lkdtm/bugs: Adjust lkdtm_HUNG_TASK() to avoid tail call optimization
authorDouglas Anderson <dianders@chromium.org>
Tue, 23 Jan 2024 00:49:35 +0000 (16:49 -0800)
committerKees Cook <keescook@chromium.org>
Thu, 1 Feb 2024 17:44:07 +0000 (09:44 -0800)
When testing with lkdtm_HUNG_TASK() and looking at the output, I
expected to see lkdtm_HUNG_TASK() in the stack crawl but it wasn't
there. Instead, the top function on at least some devices was
schedule() due to tail call optimization.

Let's do two things to help here:
1. We'll mark this as "__noreturn". On GCC at least this is documented
   to prevent tail call optimization. The docs [1] say "In order to
   preserve backtraces, GCC will never turn calls to noreturn
   functions into tail calls."
2. We'll add a BUG_ON(1) at the end which means that schedule() is no
   longer a tail call. Note that this is potentially important because
   if we _did_ end up returning from schedule() due to some weird
   issue then we'd potentially be violating the "noreturn" that we
   told the compiler about. BUG is the right thing to do here.

[1] https://gcc.gnu.org/onlinedocs/gcc/Common-Function-Attributes.html

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20240122164935.2.I26e8f68c312824fcc80c19d4e91de2d2bef958f0@changeid
Signed-off-by: Kees Cook <keescook@chromium.org>
drivers/misc/lkdtm/bugs.c

index b080eb2..d1222d3 100644 (file)
@@ -294,10 +294,11 @@ static void lkdtm_SPINLOCKUP(void)
        __release(&lock_me_up);
 }
 
-static void lkdtm_HUNG_TASK(void)
+static void __noreturn lkdtm_HUNG_TASK(void)
 {
        set_current_state(TASK_UNINTERRUPTIBLE);
        schedule();
+       BUG_ON(1);
 }
 
 static volatile unsigned int huge = INT_MAX - 2;