[PATCH net 2/5] e1000e: do not cache the hwtstamp config before it is applied

[PATCH net 2/5] e1000e: do not cache the hwtstamp config before it is applied

[Linkui Xiao]

From: Linkui Xiao <xiaolinkui@kylinos.cn>

e1000e_config_hwtstamp() stores the requested configuration in
adapter->hwtstamp_config right after the rx filter has been
translated, but the two register writes that follow are verified by
reading the registers back and each of them can fail with -EAGAIN.

When one of them does, the driver is left holding a configuration the
hardware never accepted:

 - e1000e_hwtstamp_get() hands it back to user space, so an
   application that reads the configuration back is told that
   hardware time stamping is in effect while it is not;

 - e1000e_systim_reset() re-applies it after every reset, including
   the ones the watchdog triggers, and keeps failing;

 - e1000_watchdog() starts polling TSYNCRXCTL_VALID for the cached rx
   filter, and e1000_xmit_frame() still parks skbs that ask for a Tx
   time stamp because it only looks at FLAG_HAS_HW_TIMESTAMP, so they
   are eventually dropped by e1000e_tx_hwtstamp_work() with a
   "clearing Tx timestamp hang" warning.

Cache the configuration only once both registers have been verified,
so that adapter->hwtstamp_config always describes what the hardware
is programmed with.

Fixes: 62d7e3a2d006 ("e1000e: Validate hwtstamp_config completely before applying it")
Cc: stable@vger.kernel.org
Signed-off-by: Linkui Xiao <xiaolinkui@kylinos.cn>
---
 drivers/net/ethernet/intel/e1000e/netdev.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/net/ethernet/intel/e1000e/netdev.c b/drivers/net/ethernet/intel/e1000e/netdev.c
index 35f5590228eb..7ae33273734f 100644
--- a/drivers/net/ethernet/intel/e1000e/netdev.c
+++ b/drivers/net/ethernet/intel/e1000e/netdev.c
@@ -3715,8 +3715,6 @@ static int e1000e_config_hwtstamp(struct e1000_adapter *adapter,
 		return -ERANGE;
 	}
 
-	adapter->hwtstamp_config = *config;
-
 	/* enable/disable Tx h/w time stamping */
 	regval = er32(TSYNCTXCTL);
 	regval &= ~E1000_TSYNCTXCTL_ENABLED;
@@ -3763,6 +3761,8 @@ static int e1000e_config_hwtstamp(struct e1000_adapter *adapter,
 	er32(RXSTMPH);
 	er32(TXSTMPH);
 
+	adapter->hwtstamp_config = *config;
+
 	return 0;
 }
 
-- 
2.25.1

RE: [PATCH net 2/5] e1000e: do not cache the hwtstamp config before it is applied

[Loktionov, Aleksandr]

> -----Original Message-----
> From: Linkui Xiao <xiaolinkui@126.com>
> Sent: Wednesday, October 28, 2026 10:05 AM
> To: Nguyen, Anthony L <anthony.l.nguyen@intel.com>; Kitszel, Przemyslaw
> <przemyslaw.kitszel@intel.com>; andrew+netdev@lunn.ch;
> davem@davemloft.net; edumazet@kernel.org; kuba@kernel.org;
> pabeni@redhat.com
> Cc: intel-wired-lan@lists.osuosl.org; netdev@vger.kernel.org; linux-
> kernel@vger.kernel.org; Linkui Xiao <xiaolinkui@kylinos.cn>;
> stable@vger.kernel.org
> Subject: [PATCH net 2/5] e1000e: do not cache the hwtstamp config before
> it is applied
> 
> From: Linkui Xiao <xiaolinkui@kylinos.cn>
> 
> e1000e_config_hwtstamp() stores the requested configuration in
> adapter->hwtstamp_config right after the rx filter has been
> translated, but the two register writes that follow are verified by
> reading the registers back and each of them can fail with -EAGAIN.
> 
> When one of them does, the driver is left holding a configuration the
> hardware never accepted:
> 
>  - e1000e_hwtstamp_get() hands it back to user space, so an
>    application that reads the configuration back is told that
>    hardware time stamping is in effect while it is not;
> 
>  - e1000e_systim_reset() re-applies it after every reset, including
>    the ones the watchdog triggers, and keeps failing;
> 
>  - e1000_watchdog() starts polling TSYNCRXCTL_VALID for the cached rx
>    filter, and e1000_xmit_frame() still parks skbs that ask for a Tx
>    time stamp because it only looks at FLAG_HAS_HW_TIMESTAMP, so they
>    are eventually dropped by e1000e_tx_hwtstamp_work() with a
>    "clearing Tx timestamp hang" warning.
> 
> Cache the configuration only once both registers have been verified, so
> that adapter->hwtstamp_config always describes what the hardware is
> programmed with.
> 
> Fixes: 62d7e3a2d006 ("e1000e: Validate hwtstamp_config completely before
> applying it")
> Cc: stable@vger.kernel.org
> Signed-off-by: Linkui Xiao <xiaolinkui@kylinos.cn>
> ---
>  drivers/net/ethernet/intel/e1000e/netdev.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/ethernet/intel/e1000e/netdev.c
> b/drivers/net/ethernet/intel/e1000e/netdev.c
> index 35f5590228eb..7ae33273734f 100644
> --- a/drivers/net/ethernet/intel/e1000e/netdev.c
> +++ b/drivers/net/ethernet/intel/e1000e/netdev.c
> @@ -3715,8 +3715,6 @@ static int e1000e_config_hwtstamp(struct
> e1000_adapter *adapter,
>  		return -ERANGE;
>  	}
> 
> -	adapter->hwtstamp_config = *config;
> -
>  	/* enable/disable Tx h/w time stamping */
>  	regval = er32(TSYNCTXCTL);
>  	regval &= ~E1000_TSYNCTXCTL_ENABLED;
> @@ -3763,6 +3761,8 @@ static int e1000e_config_hwtstamp(struct
> e1000_adapter *adapter,
>  	er32(RXSTMPH);
>  	er32(TXSTMPH);
> 
> +	adapter->hwtstamp_config = *config;
> +
>  	return 0;
>  }
> 
> --
> 2.25.1

