[SRU,Yakkety,1/1] net/mlx4_core: Avoid delays during VF driver device shutdown

Submitted by Joseph Salisbury on March 30, 2017, 8:35 p.m.

Details

Message ID 56328a74e135e8c904af4aa3f73e0fe2843b7500.1490903650.git.joseph.salisbury@canonical.com
State New
Headers show

Commit Message

Joseph Salisbury March 30, 2017, 8:35 p.m.
From: Jack Morgenstein <jackm@dev.mellanox.co.il>

BugLink: http://bugs.launchpad.net/bugs/1672785

Some Hypervisors detach VFs from VMs by instantly causing an FLR event
to be generated for a VF.

In the mlx4 case, this will cause that VF's comm channel to be disabled
before the VM has an opportunity to invoke the VF device's "shutdown"
method.

For such Hypervisors, there is a race condition between the VF's
shutdown method and its internal-error detection/reset thread.

The internal-error detection/reset thread (which runs every 5 seconds) also
detects a disabled comm channel. If the internal-error detection/reset
flow wins the race, we still get delays (while that flow tries repeatedly
to detect comm-channel recovery).

The cited commit fixed the command timeout problem when the
internal-error detection/reset flow loses the race.

This commit avoids the unneeded delays when the internal-error
detection/reset flow wins.

Fixes: d585df1c5ccf ("net/mlx4_core: Avoid command timeouts during VF driver device shutdown")
Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
Reported-by: Simon Xiao <sixiao@microsoft.com>
Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
(backported from commit 4cbe4dac82e423ecc9a0ba46af24a860853259f4)
Signed-off-by: Joseph Salisbury <joseph.salisbury@canonical.com>
---
 drivers/net/ethernet/mellanox/mlx4/cmd.c  | 11 +++++++++++
 drivers/net/ethernet/mellanox/mlx4/main.c | 11 +++++++++++
 include/linux/mlx4/device.h               |  1 +
 3 files changed, 23 insertions(+)

Comments

Tim Gardner March 30, 2017, 9:21 p.m.

Thadeu Lima de Souza Cascardo April 3, 2017, 9:26 a.m.
On Thu, Mar 30, 2017 at 04:35:59PM -0400, Joseph Salisbury wrote:
> From: Jack Morgenstein <jackm@dev.mellanox.co.il>
> 
> BugLink: http://bugs.launchpad.net/bugs/1672785
> 
> Some Hypervisors detach VFs from VMs by instantly causing an FLR event
> to be generated for a VF.
> 
> In the mlx4 case, this will cause that VF's comm channel to be disabled
> before the VM has an opportunity to invoke the VF device's "shutdown"
> method.
> 
> For such Hypervisors, there is a race condition between the VF's
> shutdown method and its internal-error detection/reset thread.
> 
> The internal-error detection/reset thread (which runs every 5 seconds) also
> detects a disabled comm channel. If the internal-error detection/reset
> flow wins the race, we still get delays (while that flow tries repeatedly
> to detect comm-channel recovery).
> 
> The cited commit fixed the command timeout problem when the
> internal-error detection/reset flow loses the race.
> 
> This commit avoids the unneeded delays when the internal-error
> detection/reset flow wins.
> 
> Fixes: d585df1c5ccf ("net/mlx4_core: Avoid command timeouts during VF driver device shutdown")
> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
> Reported-by: Simon Xiao <sixiao@microsoft.com>
> Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
> Signed-off-by: David S. Miller <davem@davemloft.net>
> (backported from commit 4cbe4dac82e423ecc9a0ba46af24a860853259f4)
> Signed-off-by: Joseph Salisbury <joseph.salisbury@canonical.com>
> ---
>  drivers/net/ethernet/mellanox/mlx4/cmd.c  | 11 +++++++++++
>  drivers/net/ethernet/mellanox/mlx4/main.c | 11 +++++++++++
>  include/linux/mlx4/device.h               |  1 +
>  3 files changed, 23 insertions(+)
> 
> diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
> index f04a423..be6906b 100644
> --- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
> +++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
> @@ -2278,6 +2278,17 @@ static int sync_toggles(struct mlx4_dev *dev)
>  		rd_toggle = swab32(readl(&priv->mfunc.comm->slave_read));
>  		if (wr_toggle == 0xffffffff || rd_toggle == 0xffffffff) {
>  			/* PCI might be offline */
> +
> +			/* If device removal has been requested,
> +			 * do not continue retrying.
> +			 */
> +			if (dev->persist->interface_state &
> +			    MLX4_INTERFACE_STATE_NOWAIT) {
> +				mlx4_warn(dev,
> +					  "communication channel is offline\n");
> +				return -EIO;
> +			}
> +
>  			msleep(100);
>  			wr_toggle = swab32(readl(&priv->mfunc.comm->
>  					   slave_write));
> diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
> index 7183ac4..ea5e362 100644
> --- a/drivers/net/ethernet/mellanox/mlx4/main.c
> +++ b/drivers/net/ethernet/mellanox/mlx4/main.c
> @@ -1916,6 +1916,14 @@ static int mlx4_comm_check_offline(struct mlx4_dev *dev)
>  			       (u32)(1 << COMM_CHAN_OFFLINE_OFFSET));
>  		if (!offline_bit)
>  			return 0;
> +
> +		/* If device removal has been requested,
> +		 * do not continue retrying.
> +		 */
> +		if (dev->persist->interface_state &
> +		    MLX4_INTERFACE_STATE_NOWAIT)
> +			break;
> +
>  		/* There are cases as part of AER/Reset flow that PF needs
>  		 * around 100 msec to load. We therefore sleep for 100 msec
>  		 * to allow other tasks to make use of that CPU during this
> @@ -3930,6 +3938,9 @@ static void mlx4_remove_one(struct pci_dev *pdev)
>  	struct devlink *devlink = priv_to_devlink(priv);
>  	int active_vfs = 0;
>  
> +	if (mlx4_is_slave(dev))
> +		persist->interface_state |= MLX4_INTERFACE_STATE_NOWAIT;
> +
>  	mutex_lock(&persist->interface_state_mutex);
>  	persist->interface_state |= MLX4_INTERFACE_STATE_DELETION;
>  	mutex_unlock(&persist->interface_state_mutex);
> diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
> index 42da355..d49f11d 100644
> --- a/include/linux/mlx4/device.h
> +++ b/include/linux/mlx4/device.h
> @@ -468,6 +468,7 @@ enum {
>  	MLX4_INTERFACE_STATE_UP		= 1 << 0,
>  	MLX4_INTERFACE_STATE_DELETION	= 1 << 1,
>  	MLX4_INTERFACE_STATE_SHUTDOWN	= 1 << 2,
> +	MLX4_INTERFACE_STATE_NOWAIT	= 1 << 2,

So, NOWAIT and SHUTDOWN are defined to the same value, so it might be
necessary to review it that is going to work as expected. Maybe doing
the shutdown removal is a path to consider (commit
b4353708f5a1c084fd73f1b6fd243b142157b173).

Cascardo.

>  };
>  
>  #define MSTR_SM_CHANGE_MASK (MLX4_EQ_PORT_INFO_MSTR_SM_SL_CHANGE_MASK | \
> -- 
> 2.7.4
> 
> 
> -- 
> kernel-team mailing list
> kernel-team@lists.ubuntu.com
> https://lists.ubuntu.com/mailman/listinfo/kernel-team
Joshua R. Poulson April 3, 2017, 12:11 p.m.
We're engaged to test the driver in the intended scenario and code
path. We have a stable repro for the bug that this fixes.

