[PATCH] D155190: [amdgpu][lds] Remove recalculation of LDS frame from backend

Jon Chesterfield via Phabricator via llvm-commits llvm-commits at lists.llvm.org
Thu Jul 13 13:10:05 PDT 2023


JonChesterfield added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPUMachineFunction.cpp:87
+      if (ObjectStart != alignTo(ObjectStart, Alignment)) {
+        report_fatal_error("Absolute address LDS variable inconsistent with "
+                           "variable alignment");
----------------
arsenm wrote:
> Probably should go through DiagnosticInfo. I don't know why we don't have sensible predefined types for this sort of thing.   DK_Lowering looks promising but is buried in one specific user
Maybe, it's not user facing. Currently the only way to reach this is via a bug in LowerModuleLDS. Could be an assert instead, or just delete the checks. 

Equivalent to the fatal errors in the allocateKnownAddressLDSGlobal which this patch deletes.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D155190/new/

https://reviews.llvm.org/D155190



More information about the llvm-commits mailing list