Hi Alexander,
alex.aring@xxxxxxxxx wrote on Sun, 23 Jan 2022 18:14:12 -0500:
> Hi,
>
> On Sun, 23 Jan 2022 at 17:41, Alexander Aring <alex.aring@xxxxxxxxx> wrote:
> >
> > Hi,
> >
> > On Sun, 23 Jan 2022 at 15:43, Alexander Aring <alex.aring@xxxxxxxxx> wrote:
> > >
> > > Hi,
> > >
> > > On Thu, 20 Jan 2022 at 06:21, Miquel Raynal <miquel.raynal@xxxxxxxxxxx> wrote:
> > > >
> > > > Upon error the ieee802154_xmit_complete() helper is not called. Only
> > > > ieee802154_wake_queue() is called manually. We then leak the skb
> > > > structure.
> > > >
> > > > Free the skb structure upon error before returning.
> > > >
> > > > There is no Fixes tag applying here, many changes have been made on this
> > > > area and the issue kind of always existed.
> > > >
> > > > Signed-off-by: Miquel Raynal <miquel.raynal@xxxxxxxxxxx>
> > > > ---
> > > > drivers/net/ieee802154/at86rf230.c | 1 +
> > > > 1 file changed, 1 insertion(+)
> > > >
> > > > diff --git a/drivers/net/ieee802154/at86rf230.c b/drivers/net/ieee802154/at86rf230.c
> > > > index 7d67f41387f5..0746150f78cf 100644
> > > > --- a/drivers/net/ieee802154/at86rf230.c
> > > > +++ b/drivers/net/ieee802154/at86rf230.c
> > > > @@ -344,6 +344,7 @@ at86rf230_async_error_recover_complete(void *context)
> > > > kfree(ctx);
> > > >
> > > > ieee802154_wake_queue(lp->hw);
> > > > + dev_kfree_skb_any(lp->tx_skb);
> > >
> > > as I said in other mails there is more broken, we need a:
> > >
> > > if (lp->is_tx) {
> > > ieee802154_wake_queue(lp->hw);
> > > dev_kfree_skb_any(lp->tx_skb);
> > > lp->is_tx = 0;
> > > }
> > >
> > > in at86rf230_async_error_recover().
> > >
> > s/at86rf230_async_error_recover/at86rf230_async_error_recover_complete/
> >
> > move the is_tx = 0 out of at86rf230_async_error_recover().
>
> Sorry, still seeing an issue here.
>
> We cannot move is_tx = 0 out of at86rf230_async_error_recover()
> because switching to RX_AACK_ON races with a new interrupt and is_tx
> is not correct anymore. We need something new like "was_tx" to
> remember that it was a tx case for the error handling in
> at86rf230_async_error_recover_complete().
It wasn't easy to catch...
I've added a was_tx boolean which is set at the same time is_tx is
reset. Then, in the complete handler, if was_tx was set we reset it and
run the kfree/wake calls. I believe this should sort it out.
Thanks,
Miquèl
