From: Rafael Espindola <rafael.espindola@gmail.com>
Date: Thu, 16 Jan 2014 04:50:58 +0000 (+0000)
Subject: Prevent calls to __jit_debug_register_code from being optimized out.
X-Git-Url: http://demsky.eecs.uci.edu/git/?a=commitdiff_plain;h=59c2fe1af1bc2b3107140668c9237aef9d3ad392;p=oota-llvm.git

Prevent calls to __jit_debug_register_code from being optimized out.

Patch by Andrew MacPherson. I just tweaked the comment.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@199350 91177308-0d34-0410-b5e6-96231b3b80d8
---

diff --git a/lib/ExecutionEngine/RuntimeDyld/GDBRegistrar.cpp b/lib/ExecutionEngine/RuntimeDyld/GDBRegistrar.cpp
index 603c526d06e..60adc9b383d 100644
--- a/lib/ExecutionEngine/RuntimeDyld/GDBRegistrar.cpp
+++ b/lib/ExecutionEngine/RuntimeDyld/GDBRegistrar.cpp
@@ -47,7 +47,11 @@ extern "C" {
   struct jit_descriptor __jit_debug_descriptor = { 1, 0, 0, 0 };
 
   // Debuggers puts a breakpoint in this function.
-  LLVM_ATTRIBUTE_NOINLINE void __jit_debug_register_code() { }
+  LLVM_ATTRIBUTE_NOINLINE void __jit_debug_register_code() {
+    // The noinline and the asm prevent calls to this function from being
+    // optimized out.
+    asm volatile("":::"memory");
+  }
 
 }