diff mbox series

[V2] mpt3sas: support target smid for [abort|query] task

Message ID 20190714034415epcms2p25f9787cb71993a30f58524d2f355b543@epcms2p2 (mailing list archive)
State Mainlined
Commit 8f55c307ab577af74efab066ba1e45b0e51e4bae
Headers show
Series [V2] mpt3sas: support target smid for [abort|query] task | expand

Commit Message

Minwoo Im July 14, 2019, 3:44 a.m. UTC
We can request task management IOCTL command(MPI2_FUNCTION_SCSI_TASK_MGMT)
to /dev/mpt3ctl.  If the given task_type is either abort task or query
task, it may need a field named "Initiator Port Transfer Tag to Manage"
in the IU.

Current code does not support to check target IPTT tag from the
tm_request.  This patch introduces to check TaskMID given from the
userspace as a target tag.  We have a rule of relationship between
(struct request *req->tag) and smid in mpt3sas_base.c:

3318 u16
3319 mpt3sas_base_get_smid_scsiio(struct MPT3SAS_ADAPTER *ioc, u8 cb_idx,
3320         struct scsi_cmnd *scmd)
3321 {
3322         struct scsiio_tracker *request = scsi_cmd_priv(scmd);
3323         unsigned int tag = scmd->request->tag;
3324         u16 smid;
3325
3326         smid = tag + 1;

So if we want to abort a request tagged #X, then we can pass (X + 1) to
this IOCTL handler.  Otherwise, user space just can pass 0 TaskMID to
abort the first outstanding smid which is legacy behaviour.

Cc: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Cc: Suganath Prabu Subramani <suganath-prabu.subramani@broadcom.com>
Cc: Sathya Prakash <sathya.prakash@broadcom.com>
Cc: James E.J. Bottomley <jejb@linux.ibm.com>
Cc: Martin K. Petersen <martin.petersen@oracle.com>
Cc: MPT-FusionLinux.pdl@broadcom.com
Signed-off-by: Minwoo Im <minwoo.im@samsung.com>
---
 drivers/scsi/mpt3sas/mpt3sas_ctl.c | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)

Comments

Hannes Reinecke July 15, 2019, 6:13 a.m. UTC | #1
On 7/14/19 5:44 AM, Minwoo Im wrote:
> We can request task management IOCTL command(MPI2_FUNCTION_SCSI_TASK_MGMT)
> to /dev/mpt3ctl.  If the given task_type is either abort task or query
> task, it may need a field named "Initiator Port Transfer Tag to Manage"
> in the IU.
> 
> Current code does not support to check target IPTT tag from the
> tm_request.  This patch introduces to check TaskMID given from the
> userspace as a target tag.  We have a rule of relationship between
> (struct request *req->tag) and smid in mpt3sas_base.c:
> 
> 3318 u16
> 3319 mpt3sas_base_get_smid_scsiio(struct MPT3SAS_ADAPTER *ioc, u8 cb_idx,
> 3320         struct scsi_cmnd *scmd)
> 3321 {
> 3322         struct scsiio_tracker *request = scsi_cmd_priv(scmd);
> 3323         unsigned int tag = scmd->request->tag;
> 3324         u16 smid;
> 3325
> 3326         smid = tag + 1;
> 
> So if we want to abort a request tagged #X, then we can pass (X + 1) to
> this IOCTL handler.  Otherwise, user space just can pass 0 TaskMID to
> abort the first outstanding smid which is legacy behaviour.
> 
> Cc: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
> Cc: Suganath Prabu Subramani <suganath-prabu.subramani@broadcom.com>
> Cc: Sathya Prakash <sathya.prakash@broadcom.com>
> Cc: James E.J. Bottomley <jejb@linux.ibm.com>
> Cc: Martin K. Petersen <martin.petersen@oracle.com>
> Cc: MPT-FusionLinux.pdl@broadcom.com
> Signed-off-by: Minwoo Im <minwoo.im@samsung.com>
> ---
>  drivers/scsi/mpt3sas/mpt3sas_ctl.c | 12 ++++++++++--
>  1 file changed, 10 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/scsi/mpt3sas/mpt3sas_ctl.c b/drivers/scsi/mpt3sas/mpt3sas_ctl.c
> index b2bb47c14d35..f6b8fd90610a 100644
> --- a/drivers/scsi/mpt3sas/mpt3sas_ctl.c
> +++ b/drivers/scsi/mpt3sas/mpt3sas_ctl.c
> @@ -596,8 +596,16 @@ _ctl_set_task_mid(struct MPT3SAS_ADAPTER *ioc, struct mpt3_ioctl_command *karg,
>  		if (priv_data->sas_target->handle != handle)
>  			continue;
>  		st = scsi_cmd_priv(scmd);
> -		tm_request->TaskMID = cpu_to_le16(st->smid);
> -		found = 1;
> +
> +		/*
> +		 * If the given TaskMID from the user space is zero, then the
> +		 * first outstanding smid will be picked up.  Otherwise,
> +		 * targeted smid will be the one.
> +		 */
> +		if (!tm_request->TaskMID || tm_request->TaskMID == st->smid) {
> +			tm_request->TaskMID = cpu_to_le16(st->smid);
> +			found = 1;
> +		}
>  	}
>  
>  	if (!found) {
> 
I think this is fundamentally wrong.
ABORT_TASK is used to abort a single task, which of course has to be
known beforehand. If you don't know the task, what exactly do you hope
to achieve here? Aborting random I/O?
Or, even worse, aborting I/O the driver uses internally and corrupt the
internal workflow of the driver?

