[Mesa-dev] [PATCH] nv50/ir/ra: improve condition for short regs, unify with cond for 16-bit

[Ilia Mirkin]

Instead of the size restriction existing in two places, and potentially
being applied twice, we move this together. Ops with 16-bit register
addresses can only take a short reg, and ops with immediates can only
take a short reg.

Of course we leave the immediate 0 in place since we know that it will
be replaced by r63/r127 down the line, so don't treat zeroes as an
immediate.

Signed-off-by: Ilia Mirkin <imirkin at alum.mit.edu>
---
 src/gallium/drivers/nouveau/codegen/nv50_ir_ra.cpp | 14 +++++++-------
 1 file changed, 7 insertions(+), 7 deletions(-)

diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_ra.cpp b/src/gallium/drivers/nouveau/codegen/nv50_ir_ra.cpp
index 21cbd955537..afaf59a8cd9 100644
--- a/src/gallium/drivers/nouveau/codegen/nv50_ir_ra.cpp
+++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_ra.cpp
@@ -66,10 +66,8 @@ public:
 
    inline int getMaxAssigned(DataFile f) const { return fill[f]; }
 
-   inline unsigned int getFileSize(DataFile f, uint8_t regSize) const
+   inline unsigned int getFileSize(DataFile f) const
    {
-      if (restrictedGPR16Range && f == FILE_GPR && regSize == 2)
-         return (last[f] + 1) / 2;
       return last[f] + 1;
    }
 
@@ -813,9 +811,11 @@ GCRA::printNodeInfo() const
 static bool
 isShortRegOp(Instruction *insn)
 {
-   // Immediates are always in src1. Every other situation can be resolved by
+   // Immediates are always in src1 (except zeroes, which end up getting
+   // replaced with a zero reg). Every other situation can be resolved by
    // using a long encoding.
-   return insn->srcExists(1) && insn->src(1).getFile() == FILE_IMMEDIATE;
+   return insn->srcExists(1) && insn->src(1).getFile() == FILE_IMMEDIATE &&
+      insn->getSrc(1)->reg.data.u64;
 }
 
 // Check if this LValue is ever used in an instruction that can't be encoded
@@ -851,10 +851,10 @@ GCRA::RIG_Node::init(const RegisterSet& regs, LValue *lval)
 
    weight = std::numeric_limits<float>::infinity();
    degree = 0;
-   int size = regs.getFileSize(f, lval->reg.size);
+   int size = regs.getFileSize(f);
    // On nv50, we lose a bit of gpr encoding when there's an embedded
    // immediate.
-   if (regs.restrictedGPR16Range && f == FILE_GPR && isShortRegVal(lval))
+   if (regs.restrictedGPR16Range && f == FILE_GPR && (lval->reg.size == 2 || isShortRegVal(lval)))
       size /= 2;
    degreeLimit = size;
    degreeLimit -= relDegree[1][colors] - 1;
-- 
2.18.1