linereq_free() can be called from in the middle of errors, where the
descriptor may be NULL for few lines. Don't access uninitialized
descriptor pointer as it leads to kernel crash:
Unable to handle kernel NULL pointer dereference at virtual address 0000000000000008
[...]
Call trace:
linereq_free+0x54/0xb8
linereq_create+0x424/0x570
gpio_ioctl+0x94/0x640
__arm64_sys_ioctl+0xac/0xf0
invoke_syscall+0x44/0x100
el0_svc_common.constprop.3+0x6c/0xf0
do_el0_svc+0x2c/0xb8
el0_svc+0x20/0x60
el0t_64_sync_handler+0x98/0xc0
el0t_64_sync+0x170/0x174
Fixes: 2068339a6c35 ("gpiolib: cdev: Add hardware timestamp clock type")
Signed-off-by: Viresh Kumar <[email protected]>
---
drivers/gpio/gpiolib-cdev.c | 8 +++++---
1 file changed, 5 insertions(+), 3 deletions(-)
diff --git a/drivers/gpio/gpiolib-cdev.c b/drivers/gpio/gpiolib-cdev.c
index f5aa5f93342a..d3d1b5aed282 100644
--- a/drivers/gpio/gpiolib-cdev.c
+++ b/drivers/gpio/gpiolib-cdev.c
@@ -1460,11 +1460,13 @@ static ssize_t linereq_read(struct file *file,
static void linereq_free(struct linereq *lr)
{
unsigned int i;
- bool hte;
+ bool hte = false;
for (i = 0; i < lr->num_lines; i++) {
- hte = !!test_bit(FLAG_EVENT_CLOCK_HTE,
- &lr->lines[i].desc->flags);
+ if (lr->lines[i].desc) {
+ hte = !!test_bit(FLAG_EVENT_CLOCK_HTE,
+ &lr->lines[i].desc->flags);
+ }
edge_detector_stop(&lr->lines[i], hte);
if (lr->lines[i].desc)
gpiod_free(lr->lines[i].desc);
--
2.31.1.272.g89b43f80a514
On Fri, Jul 8, 2022 at 1:25 PM Viresh Kumar <[email protected]> wrote:
>
> linereq_free() can be called from in the middle of errors, where the
> descriptor may be NULL for few lines. Don't access uninitialized
> descriptor pointer as it leads to kernel crash:
>
> Unable to handle kernel NULL pointer dereference at virtual address 0000000000000008
>
> [...]
>
> Call trace:
> linereq_free+0x54/0xb8
> linereq_create+0x424/0x570
> gpio_ioctl+0x94/0x640
> __arm64_sys_ioctl+0xac/0xf0
> invoke_syscall+0x44/0x100
> el0_svc_common.constprop.3+0x6c/0xf0
> do_el0_svc+0x2c/0xb8
> el0_svc+0x20/0x60
> el0t_64_sync_handler+0x98/0xc0
> el0t_64_sync+0x170/0x174
>
> Fixes: 2068339a6c35 ("gpiolib: cdev: Add hardware timestamp clock type")
> Signed-off-by: Viresh Kumar <[email protected]>
> ---
> drivers/gpio/gpiolib-cdev.c | 8 +++++---
> 1 file changed, 5 insertions(+), 3 deletions(-)
>
> diff --git a/drivers/gpio/gpiolib-cdev.c b/drivers/gpio/gpiolib-cdev.c
> index f5aa5f93342a..d3d1b5aed282 100644
> --- a/drivers/gpio/gpiolib-cdev.c
> +++ b/drivers/gpio/gpiolib-cdev.c
> @@ -1460,11 +1460,13 @@ static ssize_t linereq_read(struct file *file,
> static void linereq_free(struct linereq *lr)
> {
> unsigned int i;
> - bool hte;
> + bool hte = false;
>
> for (i = 0; i < lr->num_lines; i++) {
> - hte = !!test_bit(FLAG_EVENT_CLOCK_HTE,
> - &lr->lines[i].desc->flags);
> + if (lr->lines[i].desc) {
> + hte = !!test_bit(FLAG_EVENT_CLOCK_HTE,
> + &lr->lines[i].desc->flags);
> + }
> edge_detector_stop(&lr->lines[i], hte);
> if (lr->lines[i].desc)
> gpiod_free(lr->lines[i].desc);
> --
> 2.31.1.272.g89b43f80a514
>
Hey Viresh!
Kent beat you to it with commit c8e27a4a5136e7230f9e4ffcf132705bf56864cc.
Bart
On Fri, Jul 8, 2022 at 3:32 PM Bartosz Golaszewski <[email protected]> wrote:
> On Fri, Jul 8, 2022 at 1:25 PM Viresh Kumar <[email protected]> wrote:
...
> Kent beat you to it with commit c8e27a4a5136e7230f9e4ffcf132705bf56864cc.
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=c8e27a4a5136e7230f9e4ffcf132705bf56864cc
Notice: this object is not reachable from any branch.
Rebased?
--
With Best Regards,
Andy Shevchenko
On Fri, Jul 8, 2022 at 5:14 PM Andy Shevchenko
<[email protected]> wrote:
>
> On Fri, Jul 8, 2022 at 3:32 PM Bartosz Golaszewski <[email protected]> wrote:
> > On Fri, Jul 8, 2022 at 1:25 PM Viresh Kumar <[email protected]> wrote:
>
> ...
>
> > Kent beat you to it with commit c8e27a4a5136e7230f9e4ffcf132705bf56864cc.
>
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=c8e27a4a5136e7230f9e4ffcf132705bf56864cc
> Notice: this object is not reachable from any branch.
>
> Rebased?
>
> --
> With Best Regards,
> Andy Shevchenko
It's in next and I just sent it out to Torvalds too.
Bart