Thanks, --jrp

On Mon, Apr 3, 2017 at 2:26 AM, Thadeu Lima de Souza Cascardo
<cascardo@canonical.com> wrote:
> On Thu, Mar 30, 2017 at 04:35:59PM -0400, Joseph Salisbury wrote:
>> From: Jack Morgenstein <jackm@dev.mellanox.co.il>
>>
>> BugLink: http://bugs.launchpad.net/bugs/1672785
>>
>> Some Hypervisors detach VFs from VMs by instantly causing an FLR event
>> to be generated for a VF.
>>
>> In the mlx4 case, this will cause that VF's comm channel to be disabled
>> before the VM has an opportunity to invoke the VF device's "shutdown"
>> method.
>>
>> For such Hypervisors, there is a race condition between the VF's
>> shutdown method and its internal-error detection/reset thread.
>>
>> The internal-error detection/reset thread (which runs every 5 seconds) also
>> detects a disabled comm channel. If the internal-error detection/reset
>> flow wins the race, we still get delays (while that flow tries repeatedly
>> to detect comm-channel recovery).
>>
>> The cited commit fixed the command timeout problem when the
>> internal-error detection/reset flow loses the race.
>>
>> This commit avoids the unneeded delays when the internal-error
>> detection/reset flow wins.
>>
>> Fixes: d585df1c5ccf ("net/mlx4_core: Avoid command timeouts during VF driver device shutdown")
>> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
>> Reported-by: Simon Xiao <sixiao@microsoft.com>
>> Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
>> Signed-off-by: David S. Miller <davem@davemloft.net>
>> (backported from commit 4cbe4dac82e423ecc9a0ba46af24a860853259f4)
>> Signed-off-by: Joseph Salisbury <joseph.salisbury@canonical.com>
>> ---
>>  drivers/net/ethernet/mellanox/mlx4/cmd.c  | 11 +++++++++++
>>  drivers/net/ethernet/mellanox/mlx4/main.c | 11 +++++++++++
>>  include/linux/mlx4/device.h               |  1 +
>>  3 files changed, 23 insertions(+)
>>
>> diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
>> index f04a423..be6906b 100644
>> --- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
>> +++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
>> @@ -2278,6 +2278,17 @@ static int sync_toggles(struct mlx4_dev *dev)
>>               rd_toggle = swab32(readl(&priv->mfunc.comm->slave_read));
>>               if (wr_toggle == 0xffffffff || rd_toggle == 0xffffffff) {
>>                       /* PCI might be offline */
>> +
>> +                     /* If device removal has been requested,
>> +                      * do not continue retrying.
>> +                      */
>> +                     if (dev->persist->interface_state &
>> +                         MLX4_INTERFACE_STATE_NOWAIT) {
>> +                             mlx4_warn(dev,
>> +                                       "communication channel is offline\n");
>> +                             return -EIO;
>> +                     }
>> +
>>                       msleep(100);
>>                       wr_toggle = swab32(readl(&priv->mfunc.comm->
>>                                          slave_write));
>> diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
>> index 7183ac4..ea5e362 100644
>> --- a/drivers/net/ethernet/mellanox/mlx4/main.c
>> +++ b/drivers/net/ethernet/mellanox/mlx4/main.c
>> @@ -1916,6 +1916,14 @@ static int mlx4_comm_check_offline(struct mlx4_dev *dev)
>>                              (u32)(1 << COMM_CHAN_OFFLINE_OFFSET));
>>               if (!offline_bit)
>>                       return 0;
>> +
>> +             /* If device removal has been requested,
>> +              * do not continue retrying.
>> +              */
>> +             if (dev->persist->interface_state &
>> +                 MLX4_INTERFACE_STATE_NOWAIT)
>> +                     break;
>> +
>>               /* There are cases as part of AER/Reset flow that PF needs
>>                * around 100 msec to load. We therefore sleep for 100 msec
>>                * to allow other tasks to make use of that CPU during this
>> @@ -3930,6 +3938,9 @@ static void mlx4_remove_one(struct pci_dev *pdev)
>>       struct devlink *devlink = priv_to_devlink(priv);
>>       int active_vfs = 0;
>>
>> +     if (mlx4_is_slave(dev))
>> +             persist->interface_state |= MLX4_INTERFACE_STATE_NOWAIT;
>> +
>>       mutex_lock(&persist->interface_state_mutex);
>>       persist->interface_state |= MLX4_INTERFACE_STATE_DELETION;
>>       mutex_unlock(&persist->interface_state_mutex);
>> diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
>> index 42da355..d49f11d 100644
>> --- a/include/linux/mlx4/device.h
>> +++ b/include/linux/mlx4/device.h
>> @@ -468,6 +468,7 @@ enum {
>>       MLX4_INTERFACE_STATE_UP         = 1 << 0,
>>       MLX4_INTERFACE_STATE_DELETION   = 1 << 1,
>>       MLX4_INTERFACE_STATE_SHUTDOWN   = 1 << 2,
>> +     MLX4_INTERFACE_STATE_NOWAIT     = 1 << 2,
>
> So, NOWAIT and SHUTDOWN are defined to the same value, so it might be
> necessary to review it that is going to work as expected. Maybe doing
> the shutdown removal is a path to consider (commit
> b4353708f5a1c084fd73f1b6fd243b142157b173).
>
> Cascardo.
>
>>  };
>>
>>  #define MSTR_SM_CHANGE_MASK (MLX4_EQ_PORT_INFO_MSTR_SM_SL_CHANGE_MASK | \
>> --
>> 2.7.4
>>
>>
>> --
>> kernel-team mailing list
>> kernel-team@lists.ubuntu.com
>> https://lists.ubuntu.com/mailman/listinfo/kernel-team
>
> --
> kernel-team mailing list
> kernel-team@lists.ubuntu.com
> https://lists.ubuntu.com/mailman/listinfo/kernel-team
Thadeu Lima de Souza Cascardo April 3, 2017, 1:32 p.m.
On Mon, Apr 03, 2017 at 05:11:34AM -0700, Joshua R. Poulson wrote:
> We're engaged to test the driver in the intended scenario and code
> path. We have a stable repro for the bug that this fixes.
> 
> Thanks, --jrp
> 

Is that on Yakkety, with this Yakkety backport as is?

I am more concerned about the paths that would use SHUTDOWN, have you tested
these as well?

Thanks.
Cascardo.

> On Mon, Apr 3, 2017 at 2:26 AM, Thadeu Lima de Souza Cascardo
> <cascardo@canonical.com> wrote:
> > On Thu, Mar 30, 2017 at 04:35:59PM -0400, Joseph Salisbury wrote:
> >> From: Jack Morgenstein <jackm@dev.mellanox.co.il>
> >>
> >> BugLink: http://bugs.launchpad.net/bugs/1672785
> >>
> >> Some Hypervisors detach VFs from VMs by instantly causing an FLR event
> >> to be generated for a VF.
> >>
> >> In the mlx4 case, this will cause that VF's comm channel to be disabled
> >> before the VM has an opportunity to invoke the VF device's "shutdown"
> >> method.
> >>
> >> For such Hypervisors, there is a race condition between the VF's
> >> shutdown method and its internal-error detection/reset thread.
> >>
> >> The internal-error detection/reset thread (which runs every 5 seconds) also
> >> detects a disabled comm channel. If the internal-error detection/reset
> >> flow wins the race, we still get delays (while that flow tries repeatedly
> >> to detect comm-channel recovery).
> >>
> >> The cited commit fixed the command timeout problem when the
> >> internal-error detection/reset flow loses the race.
> >>
> >> This commit avoids the unneeded delays when the internal-error
> >> detection/reset flow wins.
> >>
> >> Fixes: d585df1c5ccf ("net/mlx4_core: Avoid command timeouts during VF driver device shutdown")
> >> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
> >> Reported-by: Simon Xiao <sixiao@microsoft.com>
> >> Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
> >> Signed-off-by: David S. Miller <davem@davemloft.net>
> >> (backported from commit 4cbe4dac82e423ecc9a0ba46af24a860853259f4)
> >> Signed-off-by: Joseph Salisbury <joseph.salisbury@canonical.com>
> >> ---
> >>  drivers/net/ethernet/mellanox/mlx4/cmd.c  | 11 +++++++++++
> >>  drivers/net/ethernet/mellanox/mlx4/main.c | 11 +++++++++++
> >>  include/linux/mlx4/device.h               |  1 +
> >>  3 files changed, 23 insertions(+)
> >>
> >> diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
> >> index f04a423..be6906b 100644
> >> --- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
> >> +++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
> >> @@ -2278,6 +2278,17 @@ static int sync_toggles(struct mlx4_dev *dev)
> >>               rd_toggle = swab32(readl(&priv->mfunc.comm->slave_read));
> >>               if (wr_toggle == 0xffffffff || rd_toggle == 0xffffffff) {
> >>                       /* PCI might be offline */
> >> +
> >> +                     /* If device removal has been requested,
> >> +                      * do not continue retrying.
> >> +                      */
> >> +                     if (dev->persist->interface_state &
> >> +                         MLX4_INTERFACE_STATE_NOWAIT) {
> >> +                             mlx4_warn(dev,
> >> +                                       "communication channel is offline\n");
> >> +                             return -EIO;
> >> +                     }
> >> +
> >>                       msleep(100);
> >>                       wr_toggle = swab32(readl(&priv->mfunc.comm->
> >>                                          slave_write));
> >> diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
> >> index 7183ac4..ea5e362 100644
> >> --- a/drivers/net/ethernet/mellanox/mlx4/main.c
> >> +++ b/drivers/net/ethernet/mellanox/mlx4/main.c
> >> @@ -1916,6 +1916,14 @@ static int mlx4_comm_check_offline(struct mlx4_dev *dev)
> >>                              (u32)(1 << COMM_CHAN_OFFLINE_OFFSET));
> >>               if (!offline_bit)
> >>                       return 0;
> >> +
> >> +             /* If device removal has been requested,
> >> +              * do not continue retrying.
> >> +              */
> >> +             if (dev->persist->interface_state &
> >> +                 MLX4_INTERFACE_STATE_NOWAIT)
> >> +                     break;
> >> +
> >>               /* There are cases as part of AER/Reset flow that PF needs
> >>                * around 100 msec to load. We therefore sleep for 100 msec
> >>                * to allow other tasks to make use of that CPU during this
> >> @@ -3930,6 +3938,9 @@ static void mlx4_remove_one(struct pci_dev *pdev)
> >>       struct devlink *devlink = priv_to_devlink(priv);
> >>       int active_vfs = 0;
> >>
> >> +     if (mlx4_is_slave(dev))
> >> +             persist->interface_state |= MLX4_INTERFACE_STATE_NOWAIT;
> >> +
> >>       mutex_lock(&persist->interface_state_mutex);
> >>       persist->interface_state |= MLX4_INTERFACE_STATE_DELETION;
> >>       mutex_unlock(&persist->interface_state_mutex);
> >> diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
> >> index 42da355..d49f11d 100644
> >> --- a/include/linux/mlx4/device.h
> >> +++ b/include/linux/mlx4/device.h
> >> @@ -468,6 +468,7 @@ enum {
> >>       MLX4_INTERFACE_STATE_UP         = 1 << 0,
> >>       MLX4_INTERFACE_STATE_DELETION   = 1 << 1,
> >>       MLX4_INTERFACE_STATE_SHUTDOWN   = 1 << 2,
> >> +     MLX4_INTERFACE_STATE_NOWAIT     = 1 << 2,
> >
> > So, NOWAIT and SHUTDOWN are defined to the same value, so it might be
> > necessary to review it that is going to work as expected. Maybe doing
> > the shutdown removal is a path to consider (commit
> > b4353708f5a1c084fd73f1b6fd243b142157b173).
> >
> > Cascardo.
> >
> >>  };
> >>
> >>  #define MSTR_SM_CHANGE_MASK (MLX4_EQ_PORT_INFO_MSTR_SM_SL_CHANGE_MASK | \
> >> --
> >> 2.7.4
> >>
> >>
> >> --
> >> kernel-team mailing list
> >> kernel-team@lists.ubuntu.com
> >> https://lists.ubuntu.com/mailman/listinfo/kernel-team
> >
> > --
> > kernel-team mailing list
> > kernel-team@lists.ubuntu.com
> > https://lists.ubuntu.com/mailman/listinfo/kernel-team
Joshua R. Poulson April 3, 2017, 1:35 p.m.
Our friends at Mellanox have, we have been testing VF remove and add
as part of SR-IOV in Azure testing. Our regression tests include
on-premises Hyper-V as well.

