Re: [PATCH] remoteproc: sysfs: fix race while updating recovery flag

[Satya Durga Srinivasu Prabhala]

On 1/30/23 12:03 AM, Mukesh Ojha wrote:
> On 1/30/2023 4:21 AM, Satya Durga Srinivasu Prabhala wrote:
> > When multiple clients try to update the recovery flag, it is
>
> Multiple user-space clients ?
Yes, on SMP systems, it is possible that there can be multiple user 
space clients (can simply be fuzzing kind of scripts) which could be 
updating the recovery flag.
> > possible that, race condition would lead to undesired results
> > as updates to recovery flag isn't protected by any mechanism
> > today. To avoid such issues, take remoteproc mutex lock before
> > updating recovery flag and release the lock once done.
>
> But your patch also adds locks for the case which does not update
> recovery flag..
Yes, was trying to cover entire function, can be restricted to only when 
recovery flag is being updated as well.
> > Signed-off-by: Satya Durga Srinivasu Prabhala <quic_satyap@xxxxxxxxxxx>
> > ---
> >   drivers/remoteproc/remoteproc_sysfs.c | 5 +++++
> >   1 file changed, 5 insertions(+)
> >
> > diff --git a/drivers/remoteproc/remoteproc_sysfs.c 
> > b/drivers/remoteproc/remoteproc_sysfs.c
> > index 8c7ea8922638..ec37176e1589 100644
> > --- a/drivers/remoteproc/remoteproc_sysfs.c
> > +++ b/drivers/remoteproc/remoteproc_sysfs.c
> > @@ -48,16 +48,21 @@ static ssize_t recovery_store(struct device *dev,
> >   {
> >       struct rproc *rproc = to_rproc(dev);
> >
> > +    mutex_lock(&rproc->lock);
> >       if (sysfs_streq(buf, "enabled")) {
> >           /* change the flag and begin the recovery process if needed */
> >           rproc->recovery_disabled = false;
> > +        mutex_unlock(&rproc->lock);
> >           rproc_trigger_recovery(rproc);
> >       } else if (sysfs_streq(buf, "disabled")) {
> >           rproc->recovery_disabled = true;
> > +        mutex_unlock(&rproc->lock);
> >       } else if (sysfs_streq(buf, "recover")) {
> >           /* begin the recovery process without changing the flag */
> > +        mutex_unlock(&rproc->lock);
>
> is it really needed for this case?
As mentioned above, was trying to cover entire function. Not really 
needed in this case as such.
> >           rproc_trigger_recovery(rproc);
> >       } else {
> > +        mutex_unlock(&rproc->lock);
>
> same here..
>
> >           return -EINVAL;
> >       }
>
> Do you also need to add lock for rproc_recovery_write in
> drivers/remoteproc/remoteproc_debugfs.c ?
Thanks, yes. Debug FS needs to be updated too.
> -Mukesh