b))
                 return false;
 
         struct v3d_qpu_instr merge = *a;

src/broadcom/qpu/qpu_instr.c

@@ -682,6 +682,16 @@ v3d_qpu_writes_tmu(const struct v3d_qpu_instr *inst)
                   v3d_qpu_magic_waddr_is_tmu(inst->alu.mul.waddr))));
 }
 
+bool
+v3d_qpu_writes_tmu_not_tmuc(const struct v3d_qpu_instr *inst)
+{
+        return v3d_qpu_writes_tmu(inst) &&
+               (!inst->alu.add.magic_write ||
+                inst->alu.add.waddr != V3D_QPU_WADDR_TMUC) &&
+               (!inst->alu.mul.magic_write ||
+                inst->alu.mul.waddr != V3D_QPU_WADDR_TMUC);
+}
+
 bool
 v3d_qpu_reads_vpm(const struct v3d_qpu_instr *inst)
 {

src/broadcom/qpu/qpu_instr.h

@@ -449,6 +449,7 @@ bool v3d_qpu_magic_waddr_loads_unif(enum v3d_qpu_waddr waddr) ATTRIBUTE_CONST;
 bool v3d_qpu_uses_tlb(const struct v3d_qpu_instr *inst) ATTRIBUTE_CONST;
 bool v3d_qpu_uses_sfu(const struct v3d_qpu_instr *inst) ATTRIBUTE_CONST;
 bool v3d_qpu_writes_tmu(const struct v3d_qpu_instr *inst) ATTRIBUTE_CONST;
+bool v3d_qpu_writes_tmu_not_tmuc(const struct v3d_qpu_instr *inst) ATTRIBUTE_CONST;
 bool v3d_qpu_writes_r3(const struct v3d_device_info *devinfo,
                        const struct v3d_qpu_instr *instr) ATTRIBUTE_CONST;
 bool v3d_qpu_writes_r4(const struct v3d_device_info *devinfo,