Thanks --jrp

On Mon, Apr 3, 2017 at 6:32 AM, Thadeu Lima de Souza Cascardo
<cascardo@canonical.com> wrote:
> On Mon, Apr 03, 2017 at 05:11:34AM -0700, Joshua R. Poulson wrote:
>> We're engaged to test the driver in the intended scenario and code
>> path. We have a stable repro for the bug that this fixes.
>>
>> Thanks, --jrp
>>
>
> Is that on Yakkety, with this Yakkety backport as is?
>
> I am more concerned about the paths that would use SHUTDOWN, have you tested
> these as well?
>
> Thanks.
> Cascardo.
>
>> On Mon, Apr 3, 2017 at 2:26 AM, Thadeu Lima de Souza Cascardo
>> <cascardo@canonical.com> wrote:
>> > On Thu, Mar 30, 2017 at 04:35:59PM -0400, Joseph Salisbury wrote:
>> >> From: Jack Morgenstein <jackm@dev.mellanox.co.il>
>> >>
>> >> BugLink: http://bugs.launchpad.net/bugs/1672785
>> >>
>> >> Some Hypervisors detach VFs from VMs by instantly causing an FLR event
>> >> to be generated for a VF.
>> >>
>> >> In the mlx4 case, this will cause that VF's comm channel to be disabled
>> >> before the VM has an opportunity to invoke the VF device's "shutdown"
>> >> method.
>> >>
>> >> For such Hypervisors, there is a race condition between the VF's
>> >> shutdown method and its internal-error detection/reset thread.
>> >>
>> >> The internal-error detection/reset thread (which runs every 5 seconds) also
>> >> detects a disabled comm channel. If the internal-error detection/reset
>> >> flow wins the race, we still get delays (while that flow tries repeatedly
>> >> to detect comm-channel recovery).
>> >>
>> >> The cited commit fixed the command timeout problem when the
>> >> internal-error detection/reset flow loses the race.
>> >>
>> >> This commit avoids the unneeded delays when the internal-error
>> >> detection/reset flow wins.
>> >>
>> >> Fixes: d585df1c5ccf ("net/mlx4_core: Avoid command timeouts during VF driver device shutdown")
>> >> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
>> >> Reported-by: Simon Xiao <sixiao@microsoft.com>
>> >> Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
>> >> Signed-off-by: David S. Miller <davem@davemloft.net>
>> >> (backported from commit 4cbe4dac82e423ecc9a0ba46af24a860853259f4)
>> >> Signed-off-by: Joseph Salisbury <joseph.salisbury@canonical.com>
>> >> ---
>> >>  drivers/net/ethernet/mellanox/mlx4/cmd.c  | 11 +++++++++++
>> >>  drivers/net/ethernet/mellanox/mlx4/main.c | 11 +++++++++++
>> >>  include/linux/mlx4/device.h               |  1 +
>> >>  3 files changed, 23 insertions(+)
>> >>
>> >> diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
>> >> index f04a423..be6906b 100644
>> >> --- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
>> >> +++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
>> >> @@ -2278,6 +2278,17 @@ static int sync_toggles(struct mlx4_dev *dev)
>> >>               rd_toggle = swab32(readl(&priv->mfunc.comm->slave_read));
>> >>               if (wr_toggle == 0xffffffff || rd_toggle == 0xffffffff) {
>> >>                       /* PCI might be offline */
>> >> +
>> >> +                     /* If device removal has been requested,
>> >> +                      * do not continue retrying.
>> >> +                      */
>> >> +                     if (dev->persist->interface_state &
>> >> +                         MLX4_INTERFACE_STATE_NOWAIT) {
>> >> +                             mlx4_warn(dev,
>> >> +                                       "communication channel is offline\n");
>> >> +                             return -EIO;
>> >> +                     }
>> >> +
>> >>                       msleep(100);
>> >>                       wr_toggle = swab32(readl(&priv->mfunc.comm->
>> >>                                          slave_write));
>> >> diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
>> >> index 7183ac4..ea5e362 100644
>> >> --- a/drivers/net/ethernet/mellanox/mlx4/main.c
>> >> +++ b/drivers/net/ethernet/mellanox/mlx4/main.c
>> >> @@ -1916,6 +1916,14 @@ static int mlx4_comm_check_offline(struct mlx4_dev *dev)
>> >>                              (u32)(1 << COMM_CHAN_OFFLINE_OFFSET));
>> >>               if (!offline_bit)
>> >>                       return 0;
>> >> +
>> >> +             /* If device removal has been requested,
>> >> +              * do not continue retrying.
>> >> +              */
>> >> +             if (dev->persist->interface_state &
>> >> +                 MLX4_INTERFACE_STATE_NOWAIT)
>> >> +                     break;
>> >> +
>> >>               /* There are cases as part of AER/Reset flow that PF needs
>> >>                * around 100 msec to load. We therefore sleep for 100 msec
>> >>                * to allow other tasks to make use of that CPU during this
>> >> @@ -3930,6 +3938,9 @@ static void mlx4_remove_one(struct pci_dev *pdev)
>> >>       struct devlink *devlink = priv_to_devlink(priv);
>> >>       int active_vfs = 0;
>> >>
>> >> +     if (mlx4_is_slave(dev))
>> >> +             persist->interface_state |= MLX4_INTERFACE_STATE_NOWAIT;
>> >> +
>> >>       mutex_lock(&persist->interface_state_mutex);
>> >>       persist->interface_state |= MLX4_INTERFACE_STATE_DELETION;
>> >>       mutex_unlock(&persist->interface_state_mutex);
>> >> diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
>> >> index 42da355..d49f11d 100644
>> >> --- a/include/linux/mlx4/device.h
>> >> +++ b/include/linux/mlx4/device.h
>> >> @@ -468,6 +468,7 @@ enum {
>> >>       MLX4_INTERFACE_STATE_UP         = 1 << 0,
>> >>       MLX4_INTERFACE_STATE_DELETION   = 1 << 1,
>> >>       MLX4_INTERFACE_STATE_SHUTDOWN   = 1 << 2,
>> >> +     MLX4_INTERFACE_STATE_NOWAIT     = 1 << 2,
>> >
>> > So, NOWAIT and SHUTDOWN are defined to the same value, so it might be
>> > necessary to review it that is going to work as expected. Maybe doing
>> > the shutdown removal is a path to consider (commit
>> > b4353708f5a1c084fd73f1b6fd243b142157b173).
>> >
>> > Cascardo.
>> >
>> >>  };
>> >>
>> >>  #define MSTR_SM_CHANGE_MASK (MLX4_EQ_PORT_INFO_MSTR_SM_SL_CHANGE_MASK | \
>> >> --
>> >> 2.7.4
>> >>
>> >>
>> >> --
>> >> kernel-team mailing list
>> >> kernel-team@lists.ubuntu.com
>> >> https://lists.ubuntu.com/mailman/listinfo/kernel-team
>> >
>> > --
>> > kernel-team mailing list
>> > kernel-team@lists.ubuntu.com
>> > https://lists.ubuntu.com/mailman/listinfo/kernel-team
Joseph Salisbury April 3, 2017, 1:45 p.m.
I have a Yakkety test kernel now available that included commit
b4353708f5a1c084fd73f1b6fd243b142157b173 and commit
4cbe4dac82e423ecc9a0ba46af24a860853259f4.  Having commit b4353708f5a1c0
removes MLX4_INTERFACE_STATE_SHUTDOWN and all references to it.  It
allows commit 4cbe4dac82e423ecc9a0ba46af24a860853259f4 to pick cleanly
and not require a backport.
 
