Commit aa1f1a639621 ("lib/syscall: Pin the task stack in collect_syscall()") added logic to handle a process stack not existing, but left sp and pc uninitialized, which can be later reported via /proc/$pid/syscall for zombie processes, potentially exposing kernel memory to userspace.
Zombie /proc/$pid/syscall before: -1 0xffffffff9a060100 0xffff92f42d6ad900 Zombie /proc/$pid/syscall after: -1 0x0 0x0 Reported-by: Robert Święcki <rob...@swiecki.net> Fixes: aa1f1a639621 ("lib/syscall: Pin the task stack in collect_syscall()") Cc: sta...@vger.kernel.org # v4.9+ Signed-off-by: Kees Cook <keesc...@chromium.org> --- lib/syscall.c | 1 + 1 file changed, 1 insertion(+) diff --git a/lib/syscall.c b/lib/syscall.c index 17d5ff5fa6a3..2c6cd1b5c3ea 100644 --- a/lib/syscall.c +++ b/lib/syscall.c @@ -12,6 +12,7 @@ static int collect_syscall(struct task_struct *target, long *callno, if (!try_get_task_stack(target)) { /* Task has no stack, so the task isn't in a syscall. */ + *sp = *pc = 0; *callno = -1; return 0; } -- 2.7.4 -- Kees Cook Pixel Security