Reviewed-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>

Re: [PATCH net 2/5] e1000e: do not cache the hwtstamp config before it is applied

[Vadim Fedorenko]

On 28/10/2026 09:04, Linkui Xiao wrote:
> From: Linkui Xiao <xiaolinkui@kylinos.cn>
> 
> e1000e_config_hwtstamp() stores the requested configuration in
> adapter->hwtstamp_config right after the rx filter has been
> translated, but the two register writes that follow are verified by
> reading the registers back and each of them can fail with -EAGAIN.
> 
> When one of them does, the driver is left holding a configuration the
> hardware never accepted:
> 
>   - e1000e_hwtstamp_get() hands it back to user space, so an
>     application that reads the configuration back is told that
>     hardware time stamping is in effect while it is not;
> 
>   - e1000e_systim_reset() re-applies it after every reset, including
>     the ones the watchdog triggers, and keeps failing;
> 
>   - e1000_watchdog() starts polling TSYNCRXCTL_VALID for the cached rx
>     filter, and e1000_xmit_frame() still parks skbs that ask for a Tx
>     time stamp because it only looks at FLAG_HAS_HW_TIMESTAMP, so they
>     are eventually dropped by e1000e_tx_hwtstamp_work() with a
>     "clearing Tx timestamp hang" warning.
> 
> Cache the configuration only once both registers have been verified,
> so that adapter->hwtstamp_config always describes what the hardware
> is programmed with.
> 
> Fixes: 62d7e3a2d006 ("e1000e: Validate hwtstamp_config completely before applying it")
> Cc: stable@vger.kernel.org
> Signed-off-by: Linkui Xiao <xiaolinkui@kylinos.cn>
> ---
>   drivers/net/ethernet/intel/e1000e/netdev.c | 4 ++--
>   1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/ethernet/intel/e1000e/netdev.c b/drivers/net/ethernet/intel/e1000e/netdev.c
> index 35f5590228eb..7ae33273734f 100644
> --- a/drivers/net/ethernet/intel/e1000e/netdev.c
> +++ b/drivers/net/ethernet/intel/e1000e/netdev.c
> @@ -3715,8 +3715,6 @@ static int e1000e_config_hwtstamp(struct e1000_adapter *adapter,
>   		return -ERANGE;
>   	}
>   
> -	adapter->hwtstamp_config = *config;
> -
>   	/* enable/disable Tx h/w time stamping */
>   	regval = er32(TSYNCTXCTL);
>   	regval &= ~E1000_TSYNCTXCTL_ENABLED;
> @@ -3763,6 +3761,8 @@ static int e1000e_config_hwtstamp(struct e1000_adapter *adapter,
>   	er32(RXSTMPH);
>   	er32(TXSTMPH);
>   
> +	adapter->hwtstamp_config = *config;
> +
>   	return 0;
>   }
>   

Reviewed-by: Vadim Fedorenko <vadim.fedorenko@linux.dev>