On Jan 12, 2023 / 22:18, Shin'ichiro Kawasaki wrote:
> On Jan 12, 2023 / 19:53, Yu Kuai wrote:
> > Hi,
> >
> > 在 2023/01/12 19:47, Yu Kuai 写道:
> > > Thanks for the report, is the problem easy to reporduce? If so, can you
> > > try the following patch?
> > >
> > > diff --git a/block/bfq-cgroup.c b/block/bfq-cgroup.c
> > > index 1b2829e99dad..81d2f401fa3f 100644
> > > --- a/block/bfq-cgroup.c
> > > +++ b/block/bfq-cgroup.c
> > > @@ -771,8 +771,8 @@ static void __bfq_bic_change_cgroup(struct bfq_data
> > > *bfqd,
> > > * request from the old cgroup.
> > > */
> > > bfq_put_cooperator(sync_bfqq);
> > > - bfq_release_process_ref(bfqd, sync_bfqq);
> > > bic_set_bfqq(bic, NULL, true);
> > > + bfq_release_process_ref(bfqd, sync_bfqq);
> > > }
> > > }
> > > }
> > >
> > Just in case you hit the problem in another place, please using the
> > following patch:
> >
> > diff --git a/block/bfq-cgroup.c b/block/bfq-cgroup.c
> > index 1b2829e99dad..81d2f401fa3f 100644
> > --- a/block/bfq-cgroup.c
> > +++ b/block/bfq-cgroup.c
> > @@ -771,8 +771,8 @@ static void __bfq_bic_change_cgroup(struct bfq_data
> > *bfqd,
> > * request from the old cgroup.
> > */
> > bfq_put_cooperator(sync_bfqq);
> > - bfq_release_process_ref(bfqd, sync_bfqq);
> > bic_set_bfqq(bic, NULL, true);
> > + bfq_release_process_ref(bfqd, sync_bfqq);
> > }
> > }
> > }
> > diff --git a/block/bfq-iosched.c b/block/bfq-iosched.c
> > index 16f43bbc575a..687285612e57 100644
> > --- a/block/bfq-iosched.c
> > +++ b/block/bfq-iosched.c
> > @@ -5425,9 +5425,10 @@ static void bfq_check_ioprio_change(struct bfq_io_cq
> > *bic, struct bio *bio)
> >
> > bfqq = bic_to_bfqq(bic, false);
> > if (bfqq) {
> > - bfq_release_process_ref(bfqd, bfqq);
> > + struct bfq_queue *old_bfqq = bfqq;
> > bfqq = bfq_get_queue(bfqd, bio, false, bic, true);
> > bic_set_bfqq(bic, bfqq, false);
> > + bfq_release_process_ref(bfqd, old_bfqq);
> > }
> >
> > bfqq = bic_to_bfqq(bic, true);
> >
>
> Ah, I've just noticed this e-mail. Will test this patch tomorrow.
This second trial patch also avoided the KASAN uaf message. I repeated the
system boot and ssh login 6 times and did not observe the failure. Looks good.
--
Shin'ichiro Kawasaki
