[PATCH] D82562: Implement AVX ABI Warning/error
Craig Topper via Phabricator via cfe-commits
cfe-commits at lists.llvm.org
Sun Jun 28 16:06:27 PDT 2020
craig.topper added inline comments.
================
Comment at: clang/include/clang/Basic/DiagnosticFrontendKinds.td:246
+ "enabled changes the ABI">,
+ InGroup<DiagGroup<"avx-abi">>;
+def err_avx_calling_convention : Error<warn_avx_calling_convention.Text>;
----------------
Should this be -Wpsabi to match gcc? I think that's what gcc called it.
================
Comment at: clang/lib/CodeGen/CGCall.cpp:4248
const Decl *TargetDecl = Callee.getAbstractInfo().getCalleeDecl().getDecl();
- if (const FunctionDecl *FD = dyn_cast_or_null<FunctionDecl>(TargetDecl))
+ if (const FunctionDecl *FD = dyn_cast_or_null<FunctionDecl>(TargetDecl)) {
// We can only guarantee that a function is called from the correct
----------------
Wow that was a big comment for no curly's.
================
Comment at: clang/lib/CodeGen/CGCall.cpp:4259
+ // Some architectures (such as x86_64) have the ABI changed based on
+ // attribute-target/features. Give them a chance to diagnose.
----------------
nit should probably be x86-64. I think we only use underscore when we're already using - for something else like in the triple.
================
Comment at: clang/lib/CodeGen/TargetInfo.cpp:2493
+ SourceLocation CallLoc,
+ llvm::StringMap<bool> &CallerMap,
+ llvm::StringMap<bool> &CalleeMap, QualType Ty,
----------------
Should these maps be const? Which means you have to use find to do the lookups.
================
Comment at: clang/lib/CodeGen/TargetInfo.cpp:2554
+ if (checkAVXParam(CGM.getDiags(), CGM.getContext(), CallLoc, CallerMap,
+ CalleeMap, Ty, /*isArgument*/ true))
+ return;
----------------
isArgument->IsArgument
================
Comment at: clang/lib/CodeGen/TargetInfo.cpp:2567
+ CalleeMap, Callee->getReturnType(),
+ /*isArgument*/ false);
+ }
----------------
Same here
Repository:
rC Clang
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D82562/new/
https://reviews.llvm.org/D82562
More information about the cfe-commits
mailing list