Skip to contentSkip to conversation

iio: buffer: hw-consumer: fix use-after-free in error path

PATCH Applied Felix Gu 6 participants 9 messages Apr 30 – May 5 IIO

lore
Share & tools
lore

Felix Gu

author
#0 lore raw

PATCH v2 1/1 iio: buffer: hw-consumer: fix use-after-free in error path

In the err_put_buffers cleanup path of iio_hw_consumer_alloc(), the code was using list_for_each_entry() to iterate through buffers while calling iio_buffer_put() which can free the current buffer if refcount drops to 0. The list_for_each_entry() loop macro then evaluates buf->head.next to continue iteration, accessing the freed buffer.

Fix this by using list_for_each_entry_safe().

Fixes 48b66f8f936f ("iio: Add hardware consumer buffer support")

---
Changes in v2:
- Fix Andy's comment.
- Add Reported-by tag.
- Link to v1: https://lore.kernel.org/r/[email protected]

---
 drivers/iio/buffer/industrialio-hw-consumer.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/iio/buffer/industrialio-hw-consumer.c +2 -2
index 24d7df603760..700528c9a0a4 100644
@@ -85,7 +85,7 @@ static struct hw_consumer_buffer *iio_hw_consumer_get_buffer(
  */
 struct iio_hw_consumer *iio_hw_consumer_alloc(struct device *dev)
 {
-	struct hw_consumer_buffer *buf;
+	struct hw_consumer_buffer *buf, *tmp;
 	struct iio_hw_consumer *hwc;
 	struct iio_channel *chan;
 	int ret;
@@ -116,7 +116,7 @@ struct iio_hw_consumer *iio_hw_consumer_alloc(struct device *dev)
 	return hwc;
 
 err_put_buffers:
-	list_for_each_entry(buf, &hwc->buffers, head)
+	list_for_each_entry_safe(buf, tmp, &hwc->buffers, head)
 		iio_buffer_put(&buf->buffer);
 	iio_channel_release_all(hwc->channels);
 err_free_hwc:

---
base-commit: 7080e32d3f09d8688c4a87d81bdcc71f7f606b16
change-id: 20260428-iio-buf-4c8559020a54

Best regards,

Signature · 1 line
Felix Gu <[email protected]>
Reviewed-by on v2 1/1
#1 lore raw
Quoted 3 lines · Felix Gu

In the err_put_buffers cleanup path of iio_hw_consumer_alloc(), the code was using list_for_each_entry() to iterate through buffers while calling iio_buffer_put() which can free the current buffer if refcount drops to 0=

.

Quoted 9 lines

The list_for_each_entry() loop macro then evaluates buf->head.next to continue iteration, accessing the freed buffer.

Fix this by using list_for_each_entry_safe().

Fixes 48b66f8f936f ("iio: Add hardware consumer buffer support")
Quoted 9 lines

---
Changes in v2:
- Fix Andy's comment.
- Add Reported-by tag.
- Link to v1: https://lore.kernel.org/r/[email protected]

---
 drivers/iio/buffer/industrialio-hw-consumer.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

Looks fine to me.

best regards,
maxwell

Quoted 3 lines

Best regards,
--=
Felix Gu <[email protected]>

maintainer Applied
#2 lore raw
Quoted 24 lines · On Thu, 30 Apr 2026 20:27:27 -0500 "Maxwell Doose" <[email protected]> wrote:
Quoted 3 lines · Felix Gu

In the err_put_buffers cleanup path of iio_hw_consumer_alloc(), the code was using list_for_each_entry() to iterate through buffers while calling iio_buffer_put() which can free the current buffer if refcount drops to 0=

.

Quoted 9 lines

The list_for_each_entry() loop macro then evaluates buf->head.next to continue iteration, accessing the freed buffer.

Fix this by using list_for_each_entry_safe().

Fixes 48b66f8f936f ("iio: Add hardware consumer buffer support")
Quoted 9 lines

---
Changes in v2:
- Fix Andy's comment.
- Add Reported-by tag.
- Link to v1: https://lore.kernel.org/r/[email protected]

---
 drivers/iio/buffer/industrialio-hw-consumer.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)
Quoted 6 lines

Looks fine to me.

best regards,
maxwell

Applied to the fixes-togreg branch of iio.git and marked for stable.

Thanks

J

Quoted 7 lines
Quoted 6 lines

Best regards,
--=
Felix Gu <[email protected]>

Reviewed-by on v2 1/1
#3 lore raw
In reply to Felix Gu
Quoted 12 lines · Felix Gu

In the err_put_buffers cleanup path of iio_hw_consumer_alloc(), the code was using list_for_each_entry() to iterate through buffers while calling iio_buffer_put() which can free the current buffer if refcount drops to 0. The list_for_each_entry() loop macro then evaluates buf->head.next to continue iteration, accessing the freed buffer.

Fix this by using list_for_each_entry_safe().

