* [PATCH net v4] ravb: Fix races between ravb_tx_timeout_work() and net related ops
@ 2023-11-27 12:24 Yoshihiro Shimoda
2023-11-27 16:27 ` Sergey Shtylyov
2023-11-29 4:10 ` patchwork-bot+netdevbpf
0 siblings, 2 replies; 4+ messages in thread
From: Yoshihiro Shimoda @ 2023-11-27 12:24 UTC (permalink / raw)
To: s.shtylyov, davem, edumazet, kuba, pabeni
Cc: netdev, linux-renesas-soc, Yoshihiro Shimoda
Fix races between ravb_tx_timeout_work() and functions of net_device_ops
and ethtool_ops by using rtnl_trylock() and rtnl_unlock(). Note that
since ravb_close() is under the rtnl lock and calls cancel_work_sync(),
ravb_tx_timeout_work() should calls rtnl_trylock(). Otherwise, a deadlock
may happen in ravb_tx_timeout_work() like below:
CPU0 CPU1
ravb_tx_timeout()
schedule_work()
...
__dev_close_many()
// Under rtnl lock
ravb_close()
cancel_work_sync()
// Waiting
ravb_tx_timeout_work()
rtnl_lock()
// This is possible to cause a deadlock
If rtnl_trylock() fails, rescheduling the work with sleep for 1 msec.
Fixes: c156633f1353 ("Renesas Ethernet AVB driver proper")
Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
---
Changes from v3:
https://lore.kernel.org/all/20231115022644.2316961-1-yoshihiro.shimoda.uh@renesas.com/
- Based on v2 patch. This means that delayed work is not used.
- Add rescheduling the work if rtnl_trylock() fails.
- Drop Reviewed-by tags because implementation was changed.
Changes from v2:
https://lore.kernel.org/netdev/20231019113308.1133944-1-yoshihiro.shimoda.uh@renesas.com/
- Add rescheduling if rtnl_trylock() fails and the netif is still running
and update commit description for it.
- Add Reviewed-by tags.
drivers/net/ethernet/renesas/ravb_main.c | 11 ++++++++++-
1 file changed, 10 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ethernet/renesas/ravb_main.c b/drivers/net/ethernet/renesas/ravb_main.c
index c70cff80cc99..7c007ecd3ff6 100644
--- a/drivers/net/ethernet/renesas/ravb_main.c
+++ b/drivers/net/ethernet/renesas/ravb_main.c
@@ -1874,6 +1874,12 @@ static void ravb_tx_timeout_work(struct work_struct *work)
struct net_device *ndev = priv->ndev;
int error;
+ if (!rtnl_trylock()) {
+ usleep_range(1000, 2000);
+ schedule_work(&priv->work);
+ return;
+ }
+
netif_tx_stop_all_queues(ndev);
/* Stop PTP Clock driver */
@@ -1907,7 +1913,7 @@ static void ravb_tx_timeout_work(struct work_struct *work)
*/
netdev_err(ndev, "%s: ravb_dmac_init() failed, error %d\n",
__func__, error);
- return;
+ goto out_unlock;
}
ravb_emac_init(ndev);
@@ -1917,6 +1923,9 @@ static void ravb_tx_timeout_work(struct work_struct *work)
ravb_ptp_init(ndev, priv->pdev);
netif_tx_start_all_queues(ndev);
+
+out_unlock:
+ rtnl_unlock();
}
/* Packet transmit function for Ethernet AVB */
--
2.25.1
^ permalink raw reply related [flat|nested] 4+ messages in thread
* Re: [PATCH net v4] ravb: Fix races between ravb_tx_timeout_work() and net related ops
2023-11-27 12:24 [PATCH net v4] ravb: Fix races between ravb_tx_timeout_work() and net related ops Yoshihiro Shimoda
@ 2023-11-27 16:27 ` Sergey Shtylyov
2023-11-27 23:58 ` Yoshihiro Shimoda
2023-11-29 4:10 ` patchwork-bot+netdevbpf
1 sibling, 1 reply; 4+ messages in thread
From: Sergey Shtylyov @ 2023-11-27 16:27 UTC (permalink / raw)
To: Yoshihiro Shimoda, davem, edumazet, kuba, pabeni
Cc: netdev, linux-renesas-soc
On 11/27/23 3:24 PM, Yoshihiro Shimoda wrote:
> Fix races between ravb_tx_timeout_work() and functions of net_device_ops
> and ethtool_ops by using rtnl_trylock() and rtnl_unlock(). Note that
> since ravb_close() is under the rtnl lock and calls cancel_work_sync(),
> ravb_tx_timeout_work() should calls rtnl_trylock(). Otherwise, a deadlock
> may happen in ravb_tx_timeout_work() like below:
>
> CPU0 CPU1
> ravb_tx_timeout()
> schedule_work()
> ...
> __dev_close_many()
> // Under rtnl lock
> ravb_close()
> cancel_work_sync()
> // Waiting
> ravb_tx_timeout_work()
> rtnl_lock()
> // This is possible to cause a deadlock
>
> If rtnl_trylock() fails, rescheduling the work with sleep for 1 msec.
>
> Fixes: c156633f1353 ("Renesas Ethernet AVB driver proper")
> Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>
[...]
> diff --git a/drivers/net/ethernet/renesas/ravb_main.c b/drivers/net/ethernet/renesas/ravb_main.c
> index c70cff80cc99..7c007ecd3ff6 100644
> --- a/drivers/net/ethernet/renesas/ravb_main.c
> +++ b/drivers/net/ethernet/renesas/ravb_main.c
> @@ -1874,6 +1874,12 @@ static void ravb_tx_timeout_work(struct work_struct *work)
> struct net_device *ndev = priv->ndev;
> int error;
>
> + if (!rtnl_trylock()) {
> + usleep_range(1000, 2000);
Why not msleep(1) though?