Xenial did not require b4353708f5a(It reverts commit 9d769311805 which
was added in v4.7-rc6) and it was a clean pick in Xenial.  Zesty already
has commit b4353708f5a.  I think both commits should be added to
Yakkety, not just 4cbe4dac82e4.

I'll post a link to the v2 test kernel in the bug and resubmit a v2 SRU
once everyone agrees thats the way to go.

Thanks,

Joe



On 04/03/2017 02:35 PM, Joshua R. Poulson wrote:
> Our friends at Mellanox have, we have been testing VF remove and add
> as part of SR-IOV in Azure testing. Our regression tests include
> on-premises Hyper-V as well.
>
> Thanks --jrp
>
> On Mon, Apr 3, 2017 at 6:32 AM, Thadeu Lima de Souza Cascardo
> <cascardo@canonical.com> wrote:
>> On Mon, Apr 03, 2017 at 05:11:34AM -0700, Joshua R. Poulson wrote:
>>> We're engaged to test the driver in the intended scenario and code
>>> path. We have a stable repro for the bug that this fixes.
>>>
>>> Thanks, --jrp
>>>
>> Is that on Yakkety, with this Yakkety backport as is?
>>
>> I am more concerned about the paths that would use SHUTDOWN, have you tested
>> these as well?
>>
>> Thanks.
>> Cascardo.
>>
>>> On Mon, Apr 3, 2017 at 2:26 AM, Thadeu Lima de Souza Cascardo
>>> <cascardo@canonical.com> wrote:
>>>> On Thu, Mar 30, 2017 at 04:35:59PM -0400, Joseph Salisbury wrote:
>>>>> From: Jack Morgenstein <jackm@dev.mellanox.co.il>
>>>>>
>>>>> BugLink: http://bugs.launchpad.net/bugs/1672785
>>>>>
>>>>> Some Hypervisors detach VFs from VMs by instantly causing an FLR event
>>>>> to be generated for a VF.
>>>>>
>>>>> In the mlx4 case, this will cause that VF's comm channel to be disabled
>>>>> before the VM has an opportunity to invoke the VF device's "shutdown"
>>>>> method.
>>>>>
>>>>> For such Hypervisors, there is a race condition between the VF's
>>>>> shutdown method and its internal-error detection/reset thread.
>>>>>
>>>>> The internal-error detection/reset thread (which runs every 5 seconds) also
>>>>> detects a disabled comm channel. If the internal-error detection/reset
>>>>> flow wins the race, we still get delays (while that flow tries repeatedly
>>>>> to detect comm-channel recovery).
>>>>>
>>>>> The cited commit fixed the command timeout problem when the
>>>>> internal-error detection/reset flow loses the race.
>>>>>
>>>>> This commit avoids the unneeded delays when the internal-error
>>>>> detection/reset flow wins.
>>>>>
>>>>> Fixes: d585df1c5ccf ("net/mlx4_core: Avoid command timeouts during VF driver device shutdown")
>>>>> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
>>>>> Reported-by: Simon Xiao <sixiao@microsoft.com>
>>>>> Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
>>>>> Signed-off-by: David S. Miller <davem@davemloft.net>
>>>>> (backported from commit 4cbe4dac82e423ecc9a0ba46af24a860853259f4)
>>>>> Signed-off-by: Joseph Salisbury <joseph.salisbury@canonical.com>
>>>>> ---
>>>>>  drivers/net/ethernet/mellanox/mlx4/cmd.c  | 11 +++++++++++
>>>>>  drivers/net/ethernet/mellanox/mlx4/main.c | 11 +++++++++++
>>>>>  include/linux/mlx4/device.h               |  1 +
>>>>>  3 files changed, 23 insertions(+)
>>>>>
>>>>> diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
>>>>> index f04a423..be6906b 100644
>>>>> --- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
>>>>> +++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
>>>>> @@ -2278,6 +2278,17 @@ static int sync_toggles(struct mlx4_dev *dev)
>>>>>               rd_toggle = swab32(readl(&priv->mfunc.comm->slave_read));
>>>>>               if (wr_toggle == 0xffffffff || rd_toggle == 0xffffffff) {
>>>>>                       /* PCI might be offline */
>>>>> +
>>>>> +                     /* If device removal has been requested,
>>>>> +                      * do not continue retrying.
>>>>> +                      */
>>>>> +                     if (dev->persist->interface_state &
>>>>> +                         MLX4_INTERFACE_STATE_NOWAIT) {
>>>>> +                             mlx4_warn(dev,
>>>>> +                                       "communication channel is offline\n");
>>>>> +                             return -EIO;
>>>>> +                     }
>>>>> +
>>>>>                       msleep(100);
>>>>>                       wr_toggle = swab32(readl(&priv->mfunc.comm->
>>>>>                                          slave_write));
>>>>> diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
>>>>> index 7183ac4..ea5e362 100644
>>>>> --- a/drivers/net/ethernet/mellanox/mlx4/main.c
>>>>> +++ b/drivers/net/ethernet/mellanox/mlx4/main.c
>>>>> @@ -1916,6 +1916,14 @@ static int mlx4_comm_check_offline(struct mlx4_dev *dev)
>>>>>                              (u32)(1 << COMM_CHAN_OFFLINE_OFFSET));
>>>>>               if (!offline_bit)
>>>>>                       return 0;
>>>>> +
>>>>> +             /* If device removal has been requested,
>>>>> +              * do not continue retrying.
>>>>> +              */
>>>>> +             if (dev->persist->interface_state &
>>>>> +                 MLX4_INTERFACE_STATE_NOWAIT)
>>>>> +                     break;
>>>>> +
>>>>>               /* There are cases as part of AER/Reset flow that PF needs
>>>>>                * around 100 msec to load. We therefore sleep for 100 msec
>>>>>                * to allow other tasks to make use of that CPU during this
>>>>> @@ -3930,6 +3938,9 @@ static void mlx4_remove_one(struct pci_dev *pdev)
>>>>>       struct devlink *devlink = priv_to_devlink(priv);
>>>>>       int active_vfs = 0;
>>>>>
>>>>> +     if (mlx4_is_slave(dev))
>>>>> +             persist->interface_state |= MLX4_INTERFACE_STATE_NOWAIT;
>>>>> +
>>>>>       mutex_lock(&persist->interface_state_mutex);
>>>>>       persist->interface_state |= MLX4_INTERFACE_STATE_DELETION;
>>>>>       mutex_unlock(&persist->interface_state_mutex);
>>>>> diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
>>>>> index 42da355..d49f11d 100644
>>>>> --- a/include/linux/mlx4/device.h
>>>>> +++ b/include/linux/mlx4/device.h
>>>>> @@ -468,6 +468,7 @@ enum {
>>>>>       MLX4_INTERFACE_STATE_UP         = 1 << 0,
>>>>>       MLX4_INTERFACE_STATE_DELETION   = 1 << 1,
>>>>>       MLX4_INTERFACE_STATE_SHUTDOWN   = 1 << 2,
>>>>> +     MLX4_INTERFACE_STATE_NOWAIT     = 1 << 2,
>>>> So, NOWAIT and SHUTDOWN are defined to the same value, so it might be
>>>> necessary to review it that is going to work as expected. Maybe doing
>>>> the shutdown removal is a path to consider (commit
>>>> b4353708f5a1c084fd73f1b6fd243b142157b173).
>>>>
>>>> Cascardo.
>>>>
>>>>>  };
>>>>>
>>>>>  #define MSTR_SM_CHANGE_MASK (MLX4_EQ_PORT_INFO_MSTR_SM_SL_CHANGE_MASK | \
>>>>> --
>>>>> 2.7.4
>>>>>
>>>>>
>>>>> --
>>>>> kernel-team mailing list
>>>>> kernel-team@lists.ubuntu.com
>>>>> https://lists.ubuntu.com/mailman/listinfo/kernel-team
>>>> --
>>>> kernel-team mailing list
>>>> kernel-team@lists.ubuntu.com
>>>> https://lists.ubuntu.com/mailman/listinfo/kernel-team
Joshua R. Poulson April 3, 2017, 2:03 p.m.
I'll test it as soon as I see it, thanks! --jrp