We should simply disallow any ABORT TASK from userspace if the TaskMID
is zero. And I would even argue to disabllow ABORT TASK from userspace
completely, as the smid is never relayed to userland, and as such the
user cannot know which task should be aborted.

Cheers,

Hannes
Minwoo Im July 16, 2019, 5:44 a.m. UTC | #2
> I think this is fundamentally wrong.
> ABORT_TASK is used to abort a single task, which of course has to be
> known beforehand. If you don't know the task, what exactly do you hope
> to achieve here? Aborting random I/O?
> Or, even worse, aborting I/O the driver uses internally and corrupt the
> internal workflow of the driver?

This patch is nothing but a case-addition to the existing code.  I also
have a doubt here why the first picked SMID should be aborted/queried,
but not for this time in this patch.

> 
> We should simply disallow any ABORT TASK from userspace if the TaskMID
> is zero. And I would even argue to disabllow ABORT TASK from userspace
> completely, as the smid is never relayed to userland, and as such the
> user cannot know which task should be aborted.

System administrator might want to query task or abort task if something
happens based on the tag in block layer via debugfs or some logs.
You're right that userspaces has nothing to do with the tag generation
which will be held inside block layer.  But some of administrator would
know the relationship between smid and tag which can be found at debugfs
or the logs.

I'm not sure if it's okay to be picked, but if we can request TMF for the
Targeted smid to the HBA firmware, it would be great to test devices or
Figure out what happened in the target device.

Thanks,
Minwoo Im July 17, 2019, 12:12 p.m. UTC | #3
On 19-07-15 08:13:36, Hannes Reinecke wrote:
> I think this is fundamentally wrong.
> ABORT_TASK is used to abort a single task, which of course has to be
> known beforehand. If you don't know the task, what exactly do you hope
> to achieve here? Aborting random I/O?
> Or, even worse, aborting I/O the driver uses internally and corrupt the
> internal workflow of the driver?
> 
> We should simply disallow any ABORT TASK from userspace if the TaskMID
> is zero. And I would even argue to disabllow ABORT TASK from userspace
> completely, as the smid is never relayed to userland, and as such the
> user cannot know which task should be aborted.
> 
> Cheers,
> 
> Hannes

