Skip to contentSkip to conversation

octeon_ep: fixes for error and remove paths

PATCH Applied Michal Schmidt 3 participants 10 messages Aug 10, 2023 – Aug 15, 2023

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

cover octeon_ep: fixes for error and remove paths

I have an Octeon card that's misconfigured in a way that exposes a couple of bugs in the octeon_ep driver's error paths. It can reproduce the issues that patches 1 & 4 are fixing. Patches 2 & 3 are a result of reviewing the nearby code.

Michal Schmidt (4):
octeon_ep: fix timeout value for waiting on mbox response
octeon_ep: cancel tx_timeout_task later in remove sequence
octeon_ep: cancel ctrl_mbox_task after intr_poll_task
octeon_ep: cancel queued works in probe error path

 drivers/net/ethernet/marvell/octeon_ep/octep_ctrl_net.c | 2 +-
 drivers/net/ethernet/marvell/octeon_ep/octep_main.c     | 9 +++++----
 2 files changed, 6 insertions(+), 5 deletions(-)
Signature · 1 line
2.41.0
#1 lore raw

PATCH 1/4 octeon_ep: fix timeout value for waiting on mbox response

The intention was to wait up to 500 ms for the mbox response. The third argument to wait_event_interruptible_timeout() is supposed to be the timeout duration. The driver mistakenly passed absolute time instead.

Fixes 577f0d1b1c5f ("octeon_ep: add separate mailbox command and response queues")
---
 drivers/net/ethernet/marvell/octeon_ep/octep_ctrl_net.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/marvell/octeon_ep/octep_ctrl_net.c +1 -1
index 1cc6af2feb38..565320ec24f8 100644
@@ -55,7 +55,7 @@ static int octep_send_mbox_req(struct octep_device *oct,
 	list_add_tail(&d->list, &oct->ctrl_req_wait_list);
 	ret = wait_event_interruptible_timeout(oct->ctrl_req_wait_q,
 					       (d->done != 0),
-					       jiffies + msecs_to_jiffies(500));
+					       msecs_to_jiffies(500));
 	list_del(&d->list);
 	if (ret == 0 || ret == 1)
 		return -EAGAIN;
Signature · 1 line
2.41.0
reviewer Reviewed-by on v1 1/4
#2 lore raw

Re: [PATCH net 1/4] octeon_ep: fix timeout value for waiting on mbox response

Quoted 7 lines · Michal Schmidt

The intention was to wait up to 500 ms for the mbox response. The third argument to wait_event_interruptible_timeout() is supposed to be the timeout duration. The driver mistakenly passed absolute time instead.

Fixes 577f0d1b1c5f ("octeon_ep: add separate mailbox command and response queues")
#3 lore raw

PATCH 2/4 octeon_ep: cancel tx_timeout_task later in remove sequence

In reply to Michal Schmidt

tx_timeout_task is canceled too early when removing the driver. Nothing prevents .ndo_tx_timeout from triggering and queuing the work again.

Better cancel it after the netdev is unregistered.
It's harmless for octep_tx_timeout_task to run in the window between the unregistration and cancelation, because it checks netif_running.

Fixes 862cd659a6fb ("octeon_ep: Add driver framework and device initialization")
---
 drivers/net/ethernet/marvell/octeon_ep/octep_main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/marvell/octeon_ep/octep_main.c +1 -1
index 43eb6e871351..d8066bff5f7b 100644
@@ -1200,12 +1200,12 @@ static void octep_remove(struct pci_dev *pdev)
 	if (!oct)
 		return;
 
-	cancel_work_sync(&oct->tx_timeout_task);
 	cancel_work_sync(&oct->ctrl_mbox_task);
 	netdev = oct->netdev;
 	if (netdev->reg_state == NETREG_REGISTERED)
 		unregister_netdev(netdev);
 
+	cancel_work_sync(&oct->tx_timeout_task);
 	oct->poll_non_ioq_intr = false;
 	cancel_delayed_work_sync(&oct->intr_poll_task);
 	octep_device_cleanup(oct);
Signature · 1 line
2.41.0
reviewer
#4 lore raw

Re: [PATCH net 2/4] octeon_ep: cancel tx_timeout_task later in remove sequence

Quoted 1 line · Michal Schmidt

tx_timeout_task is canceled too early when removing the driver. Nothing

nit: canceled -> cancelled

also elsewhere in this patchset

./checkpatch.pl --codespell is your friend here

...

#5 lore raw

Re: [PATCH net 2/4] octeon_ep: cancel tx_timeout_task later in remove sequence

Quoted 11 lines · Simon Horman
Quoted 1 line · Michal Schmidt

tx_timeout_task is canceled too early when removing the driver. Nothing

nit: canceled -> cancelled

also elsewhere in this patchset

./checkpatch.pl --codespell is your friend here

...

Hi Simon,
thank you for the review.

I think both spellings are valid.
https://www.grammarly.com/blog/canceled-vs-cancelled/

Do you want me to resend the patchset for this?

Michal

reviewer
#6 lore raw

Re: [PATCH net 2/4] octeon_ep: cancel tx_timeout_task later in remove sequence

Quoted 11 lines · Michal Schmidt
Quoted 10 lines · Simon Horman
Quoted 1 line · Michal Schmidt

tx_timeout_task is canceled too early when removing the driver. Nothing

nit: canceled -> cancelled

also elsewhere in this patchset

./checkpatch.pl --codespell is your friend here

Quoted 7 lines
Quoted 1 line

...

