[PATCH] R600/SI: Fix config value for number of gprs

Tom Stellard thomas.stellard at amd.com
Fri Sep 19 11:33:37 PDT 2014


In r217636, the value stored in KernelInfo.Num[VS]GPRSs was changed from
the highest GPR index used to the number of gprs in order to be
consistent with the name of the variable.

The code writing the config values still assumed that the value in this
variable was the highest GPR index used, which caused the compiler to
over report the number of GPRs being used.

https://bugs.freedesktop.org/show_bug.cgi?id=84089
---
 lib/Target/R600/AMDGPUAsmPrinter.cpp | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/lib/Target/R600/AMDGPUAsmPrinter.cpp b/lib/Target/R600/AMDGPUAsmPrinter.cpp
index 2755af2..b736afc 100644
--- a/lib/Target/R600/AMDGPUAsmPrinter.cpp
+++ b/lib/Target/R600/AMDGPUAsmPrinter.cpp
@@ -389,12 +389,15 @@ void AMDGPUAsmPrinter::EmitProgramInfoSI(const MachineFunction &MF,
     RoundUpToAlignment(KernelInfo.ScratchSize * STM.getWavefrontSize(),
                        1 << ScratchAlignShift) >> ScratchAlignShift;
 
+  unsigned VGPRBlocks = (KernelInfo.NumVGPR - 1) / 4;
+  unsigned SGPRBlocks = (KernelInfo.NumSGPR - 1) / 8;
+
   if (MFI->getShaderType() == ShaderType::COMPUTE) {
     OutStreamer.EmitIntValue(R_00B848_COMPUTE_PGM_RSRC1, 4);
 
     const uint32_t ComputePGMRSrc1 =
-      S_00B848_VGPRS(KernelInfo.NumVGPR / 4) |
-      S_00B848_SGPRS(KernelInfo.NumSGPR / 8) |
+      S_00B848_VGPRS(VGPRBlocks) |
+      S_00B848_SGPRS(SGPRBlocks) |
       S_00B848_PRIORITY(KernelInfo.Priority) |
       S_00B848_FLOAT_MODE(KernelInfo.FloatMode) |
       S_00B848_PRIV(KernelInfo.Priv) |
@@ -418,8 +421,8 @@ void AMDGPUAsmPrinter::EmitProgramInfoSI(const MachineFunction &MF,
     // 0" comment but I don't see a corresponding field in the register spec.
   } else {
     OutStreamer.EmitIntValue(RsrcReg, 4);
-    OutStreamer.EmitIntValue(S_00B028_VGPRS(KernelInfo.NumVGPR / 4) |
-                             S_00B028_SGPRS(KernelInfo.NumSGPR / 8), 4);
+    OutStreamer.EmitIntValue(S_00B028_VGPRS(VGPRBlocks) |
+                             S_00B028_SGPRS(SGPRBlocks), 4);
   }
 
   if (MFI->getShaderType() == ShaderType::PIXEL) {
-- 
1.8.5.5




More information about the llvm-commits mailing list