Re: [PATCH REPOST] mfd: dln2: use irqsave() in USB's complete callback

[Johan Hovold]

On Mon, Jul 02, 2018 at 12:07:22AM +0200, Octavian Purdila wrote:
> On Sun, Jul 1, 2018 at 5:41 PM, Sebastian Andrzej Siewior
> <bigeasy@xxxxxxxxxxxxx> wrote:
> > The USB completion callback does not disable interrupts while acquiring
> > the lock. We want to remove the local_irq_disable() invocation from
> > __usb_hcd_giveback_urb() and therefore it is required for the callback
> > handler to disable the interrupts while acquiring the lock.
> > The callback may be invoked either in IRQ or BH context depending on the
> > USB host controller.
> > Use the _irqsave() variant of the locking primitives.
> >
> > Cc: Lee Jones <lee.jones@xxxxxxxxxx>
> > Cc: Octavian Purdila <octavian.purdila@xxxxxxxxx>
> > Signed-off-by: Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx>
> > ---
> >  drivers/mfd/dln2.c | 5 +++--
> >  1 file changed, 3 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/mfd/dln2.c b/drivers/mfd/dln2.c
> > index 704e189ca162..1ef3c92f32e0 100644
> > --- a/drivers/mfd/dln2.c
> > +++ b/drivers/mfd/dln2.c
> > @@ -194,6 +194,7 @@ static bool dln2_transfer_complete(struct dln2_dev *dln2, struct urb *urb,
> >         struct device *dev = &dln2->interface->dev;
> >         struct dln2_mod_rx_slots *rxs = &dln2->mod_rx_slots[handle];
> >         struct dln2_rx_context *rxc;
> > +       unsigned long flags;
> >         bool valid_slot = false;
> >
> >         if (rx_slot >= DLN2_MAX_RX_SLOTS)
> > @@ -206,13 +207,13 @@ static bool dln2_transfer_complete(struct dln2_dev *dln2, struct urb *urb,
> >          * context elsewhere in this driver. This function (or its callers) are
> >          * also not exported to other modules.
> >          */
> > -       spin_lock(&rxs->lock);
> > +       spin_lock_irqsave(&rxs->lock, flags);
> 
> 
> I don't think disabling the IRQ is necessary, please see the comment
> above and also this discussion:
> 
> https://www.spinics.net/lists/linux-usb/msg115214.html

It hasn't been necessary so far, but we want to get rid of the disabling
of interrupts before calling the completion handler so this patch is
correct.

The comment above the lock is not correct as it stands, and definitely
is not after the conversion, so Sebastian, could you please drop it as
part of this change?

With that you can add my:

Reviewed-by: Johan Hovold <johan@xxxxxxxxxx>

Thanks,
Johan