| Message ID | caf6a487-ef8c-4570-88f9-f47a659faf33@gmail.com (mailing list archive) |
|---|---|
| State | Accepted |
| Commit | 59d395ed606d8df14615712b0cdcdadb2d962175 |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | [net] r8169: fix deadlock on RTL8125 in jumbo mtu mode | expand |
On 26.11.2023 19:36, Heiner Kallweit wrote: > The original change results in a deadlock if jumbo mtu mode is used. > Reason is that the phydev lock is held when rtl_reset_work() is called > here, and rtl_jumbo_config() calls phy_start_aneg() which also tries > to acquire the phydev lock. Fix this by calling rtl_reset_work() > asynchronously. > > Fixes: 621735f59064 ("r8169: fix rare issue with broken rx after link-down on RTL8125") > Reported-by: Ian Chen <free122448@hotmail.com> > Tested-by: Ian Chen <free122448@hotmail.com> > Cc: stable@vger.kernel.org > Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com> > --- > drivers/net/ethernet/realtek/r8169_main.c | 5 ++++- > 1 file changed, 4 insertions(+), 1 deletion(-) > > diff --git a/drivers/net/ethernet/realtek/r8169_main.c b/drivers/net/ethernet/realtek/r8169_main.c > index 0ee3579ce..e32cc3279 100644 > --- a/drivers/net/ethernet/realtek/r8169_main.c > +++ b/drivers/net/ethernet/realtek/r8169_main.c > @@ -575,6 +575,7 @@ struct rtl8169_tc_offsets { > enum rtl_flag { > RTL_FLAG_TASK_ENABLED = 0, > RTL_FLAG_TASK_RESET_PENDING, > + RTL_FLAG_TASK_RESET_NO_QUEUE_WAKE, > RTL_FLAG_TASK_TX_TIMEOUT, > RTL_FLAG_MAX > }; > @@ -4494,6 +4495,8 @@ static void rtl_task(struct work_struct *work) > reset: > rtl_reset_work(tp); > netif_wake_queue(tp->dev); > + } else if (test_and_clear_bit(RTL_FLAG_TASK_RESET_NO_QUEUE_WAKE, tp->wk.flags)) { > + rtl_reset_work(tp); > } > out_unlock: > rtnl_unlock(); > @@ -4527,7 +4530,7 @@ static void r8169_phylink_handler(struct net_device *ndev) > } else { > /* In few cases rx is broken after link-down otherwise */ > if (rtl_is_8125(tp)) > - rtl_reset_work(tp); > + rtl_schedule_task(tp, RTL_FLAG_TASK_RESET_NO_QUEUE_WAKE); > pm_runtime_idle(d); > } > I noticed there's a potential issue with my approach. So I have to rework this, please do not apply.
On 26.11.2023 21:14, Heiner Kallweit wrote: > On 26.11.2023 19:36, Heiner Kallweit wrote: >> The original change results in a deadlock if jumbo mtu mode is used. >> Reason is that the phydev lock is held when rtl_reset_work() is called >> here, and rtl_jumbo_config() calls phy_start_aneg() which also tries >> to acquire the phydev lock. Fix this by calling rtl_reset_work() >> asynchronously. >> >> Fixes: 621735f59064 ("r8169: fix rare issue with broken rx after link-down on RTL8125") >> Reported-by: Ian Chen <free122448@hotmail.com> >> Tested-by: Ian Chen <free122448@hotmail.com> >> Cc: stable@vger.kernel.org >> Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com> >> --- >> drivers/net/ethernet/realtek/r8169_main.c | 5 ++++- >> 1 file changed, 4 insertions(+), 1 deletion(-) >> >> diff --git a/drivers/net/ethernet/realtek/r8169_main.c b/drivers/net/ethernet/realtek/r8169_main.c >> index 0ee3579ce..e32cc3279 100644 >> --- a/drivers/net/ethernet/realtek/r8169_main.c >> +++ b/drivers/net/ethernet/realtek/r8169_main.c >> @@ -575,6 +575,7 @@ struct rtl8169_tc_offsets { >> enum rtl_flag { >> RTL_FLAG_TASK_ENABLED = 0, >> RTL_FLAG_TASK_RESET_PENDING, >> + RTL_FLAG_TASK_RESET_NO_QUEUE_WAKE, >> RTL_FLAG_TASK_TX_TIMEOUT, >> RTL_FLAG_MAX >> }; >> @@ -4494,6 +4495,8 @@ static void rtl_task(struct work_struct *work) >> reset: >> rtl_reset_work(tp); >> netif_wake_queue(tp->dev); >> + } else if (test_and_clear_bit(RTL_FLAG_TASK_RESET_NO_QUEUE_WAKE, tp->wk.flags)) { >> + rtl_reset_work(tp); >> } >> out_unlock: >> rtnl_unlock(); >> @@ -4527,7 +4530,7 @@ static void r8169_phylink_handler(struct net_device *ndev) >> } else { >> /* In few cases rx is broken after link-down otherwise */ >> if (rtl_is_8125(tp)) >> - rtl_reset_work(tp); >> + rtl_schedule_task(tp, RTL_FLAG_TASK_RESET_NO_QUEUE_WAKE); >> pm_runtime_idle(d); >> } >> > > I noticed there's a potential issue with my approach. > So I have to rework this, please do not apply. > The mentioned issue has an impact on patch "r8169: prevent potential deadlock in rtl8169_close" only. The patch here can remain as it is and is ready to be applied. Sorry for the noise.
