* [PATCH RESEND] gpiolib: hold gpio devices lock until ->descs array is initialised
@ 2020-01-21 0:12 Dan Callaghan
2020-01-23 15:26 ` Linus Walleij
0 siblings, 1 reply; 2+ messages in thread
From: Dan Callaghan @ 2020-01-21 0:12 UTC (permalink / raw)
To: Linus Walleij, Bartosz Golaszewski, open list:GPIO SUBSYSTEM, open list
Cc: Dan Callaghan
If a driver consuming the GPIO chip is being probed at the same time as
the GPIO driver is registering the chip, it is possible for the
consuming driver to see the ->descs array in an uninitialised state.
For example, the gpio-keys-polled driver can fail like this:
kernel: gpiod_request: invalid GPIO (no device)
kernel: gpio-keys-polled PRP0001:07: failed to get gpio: -22
kernel: gpio-keys-polled: probe of PRP0001:07 failed with error -22
This patch makes gpiochip_add() hold the lock protecting gpio_devices
until it has finished setting desc->gdev on the newly inserted list
entry.
Signed-off-by: Dan Callaghan <dan.callaghan@opengear.com>
---
Resending this one because I failed to cc the maintainers on the
original posting, sorry about that.
drivers/gpio/gpiolib.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/drivers/gpio/gpiolib.c b/drivers/gpio/gpiolib.c
index 78a16e42f222..33c72fb12bb5 100644
--- a/drivers/gpio/gpiolib.c
+++ b/drivers/gpio/gpiolib.c
@@ -1495,11 +1495,11 @@ int gpiochip_add_data_with_key(struct gpio_chip *chip, void *data,
goto err_free_label;
}
- spin_unlock_irqrestore(&gpio_lock, flags);
-
for (i = 0; i < chip->ngpio; i++)
gdev->descs[i].gdev = gdev;
+ spin_unlock_irqrestore(&gpio_lock, flags);
+
#ifdef CONFIG_PINCTRL
INIT_LIST_HEAD(&gdev->pin_ranges);
#endif
--
2.21.0
^ permalink raw reply related [flat|nested] 2+ messages in thread
* Re: [PATCH RESEND] gpiolib: hold gpio devices lock until ->descs array is initialised
2020-01-21 0:12 [PATCH RESEND] gpiolib: hold gpio devices lock until ->descs array is initialised Dan Callaghan
@ 2020-01-23 15:26 ` Linus Walleij
0 siblings, 0 replies; 2+ messages in thread
From: Linus Walleij @ 2020-01-23 15:26 UTC (permalink / raw)
To: Dan Callaghan; +Cc: Bartosz Golaszewski, open list:GPIO SUBSYSTEM, open list
On Tue, Jan 21, 2020 at 1:13 AM Dan Callaghan
<dan.callaghan@opengear.com> wrote:
> If a driver consuming the GPIO chip is being probed at the same time as
> the GPIO driver is registering the chip, it is possible for the
> consuming driver to see the ->descs array in an uninitialised state.
> For example, the gpio-keys-polled driver can fail like this:
>
> kernel: gpiod_request: invalid GPIO (no device)
> kernel: gpio-keys-polled PRP0001:07: failed to get gpio: -22
> kernel: gpio-keys-polled: probe of PRP0001:07 failed with error -22
>
> This patch makes gpiochip_add() hold the lock protecting gpio_devices
> until it has finished setting desc->gdev on the newly inserted list
> entry.
>
> Signed-off-by: Dan Callaghan <dan.callaghan@opengear.com>
> ---
> Resending this one because I failed to cc the maintainers on the
> original posting, sorry about that.
This makes a lot of sense, I'm impressed that you managed to
provoke this error!
Patch applied.
Yours,
Linus Walleij
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2020-01-23 15:26 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-01-21 0:12 [PATCH RESEND] gpiolib: hold gpio devices lock until ->descs array is initialised Dan Callaghan
2020-01-23 15:26 ` Linus Walleij
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).