On Thu, Mar 11, 2021 at 10:13 PM Christoph Hellwig [off-list ref] wrote:

On Thu, Mar 11, 2021 at 04:49:08PM -0800, Sami Tolvanen wrote:

quoted

With CONFIG_CFI_CLANG, a callback function passed to
__kthread_queue_delayed_work from a module points to a jump table
entry defined in the module instead of the one used in the core
kernel, which breaks function address equality in this check:

  WARN_ON_ONCE(timer->function != kthread_delayed_work_timer_fn);

Disable the warning when CFI and modules are enabled.

Signed-off-by: Sami Tolvanen <samitolvanen@google.com>
---
 kernel/kthread.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/kernel/kthread.c b/kernel/kthread.c
index 1578973c5740..af5fee350586 100644
--- a/kernel/kthread.c
+++ b/kernel/kthread.c

@@ -963,7 +963,13 @@ static void __kthread_queue_delayed_work(struct kthread_worker *worker,
      struct timer_list *timer = &dwork->timer;
      struct kthread_work *work = &dwork->work;

-     WARN_ON_ONCE(timer->function != kthread_delayed_work_timer_fn);
+     /*
+      * With CFI, timer->function can point to a jump table entry in a module,

you keep spewing this comment line that has exactly 81 characters and
thus badly messes up read it with a normal termina everywhere.

Maybe instead of fixing that in ever duplication hide the whole check in
a well documented helper (which would have to be a macro due to the
typing involved).

Sure, that sounds cleaner. I'll add a helper macro in v2.

Sami