[clang] [llvm] [X86] Add Support for X86 TLSDESC Relocations (PR #83136)

[Roland McGrath via cfe-commits]

================
@@ -0,0 +1,165 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
+; RUN: llc < %s -mtriple=i686-unknown-unknown --relocation-model=pic -enable-tlsdesc | FileCheck %s --check-prefix=X86
+; RUN: llc < %s -mtriple=x86_64-pc-linux-gnux32 --relocation-model=pic -enable-tlsdesc | FileCheck %s --check-prefix=X32
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown --relocation-model=pic -enable-tlsdesc | FileCheck %s --check-prefix=X64
+
+ at x = thread_local global i32 0, align 4
+ at y = internal thread_local global i32 0, align 4
----------------
frobtech wrote:

That example allows the variables to be optimized out entirely because there's no way they can ever be modified. https://godbolt.org/z/63oz4xrj1 shows an example where it's generating LD model accesses.

Note that in some cases I've seen GCC generate a local anchor symbol instead of using `_TLS_MODULE_BASE_`. That is less optimal than using `_TLS_MODULE_BASE_`, because multiple uses of `_TLS_MODULE_BASE_` across multiple TUs can be resolved by the linker to a single GOT slot (pair), whereas a local anchor symbol will always have its own GOT slot (pair) that won't be shared with any other TU.

https://github.com/llvm/llvm-project/pull/83136