[llvm] [AArch64] Don't tail call memset if it would convert to a bzero. (PR #98969)

[Eli Friedman via llvm-commits]

================
@@ -685,7 +687,10 @@ bool llvm::returnTypeIsEligibleForTailCall(const Function *F,
          (IID == Intrinsic::memmove &&
           TLI.getLibcallName(RTLIB::MEMMOVE) == StringRef("memmove")) ||
          (IID == Intrinsic::memset &&
-          TLI.getLibcallName(RTLIB::MEMSET) == StringRef("memset"))) &&
+          TLI.getLibcallName(RTLIB::MEMSET) == StringRef("memset") &&
+          (!isa<ConstantInt>(Call->getOperand(1)) ||
+           !cast<ConstantInt>(Call->getOperand(1))->isZero() ||
+           !TLI.getLibcallName(RTLIB::BZERO)))) &&
----------------
efriedma-quic wrote:

This is sort of ugly... if this code doesn't precisely match SelectionDAG::getMemset, we miscompile.

Can we rearrange the code so that the caller of isInTailCallPosition() does this computation?  That could be directly adjacent to the code in getMemset that actually makes the decision of whether to call memset or bzero.  We then just pass that into isInTailCallPosition() as a argument.

https://github.com/llvm/llvm-project/pull/98969