On Mon, Apr 3, 2017 at 6:45 AM, Joseph Salisbury
<joseph.salisbury@canonical.com> wrote:
> I have a Yakkety test kernel now available that included commit
> b4353708f5a1c084fd73f1b6fd243b142157b173 and commit
> 4cbe4dac82e423ecc9a0ba46af24a860853259f4.  Having commit b4353708f5a1c0
> removes MLX4_INTERFACE_STATE_SHUTDOWN and all references to it.  It
> allows commit 4cbe4dac82e423ecc9a0ba46af24a860853259f4 to pick cleanly
> and not require a backport.
>
> Xenial did not require b4353708f5a(It reverts commit 9d769311805 which
> was added in v4.7-rc6) and it was a clean pick in Xenial.  Zesty already
> has commit b4353708f5a.  I think both commits should be added to
> Yakkety, not just 4cbe4dac82e4.
>
> I'll post a link to the v2 test kernel in the bug and resubmit a v2 SRU
> once everyone agrees thats the way to go.
>
> Thanks,
>
> Joe
>
>
>
> On 04/03/2017 02:35 PM, Joshua R. Poulson wrote:
>> Our friends at Mellanox have, we have been testing VF remove and add
>> as part of SR-IOV in Azure testing. Our regression tests include
>> on-premises Hyper-V as well.
>>
>> Thanks --jrp
>>
>> On Mon, Apr 3, 2017 at 6:32 AM, Thadeu Lima de Souza Cascardo
>> <cascardo@canonical.com> wrote:
>>> On Mon, Apr 03, 2017 at 05:11:34AM -0700, Joshua R. Poulson wrote:
>>>> We're engaged to test the driver in the intended scenario and code
>>>> path. We have a stable repro for the bug that this fixes.
>>>>
>>>> Thanks, --jrp
>>>>
>>> Is that on Yakkety, with this Yakkety backport as is?
>>>
>>> I am more concerned about the paths that would use SHUTDOWN, have you tested
>>> these as well?
>>>
>>> Thanks.
>>> Cascardo.
>>>
>>>> On Mon, Apr 3, 2017 at 2:26 AM, Thadeu Lima de Souza Cascardo
>>>> <cascardo@canonical.com> wrote:
>>>>> On Thu, Mar 30, 2017 at 04:35:59PM -0400, Joseph Salisbury wrote:
>>>>>> From: Jack Morgenstein <jackm@dev.mellanox.co.il>
>>>>>>
>>>>>> BugLink: http://bugs.launchpad.net/bugs/1672785
>>>>>>
>>>>>> Some Hypervisors detach VFs from VMs by instantly causing an FLR event
>>>>>> to be generated for a VF.
>>>>>>
>>>>>> In the mlx4 case, this will cause that VF's comm channel to be disabled
>>>>>> before the VM has an opportunity to invoke the VF device's "shutdown"
>>>>>> method.
>>>>>>
>>>>>> For such Hypervisors, there is a race condition between the VF's
>>>>>> shutdown method and its internal-error detection/reset thread.
>>>>>>
>>>>>> The internal-error detection/reset thread (which runs every 5 seconds) also
>>>>>> detects a disabled comm channel. If the internal-error detection/reset
>>>>>> flow wins the race, we still get delays (while that flow tries repeatedly
>>>>>> to detect comm-channel recovery).
>>>>>>
>>>>>> The cited commit fixed the command timeout problem when the
>>>>>> internal-error detection/reset flow loses the race.
>>>>>>
>>>>>> This commit avoids the unneeded delays when the internal-error
>>>>>> detection/reset flow wins.
>>>>>>
>>>>>> Fixes: d585df1c5ccf ("net/mlx4_core: Avoid command timeouts during VF driver device shutdown")
>>>>>> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
>>>>>> Reported-by: Simon Xiao <sixiao@microsoft.com>
>>>>>> Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
>>>>>> Signed-off-by: David S. Miller <davem@davemloft.net>
>>>>>> (backported from commit 4cbe4dac82e423ecc9a0ba46af24a860853259f4)
>>>>>> Signed-off-by: Joseph Salisbury <joseph.salisbury@canonical.com>
>>>>>> ---
>>>>>>  drivers/net/ethernet/mellanox/mlx4/cmd.c  | 11 +++++++++++
>>>>>>  drivers/net/ethernet/mellanox/mlx4/main.c | 11 +++++++++++
>>>>>>  include/linux/mlx4/device.h               |  1 +
>>>>>>  3 files changed, 23 insertions(+)
>>>>>>
>>>>>> diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
>>>>>> index f04a423..be6906b 100644
>>>>>> --- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
>>>>>> +++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
>>>>>> @@ -2278,6 +2278,17 @@ static int sync_toggles(struct mlx4_dev *dev)
>>>>>>               rd_toggle = swab32(readl(&priv->mfunc.comm->slave_read));
>>>>>>               if (wr_toggle == 0xffffffff || rd_toggle == 0xffffffff) {
>>>>>>                       /* PCI might be offline */
>>>>>> +
>>>>>> +                     /* If device removal has been requested,
>>>>>> +                      * do not continue retrying.
>>>>>> +                      */
>>>>>> +                     if (dev->persist->interface_state &
>>>>>> +                         MLX4_INTERFACE_STATE_NOWAIT) {
>>>>>> +                             mlx4_warn(dev,
>>>>>> +                                       "communication channel is offline\n");
>>>>>> +                             return -EIO;
>>>>>> +                     }
>>>>>> +
>>>>>>                       msleep(100);
>>>>>>                       wr_toggle = swab32(readl(&priv->mfunc.comm->
>>>>>>                                          slave_write));
>>>>>> diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
>>>>>> index 7183ac4..ea5e362 100644
>>>>>> --- a/drivers/net/ethernet/mellanox/mlx4/main.c
>>>>>> +++ b/drivers/net/ethernet/mellanox/mlx4/main.c
>>>>>> @@ -1916,6 +1916,14 @@ static int mlx4_comm_check_offline(struct mlx4_dev *dev)
>>>>>>                              (u32)(1 << COMM_CHAN_OFFLINE_OFFSET));
>>>>>>               if (!offline_bit)
>>>>>>                       return 0;
>>>>>> +
>>>>>> +             /* If device removal has been requested,
>>>>>> +              * do not continue retrying.
>>>>>> +              */
>>>>>> +             if (dev->persist->interface_state &
>>>>>> +                 MLX4_INTERFACE_STATE_NOWAIT)
>>>>>> +                     break;
>>>>>> +
>>>>>>               /* There are cases as part of AER/Reset flow that PF needs
>>>>>>                * around 100 msec to load. We therefore sleep for 100 msec
>>>>>>                * to allow other tasks to make use of that CPU during this
>>>>>> @@ -3930,6 +3938,9 @@ static void mlx4_remove_one(struct pci_dev *pdev)
>>>>>>       struct devlink *devlink = priv_to_devlink(priv);
>>>>>>       int active_vfs = 0;
>>>>>>
>>>>>> +     if (mlx4_is_slave(dev))
>>>>>> +             persist->interface_state |= MLX4_INTERFACE_STATE_NOWAIT;
>>>>>> +
>>>>>>       mutex_lock(&persist->interface_state_mutex);
>>>>>>       persist->interface_state |= MLX4_INTERFACE_STATE_DELETION;
>>>>>>       mutex_unlock(&persist->interface_state_mutex);
>>>>>> diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
>>>>>> index 42da355..d49f11d 100644
>>>>>> --- a/include/linux/mlx4/device.h
>>>>>> +++ b/include/linux/mlx4/device.h
>>>>>> @@ -468,6 +468,7 @@ enum {
>>>>>>       MLX4_INTERFACE_STATE_UP         = 1 << 0,
>>>>>>       MLX4_INTERFACE_STATE_DELETION   = 1 << 1,
>>>>>>       MLX4_INTERFACE_STATE_SHUTDOWN   = 1 << 2,
>>>>>> +     MLX4_INTERFACE_STATE_NOWAIT     = 1 << 2,
>>>>> So, NOWAIT and SHUTDOWN are defined to the same value, so it might be
>>>>> necessary to review it that is going to work as expected. Maybe doing
>>>>> the shutdown removal is a path to consider (commit
>>>>> b4353708f5a1c084fd73f1b6fd243b142157b173).
>>>>>
>>>>> Cascardo.
>>>>>
>>>>>>  };
>>>>>>
>>>>>>  #define MSTR_SM_CHANGE_MASK (MLX4_EQ_PORT_INFO_MSTR_SM_SL_CHANGE_MASK | \
>>>>>> --
>>>>>> 2.7.4
>>>>>>
>>>>>>
>>>>>> --
>>>>>> kernel-team mailing list
>>>>>> kernel-team@lists.ubuntu.com
>>>>>> https://lists.ubuntu.com/mailman/listinfo/kernel-team
>>>>> --
>>>>> kernel-team mailing list
>>>>> kernel-team@lists.ubuntu.com
>>>>> https://lists.ubuntu.com/mailman/listinfo/kernel-team
>
>
>
Joseph Salisbury April 3, 2017, 2:15 p.m.
Thanks, Josh.  An explanation of the new test kernel and a link can be
found here:
https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1672785/comments/7

