With CONFIG_CFI_CLANG, a callback function passed to
__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 != delayed_work_timer_fn);
Use WARN_ON_FUNCTION_MISMATCH() instead to disable the warning
when CFI and modules are both enabled.
Signed-off-by: Sami Tolvanen <[email protected]>
---
kernel/workqueue.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/kernel/workqueue.c b/kernel/workqueue.c
index 0d150da252e8..03fe07d2f39f 100644
--- a/kernel/workqueue.c
+++ b/kernel/workqueue.c
@@ -1630,7 +1630,7 @@ static void __queue_delayed_work(int cpu, struct workqueue_struct *wq,
struct work_struct *work = &dwork->work;
WARN_ON_ONCE(!wq);
- WARN_ON_ONCE(timer->function != delayed_work_timer_fn);
+ WARN_ON_FUNCTION_MISMATCH(timer->function, delayed_work_timer_fn);
WARN_ON_ONCE(timer_pending(timer));
WARN_ON_ONCE(!list_empty(&work->entry));
--
2.31.0.291.g576ba9dcdaf-goog
On Thu, Mar 18, 2021 at 10:11 AM Sami Tolvanen <[email protected]> wrote:
>
> With CONFIG_CFI_CLANG, a callback function passed to
> __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 != delayed_work_timer_fn);
>
> Use WARN_ON_FUNCTION_MISMATCH() instead to disable the warning
> when CFI and modules are both enabled.
Does __cficanonical help with such comparisons? Or would that be a
very invasive change, if the concern was to try to keep these checks
in place for CONFIG_CFI_CLANG?
>
> Signed-off-by: Sami Tolvanen <[email protected]>
> ---
> kernel/workqueue.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/kernel/workqueue.c b/kernel/workqueue.c
> index 0d150da252e8..03fe07d2f39f 100644
> --- a/kernel/workqueue.c
> +++ b/kernel/workqueue.c
> @@ -1630,7 +1630,7 @@ static void __queue_delayed_work(int cpu, struct workqueue_struct *wq,
> struct work_struct *work = &dwork->work;
>
> WARN_ON_ONCE(!wq);
> - WARN_ON_ONCE(timer->function != delayed_work_timer_fn);
> + WARN_ON_FUNCTION_MISMATCH(timer->function, delayed_work_timer_fn);
> WARN_ON_ONCE(timer_pending(timer));
> WARN_ON_ONCE(!list_empty(&work->entry));
>
> --
> 2.31.0.291.g576ba9dcdaf-goog
>
--
Thanks,
~Nick Desaulniers
On Thu, Mar 18, 2021 at 11:50 AM Nick Desaulniers
<[email protected]> wrote:
>
> On Thu, Mar 18, 2021 at 10:11 AM Sami Tolvanen <[email protected]> wrote:
> >
> > With CONFIG_CFI_CLANG, a callback function passed to
> > __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 != delayed_work_timer_fn);
> >
> > Use WARN_ON_FUNCTION_MISMATCH() instead to disable the warning
> > when CFI and modules are both enabled.
>
> Does __cficanonical help with such comparisons? Or would that be a
> very invasive change, if the concern was to try to keep these checks
> in place for CONFIG_CFI_CLANG?
The last time I checked, Clang ignored the __cficanonical attribute in
header files, which means it would still generate a local jump table
entry in each module for such functions, and the comparison here would
fail. We could avoid the issue by using __cficanonical for the
callback function *and* using __va_function() when we take the
function address in modules, but that feels way too invasive for this
particular use case.
Sami