Rollup merge of #139700 - EnzymeAD:autodiff-flags, r=oli-obk

Autodiff flags

Interestingly, it seems that some other projects have conflicts with exactly the same LLVM optimization passes as autodiff.
At least `LLVMRustOptimize` has exactly the flags that we need to disable problematic opt passes.

This PR enables us to compile code where users differentiate two identical functions in the same module. This has been especially common in test cases, but it's not impossible to encounter in the wild.

It also enables two new flags for testing/debugging. I consider writing an MCP to upgrade PrintPasses to be a standalone -Z flag, since it is *not* the same as `-Z print-llvm-passes`, which IMHO gives less useful output. A discussion can be found here: [#t-compiler/llvm > Print llvm passes. @ 💬](https://rust-lang.zulipchat.com/#narrow/channel/187780-t-compiler.2Fllvm/topic/Print.20llvm.20passes.2E/near/511533038)

Finally, it improves `PrintModBefore` and `PrintModAfter`. They used to work reliable, but now we just schedule enzyme as part of an existing ModulePassManager (MPM). Since Enzyme is last in the MPM scheduling, PrintModBefore became very inaccurate. It used to print the input module, which we gave to the Enzyme and was great to create llvm-ir reproducer. However, lately the MPM would run the whole `default<O3>` pipeline, which heavily modifies the llvm module, before we pass it to Enzyme. That made it impossible to use the flag to create llvm-ir reproducers for Enzyme bugs. We now schedule a PrintModule pass just before Enzyme, solving this problem.

Based on the PrintPass output, it also _seems_ like changing `registerEnzymeAndPassPipeline(PB, true);` to `registerEnzymeAndPassPipeline(PB, false);` has no effect. In theory, the bool should tell Enzyme to schedule some helpful passes in the PassBuilder. However, since it doesn't do anything and I'm not 100% sure anymore on whether we really need it, I'll just disable it for now and postpone investigations.

r? ``@oli-obk``

closes #139471

Tracking:

- https://github.com/rust-lang/rust/issues/124509

1 files changed, 15 insertions, 1 deletions

diff --git a/compiler/rustc_codegen_llvm/src/back/write.rs b/compiler/rustc_codegen_llvm/src/back/write.rs
index 18d221d232e..4ac77c8f7f1 100644
--- a/compiler/rustc_codegen_llvm/src/back/write.rs
+++ b/compiler/rustc_codegen_llvm/src/back/write.rs
@@ -572,6 +572,10 @@ pub(crate) unsafe fn llvm_optimize(
 
     let consider_ad = cfg!(llvm_enzyme) && config.autodiff.contains(&config::AutoDiff::Enable);
     let run_enzyme = autodiff_stage == AutodiffStage::DuringAD;
+    let print_before_enzyme = config.autodiff.contains(&config::AutoDiff::PrintModBefore);
+    let print_after_enzyme = config.autodiff.contains(&config::AutoDiff::PrintModAfter);
+    let print_passes = config.autodiff.contains(&config::AutoDiff::PrintPasses);
+    let merge_functions;
     let unroll_loops;
     let vectorize_slp;
     let vectorize_loop;
@@ -579,13 +583,20 @@ pub(crate) unsafe fn llvm_optimize(
     // When we build rustc with enzyme/autodiff support, we want to postpone size-increasing
     // optimizations until after differentiation. Our pipeline is thus: (opt + enzyme), (full opt).
     // We therefore have two calls to llvm_optimize, if autodiff is used.
+    //
+    // We also must disable merge_functions, since autodiff placeholder/dummy bodies tend to be
+    // identical. We run opts before AD, so there is a chance that LLVM will merge our dummies.
+    // In that case, we lack some dummy bodies and can't replace them with the real AD code anymore.
+    // We then would need to abort compilation. This was especially common in test cases.
     if consider_ad && autodiff_stage != AutodiffStage::PostAD {
+        merge_functions = false;
         unroll_loops = false;
         vectorize_slp = false;
         vectorize_loop = false;
     } else {
         unroll_loops =
             opt_level != config::OptLevel::Size && opt_level != config::OptLevel::SizeMin;
+        merge_functions = config.merge_functions;
         vectorize_slp = config.vectorize_slp;
         vectorize_loop = config.vectorize_loop;
     }
@@ -663,13 +674,16 @@ pub(crate) unsafe fn llvm_optimize(
             thin_lto_buffer,
             config.emit_thin_lto,
             config.emit_thin_lto_summary,
-            config.merge_functions,
+            merge_functions,
             unroll_loops,
             vectorize_slp,
             vectorize_loop,
             config.no_builtins,
             config.emit_lifetime_markers,
             run_enzyme,
+            print_before_enzyme,
+            print_after_enzyme,
+            print_passes,
             sanitizer_options.as_ref(),
             pgo_gen_path.as_ref().map_or(std::ptr::null(), |s| s.as_ptr()),
             pgo_use_path.as_ref().map_or(std::ptr::null(), |s| s.as_ptr()),