[lld] Change --lto-emit-llvm to use the pre-codegen module

[jhuber6]

Summary:
Currently the --lto-emit-llvm option writes out the
post-internalization bitcode. This is the bitcode before any
optimizations or other pipelines have been run on it. This patch changes
that to use the pre-codegen module, which is the state of the LLVM-IR
after the optimizations have been run.

I believe that this makes sense as the --lto-emit-llvm option seems to
imply that we should emit the final output of the LLVM pass as if it
were the desired output. This should include optimizations at the
requested optimization level. My main motivation for this change is to
be able to use this to link several LLVM-IR files into a single one that
I can then pass back to ld.lld later (for JIT purposes).

[llvmbot]

@llvm/pr-subscribers-lld

@llvm/pr-subscribers-lld-elf

Author: Joseph Huber (jhuber6)

Changes

Summary:
Currently the --lto-emit-llvm option writes out the
post-internalization bitcode. This is the bitcode before any
optimizations or other pipelines have been run on it. This patch changes
that to use the pre-codegen module, which is the state of the LLVM-IR
after the optimizations have been run.

I believe that this makes sense as the --lto-emit-llvm option seems to
imply that we should emit the final output of the LLVM pass as if it
were the desired output. This should include optimizations at the
requested optimization level. My main motivation for this change is to
be able to use this to link several LLVM-IR files into a single one that
I can then pass back to ld.lld later (for JIT purposes).

---

Full diff: https://github.com/llvm/llvm-project/pull/97480.diff

2 Files Affected:

• (modified) lld/ELF/LTO.cpp (+1-1)
• (modified) lld/test/ELF/lto/emit-llvm.ll (+4-2)

diff --git a/lld/ELF/LTO.cpp b/lld/ELF/LTO.cpp
index 3d92007469263..935d0a9eab9ee 100644
--- a/lld/ELF/LTO.cpp
+++ b/lld/ELF/LTO.cpp
@@ -147,7 +147,7 @@ static lto::Config createConfig() {
   c.PGOWarnMismatch = config->ltoPGOWarnMismatch;
 
   if (config->emitLLVM) {
-    c.PostInternalizeModuleHook = [](size_t task, const Module &m) {
+    c.PreCodeGenModuleHook = [](size_t task, const Module &m) {
       if (std::unique_ptr<raw_fd_ostream> os =
               openLTOOutputFile(config->outputFile))
         WriteBitcodeToFile(m, *os, false);
diff --git a/lld/test/ELF/lto/emit-llvm.ll b/lld/test/ELF/lto/emit-llvm.ll
index 01f5a056e0c0d..37488016a4bc2 100644
--- a/lld/test/ELF/lto/emit-llvm.ll
+++ b/lld/test/ELF/lto/emit-llvm.ll
@@ -9,11 +9,13 @@
 ; RUN: ld.lld --plugin-opt=emit-llvm -mllvm -bitcode-flush-threshold=0 -o /dev/null %t.o
 ; RUN: ld.lld --lto-emit-llvm -mllvm -bitcode-flush-threshold=0 -o /dev/null %t.o
 
-; CHECK: define internal void @main()
+; CHECK: define hidden void @main()
 
 target datalayout = "e-m:e-p270:32:32-p271:32:32-p272:64:64-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64-unknown-linux-gnu"
 
-define void @main() {
+@llvm.compiler.used = appending global [1 x ptr] [ptr @main], section "llvm.metadata"
+
+define hidden void @main() {
   ret void
 }

[MaskRay]

PreCodeGenModuleHook makes sense as that is similar to clang cc1 -emit-llvm and -emit-llvm-bc for ThinLTO backend compiles.

[MaskRay]

The test doesn't change for PostInternal / PreCodegen. Use a different one? Perhaps two modules are needed to show a difference.

[jhuber6]

It optimizes it out completely, so I just made it used so it sticks around. Could make it slightly more complicated if you want. But, I think the test is mostly just to show "Yes it outputs something"