[clang] [llvm] [CIR][AMDGPU] Lower cir.offload.printf to AMDGPU printf (PR #231563)

Steffen Larsen via cfe-commits cfe-commits at lists.llvm.org
Fri Oct 30 03:04:00 PDT 2026


================
@@ -3232,7 +3232,7 @@ RValue CIRGenFunction::emitBuiltinExpr(const GlobalDecl &gd, unsigned builtinID,
       if ((getTarget().getTriple().isAMDGCN() ||
            getTarget().getTriple().isSPIRV()) &&
           getLangOpts().HIP)
-        return errorBuiltinNYI(*this, e, builtinID);
+        return RValue::get(emitDevicePrintfCallExpr(e));
----------------
steffenlarsen wrote:

Even better, all paths inside the outer condition should go through this path, so I've simplified it.

https://github.com/llvm/llvm-project/pull/231563


More information about the cfe-commits mailing list