Sreekanth,

Could you please give some thoughts about what Hannes said?
Sreekanth Reddy July 23, 2019, 11:27 a.m. UTC | #4
On Mon, Jul 15, 2019 at 11:43 AM Hannes Reinecke <hare@suse.de> wrote:
>
> On 7/14/19 5:44 AM, Minwoo Im wrote:
> > We can request task management IOCTL command(MPI2_FUNCTION_SCSI_TASK_MGMT)
> > to /dev/mpt3ctl.  If the given task_type is either abort task or query
> > task, it may need a field named "Initiator Port Transfer Tag to Manage"
> > in the IU.
> >
> > Current code does not support to check target IPTT tag from the
> > tm_request.  This patch introduces to check TaskMID given from the
> > userspace as a target tag.  We have a rule of relationship between
> > (struct request *req->tag) and smid in mpt3sas_base.c:
> >
> > 3318 u16
> > 3319 mpt3sas_base_get_smid_scsiio(struct MPT3SAS_ADAPTER *ioc, u8 cb_idx,
> > 3320         struct scsi_cmnd *scmd)
> > 3321 {
> > 3322         struct scsiio_tracker *request = scsi_cmd_priv(scmd);
> > 3323         unsigned int tag = scmd->request->tag;
> > 3324         u16 smid;
> > 3325
> > 3326         smid = tag + 1;
> >
> > So if we want to abort a request tagged #X, then we can pass (X + 1) to
> > this IOCTL handler.  Otherwise, user space just can pass 0 TaskMID to
> > abort the first outstanding smid which is legacy behaviour.
> >
> > Cc: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
> > Cc: Suganath Prabu Subramani <suganath-prabu.subramani@broadcom.com>
> > Cc: Sathya Prakash <sathya.prakash@broadcom.com>
> > Cc: James E.J. Bottomley <jejb@linux.ibm.com>
> > Cc: Martin K. Petersen <martin.petersen@oracle.com>
> > Cc: MPT-FusionLinux.pdl@broadcom.com
> > Signed-off-by: Minwoo Im <minwoo.im@samsung.com>
> > ---
> >  drivers/scsi/mpt3sas/mpt3sas_ctl.c | 12 ++++++++++--
> >  1 file changed, 10 insertions(+), 2 deletions(-)
> >
> > diff --git a/drivers/scsi/mpt3sas/mpt3sas_ctl.c b/drivers/scsi/mpt3sas/mpt3sas_ctl.c
> > index b2bb47c14d35..f6b8fd90610a 100644
> > --- a/drivers/scsi/mpt3sas/mpt3sas_ctl.c
> > +++ b/drivers/scsi/mpt3sas/mpt3sas_ctl.c
> > @@ -596,8 +596,16 @@ _ctl_set_task_mid(struct MPT3SAS_ADAPTER *ioc, struct mpt3_ioctl_command *karg,
> >               if (priv_data->sas_target->handle != handle)
> >                       continue;
> >               st = scsi_cmd_priv(scmd);
> > -             tm_request->TaskMID = cpu_to_le16(st->smid);
> > -             found = 1;
> > +
> > +             /*
> > +              * If the given TaskMID from the user space is zero, then the
> > +              * first outstanding smid will be picked up.  Otherwise,
> > +              * targeted smid will be the one.
> > +              */
> > +             if (!tm_request->TaskMID || tm_request->TaskMID == st->smid) {
> > +                     tm_request->TaskMID = cpu_to_le16(st->smid);
> > +                     found = 1;
> > +             }
> >       }
> >
> >       if (!found) {
> >
> I think this is fundamentally wrong.
> ABORT_TASK is used to abort a single task, which of course has to be
> known beforehand. If you don't know the task, what exactly do you hope
> to achieve here? Aborting random I/O?
> Or, even worse, aborting I/O the driver uses internally and corrupt the
> internal workflow of the driver?
>
> We should simply disallow any ABORT TASK from userspace if the TaskMID
> is zero. And I would even argue to disabllow ABORT TASK from userspace
> completely, as the smid is never relayed to userland, and as such the
> user cannot know which task should be aborted.

Hannes,

This interface was added long time back in mpt2sas driver and I don't
have exact reason of adding this interface at that time.
But I know that this interface is still used by BRCM test team & few
customers only for some functionality and regression testing.