> + schedule_work(&priv->work);
> + return;
> + }
> +
> netif_tx_stop_all_queues(ndev);
>
> /* Stop PTP Clock driver */
[...]
MBR, Sergey
^ permalink raw reply [flat|nested] 4+ messages in thread
* RE: [PATCH net v4] ravb: Fix races between ravb_tx_timeout_work() and net related ops
2023-11-27 16:27 ` Sergey Shtylyov
@ 2023-11-27 23:58 ` Yoshihiro Shimoda
0 siblings, 0 replies; 4+ messages in thread
From: Yoshihiro Shimoda @ 2023-11-27 23:58 UTC (permalink / raw)
To: Sergey Shtylyov, davem@davemloft.net, edumazet@google.com,
kuba@kernel.org, pabeni@redhat.com
Cc: netdev@vger.kernel.org, linux-renesas-soc@vger.kernel.org
Hi Sergey,
> From: Sergey Shtylyov, Sent: Tuesday, November 28, 2023 1:28 AM
>
> On 11/27/23 3:24 PM, Yoshihiro Shimoda wrote:
>
> > Fix races between ravb_tx_timeout_work() and functions of net_device_ops
> > and ethtool_ops by using rtnl_trylock() and rtnl_unlock(). Note that
> > since ravb_close() is under the rtnl lock and calls cancel_work_sync(),
> > ravb_tx_timeout_work() should calls rtnl_trylock(). Otherwise, a deadlock
> > may happen in ravb_tx_timeout_work() like below:
> >
> > CPU0 CPU1
> > ravb_tx_timeout()
> > schedule_work()
> > ...
> > __dev_close_many()
> > // Under rtnl lock
> > ravb_close()
> > cancel_work_sync()
> > // Waiting
> > ravb_tx_timeout_work()
> > rtnl_lock()
> > // This is possible to cause a deadlock
> >
> > If rtnl_trylock() fails, rescheduling the work with sleep for 1 msec.
> >
> > Fixes: c156633f1353 ("Renesas Ethernet AVB driver proper")
> > Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
>
> Reviewed-by: Sergey Shtylyov <s.shtylyov@omp.ru>
Thank you very much for your review!
> [...]
>
> > diff --git a/drivers/net/ethernet/renesas/ravb_main.c b/drivers/net/ethernet/renesas/ravb_main.c
> > index c70cff80cc99..7c007ecd3ff6 100644
> > --- a/drivers/net/ethernet/renesas/ravb_main.c
> > +++ b/drivers/net/ethernet/renesas/ravb_main.c
> > @@ -1874,6 +1874,12 @@ static void ravb_tx_timeout_work(struct work_struct *work)
> > struct net_device *ndev = priv->ndev;
> > int error;
> >
> > + if (!rtnl_trylock()) {
> > + usleep_range(1000, 2000);
>
> Why not msleep(1) though?
It's for the following guideline:
https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/Documentation/timers/timers-howto.rst?h=v6.7-rc3
-----
SLEEPING FOR ~USECS OR SMALL MSECS ( 10us - 20ms):
* Use usleep_range
-----
Best regards,
Yoshihiro Shimoda
> > + schedule_work(&priv->work);
> > + return;
> > + }
> > +
> > netif_tx_stop_all_queues(ndev);
> >
> > /* Stop PTP Clock driver */
> [...]
>
> MBR, Sergey
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [PATCH net v4] ravb: Fix races between ravb_tx_timeout_work() and net related ops
2023-11-27 12:24 [PATCH net v4] ravb: Fix races between ravb_tx_timeout_work() and net related ops Yoshihiro Shimoda
2023-11-27 16:27 ` Sergey Shtylyov
@ 2023-11-29 4:10 ` patchwork-bot+netdevbpf
1 sibling, 0 replies; 4+ messages in thread
From: patchwork-bot+netdevbpf @ 2023-11-29 4:10 UTC (permalink / raw)
To: Yoshihiro Shimoda
Cc: s.shtylyov, davem, edumazet, kuba, pabeni, netdev,
linux-renesas-soc
Hello:
This patch was applied to netdev/net.git (main)
by Jakub Kicinski <kuba@kernel.org>:
On Mon, 27 Nov 2023 21:24:20 +0900 you wrote:
> Fix races between ravb_tx_timeout_work() and functions of net_device_ops
> and ethtool_ops by using rtnl_trylock() and rtnl_unlock(). Note that
> since ravb_close() is under the rtnl lock and calls cancel_work_sync(),
> ravb_tx_timeout_work() should calls rtnl_trylock(). Otherwise, a deadlock
> may happen in ravb_tx_timeout_work() like below:
>
> CPU0 CPU1
> ravb_tx_timeout()
> schedule_work()
> ...
> __dev_close_many()
> // Under rtnl lock
> ravb_close()
> cancel_work_sync()
> // Waiting
> ravb_tx_timeout_work()
> rtnl_lock()
> // This is possible to cause a deadlock
>
> [...]
Here is the summary with links:
- [net,v4] ravb: Fix races between ravb_tx_timeout_work() and net related ops
https://git.kernel.org/netdev/net/c/9870257a0a33
You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2023-11-29 4:10 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2023-11-27 12:24 [PATCH net v4] ravb: Fix races between ravb_tx_timeout_work() and net related ops Yoshihiro Shimoda
2023-11-27 16:27 ` Sergey Shtylyov
2023-11-27 23:58 ` Yoshihiro Shimoda
2023-11-29 4:10 ` patchwork-bot+netdevbpf
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).