On Fri, Aug 18, 2023 at 02:59:13PM +0800, Chengming Zhou wrote:
> Hi,
>
> On 2023/8/18 09:52, Ming Lei wrote:
> > group_cpus_evenly() could be part of storage driver's error handler,
> > such as nvme driver, when may happen during CPU hotplug, in which
> > storage queue has to drain its pending IOs because all CPUs associated
> > with the queue are offline and the queue is becoming inactive. And
> > handling IO needs error handler to provide forward progress.
> >
> > Then dead lock is caused:
> >
> > 1) inside CPU hotplug handler, CPU hotplug lock is held, and blk-mq's
> > handler is waiting for inflight IO
> >
> > 2) error handler is waiting for CPU hotplug lock
> >
> > 3) inflight IO can't be completed in blk-mq's CPU hotplug handler because
> > error handling can't provide forward progress.
> >
> > Solve the deadlock by not holding CPU hotplug lock in group_cpus_evenly(),
> > in which two stage spreads are taken: 1) the 1st stage is over all present
> > CPUs; 2) the end stage is over all other CPUs.
> >
> > Turns out the two stage spread just needs consistent 'cpu_present_mask', and
> > remove the CPU hotplug lock by storing it into one local cache. This way
> > doesn't change correctness, because all CPUs are still covered.
> >
> > Cc: Keith Busch <kbusch@xxxxxxxxxx>
> > Cc: linux-nvme@xxxxxxxxxxxxxxxxxxx
> > Cc: linux-block@xxxxxxxxxxxxxxx
> > Reported-by: Yi Zhang <yi.zhang@xxxxxxxxxx>
> > Reported-by: Guangwu Zhang <guazhang@xxxxxxxxxx>
> > Tested-by: Guangwu Zhang <guazhang@xxxxxxxxxx>
> > Signed-off-by: Ming Lei <ming.lei@xxxxxxxxxx>
> > ---
> > V2:
> > - fix "Cc: block list"
> > - add tested-by tag
> >
> > lib/group_cpus.c | 22 ++++++++++++++++------
> > 1 file changed, 16 insertions(+), 6 deletions(-)
> >
> > diff --git a/lib/group_cpus.c b/lib/group_cpus.c
> > index aa3f6815bb12..15006e79196f 100644
> > --- a/lib/group_cpus.c
> > +++ b/lib/group_cpus.c
> > @@ -348,6 +348,7 @@ struct cpumask *group_cpus_evenly(unsigned int numgrps)
> > {
> > unsigned int curgrp = 0, nr_present = 0, nr_others = 0;
> > cpumask_var_t *node_to_cpumask;
> > + cpumask_var_t local_cpu_present_mask;
> > cpumask_var_t nmsk, npresmsk;
> > int ret = -ENOMEM;
> > struct cpumask *masks = NULL;
> > @@ -355,6 +356,16 @@ struct cpumask *group_cpus_evenly(unsigned int numgrps)
> > if (!zalloc_cpumask_var(&nmsk, GFP_KERNEL))
> > return NULL;
> >
> > + if (!zalloc_cpumask_var(&local_cpu_present_mask, GFP_KERNEL))
> > + goto fail_local_pres_mask;
> > +
> > + /*
> > + * Make a local cache of 'cpu_present_mask', so the two stages
> > + * spread can observe consistent 'cpu_present_mask' without holding
> > + * cpu hotplug lock.
> > + */
> > + cpumask_copy(local_cpu_present_mask, cpu_present_mask);
> > +
>
> Maybe we can reuse npresmsk instead of allocating another cpumask?
> In the first stage: npresmsk = cpu_present_mask
> In the second stage: npresmsk = cpu_possible_mask & ~npresmsk
Good idea!
Thanks,
Ming
