[llvm] [BOLT] Skip FDE emission for patch functions (PR #136224)

via llvm-commits llvm-commits at lists.llvm.org
Thu Apr 17 16:31:18 PDT 2025


llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-bolt

Author: Maksim Panchenko (maksfb)

<details>
<summary>Changes</summary>

Patch functions are used to fix instructions in the original code, i.e., they are not functions in a traditional sense, but rather pieces of emitted code that are embedded into real functions.

We used to emit FDEs for all functions, including patch functions. However, FDEs for patches are not only unnecessary, but they can lead to problems with libraries and runtimes that consume FDEs, e.g. C++ exception handling runtime.

---
Full diff: https://github.com/llvm/llvm-project/pull/136224.diff


2 Files Affected:

- (modified) bolt/lib/Core/BinaryEmitter.cpp (+2-2) 
- (modified) bolt/test/AArch64/lite-mode.s (+7) 


``````````diff
diff --git a/bolt/lib/Core/BinaryEmitter.cpp b/bolt/lib/Core/BinaryEmitter.cpp
index d3ed9e8088bfa..3414fc938719b 100644
--- a/bolt/lib/Core/BinaryEmitter.cpp
+++ b/bolt/lib/Core/BinaryEmitter.cpp
@@ -374,7 +374,7 @@ bool BinaryEmitter::emitFunction(BinaryFunction &Function,
   }
 
   // Emit CFI start
-  if (Function.hasCFI()) {
+  if (Function.hasCFI() && !Function.isPatch()) {
     Streamer.emitCFIStartProc(/*IsSimple=*/false);
     if (Function.getPersonalityFunction() != nullptr)
       Streamer.emitCFIPersonality(Function.getPersonalityFunction(),
@@ -421,7 +421,7 @@ bool BinaryEmitter::emitFunction(BinaryFunction &Function,
     Streamer.emitBytes(BC.MIB->getTrapFillValue());
 
   // Emit CFI end
-  if (Function.hasCFI())
+  if (Function.hasCFI() && !Function.isPatch())
     Streamer.emitCFIEndProc();
 
   MCSymbol *EndSymbol = Function.getFunctionEndLabel(FF.getFragmentNum());
diff --git a/bolt/test/AArch64/lite-mode.s b/bolt/test/AArch64/lite-mode.s
index d81206089aaef..a71edbe034669 100644
--- a/bolt/test/AArch64/lite-mode.s
+++ b/bolt/test/AArch64/lite-mode.s
@@ -11,6 +11,13 @@
 # RUN: llvm-objdump -d --disassemble-symbols=cold_function %t.bolt \
 # RUN:   | FileCheck %s
 
+
+## Verify that the number of FDEs matches the number of functions in the output
+## binary. There are three original functions and two optimized.
+# RUN: llvm-readelf -u %t.bolt | grep -wc FDE \
+# RUN:   | FileCheck --check-prefix=CHECK-FDE %s
+# CHECK-FDE: 5
+
 ## In lite mode, optimized code will be separated from the original .text by
 ## over 128MB, making it impossible for call/bl instructions in cold functions
 ## to reach optimized functions directly.

``````````

</details>


https://github.com/llvm/llvm-project/pull/136224


More information about the llvm-commits mailing list