[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <YfQZT3NokrgF5OQA@smile.fi.intel.com>
Date: Fri, 28 Jan 2022 18:26:55 +0200
From: Andy Shevchenko <andriy.shevchenko@...ux.intel.com>
To: Florian Fainelli <f.fainelli@...il.com>
Cc: linux-kernel@...r.kernel.org,
Linus Walleij <linus.walleij@...aro.org>,
Ray Jui <rjui@...adcom.com>,
Scott Branden <sbranden@...adcom.com>,
"maintainer:BROADCOM BCM281XX/BCM11XXX/BCM216XX ARM ARCHITE..."
<bcm-kernel-feedback-list@...adcom.com>,
Nicolas Saenz Julienne <nsaenz@...nel.org>,
Phil Elwell <phil@...pberrypi.com>,
Krzysztof Kozlowski <krzysztof.kozlowski@...onical.com>,
Hans Verkuil <hverkuil-cisco@...all.nl>,
Jason Wang <wangborong@...rlc.com>,
Marc Zyngier <maz@...nel.org>,
"open list:PIN CONTROL SUBSYSTEM" <linux-gpio@...r.kernel.org>,
"moderated list:BROADCOM BCM2711/BCM2835 ARM ARCHITECTURE"
<linux-rpi-kernel@...ts.infradead.org>,
"moderated list:BROADCOM BCM2711/BCM2835 ARM ARCHITECTURE"
<linux-arm-kernel@...ts.infradead.org>
Subject: Re: [PATCH] pinctrl: bcm2835: Fix a few error paths
On Fri, Jan 28, 2022 at 08:12:02AM -0800, Florian Fainelli wrote:
> On 1/28/2022 6:35 AM, Andy Shevchenko wrote:
> > On Thu, Jan 27, 2022 at 01:50:31PM -0800, Florian Fainelli wrote:
> > > After commit 266423e60ea1 ("pinctrl: bcm2835: Change init order for gpio
> > > hogs") a few error paths would not unwind properly the registration of
> > > gpio ranges. Correct that by assigning a single error label and goto it
> > > whenever we encounter a fatal error.
> >
> > > 1 file changed, 15 insertions(+), 8 deletions(-)
> >
> > While this seems legit per se, my eyes caught this:
> >
> >
> > > if (!girq->parents) {
> > > - pinctrl_remove_gpio_range(pc->pctl_dev, &pc->gpio_range);
> > > - return -ENOMEM;
> > > + err = -ENOMEM;
> > > + goto out_remove;
> >
> > Non-devm....
> >
> > > }
> > > if (is_7211) {
> > > pc->wake_irq = devm_kcalloc(dev, BCM2835_NUM_IRQS,
> > > sizeof(*pc->wake_irq),
> > > GFP_KERNEL);
> >
> > ...followed by devm.
> >
> > It means more ordering bugs in the ->remove() and error path are lurking
> > around. Can you double check and be sure that we do not have a case where
> > non-devm registration code followed by devm?
>
> It seems to me like we are fine with the patch as is, because:
>
> - girq->parents is allocated with devm
> - pc->wake_irq is allocated with devm
> - name is allocated with devm
>
> and those are the only variables being allocated for which we also process
> an error handling path.
Okay, thanks.
My worries that it might be the case when the GPIO ranges have been removed by
explicit call in ->remove() followed by some interrupt or so and oops or
misbehaviour because of that.
--
With Best Regards,
Andy Shevchenko
Powered by blists - more mailing lists