Rebase and address comments.

duplicate existing diff

Don't mangle ParmVarDecl.

rebase and address comment.

@rsmith Are you okay with this patch?

Add 19.20 version. Only add mangled type for version 19. 20 or later to be abi-compatible with https://godbolt.org/z/Bhc__A

• Only add mangled type for auto template parameter, whose argument kind is Integral.
• MSVC mangled name has 1 more letter M before the mangled type, but I don't do it here because I think the M is from mangled qualifier and the function mangleQualifiers is incomplete yet.

landed

update test case.

rename parameter uses to um

Check LambdaContextDecl is not ParamVarDecl.

Update test cases.

Remove check for context of LambdaContextDecl.
Update test cases.

In D79895#2042992, @xbolva00 wrote:

Can you provide some compile time data with warning enabled/disabled?

Fix typo.
Diagnose self-init warning if the self-init variable is used as const reference later.

Change some CHECK to CHECK-NEXT where the call to bar is first instruction in blocks.

Update comment for NoMerge flag

Ping

Since the new warning is controlled by -Wuninitialized, I disabled it in existing test case and added a separate test case for -Wuninitialized-const-reference.

Use another mapvector constRefUses to store uninitialized const reference uses so that the new warning can be easily disable by -Wno-initialized-const-reference

Add new warning as subgroup of Uninitialized.

In D79537#2035041, @rnk wrote:

After this lands, can you follow this up by migrating the sanitizer passes to using the attribute instead of using empty inline asm blobs? Send the patch to @vitalybuka. This is the code I'm thinking of:
https://github.com/llvm/llvm-project/blob/028bfdd8913616f7a3e57e8ef5c2a9990e528ff0/llvm/lib/Transforms/Instrumentation/AddressSanitizer.cpp#L1626
https://github.com/llvm/llvm-project/blob/d6695e18763a05b30cb336c18157175277da8f4b/llvm/lib/Transforms/Instrumentation/MemorySanitizer.cpp#L864
... and any similar uses.

Run check-asan, check-msan, etc to run the sanitizer integration tests, and confirm that they pass. You can run check-all to run everything and be sure as well.

Zero Initialize NoMerge in CallSiteDbgInfo.
Put Check in addNoMergeSiteInfo to avoid unnecessary hash table insertion.

Handle nomerge flag in LowerCall at targets AArch64, ARM, PowerPC, RISCV, and add test cases for each target.

add nomerge case in CodeExtractor.cpp

use ConstEvaluatedExprVisitor to recursively looking up for call expr

update comment.

use ConstStmtVisitor to do recursive lookup for call expression.

add testcase

To keep it simple, nomerge attribute should not be applied to decl statement. Since label statement calls ProcessDeclAttributeList to handle attributes, label statement should not have nomerge attribute.

Add Sema test.
Check if nomerge attribute has no arguments and statement contains call expressions.

change nomerge to statement attribute

Add test case for preventing hoisting.

Does sinking and hoisting remove the debug source location? I assumed that it wouldn't, but now after all the smooth stepping work, maybe it does.

Yes, both sinking and hoisting in simplifycfg remove the debug source location by Instruction::applyMergedLocation
sinking: https://github.com/llvm/llvm-project/blob/master/llvm/lib/Transforms/Utils/SimplifyCFG.cpp#L1646
hoisting: https://github.com/llvm/llvm-project/blob/master/llvm/lib/Transforms/Utils/SimplifyCFG.cpp#L1338

Split the the original differential to 2 smaller ones. This one add nomerge function attribute in IR level. Another one will add the attribute to let frontend recognize it.

Add check in LICM to prevent sink or hoist on nomerge call

In D78659#1997528, @lebedev.ri wrote:

1. Tests missing
2. Why isn't noinline sufficient, why a whole new attribue?