Hello: This patch was applied to netdev/net.git (main) by Paolo Abeni <pabeni@redhat.com>: On Sun, 26 Nov 2023 19:36:46 +0100 you wrote: > The original change results in a deadlock if jumbo mtu mode is used. > Reason is that the phydev lock is held when rtl_reset_work() is called > here, and rtl_jumbo_config() calls phy_start_aneg() which also tries > to acquire the phydev lock. Fix this by calling rtl_reset_work() > asynchronously. > > Fixes: 621735f59064 ("r8169: fix rare issue with broken rx after link-down on RTL8125") > Reported-by: Ian Chen <free122448@hotmail.com> > Tested-by: Ian Chen <free122448@hotmail.com> > Cc: stable@vger.kernel.org > Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com> > > [...] Here is the summary with links: - [net] r8169: fix deadlock on RTL8125 in jumbo mtu mode https://git.kernel.org/netdev/net/c/59d395ed606d You are awesome, thank you!
diff --git a/drivers/net/ethernet/realtek/r8169_main.c b/drivers/net/ethernet/realtek/r8169_main.c index 0ee3579ce..e32cc3279 100644 --- a/drivers/net/ethernet/realtek/r8169_main.c +++ b/drivers/net/ethernet/realtek/r8169_main.c @@ -575,6 +575,7 @@ struct rtl8169_tc_offsets { enum rtl_flag { RTL_FLAG_TASK_ENABLED = 0, RTL_FLAG_TASK_RESET_PENDING, + RTL_FLAG_TASK_RESET_NO_QUEUE_WAKE, RTL_FLAG_TASK_TX_TIMEOUT, RTL_FLAG_MAX }; @@ -4494,6 +4495,8 @@ static void rtl_task(struct work_struct *work) reset: rtl_reset_work(tp); netif_wake_queue(tp->dev); + } else if (test_and_clear_bit(RTL_FLAG_TASK_RESET_NO_QUEUE_WAKE, tp->wk.flags)) { + rtl_reset_work(tp); } out_unlock: rtnl_unlock(); @@ -4527,7 +4530,7 @@ static void r8169_phylink_handler(struct net_device *ndev) } else { /* In few cases rx is broken after link-down otherwise */ if (rtl_is_8125(tp)) - rtl_reset_work(tp); + rtl_schedule_task(tp, RTL_FLAG_TASK_RESET_NO_QUEUE_WAKE); pm_runtime_idle(d); }
The original change results in a deadlock if jumbo mtu mode is used. Reason is that the phydev lock is held when rtl_reset_work() is called here, and rtl_jumbo_config() calls phy_start_aneg() which also tries to acquire the phydev lock. Fix this by calling rtl_reset_work() asynchronously. Fixes: 621735f59064 ("r8169: fix rare issue with broken rx after link-down on RTL8125") Reported-by: Ian Chen <free122448@hotmail.com> Tested-by: Ian Chen <free122448@hotmail.com> Cc: stable@vger.kernel.org Signed-off-by: Heiner Kallweit <hkallweit1@gmail.com> --- drivers/net/ethernet/realtek/r8169_main.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-)