Fixes 48b66f8f936f ("iio: Add hardware consumer buffer support")

Closes:
https://sashiko.dev/#/patchset/20260427-iio_buf-v1-1-2bbdac844647%40gmail.com

Quoted 8 lines

---
Changes in v2:
- Fix Andy's comment.
- Add Reported-by tag.
- Link to v1:
https://lore.kernel.org/r/[email protected]
---

LGTM

Quoted 33 lines

 drivers/iio/buffer/industrialio-hw-consumer.c | 4 ++--

 1 file changed, 2 insertions(+), 2 deletions(-)
drivers/iio/buffer/industrialio-hw-consumer.c +1 -1
index 24d7df603760..700528c9a0a4 100644
@@ -85,7 +85,7 @@ static struct hw_consumer_buffer *iio_hw_consumer_get_buffer(
   */
  struct iio_hw_consumer *iio_hw_consumer_alloc(struct device *dev)
  {
-	struct hw_consumer_buffer *buf;
+	struct hw_consumer_buffer *buf, *tmp;
  	struct iio_hw_consumer *hwc;
  	struct iio_channel *chan;
  	int ret;
@@ -116,7 +116,7 @@ struct iio_hw_consumer *iio_hw_consumer_alloc(struct device
 *dev)
  	return hwc;


 err_put_buffers:

drivers/iio/buffer/industrialio-hw-consumer.c +1 -1
-	list_for_each_entry(buf, &hwc->buffers, head)
+	list_for_each_entry_safe(buf, tmp, &hwc->buffers, head)

  iio_buffer_put(&buf->buffer);
  iio_channel_release_all(hwc->channels);
 err_free_hwc:

---
base-commit: 7080e32d3f09d8688c4a87d81bdcc71f7f606b16
change-id: 20260428-iio-buf-4c8559020a54

Best regards,

#4 lore raw
In reply to Felix Gu
Quoted 7 lines · Felix Gu

In the err_put_buffers cleanup path of iio_hw_consumer_alloc(), the code was using list_for_each_entry() to iterate through buffers while calling iio_buffer_put() which can free the current buffer if refcount drops to 0. The list_for_each_entry() loop macro then evaluates buf->head.next to continue iteration, accessing the freed buffer.

Quoted 5 lines

Fix this by using list_for_each_entry_safe().

Fixes 48b66f8f936f ("iio: Add hardware consumer buffer support")

FYI, Sashiko reported 2 additional regressions (neither introduced by your patches), one of them critical. Maybe another patch would be good?

https://sashiko.dev/#/patchset/20260430-iio-buf-v2-1-84c2231dea5e%40gmail.com?part=1

--
Kind regards

CJD

#5 lore raw
Quoted 10 lines · Joshua Crofts
Quoted 9 lines · Felix Gu

In the err_put_buffers cleanup path of iio_hw_consumer_alloc(), the code was using list_for_each_entry() to iterate through buffers while calling iio_buffer_put() which can free the current buffer if refcount drops to 0. The list_for_each_entry() loop macro then evaluates buf->head.next to continue iteration, accessing the freed buffer.

Fix this by using list_for_each_entry_safe().

Quoted 9 lines
Quoted 3 lines
Fixes 48b66f8f936f ("iio: Add hardware consumer buffer support")

FYI, Sashiko reported 2 additional regressions (neither introduced by your patches), one of them critical. Maybe another patch would be good?

https://sashiko.dev/#/patchset/20260430-iio-buf-v2-1-84c2231dea5e%40gmail.com?part=1

I think one of the issues raised by sashiko has an active patch:

https://lore.kernel.org/linux-iio/[email protected]/

and I think they were waiting for Nuno to review it, since they noted they were on a "wild goose chase". Thus the only issue raised by sashiko that would still need a patch would be the OOB issue, but I haven't looked at its explaination thoroughly to see if it's hallucinating.

Anyways this one should still be good to merge.

best regards,
maxwell

Quoted 4 lines

--
Kind regards

CJD

#6 lore raw

⁷

Quoted 11 lines · Maxwell Doose

I think one of the issues raised by sashiko has an active patch:

https://lore.kernel.org/linux-iio/[email protected]/

and I think they were waiting for Nuno to review it, since they noted they were on a "wild goose chase". Thus the only issue raised by sashiko that would still need a patch would be the OOB issue, but I haven't looked at its explaination thoroughly to see if it's hallucinating.

Anyways this one should still be good to merge.

Yeah this is good to merge, I was only pointing out that Sashiko reported additional errors (unrelated to this patch, it seems to keep finding more issues with the driver).

Signature · 3 lines
Kind regards

CJD
#7 lore raw
Quoted 13 lines · Joshua Crofts

⁷

Quoted 9 lines · Maxwell Doose

I think one of the issues raised by sashiko has an active patch:

https://lore.kernel.org/linux-iio/[email protected]/

and I think they were waiting for Nuno to review it, since they noted they were on a "wild goose chase". Thus the only issue raised by sashiko that would still need a patch would be the OOB issue, but I haven't looked at its explaination thoroughly to see if it's hallucinating.

Quoted 5 lines
Quoted 2 lines

Anyways this one should still be good to merge.

Yeah this is good to merge, I was only pointing out that Sashiko reported additional errors (unrelated to this patch, it seems to keep finding more issues with the driver).

Also, if anybody feels strongly about the OOB issue that sashiko raised in iio_hw_consumer_alloc(), you can probably add something like:

	if (chan->channel->scan_index < 0)
		continue;

but I feel at that point if that even happens there's something else seriously wrong. But it might be a good idea to add that check anyways, better safe than sorry. Not to mention, since this is a public function, we need to code far more defensively.

best regards,
maxwell

Reviewed-by on v2 1/1
#8 lore raw
In reply to Felix Gu
Quoted 7 lines · Felix Gu

In the err_put_buffers cleanup path of iio_hw_consumer_alloc(), the code was using list_for_each_entry() to iterate through buffers while calling iio_buffer_put() which can free the current buffer if refcount drops to 0. The list_for_each_entry() loop macro then evaluates buf->head.next to continue iteration, accessing the freed buffer.

Fix this by using list_for_each_entry_safe().

Signature · 2 lines
With Best Regards,
Andy Shevchenko