Hi Simon,
thank you for the review.

I think both spellings are valid.
https://www.grammarly.com/blog/canceled-vs-cancelled/

Thanks, I did not know that.

Quoted 1 line

Do you want me to resend the patchset for this?

No, not in light of your previous point.

Signature · 1 line
pw-bot: under-review
#7 lore raw

PATCH 3/4 octeon_ep: cancel ctrl_mbox_task after intr_poll_task

In reply to Michal Schmidt

intr_poll_task may queue ctrl_mbox_task. The function octep_poll_non_ioq_interrupts_cn93_pf does this.

When removing the driver and canceling these two works, cancel ctrl_mbox_task last to guarantee it does not run anymore.

Fixes 24d4333233b3 ("octeon_ep: poll for control messages")
---
 drivers/net/ethernet/marvell/octeon_ep/octep_main.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
drivers/net/ethernet/marvell/octeon_ep/octep_main.c +1 -1
index d8066bff5f7b..ab69b6d62509 100644
@@ -1200,7 +1200,6 @@ static void octep_remove(struct pci_dev *pdev)
 	if (!oct)
 		return;
 
-	cancel_work_sync(&oct->ctrl_mbox_task);
 	netdev = oct->netdev;
 	if (netdev->reg_state == NETREG_REGISTERED)
 		unregister_netdev(netdev);
@@ -1208,6 +1207,7 @@ static void octep_remove(struct pci_dev *pdev)
 	cancel_work_sync(&oct->tx_timeout_task);
 	oct->poll_non_ioq_intr = false;
 	cancel_delayed_work_sync(&oct->intr_poll_task);
+	cancel_work_sync(&oct->ctrl_mbox_task);
 	octep_device_cleanup(oct);
 	pci_release_mem_regions(pdev);
 	free_netdev(netdev);
Signature · 1 line
2.41.0
#8 lore raw

PATCH 4/4 octeon_ep: cancel queued works in probe error path

In reply to Michal Schmidt

If it fails to get the devices's MAC address, octep_probe exits while leaving the delayed work intr_poll_task queued. When the work later runs, it's a use after free.

Move the cancelation of intr_poll_task from octep_remove into octep_device_cleanup. This does not change anything in the octep_remove flow, but octep_device_cleanup is called also in the octep_probe error path, where the cancelation is needed.

Note that the cancelation of ctrl_mbox_task has to follow intr_poll_task's, because the ctrl_mbox_task may be queued by intr_poll_task.

Fixes 24d4333233b3 ("octeon_ep: poll for control messages")
---
 drivers/net/ethernet/marvell/octeon_ep/octep_main.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
drivers/net/ethernet/marvell/octeon_ep/octep_main.c +4 -3
index ab69b6d62509..4424de2ffd70 100644
@@ -1038,6 +1038,10 @@ static void octep_device_cleanup(struct octep_device *oct)
 {
 	int i;
 
+	oct->poll_non_ioq_intr = false;
+	cancel_delayed_work_sync(&oct->intr_poll_task);
+	cancel_work_sync(&oct->ctrl_mbox_task);
+
 	dev_info(&oct->pdev->dev, "Cleaning up Octeon Device ...\n");
 
 	for (i = 0; i < OCTEP_MAX_VF; i++) {
@@ -1205,9 +1209,6 @@ static void octep_remove(struct pci_dev *pdev)
 		unregister_netdev(netdev);
 
 	cancel_work_sync(&oct->tx_timeout_task);
-	oct->poll_non_ioq_intr = false;
-	cancel_delayed_work_sync(&oct->intr_poll_task);
-	cancel_work_sync(&oct->ctrl_mbox_task);
 	octep_device_cleanup(oct);
 	pci_release_mem_regions(pdev);
 	free_netdev(netdev);
Signature · 1 line
2.41.0
#9 lore raw
In reply to Michal Schmidt
This series was applied to netdev/net.git (main) by Jakub Kicinski <[email protected]>:

Hello:

This series was applied to netdev/net.git (main)
by Jakub Kicinski <[email protected]>:

Quoted 5 lines · On Thu, 10 Aug 2023 17:01:10 +0200 you wrote:

I have an Octeon card that's misconfigured in a way that exposes a couple of bugs in the octeon_ep driver's error paths. It can reproduce the issues that patches 1 & 4 are fixing. Patches 2 & 3 are a result of reviewing the nearby code.

Quoted 7 lines

Michal Schmidt (4):
octeon_ep: fix timeout value for waiting on mbox response
octeon_ep: cancel tx_timeout_task later in remove sequence
octeon_ep: cancel ctrl_mbox_task after intr_poll_task
octeon_ep: cancel queued works in probe error path

[...]

Here is the summary with links:
- [net,1/4] octeon_ep: fix timeout value for waiting on mbox response

    https://git.kernel.org/netdev/net/c/519b227904f0
  - [net,2/4] octeon_ep: cancel tx_timeout_task later in remove sequence
    https://git.kernel.org/netdev/net/c/28458c80006b
  - [net,3/4] octeon_ep: cancel ctrl_mbox_task after intr_poll_task
    https://git.kernel.org/netdev/net/c/607a7a45cdf3
  - [net,4/4] octeon_ep: cancel queued works in probe error path
    https://git.kernel.org/netdev/net/c/758c91078165

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html