[llvm] [SimplifyCFG] Treat umul + extract pattern as one cheap single instruction (PR #128021)

[Gábor Spaits via llvm-commits]

================
@@ -0,0 +1,129 @@
+; NOTE: Assertions have been autogenerated by utils/update_test_checks.py
+; RUN: opt -passes=simplifycfg -simplifycfg-require-and-preserve-domtree=1 -S < %s | FileCheck %s
+target triple = "riscv64-unknown-unknown-elf"
----------------
spaits wrote:

Yes it is.
If we don't specify a target, the the cost of the umul wit overflow intrinsic will be very low (It is going to be exactly 1. I have experimented with it a bit (see comment: https://github.com/llvm/llvm-project/pull/124933#issuecomment-2653407934)), and the optimization will happen regardless of my patch.

In the other test file (`llvm/test/Transforms/PhaseOrdering/X86/unsigned-multiply-overflow-check.ll`) that is fixed by this patch the target is an x86 cpu. See https://github.com/llvm/llvm-project/pull/128021/files#diff-c283e96547fa4371d4af397270274797e6dada2617eea7c72b14225adc53e5ecL10 .

https://github.com/llvm/llvm-project/pull/128021