On 04/03/2017 03:03 PM, Joshua R. Poulson wrote:
> I'll test it as soon as I see it, thanks! --jrp
>
> On Mon, Apr 3, 2017 at 6:45 AM, Joseph Salisbury
> <joseph.salisbury@canonical.com> wrote:
>> I have a Yakkety test kernel now available that included commit
>> b4353708f5a1c084fd73f1b6fd243b142157b173 and commit
>> 4cbe4dac82e423ecc9a0ba46af24a860853259f4.  Having commit b4353708f5a1c0
>> removes MLX4_INTERFACE_STATE_SHUTDOWN and all references to it.  It
>> allows commit 4cbe4dac82e423ecc9a0ba46af24a860853259f4 to pick cleanly
>> and not require a backport.
>>
>> Xenial did not require b4353708f5a(It reverts commit 9d769311805 which
>> was added in v4.7-rc6) and it was a clean pick in Xenial.  Zesty already
>> has commit b4353708f5a.  I think both commits should be added to
>> Yakkety, not just 4cbe4dac82e4.
>>
>> I'll post a link to the v2 test kernel in the bug and resubmit a v2 SRU
>> once everyone agrees thats the way to go.
>>
>> Thanks,
>>
>> Joe
>>
>>
>>
>> On 04/03/2017 02:35 PM, Joshua R. Poulson wrote:
>>> Our friends at Mellanox have, we have been testing VF remove and add
>>> as part of SR-IOV in Azure testing. Our regression tests include
>>> on-premises Hyper-V as well.
>>>
>>> Thanks --jrp
>>>
>>> On Mon, Apr 3, 2017 at 6:32 AM, Thadeu Lima de Souza Cascardo
>>> <cascardo@canonical.com> wrote:
>>>> On Mon, Apr 03, 2017 at 05:11:34AM -0700, Joshua R. Poulson wrote:
>>>>> We're engaged to test the driver in the intended scenario and code
>>>>> path. We have a stable repro for the bug that this fixes.
>>>>>
>>>>> Thanks, --jrp
>>>>>
>>>> Is that on Yakkety, with this Yakkety backport as is?
>>>>
>>>> I am more concerned about the paths that would use SHUTDOWN, have you tested
>>>> these as well?
>>>>
>>>> Thanks.
>>>> Cascardo.
>>>>
>>>>> On Mon, Apr 3, 2017 at 2:26 AM, Thadeu Lima de Souza Cascardo
>>>>> <cascardo@canonical.com> wrote:
>>>>>> On Thu, Mar 30, 2017 at 04:35:59PM -0400, Joseph Salisbury wrote:
>>>>>>> From: Jack Morgenstein <jackm@dev.mellanox.co.il>
>>>>>>>
>>>>>>> BugLink: http://bugs.launchpad.net/bugs/1672785
>>>>>>>
>>>>>>> Some Hypervisors detach VFs from VMs by instantly causing an FLR event
>>>>>>> to be generated for a VF.
>>>>>>>
>>>>>>> In the mlx4 case, this will cause that VF's comm channel to be disabled
>>>>>>> before the VM has an opportunity to invoke the VF device's "shutdown"
>>>>>>> method.
>>>>>>>
>>>>>>> For such Hypervisors, there is a race condition between the VF's
>>>>>>> shutdown method and its internal-error detection/reset thread.
>>>>>>>
>>>>>>> The internal-error detection/reset thread (which runs every 5 seconds) also
>>>>>>> detects a disabled comm channel. If the internal-error detection/reset
>>>>>>> flow wins the race, we still get delays (while that flow tries repeatedly
>>>>>>> to detect comm-channel recovery).
>>>>>>>
>>>>>>> The cited commit fixed the command timeout problem when the
>>>>>>> internal-error detection/reset flow loses the race.
>>>>>>>
>>>>>>> This commit avoids the unneeded delays when the internal-error
>>>>>>> detection/reset flow wins.
>>>>>>>
>>>>>>> Fixes: d585df1c5ccf ("net/mlx4_core: Avoid command timeouts during VF driver device shutdown")
>>>>>>> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
>>>>>>> Reported-by: Simon Xiao <sixiao@microsoft.com>
>>>>>>> Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
>>>>>>> Signed-off-by: David S. Miller <davem@davemloft.net>
>>>>>>> (backported from commit 4cbe4dac82e423ecc9a0ba46af24a860853259f4)
>>>>>>> Signed-off-by: Joseph Salisbury <joseph.salisbury@canonical.com>
>>>>>>> ---
>>>>>>>  drivers/net/ethernet/mellanox/mlx4/cmd.c  | 11 +++++++++++
>>>>>>>  drivers/net/ethernet/mellanox/mlx4/main.c | 11 +++++++++++
>>>>>>>  include/linux/mlx4/device.h               |  1 +
>>>>>>>  3 files changed, 23 insertions(+)
>>>>>>>
>>>>>>> diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
>>>>>>> index f04a423..be6906b 100644
>>>>>>> --- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
>>>>>>> +++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
>>>>>>> @@ -2278,6 +2278,17 @@ static int sync_toggles(struct mlx4_dev *dev)
>>>>>>>               rd_toggle = swab32(readl(&priv->mfunc.comm->slave_read));
>>>>>>>               if (wr_toggle == 0xffffffff || rd_toggle == 0xffffffff) {
>>>>>>>                       /* PCI might be offline */
>>>>>>> +
>>>>>>> +                     /* If device removal has been requested,
>>>>>>> +                      * do not continue retrying.
>>>>>>> +                      */
>>>>>>> +                     if (dev->persist->interface_state &
>>>>>>> +                         MLX4_INTERFACE_STATE_NOWAIT) {
>>>>>>> +                             mlx4_warn(dev,
>>>>>>> +                                       "communication channel is offline\n");
>>>>>>> +                             return -EIO;
>>>>>>> +                     }
>>>>>>> +
>>>>>>>                       msleep(100);
>>>>>>>                       wr_toggle = swab32(readl(&priv->mfunc.comm->
>>>>>>>                                          slave_write));
>>>>>>> diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
>>>>>>> index 7183ac4..ea5e362 100644
>>>>>>> --- a/drivers/net/ethernet/mellanox/mlx4/main.c
>>>>>>> +++ b/drivers/net/ethernet/mellanox/mlx4/main.c
>>>>>>> @@ -1916,6 +1916,14 @@ static int mlx4_comm_check_offline(struct mlx4_dev *dev)
>>>>>>>                              (u32)(1 << COMM_CHAN_OFFLINE_OFFSET));
>>>>>>>               if (!offline_bit)
>>>>>>>                       return 0;
>>>>>>> +
>>>>>>> +             /* If device removal has been requested,
>>>>>>> +              * do not continue retrying.
>>>>>>> +              */
>>>>>>> +             if (dev->persist->interface_state &
>>>>>>> +                 MLX4_INTERFACE_STATE_NOWAIT)
>>>>>>> +                     break;
>>>>>>> +
>>>>>>>               /* There are cases as part of AER/Reset flow that PF needs
>>>>>>>                * around 100 msec to load. We therefore sleep for 100 msec
>>>>>>>                * to allow other tasks to make use of that CPU during this
>>>>>>> @@ -3930,6 +3938,9 @@ static void mlx4_remove_one(struct pci_dev *pdev)
>>>>>>>       struct devlink *devlink = priv_to_devlink(priv);
>>>>>>>       int active_vfs = 0;
>>>>>>>
>>>>>>> +     if (mlx4_is_slave(dev))
>>>>>>> +             persist->interface_state |= MLX4_INTERFACE_STATE_NOWAIT;
>>>>>>> +
>>>>>>>       mutex_lock(&persist->interface_state_mutex);
>>>>>>>       persist->interface_state |= MLX4_INTERFACE_STATE_DELETION;
>>>>>>>       mutex_unlock(&persist->interface_state_mutex);
>>>>>>> diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
>>>>>>> index 42da355..d49f11d 100644
>>>>>>> --- a/include/linux/mlx4/device.h
>>>>>>> +++ b/include/linux/mlx4/device.h
>>>>>>> @@ -468,6 +468,7 @@ enum {
>>>>>>>       MLX4_INTERFACE_STATE_UP         = 1 << 0,
>>>>>>>       MLX4_INTERFACE_STATE_DELETION   = 1 << 1,
>>>>>>>       MLX4_INTERFACE_STATE_SHUTDOWN   = 1 << 2,
>>>>>>> +     MLX4_INTERFACE_STATE_NOWAIT     = 1 << 2,
>>>>>> So, NOWAIT and SHUTDOWN are defined to the same value, so it might be
>>>>>> necessary to review it that is going to work as expected. Maybe doing
>>>>>> the shutdown removal is a path to consider (commit
>>>>>> b4353708f5a1c084fd73f1b6fd243b142157b173).
>>>>>>
>>>>>> Cascardo.
>>>>>>
>>>>>>>  };
>>>>>>>
>>>>>>>  #define MSTR_SM_CHANGE_MASK (MLX4_EQ_PORT_INFO_MSTR_SM_SL_CHANGE_MASK | \
>>>>>>> --
>>>>>>> 2.7.4
>>>>>>>
>>>>>>>
>>>>>>> --
>>>>>>> kernel-team mailing list
>>>>>>> kernel-team@lists.ubuntu.com
>>>>>>> https://lists.ubuntu.com/mailman/listinfo/kernel-team
>>>>>> --
>>>>>> kernel-team mailing list
>>>>>> kernel-team@lists.ubuntu.com
>>>>>> https://lists.ubuntu.com/mailman/listinfo/kernel-team
>>
>>
Joshua R. Poulson April 3, 2017, 4:09 p.m.
It's in my queue. --jrp

