Re: [PATCH RT 10/30] hrtimer: Prevent using hrtimer_grab_expiry_lock() on migration_base

From: Rasmus Villemoes
Date: Tue Apr 28 2020 - 02:51:48 EST


On 27/04/2020 21.26, Tom Zanussi wrote:
> On Mon, 2020-04-27 at 15:06 -0400, Steven Rostedt wrote:
>> On Mon, 27 Apr 2020 15:10:00 +0200
>> Rasmus Villemoes <rasmus.villemoes@xxxxxxxxx> wrote:
>>
>>> Reverting
>>>
>>> b1a471ec4df1 - hrtimer: Prevent using hrtimer_grab_expiry_lock() on
>>> migration_base
>>> 40aae5708e7a - hrtimer: Add a missing bracket and hide
>>> `migration_base'
>>> on !SMP
>>>
>>> on top of v4.19.94-rt39 makes that problem go away, i.e. the board
>>> reboots as expected.
>>>
>> Thanks Rasmus for looking into this. Tom now maintains 4.19-rt.
>>
>> Tom, care to pull in these patches on top of 4.19-rt?
>>
>
> Those patches are already in 4.19-rt - he's saying that reverting them
> fixes the problem.
>
> I'm guessing that the assumption of base or base->cpu_base always being
> non-NULL in those patches might be wrong. If so, the below patch
> should fix the problem:
>
> Subject: [PATCH] hrtimer: Add back base and base->cpu_base checks in
> hrtimer_grab_expiry_lock()
>
> 4.19 commit b1a471ec4df1 [hrtimer: Prevent using
> hrtimer_grab_expiry_lock() on migration_base] removed the NULL checks
> for timer->base and timer->base->cpu_base on the assumption that
> they're always non-NULL. That assumption is apparently not to be
> true, so add the checks back.
>
> Signed-off-by: Tom Zanussi <zanussi@xxxxxxxxxx>
> ---
> kernel/time/hrtimer.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/kernel/time/hrtimer.c b/kernel/time/hrtimer.c
> index e54a95de8b79..6f20cf23008b 100644
> --- a/kernel/time/hrtimer.c
> +++ b/kernel/time/hrtimer.c
> @@ -953,7 +953,7 @@ void hrtimer_grab_expiry_lock(const struct hrtimer *timer)
> {
> struct hrtimer_clock_base *base = READ_ONCE(timer->base);
>
> - if (timer->is_soft && is_migration_base(base)) {
> + if (timer->is_soft && base && base->cpu_base && is_migration_base(base)) {

I'm sorry, but no, I don't think that can be it. For !SMP (my case),
is_migration_base() is always false, so with or without the above, the
whole if() is false. Also, the symptoms do not look like a NULL pointer
deref, but more like a dead (or live) lock - so I'm guessing (and that's
just a wild guess) that the lock/unlock sequence is needed to give some
other thread a priority boost to make the whole machine make forward
progress.

Rasmus