Thanks,
Sreekanth

>
> Cheers,
>
> Hannes
> --
> Dr. Hannes Reinecke                Teamlead Storage & Networking
> hare@suse.de                                   +49 911 74053 688
> SUSE LINUX GmbH, Maxfeldstr. 5, 90409 Nürnberg
> GF: Felix Imendörffer, Mary Higgins, Sri Rasiah
> HRB 21284 (AG Nürnberg)
Minwoo Im July 23, 2019, 1:33 p.m. UTC | #5
On 19-07-23 16:57:49, Sreekanth Reddy wrote:
> On Mon, Jul 15, 2019 at 11:43 AM Hannes Reinecke <hare@suse.de> wrote:
> >
> > On 7/14/19 5:44 AM, Minwoo Im wrote:
> > > We can request task management IOCTL command(MPI2_FUNCTION_SCSI_TASK_MGMT)
> > > to /dev/mpt3ctl.  If the given task_type is either abort task or query
> > > task, it may need a field named "Initiator Port Transfer Tag to Manage"
> > > in the IU.
> > >
> > > Current code does not support to check target IPTT tag from the
> > > tm_request.  This patch introduces to check TaskMID given from the
> > > userspace as a target tag.  We have a rule of relationship between
> > > (struct request *req->tag) and smid in mpt3sas_base.c:
> > >
> > > 3318 u16
> > > 3319 mpt3sas_base_get_smid_scsiio(struct MPT3SAS_ADAPTER *ioc, u8 cb_idx,
> > > 3320         struct scsi_cmnd *scmd)
> > > 3321 {
> > > 3322         struct scsiio_tracker *request = scsi_cmd_priv(scmd);
> > > 3323         unsigned int tag = scmd->request->tag;
> > > 3324         u16 smid;
> > > 3325
> > > 3326         smid = tag + 1;
> > >
> > > So if we want to abort a request tagged #X, then we can pass (X + 1) to
> > > this IOCTL handler.  Otherwise, user space just can pass 0 TaskMID to
> > > abort the first outstanding smid which is legacy behaviour.
> > >
> > > Cc: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
> > > Cc: Suganath Prabu Subramani <suganath-prabu.subramani@broadcom.com>
> > > Cc: Sathya Prakash <sathya.prakash@broadcom.com>
> > > Cc: James E.J. Bottomley <jejb@linux.ibm.com>
> > > Cc: Martin K. Petersen <martin.petersen@oracle.com>
> > > Cc: MPT-FusionLinux.pdl@broadcom.com
> > > Signed-off-by: Minwoo Im <minwoo.im@samsung.com>
> > > ---
> > >  drivers/scsi/mpt3sas/mpt3sas_ctl.c | 12 ++++++++++--
> > >  1 file changed, 10 insertions(+), 2 deletions(-)
> > >
> > > diff --git a/drivers/scsi/mpt3sas/mpt3sas_ctl.c b/drivers/scsi/mpt3sas/mpt3sas_ctl.c
> > > index b2bb47c14d35..f6b8fd90610a 100644
> > > --- a/drivers/scsi/mpt3sas/mpt3sas_ctl.c
> > > +++ b/drivers/scsi/mpt3sas/mpt3sas_ctl.c
> > > @@ -596,8 +596,16 @@ _ctl_set_task_mid(struct MPT3SAS_ADAPTER *ioc, struct mpt3_ioctl_command *karg,
> > >               if (priv_data->sas_target->handle != handle)
> > >                       continue;
> > >               st = scsi_cmd_priv(scmd);
> > > -             tm_request->TaskMID = cpu_to_le16(st->smid);
> > > -             found = 1;
> > > +
> > > +             /*
> > > +              * If the given TaskMID from the user space is zero, then the
> > > +              * first outstanding smid will be picked up.  Otherwise,
> > > +              * targeted smid will be the one.
> > > +              */
> > > +             if (!tm_request->TaskMID || tm_request->TaskMID == st->smid) {
> > > +                     tm_request->TaskMID = cpu_to_le16(st->smid);
> > > +                     found = 1;
> > > +             }
> > >       }
> > >
> > >       if (!found) {
> > >
> > I think this is fundamentally wrong.
> > ABORT_TASK is used to abort a single task, which of course has to be
> > known beforehand. If you don't know the task, what exactly do you hope
> > to achieve here? Aborting random I/O?
> > Or, even worse, aborting I/O the driver uses internally and corrupt the
> > internal workflow of the driver?
> >
> > We should simply disallow any ABORT TASK from userspace if the TaskMID
> > is zero. And I would even argue to disabllow ABORT TASK from userspace
> > completely, as the smid is never relayed to userland, and as such the
> > user cannot know which task should be aborted.
> 
> Hannes,
> 
> This interface was added long time back in mpt2sas driver and I don't
> have exact reason of adding this interface at that time.
> But I know that this interface is still used by BRCM test team & few
> customers only for some functionality and regression testing.

Actually I have sent this patch for the testing purpose for the SAS storage
that I'm working with now.  Some of test platform could figure out which
command has to be aborted or queried via debugfs and information from the
device itself with some other methods.  If the mpt3sas driver supports
the targeted TMF for a single command, then it would be great for the
testing.

Thanks,
	Minwoo Im
Minwoo Im July 25, 2019, 4:53 p.m. UTC | #6
On 19-07-23 16:57:49, Sreekanth Reddy wrote:
> On Mon, Jul 15, 2019 at 11:43 AM Hannes Reinecke <hare@suse.de> wrote:
> > I think this is fundamentally wrong.
> > ABORT_TASK is used to abort a single task, which of course has to be
> > known beforehand. If you don't know the task, what exactly do you hope
> > to achieve here? Aborting random I/O?
> > Or, even worse, aborting I/O the driver uses internally and corrupt the
> > internal workflow of the driver?
> >
> > We should simply disallow any ABORT TASK from userspace if the TaskMID
> > is zero. And I would even argue to disabllow ABORT TASK from userspace
> > completely, as the smid is never relayed to userland, and as such the
> > user cannot know which task should be aborted.
> 
> Hannes,
> 
> This interface was added long time back in mpt2sas driver and I don't
> have exact reason of adding this interface at that time.
> But I know that this interface is still used by BRCM test team & few
> customers only for some functionality and regression testing.

