Commit 852d9679 authored Mar 17, 2018 by Andy Lutomirski Committed by Greg Kroah-Hartman Mar 28, 2018

selftests/x86/ptrace_syscall: Fix for yet more glibc interference



commit 4b0b37d4cc54b21a6ecad7271cbc850555869c62 upstream.

glibc keeps getting cleverer, and my version now turns raise() into
more than one syscall.  Since the test relies on ptrace seeing an
exact set of syscalls, this breaks the test.  Replace raise(SIGSTOP)
with syscall(SYS_tgkill, ...) to force glibc to get out of our way.

Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: linux-kselftest@vger.kernel.org
Cc: stable@vger.kernel.org
Link: http://lkml.kernel.org/r/bc80338b453afa187bc5f895bd8e2c8d6e264da2.1521300271.git.luto@kernel.org


Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

parent 5a29ce2b

tools/testing/selftests/x86/ptrace_syscall.c

+6 −2

Original line number	Diff line number	Diff line
		@@ -183,8 +183,10 @@ static void test_ptrace_syscall_restart(void)
		if (ptrace(PTRACE_TRACEME, 0, 0, 0) != 0)
		err(1, "PTRACE_TRACEME");

		pid_t pid = getpid(), tid = syscall(SYS_gettid);

		printf("\tChild will make one syscall\n");
		raise(SIGSTOP);
		syscall(SYS_tgkill, pid, tid, SIGSTOP);

		syscall(SYS_gettid, 10, 11, 12, 13, 14, 15);
		_exit(0);
		@@ -301,9 +303,11 @@ static void test_restart_under_ptrace(void)
		if (ptrace(PTRACE_TRACEME, 0, 0, 0) != 0)
		err(1, "PTRACE_TRACEME");

		pid_t pid = getpid(), tid = syscall(SYS_gettid);

		printf("\tChild will take a nap until signaled\n");
		setsigign(SIGUSR1, SA_RESTART);
		raise(SIGSTOP);
		syscall(SYS_tgkill, pid, tid, SIGSTOP);

		syscall(SYS_pause, 0, 0, 0, 0, 0, 0);
		_exit(0);