[PATCH] D80434: [AMDGPU] Reject moving PHI to VALU if the only VGPR input originated from move immediate
Stanislav Mekhanoshin via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Tue May 26 10:49:58 PDT 2020
rampitec added inline comments.
================
Comment at: llvm/test/CodeGen/AMDGPU/phi-vgpr-input-moveimm.mir:36
+---
+# GCN_LABEL: bb.2:
+# GCN-NOT: %{{[0-9]+}}:sreg_64 = PHI %{{[0-9]+}}, %bb.3, %{{[0-9]+}}, %bb.1
----------------
Please fix the label to be a function name.
================
Comment at: llvm/test/CodeGen/AMDGPU/phi-vgpr-input-moveimm.mir:69
+---
+# GCN_LABEL: bb.2:
+# GCN-NOT: %{{[0-9]+}}:sreg_32 = PHI %{{[0-9]+}}, %bb.3, %{{[0-9]+}}, %bb.1
----------------
Here too.
================
Comment at: llvm/test/CodeGen/AMDGPU/phi-vgpr-input-moveimm.mir:72
+# GCN: %{{[0-9]+}}:vgpr_32 = PHI %{{[0-9]+}}, %bb.3, %{{[0-9]+}}, %bb.1
+name: phi_moveimm_bad opcode_input
+tracksRegLiveness: true
----------------
Does it compile with a white space in function's name?
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D80434/new/
https://reviews.llvm.org/D80434
More information about the llvm-commits
mailing list