On Mon, Apr 3, 2017 at 7:15 AM, Joseph Salisbury
<joseph.salisbury@canonical.com> wrote:
> Thanks, Josh.  An explanation of the new test kernel and a link can be
> found here:
> https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1672785/comments/7
>
> On 04/03/2017 03:03 PM, Joshua R. Poulson wrote:
>> I'll test it as soon as I see it, thanks! --jrp
>>
>> On Mon, Apr 3, 2017 at 6:45 AM, Joseph Salisbury
>> <joseph.salisbury@canonical.com> wrote:
>>> I have a Yakkety test kernel now available that included commit
>>> b4353708f5a1c084fd73f1b6fd243b142157b173 and commit
>>> 4cbe4dac82e423ecc9a0ba46af24a860853259f4.  Having commit b4353708f5a1c0
>>> removes MLX4_INTERFACE_STATE_SHUTDOWN and all references to it.  It
>>> allows commit 4cbe4dac82e423ecc9a0ba46af24a860853259f4 to pick cleanly
>>> and not require a backport.
>>>
>>> Xenial did not require b4353708f5a(It reverts commit 9d769311805 which
>>> was added in v4.7-rc6) and it was a clean pick in Xenial.  Zesty already
>>> has commit b4353708f5a.  I think both commits should be added to
>>> Yakkety, not just 4cbe4dac82e4.
>>>
>>> I'll post a link to the v2 test kernel in the bug and resubmit a v2 SRU
>>> once everyone agrees thats the way to go.
>>>
>>> Thanks,
>>>
>>> Joe
>>>
>>>
>>>
>>> On 04/03/2017 02:35 PM, Joshua R. Poulson wrote:
>>>> Our friends at Mellanox have, we have been testing VF remove and add
>>>> as part of SR-IOV in Azure testing. Our regression tests include
>>>> on-premises Hyper-V as well.
>>>>
>>>> Thanks --jrp
>>>>
>>>> On Mon, Apr 3, 2017 at 6:32 AM, Thadeu Lima de Souza Cascardo
>>>> <cascardo@canonical.com> wrote:
>>>>> On Mon, Apr 03, 2017 at 05:11:34AM -0700, Joshua R. Poulson wrote:
>>>>>> We're engaged to test the driver in the intended scenario and code
>>>>>> path. We have a stable repro for the bug that this fixes.
>>>>>>
>>>>>> Thanks, --jrp
>>>>>>
>>>>> Is that on Yakkety, with this Yakkety backport as is?
>>>>>
>>>>> I am more concerned about the paths that would use SHUTDOWN, have you tested
>>>>> these as well?
>>>>>
>>>>> Thanks.
>>>>> Cascardo.
>>>>>
>>>>>> On Mon, Apr 3, 2017 at 2:26 AM, Thadeu Lima de Souza Cascardo
>>>>>> <cascardo@canonical.com> wrote:
>>>>>>> On Thu, Mar 30, 2017 at 04:35:59PM -0400, Joseph Salisbury wrote:
>>>>>>>> From: Jack Morgenstein <jackm@dev.mellanox.co.il>
>>>>>>>>
>>>>>>>> BugLink: http://bugs.launchpad.net/bugs/1672785
>>>>>>>>
>>>>>>>> Some Hypervisors detach VFs from VMs by instantly causing an FLR event
>>>>>>>> to be generated for a VF.
>>>>>>>>
>>>>>>>> In the mlx4 case, this will cause that VF's comm channel to be disabled
>>>>>>>> before the VM has an opportunity to invoke the VF device's "shutdown"
>>>>>>>> method.
>>>>>>>>
>>>>>>>> For such Hypervisors, there is a race condition between the VF's
>>>>>>>> shutdown method and its internal-error detection/reset thread.
>>>>>>>>
>>>>>>>> The internal-error detection/reset thread (which runs every 5 seconds) also
>>>>>>>> detects a disabled comm channel. If the internal-error detection/reset
>>>>>>>> flow wins the race, we still get delays (while that flow tries repeatedly
>>>>>>>> to detect comm-channel recovery).
>>>>>>>>
>>>>>>>> The cited commit fixed the command timeout problem when the
>>>>>>>> internal-error detection/reset flow loses the race.
>>>>>>>>
>>>>>>>> This commit avoids the unneeded delays when the internal-error
>>>>>>>> detection/reset flow wins.
>>>>>>>>
>>>>>>>> Fixes: d585df1c5ccf ("net/mlx4_core: Avoid command timeouts during VF driver device shutdown")
>>>>>>>> Signed-off-by: Jack Morgenstein <jackm@dev.mellanox.co.il>
>>>>>>>> Reported-by: Simon Xiao <sixiao@microsoft.com>
>>>>>>>> Signed-off-by: Tariq Toukan <tariqt@mellanox.com>
>>>>>>>> Signed-off-by: David S. Miller <davem@davemloft.net>
>>>>>>>> (backported from commit 4cbe4dac82e423ecc9a0ba46af24a860853259f4)
>>>>>>>> Signed-off-by: Joseph Salisbury <joseph.salisbury@canonical.com>
>>>>>>>> ---
>>>>>>>>  drivers/net/ethernet/mellanox/mlx4/cmd.c  | 11 +++++++++++
>>>>>>>>  drivers/net/ethernet/mellanox/mlx4/main.c | 11 +++++++++++
>>>>>>>>  include/linux/mlx4/device.h               |  1 +
>>>>>>>>  3 files changed, 23 insertions(+)
>>>>>>>>
>>>>>>>> diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
>>>>>>>> index f04a423..be6906b 100644
>>>>>>>> --- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
>>>>>>>> +++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
>>>>>>>> @@ -2278,6 +2278,17 @@ static int sync_toggles(struct mlx4_dev *dev)
>>>>>>>>               rd_toggle = swab32(readl(&priv->mfunc.comm->slave_read));
>>>>>>>>               if (wr_toggle == 0xffffffff || rd_toggle == 0xffffffff) {
>>>>>>>>                       /* PCI might be offline */
>>>>>>>> +
>>>>>>>> +                     /* If device removal has been requested,
>>>>>>>> +                      * do not continue retrying.
>>>>>>>> +                      */
>>>>>>>> +                     if (dev->persist->interface_state &
>>>>>>>> +                         MLX4_INTERFACE_STATE_NOWAIT) {
>>>>>>>> +                             mlx4_warn(dev,
>>>>>>>> +                                       "communication channel is offline\n");
>>>>>>>> +                             return -EIO;
>>>>>>>> +                     }
>>>>>>>> +
>>>>>>>>                       msleep(100);
>>>>>>>>                       wr_toggle = swab32(readl(&priv->mfunc.comm->
>>>>>>>>                                          slave_write));
>>>>>>>> diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
>>>>>>>> index 7183ac4..ea5e362 100644
>>>>>>>> --- a/drivers/net/ethernet/mellanox/mlx4/main.c
>>>>>>>> +++ b/drivers/net/ethernet/mellanox/mlx4/main.c
>>>>>>>> @@ -1916,6 +1916,14 @@ static int mlx4_comm_check_offline(struct mlx4_dev *dev)
>>>>>>>>                              (u32)(1 << COMM_CHAN_OFFLINE_OFFSET));
>>>>>>>>               if (!offline_bit)
>>>>>>>>                       return 0;
>>>>>>>> +
>>>>>>>> +             /* If device removal has been requested,
>>>>>>>> +              * do not continue retrying.
>>>>>>>> +              */
>>>>>>>> +             if (dev->persist->interface_state &
>>>>>>>> +                 MLX4_INTERFACE_STATE_NOWAIT)
>>>>>>>> +                     break;
>>>>>>>> +
>>>>>>>>               /* There are cases as part of AER/Reset flow that PF needs
>>>>>>>>                * around 100 msec to load. We therefore sleep for 100 msec
>>>>>>>>                * to allow other tasks to make use of that CPU during this
>>>>>>>> @@ -3930,6 +3938,9 @@ static void mlx4_remove_one(struct pci_dev *pdev)
>>>>>>>>       struct devlink *devlink = priv_to_devlink(priv);
>>>>>>>>       int active_vfs = 0;
>>>>>>>>
>>>>>>>> +     if (mlx4_is_slave(dev))
>>>>>>>> +             persist->interface_state |= MLX4_INTERFACE_STATE_NOWAIT;
>>>>>>>> +
>>>>>>>>       mutex_lock(&persist->interface_state_mutex);
>>>>>>>>       persist->interface_state |= MLX4_INTERFACE_STATE_DELETION;
>>>>>>>>       mutex_unlock(&persist->interface_state_mutex);
>>>>>>>> diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
>>>>>>>> index 42da355..d49f11d 100644
>>>>>>>> --- a/include/linux/mlx4/device.h
>>>>>>>> +++ b/include/linux/mlx4/device.h
>>>>>>>> @@ -468,6 +468,7 @@ enum {
>>>>>>>>       MLX4_INTERFACE_STATE_UP         = 1 << 0,
>>>>>>>>       MLX4_INTERFACE_STATE_DELETION   = 1 << 1,
>>>>>>>>       MLX4_INTERFACE_STATE_SHUTDOWN   = 1 << 2,
>>>>>>>> +     MLX4_INTERFACE_STATE_NOWAIT     = 1 << 2,
>>>>>>> So, NOWAIT and SHUTDOWN are defined to the same value, so it might be
>>>>>>> necessary to review it that is going to work as expected. Maybe doing
>>>>>>> the shutdown removal is a path to consider (commit
>>>>>>> b4353708f5a1c084fd73f1b6fd243b142157b173).
>>>>>>>
>>>>>>> Cascardo.
>>>>>>>
>>>>>>>>  };
>>>>>>>>
>>>>>>>>  #define MSTR_SM_CHANGE_MASK (MLX4_EQ_PORT_INFO_MSTR_SM_SL_CHANGE_MASK | \
>>>>>>>> --
>>>>>>>> 2.7.4
>>>>>>>>
>>>>>>>>
>>>>>>>> --
>>>>>>>> kernel-team mailing list
>>>>>>>> kernel-team@lists.ubuntu.com
>>>>>>>> https://lists.ubuntu.com/mailman/listinfo/kernel-team
>>>>>>> --
>>>>>>> kernel-team mailing list
>>>>>>> kernel-team@lists.ubuntu.com
>>>>>>> https://lists.ubuntu.com/mailman/listinfo/kernel-team
>>>
>>>
>

