[llvm] CodeGenPrepare: Remove threadlocal_address intrinsic when cheap to recompute. (PR #87844)

[Phoebe Wang via llvm-commits]

================
@@ -0,0 +1,149 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
+; RUN: llc -o - %s | FileCheck %s
+; RUN: llc -o - -relocation-model=pic %s | FileCheck --check-prefix=CHECK_PIC %s
+target triple = "x86_64--linux-gnu"
+
+ at foo = dso_local thread_local global i32 0, align 4
+ at foo_nonlocal = thread_local global i32 0, align 4
----------------
phoebewang wrote:

Not used?

https://github.com/llvm/llvm-project/pull/87844