Sreekanth, and Hannes,

Could you please give some review points on this?

Thanks,
Sreekanth Reddy July 26, 2019, 9:49 a.m. UTC | #7
On Sun, Jul 14, 2019 at 9:14 AM Minwoo Im <minwoo.im@samsung.com> wrote:
>
> We can request task management IOCTL command(MPI2_FUNCTION_SCSI_TASK_MGMT)
> to /dev/mpt3ctl.  If the given task_type is either abort task or query
> task, it may need a field named "Initiator Port Transfer Tag to Manage"
> in the IU.
>
> Current code does not support to check target IPTT tag from the
> tm_request.  This patch introduces to check TaskMID given from the
> userspace as a target tag.  We have a rule of relationship between
> (struct request *req->tag) and smid in mpt3sas_base.c:
>
> 3318 u16
> 3319 mpt3sas_base_get_smid_scsiio(struct MPT3SAS_ADAPTER *ioc, u8 cb_idx,
> 3320         struct scsi_cmnd *scmd)
> 3321 {
> 3322         struct scsiio_tracker *request = scsi_cmd_priv(scmd);
> 3323         unsigned int tag = scmd->request->tag;
> 3324         u16 smid;
> 3325
> 3326         smid = tag + 1;
>
> So if we want to abort a request tagged #X, then we can pass (X + 1) to
> this IOCTL handler.  Otherwise, user space just can pass 0 TaskMID to
> abort the first outstanding smid which is legacy behaviour.
>
> Cc: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
> Cc: Suganath Prabu Subramani <suganath-prabu.subramani@broadcom.com>
> Cc: Sathya Prakash <sathya.prakash@broadcom.com>
> Cc: James E.J. Bottomley <jejb@linux.ibm.com>
> Cc: Martin K. Petersen <martin.petersen@oracle.com>
> Cc: MPT-FusionLinux.pdl@broadcom.com
> Signed-off-by: Minwoo Im <minwoo.im@samsung.com>

Acked-by:  Sreekanth Reddy <sreekanth.reddy@broadcom.com>

> ---
>  drivers/scsi/mpt3sas/mpt3sas_ctl.c | 12 ++++++++++--
>  1 file changed, 10 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/scsi/mpt3sas/mpt3sas_ctl.c b/drivers/scsi/mpt3sas/mpt3sas_ctl.c
> index b2bb47c14d35..f6b8fd90610a 100644
> --- a/drivers/scsi/mpt3sas/mpt3sas_ctl.c
> +++ b/drivers/scsi/mpt3sas/mpt3sas_ctl.c
> @@ -596,8 +596,16 @@ _ctl_set_task_mid(struct MPT3SAS_ADAPTER *ioc, struct mpt3_ioctl_command *karg,
>                 if (priv_data->sas_target->handle != handle)
>                         continue;
>                 st = scsi_cmd_priv(scmd);
> -               tm_request->TaskMID = cpu_to_le16(st->smid);
> -               found = 1;
> +
> +               /*
> +                * If the given TaskMID from the user space is zero, then the
> +                * first outstanding smid will be picked up.  Otherwise,
> +                * targeted smid will be the one.
> +                */
> +               if (!tm_request->TaskMID || tm_request->TaskMID == st->smid) {
> +                       tm_request->TaskMID = cpu_to_le16(st->smid);
> +                       found = 1;
> +               }
>         }
>
>         if (!found) {
> --
> 2.16.1
>
diff mbox series

Patch

diff --git a/drivers/scsi/mpt3sas/mpt3sas_ctl.c b/drivers/scsi/mpt3sas/mpt3sas_ctl.c
index b2bb47c14d35..f6b8fd90610a 100644
--- a/drivers/scsi/mpt3sas/mpt3sas_ctl.c
+++ b/drivers/scsi/mpt3sas/mpt3sas_ctl.c
@@ -596,8 +596,16 @@  _ctl_set_task_mid(struct MPT3SAS_ADAPTER *ioc, struct mpt3_ioctl_command *karg,
 		if (priv_data->sas_target->handle != handle)
 			continue;
 		st = scsi_cmd_priv(scmd);
-		tm_request->TaskMID = cpu_to_le16(st->smid);
-		found = 1;
+
+		/*
+		 * If the given TaskMID from the user space is zero, then the
+		 * first outstanding smid will be picked up.  Otherwise,
+		 * targeted smid will be the one.
+		 */
+		if (!tm_request->TaskMID || tm_request->TaskMID == st->smid) {
+			tm_request->TaskMID = cpu_to_le16(st->smid);
+			found = 1;
+		}
 	}
 
 	if (!found) {