[PATCH] D98266: [AMDGPU] Refactor AMDGPUTargetStreamer::EmitCodeEnd

[Jay Foad via Phabricator via llvm-commits]

This revision was landed with ongoing or failed builds.
This revision was automatically updated to reflect the committed changes.
Closed by commit rG288ea820cf4a: [AMDGPU] Refactor AMDGPUTargetStreamer::EmitCodeEnd (authored by foad).

Changed prior to commit:
  https://reviews.llvm.org/D98266?vs=329353&id=329404#toc

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D98266/new/

https://reviews.llvm.org/D98266

Files:
  llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUTargetStreamer.cpp


Index: llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUTargetStreamer.cpp
===================================================================
--- llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUTargetStreamer.cpp
+++ llvm/lib/Target/AMDGPU/MCTargetDesc/AMDGPUTargetStreamer.cpp
@@ -264,15 +264,21 @@
   const uint32_t Encoded_s_code_end = 0xbf9f0000;
   const uint32_t Encoded_s_nop = 0xbf800000;
   uint32_t Encoded_pad = Encoded_s_code_end;
-  unsigned FillSize = 48;
+
+  // Instruction cache line size in bytes.
+  const unsigned Log2CacheLineSize = 6;
+  const unsigned CacheLineSize = 1u << Log2CacheLineSize;
+
+  // Extra padding amount in bytes to support prefetch mode 3.
+  unsigned FillSize = 3 * CacheLineSize;
 
   if (AMDGPU::isGFX90A(STI)) {
     Encoded_pad = Encoded_s_nop;
-    FillSize = 256;
+    FillSize = 16 * CacheLineSize;
   }
 
-  OS << "\t.p2alignl 6, " << Encoded_pad << '\n';
-  OS << "\t.fill " << FillSize << ", 4, " << Encoded_pad << '\n';
+  OS << "\t.p2alignl " << Log2CacheLineSize << ", " << Encoded_pad << '\n';
+  OS << "\t.fill " << (FillSize / 4) << ", 4, " << Encoded_pad << '\n';
   return true;
 }
 
@@ -641,17 +647,23 @@
   const uint32_t Encoded_s_code_end = 0xbf9f0000;
   const uint32_t Encoded_s_nop = 0xbf800000;
   uint32_t Encoded_pad = Encoded_s_code_end;
-  unsigned FillSize = 48;
+
+  // Instruction cache line size in bytes.
+  const unsigned Log2CacheLineSize = 6;
+  const unsigned CacheLineSize = 1u << Log2CacheLineSize;
+
+  // Extra padding amount in bytes to support prefetch mode 3.
+  unsigned FillSize = 3 * CacheLineSize;
 
   if (AMDGPU::isGFX90A(STI)) {
     Encoded_pad = Encoded_s_nop;
-    FillSize = 256;
+    FillSize = 16 * CacheLineSize;
   }
 
   MCStreamer &OS = getStreamer();
   OS.PushSection();
-  OS.emitValueToAlignment(64, Encoded_pad, 4);
-  for (unsigned I = 0; I < FillSize; ++I)
+  OS.emitValueToAlignment(CacheLineSize, Encoded_pad, 4);
+  for (unsigned I = 0; I < FillSize; I += 4)
     OS.emitInt32(Encoded_pad);
   OS.PopSection();
   return true;


-------------- next part --------------
A non-text attachment was scrubbed...
Name: D98266.329404.patch
Type: text/x-patch
Size: 2031 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20210309/927d3fed/attachment.bin>