On Mon, 6 Dec 2021 at 18:40, kernel test robot <lkp@xxxxxxxxx> wrote: [...] > kernel/kcsan/kcsan_test.c: In function 'test_barrier_nothreads': > >> kernel/kcsan/kcsan_test.c:592:35: error: implicit declaration of function 'clear_bit_unlock_is_negative_byte' [-Werror=implicit-function-declaration] > 592 | KCSAN_EXPECT_READ_BARRIER(clear_bit_unlock_is_negative_byte(0, &test_var), true); > | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > kernel/kcsan/kcsan_test.c:537:17: note: in definition of macro '__KCSAN_EXPECT_BARRIER' > 537 | barrier; \ > | ^~~~~~~ > kernel/kcsan/kcsan_test.c:592:9: note: in expansion of macro 'KCSAN_EXPECT_READ_BARRIER' > 592 | KCSAN_EXPECT_READ_BARRIER(clear_bit_unlock_is_negative_byte(0, &test_var), true); > | ^~~~~~~~~~~~~~~~~~~~~~~~~ > kernel/kcsan/kcsan_test.c: In function 'test_atomic_builtins': > kernel/kcsan/kcsan_test.c:1233:17: warning: 'atomic_thread_fence' is not supported with '-fsanitize=thread' [-Wtsan] > 1233 | __atomic_thread_fence(__ATOMIC_SEQ_CST); > | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > cc1: some warnings being treated as errors FTR, like the other issue, fixed by same patch: https://lore.kernel.org/all/20211206064151.3337384-2-elver@xxxxxxxxxx/ But I'm glad to see the test robot discovered this, too. Thanks, -- Marco
