Compiler-rt changes to react to skiping promotable allocas.

[zaks.anna]

Summary

This is the compiler-rt part of the following change: "Skip promotable allocas to improve performance at -O0" http://reviews.llvm.org/D7741.

[kubamracek]

I've submitted http://reviews.llvm.org/D7809 and http://reviews.llvm.org/D7810 which properly fixes these tests (the failures are not related to the "promotable allocas" patch, but they fail on higher -O levels as well).

[samsonov]

Thanks for looking into this, Kuba!

[kubamracek]

I believe this is now unnecessary, as both http://reviews.llvm.org/D7809 and http://reviews.llvm.org/D7810 have landed.

test/asan/TestCases/alloca_instruments_all_paddings.cc

-// RUN: %clangxx_asan -O0 -mllvm -asan-instrument-allocas %s -o %t
+// RUN: %clangxx_asan -O0 -mllvm -asan-instrument-allocas -mllvm -asan-skip-promotable-allocas=0 %s -o %t
 // RUN: %run %t 2>&1
 //
 
 #include "sanitizer/asan_interface.h"
 #include <assert.h>
 
 __attribute__((noinline)) void foo(int index, int len) {
   volatile char str[len] __attribute__((aligned(32)));

test/asan/TestCases/gc-test.cc

-// RUN: %clangxx_asan %s -pthread -o %t
+// RUN: %clangxx_asan -O0 -mllvm -asan-skip-promotable-allocas=0 %s -pthread -o %t
 // RUN: env ASAN_OPTIONS=detect_stack_use_after_return=1 %run %t 2>&1 | FileCheck %s --check-prefix=CHECK1
 // RUN: env ASAN_OPTIONS=detect_stack_use_after_return=0 %run %t 2>&1 | FileCheck %s --check-prefix=CHECK0
 // REQUIRES: stable-runtime
 
 #include <assert.h>
 #include <stdio.h>
 #include <pthread.h>
 #include <sanitizer/asan_interface.h>