maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/clang/test/CodeGen/code-coverage-tsan.c

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

13 lines
558 B
C
Raw Normal View History

Add -fprofile-update={atomic,prefer-atomic,single} GCC 7 introduced -fprofile-update={atomic,prefer-atomic} (prefer-atomic is for best efforts (some targets do not support atomics)) to increment counters atomically, which is exactly what we have done with -fprofile-instr-generate (D50867) and -fprofile-arcs (b5ef137c11b1cc6ae839ee75b49233825772bdd0). This patch adds the option to clang to surface the internal options at driver level. GCC 7 also turned on -fprofile-update=prefer-atomic when -pthread is specified, but it has performance regression (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89307). So we don't follow suit. Differential Revision: https://reviews.llvm.org/D87737
2020-09-30 01:38:51 +08:00
/// -fprofile-update=atomic (implied by -fsanitize=thread) requires the
/// (potentially concurrent) counter updates to be atomic.
[gcov] Unify driver and CC1 option names for -ftest-coverage & -fprofile-arcs No need to use -femit-coverage-notes and -femit-coverage-data.
2020-10-20 13:17:32 +08:00
// RUN: %clang_cc1 %s -triple x86_64 -emit-llvm -fprofile-update=atomic -ftest-coverage -fprofile-arcs \
[gcov] Increment counters with atomicrmw if -fsanitize=thread Without this patch, `clang --coverage -fsanitize=thread` may fail spuriously because non-atomic counter increments can be detected as data races.
2020-08-29 07:32:35 +08:00
// RUN: -coverage-notes-file /dev/null -coverage-data-file /dev/null -o - | FileCheck %s
// CHECK-LABEL: void @foo()
/// Two counters are incremented by __tsan_atomic64_fetch_add.
Add -fprofile-update={atomic,prefer-atomic,single} GCC 7 introduced -fprofile-update={atomic,prefer-atomic} (prefer-atomic is for best efforts (some targets do not support atomics)) to increment counters atomically, which is exactly what we have done with -fprofile-instr-generate (D50867) and -fprofile-arcs (b5ef137c11b1cc6ae839ee75b49233825772bdd0). This patch adds the option to clang to surface the internal options at driver level. GCC 7 also turned on -fprofile-update=prefer-atomic when -pthread is specified, but it has performance regression (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=89307). So we don't follow suit. Differential Revision: https://reviews.llvm.org/D87737
2020-09-30 01:38:51 +08:00
// CHECK: atomicrmw add i64* {{.*}} @__llvm_gcov_ctr
// CHECK-NEXT: atomicrmw sub i32*
[gcov] Increment counters with atomicrmw if -fsanitize=thread Without this patch, `clang --coverage -fsanitize=thread` may fail spuriously because non-atomic counter increments can be detected as data races.
2020-08-29 07:32:35 +08:00
_Atomic(int) cnt;
void foo() { cnt--; }