[compiler-rt] [llvm] [msan] Add 32-bit platforms support (PR #109284)

[via llvm-commits]

================
@@ -0,0 +1,91 @@
+; RUN: opt < %s -S -passes=msan 2>&1 -msan-origin-base=0x40000000 -msan-and-mask=0x80000000 | FileCheck %s
+
+target datalayout = "E-m:m-i8:8:32-i16:16:32-i32:32-n32-S64"
+target triple = "arm--linux"
+
+define i32 @foo(i32 %guard, ...) {
----------------
k-kashapov wrote:

Second step is also unclear:
If we just switch the target triple to 32bit, most of the checks will be invalid. Thus, "update tests with invalid instrumentation" would mean update all tests.
Also, since this PR is the last, MemorySanitizer will not be updated, so when running tests between 2nd and 3rd PR's the tests will fail with "unsupported architecture" assert.
I believe, this is not the intended course of action. Could you please, elaborate on the steps?

https://github.com/llvm/llvm-project/pull/109284