On Tue, Dec 17, 2024 at 04:54:07PM +0800, Xu Yang wrote:
> On Mon, Dec 16, 2024 at 07:55:40PM +0100, Francesco Dolcini wrote:
> > On Thu, Dec 12, 2024 at 08:24:09PM +0800, Xu Yang wrote:
> > > With edge irq support, the ALERT event may be missed currently. The reason
> > > is that ALERT_MASK register is written before devm_request_threaded_irq().
> > > If ALERT event happens in this time gap, it will be missed and ALERT line
> > > will not recover to high level. However, we can't meet this issue with
> > > level irq. To avoid the issue, this will add a flag set_alert_mask. So
> > > ALERT_MASK can be written after devm_request_threaded_irq() is called. The
> > > behavior of tcpm_init() keeps unchanged.
> > >
> > > Fixes: 77e85107a771 ("usb: typec: tcpci: support edge irq")
> > > Cc: stable@xxxxxxxxxxxxxxx
> > > Signed-off-by: Xu Yang <xu.yang_2@xxxxxxx>
> >
> > I wonder if this should be squashed together with the first commit,
> > given you re-introduce an issue with the previous commit.
>
> No. One patch normally should do one thing. To support edge irq, commit
> 77e85107a771 cause NULL ponter issue so path 1 fix it, it also didn't
> handle irq or alert_mask correctly, then patch 2 is needed.
Sure. And you also want your commit to be bi-sectable, your first patch
introduce a bug to fix another one, and than you fix it in the second one.
In any case, Greg will tell if he wants something different here or not.
Francesco