Patch hide | download patch | download mbox

diff --git a/drivers/net/ethernet/mellanox/mlx4/cmd.c b/drivers/net/ethernet/mellanox/mlx4/cmd.c
index f04a423..be6906b 100644
--- a/drivers/net/ethernet/mellanox/mlx4/cmd.c
+++ b/drivers/net/ethernet/mellanox/mlx4/cmd.c
@@ -2278,6 +2278,17 @@  static int sync_toggles(struct mlx4_dev *dev)
 		rd_toggle = swab32(readl(&priv->mfunc.comm->slave_read));
 		if (wr_toggle == 0xffffffff || rd_toggle == 0xffffffff) {
 			/* PCI might be offline */
+
+			/* If device removal has been requested,
+			 * do not continue retrying.
+			 */
+			if (dev->persist->interface_state &
+			    MLX4_INTERFACE_STATE_NOWAIT) {
+				mlx4_warn(dev,
+					  "communication channel is offline\n");
+				return -EIO;
+			}
+
 			msleep(100);
 			wr_toggle = swab32(readl(&priv->mfunc.comm->
 					   slave_write));
diff --git a/drivers/net/ethernet/mellanox/mlx4/main.c b/drivers/net/ethernet/mellanox/mlx4/main.c
index 7183ac4..ea5e362 100644
--- a/drivers/net/ethernet/mellanox/mlx4/main.c
+++ b/drivers/net/ethernet/mellanox/mlx4/main.c
@@ -1916,6 +1916,14 @@  static int mlx4_comm_check_offline(struct mlx4_dev *dev)
 			       (u32)(1 << COMM_CHAN_OFFLINE_OFFSET));
 		if (!offline_bit)
 			return 0;
+
+		/* If device removal has been requested,
+		 * do not continue retrying.
+		 */
+		if (dev->persist->interface_state &
+		    MLX4_INTERFACE_STATE_NOWAIT)
+			break;
+
 		/* There are cases as part of AER/Reset flow that PF needs
 		 * around 100 msec to load. We therefore sleep for 100 msec
 		 * to allow other tasks to make use of that CPU during this
@@ -3930,6 +3938,9 @@  static void mlx4_remove_one(struct pci_dev *pdev)
 	struct devlink *devlink = priv_to_devlink(priv);
 	int active_vfs = 0;
 
+	if (mlx4_is_slave(dev))
+		persist->interface_state |= MLX4_INTERFACE_STATE_NOWAIT;
+
 	mutex_lock(&persist->interface_state_mutex);
 	persist->interface_state |= MLX4_INTERFACE_STATE_DELETION;
 	mutex_unlock(&persist->interface_state_mutex);
diff --git a/include/linux/mlx4/device.h b/include/linux/mlx4/device.h
index 42da355..d49f11d 100644
--- a/include/linux/mlx4/device.h
+++ b/include/linux/mlx4/device.h
@@ -468,6 +468,7 @@  enum {
 	MLX4_INTERFACE_STATE_UP		= 1 << 0,
 	MLX4_INTERFACE_STATE_DELETION	= 1 << 1,
 	MLX4_INTERFACE_STATE_SHUTDOWN	= 1 << 2,
+	MLX4_INTERFACE_STATE_NOWAIT	= 1 << 2,
 };
 
 #define MSTR_SM_CHANGE_MASK (MLX4_EQ_PORT_INFO_MSTR_SM_SL_CHANGE_MASK | \