On Wed, Oct 14, 2020 at 02:48:19PM +0100, Cristian Marussi wrote: [...]
> > > > I have pushed a version with above change [1], please check if you are > > happy with that ? > > > > [1] https://git.kernel.org/sudeep.holla/linux/c/b2cd15549b > > I agree with the need to retain _notify name, but I'm not so sure about > the above patch...which is: > I agree, I thought about it and just cooked up this as a quick solution. I will move to that, even I wasn't happy with this TBH. > diff --git a/drivers/firmware/arm_scmi/notify.c > b/drivers/firmware/arm_scmi/notify.c > index c24e427dce0d7..f60fa630dd98f 100644 > --- a/drivers/firmware/arm_scmi/notify.c > +++ b/drivers/firmware/arm_scmi/notify.c > @@ -1461,6 +1461,7 @@ static const struct scmi_notify_ops notify_ops = { > int scmi_notification_init(struct scmi_handle *handle) > { > void *gid; > + char scmi_wq_name[32]; > struct scmi_notify_instance *ni; > > gid = devres_open_group(handle->dev, NULL, GFP_KERNEL); > @@ -1474,7 +1475,8 @@ int scmi_notification_init(struct scmi_handle *handle) > ni->gid = gid; > ni->handle = handle; > > - ni->notify_wq = alloc_workqueue("scmi_notify", > + sprintf(scmi_wq_name, "%s_notify", dev_name(handle->dev)); > + ni->notify_wq = alloc_workqueue(scmi_wq_name, > WQ_UNBOUND | WQ_FREEZABLE | WQ_SYSFS, > 0); > if (!ni->notify_wq) > > ...does not expose a potential buffer overflow given that the dev_name now > comes > from the DT chosen name for this SCMI server instance ? > > I'd use snprintf and enlarge a bit the static scmi_wq_name[] to fit a max > device bane plus "_notify". > > Regards > Cristian -- Regards, Sudeep
