[AMDGPU] Add flag to prevent reruns of LowerModuleLDS (#129520)

FullLTO has to run this early before module splitting occurs otherwise
module splitting won't work as expected. There was a targeted fix for
fortran on another branch that disables the LTO run but that'd break
full LTO module splitting entirely.

Test changes are due to metadata indexes shifting.

See #122891

1 files changed, 6 insertions, 0 deletions

diff --git a/llvm/lib/Target/AMDGPU/AMDGPULowerModuleLDSPass.cpp b/llvm/lib/Target/AMDGPU/AMDGPULowerModuleLDSPass.cpp
index 34ba53c..f79069b 100644
--- a/llvm/lib/Target/AMDGPU/AMDGPULowerModuleLDSPass.cpp
+++ b/llvm/lib/Target/AMDGPU/AMDGPULowerModuleLDSPass.cpp
@@ -1038,6 +1038,12 @@ public:
   }
 
   bool runOnModule(Module &M) {
+    // Check if we've already lowered this module. The pass may run more
+    // than once in the LTO pipeline, and multiple runs aren't supported.
+    if (M.getModuleFlag("amdgpu.lowered_lds"))
+      return false;
+    M.addModuleFlag(Module::ModFlagBehavior::Error, "amdgpu.lowered_lds", 1);
+
     CallGraph CG = CallGraph(M);
     bool Changed = superAlignLDSGlobals(M);