[PATCH 2/2] [SCSI] pm8001: Fix possible racing codition.

[PATCH 2/2] [SCSI] pm8001: Fix possible racing codition.

[santosh nayak]

From: Santosh Nayak <santoshprasadnayak@gmail.com>

There is a possble racing scenario.

'process_oq' is called by two routines, as shown below.

pm8001_8001_dispatch = {
         .......

        .isr             = pm8001_chip_isr --> process_oq,// A
        .isr_process_oq  = process_oq,                   //  B
        .....
}

process_oq() --> process_one_iomb() --> mpi_sata_completion()

In 'mpi_sata_completion', "pm8001_ha->lock" is first released.
It means lock is taken before,  which is true for
the context A, as 'pm8001_ha->lock' is taken in 'pm8001_chip_isr()'

But for context B there is no lock taken before and pm8001_ha->lock
is unlocked in 'mpi_sata_completion()'. This may unlock the lock
taken in context A. Possible racing ??

If 'pm8001_ha->lock' is taken in 'process_oq()' instead of
'pm8001_chip_isr' then the above issue can be avoided.

Signed-off-by: Santosh Nayak <santoshprasadnayak@gmail.com>
---
 drivers/scsi/pm8001/pm8001_hwi.c |    6 +++---
 1 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/scsi/pm8001/pm8001_hwi.c b/drivers/scsi/pm8001/pm8001_hwi.c
index 838e3e2..6d9973b 100644
--- a/drivers/scsi/pm8001/pm8001_hwi.c
+++ b/drivers/scsi/pm8001/pm8001_hwi.c
@@ -3663,7 +3663,9 @@ static int process_oq(struct pm8001_hba_info *pm8001_ha)
 	void *pMsg1 = NULL;
 	u8 uninitialized_var(bc);
 	u32 ret = MPI_IO_STATUS_FAIL;
+	unsigned long flags;
 
+	spin_lock_irqsave(&pm8001_ha->lock, flags);
 	circularQ = &pm8001_ha->outbnd_q_tbl[0];
 	do {
 		ret = mpi_msg_consume(pm8001_ha, circularQ, &pMsg1, &bc);
@@ -3682,6 +3684,7 @@ static int process_oq(struct pm8001_hba_info *pm8001_ha)
 				break;
 		}
 	} while (1);
+	spin_unlock_irqrestore(&pm8001_ha->lock, flags);
 	return ret;
 }
 
@@ -4087,12 +4090,9 @@ static u32 pm8001_chip_is_our_interupt(struct pm8001_hba_info *pm8001_ha)
 static irqreturn_t
 pm8001_chip_isr(struct pm8001_hba_info *pm8001_ha)
 {
-	unsigned long flags;
-	spin_lock_irqsave(&pm8001_ha->lock, flags);
 	pm8001_chip_interrupt_disable(pm8001_ha);
 	process_oq(pm8001_ha);
 	pm8001_chip_interrupt_enable(pm8001_ha);
-	spin_unlock_irqrestore(&pm8001_ha->lock, flags);
 	return IRQ_HANDLED;
 }
 
-- 
1.7.4.4

RE: [PATCH 2/2] [SCSI] pm8001: Fix possible racing codition.

[Jack Wang]

Infact , process_oq is only called in pm8001_chip_isr, but this patch looks
OK for me. Move down the lock to process_oq is OK.

Thanks for fix.
Acked-by : Jack Wang <jack_wang@usish.com>

> From: Santosh Nayak <santoshprasadnayak@gmail.com>
> 
> There is a possble racing scenario.
> 
> 'process_oq' is called by two routines, as shown below.
> 
> pm8001_8001_dispatch = {
>          .......
> 
>         .isr             = pm8001_chip_isr --> process_oq,// A
>         .isr_process_oq  = process_oq,                   //  B
>         .....
> }
> 
> process_oq() --> process_one_iomb() --> mpi_sata_completion()
> 
> In 'mpi_sata_completion', "pm8001_ha->lock" is first released.
> It means lock is taken before,  which is true for
> the context A, as 'pm8001_ha->lock' is taken in 'pm8001_chip_isr()'
> 
> But for context B there is no lock taken before and pm8001_ha->lock
> is unlocked in 'mpi_sata_completion()'. This may unlock the lock
> taken in context A. Possible racing ??
> 
> If 'pm8001_ha->lock' is taken in 'process_oq()' instead of
> 'pm8001_chip_isr' then the above issue can be avoided.
> 
> Signed-off-by: Santosh Nayak <santoshprasadnayak@gmail.com>
> ---
>  drivers/scsi/pm8001/pm8001_hwi.c |    6 +++---
>  1 files changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/scsi/pm8001/pm8001_hwi.c
> b/drivers/scsi/pm8001/pm8001_hwi.c
> index 838e3e2..6d9973b 100644
> --- a/drivers/scsi/pm8001/pm8001_hwi.c
> +++ b/drivers/scsi/pm8001/pm8001_hwi.c
> @@ -3663,7 +3663,9 @@ static int process_oq(struct pm8001_hba_info
*pm8001_ha)
>  	void *pMsg1 = NULL;
>  	u8 uninitialized_var(bc);
>  	u32 ret = MPI_IO_STATUS_FAIL;
> +	unsigned long flags;
> 
> +	spin_lock_irqsave(&pm8001_ha->lock, flags);
>  	circularQ = &pm8001_ha->outbnd_q_tbl[0];
>  	do {
>  		ret = mpi_msg_consume(pm8001_ha, circularQ, &pMsg1, &bc);
> @@ -3682,6 +3684,7 @@ static int process_oq(struct pm8001_hba_info
*pm8001_ha)
>  				break;
>  		}
>  	} while (1);
> +	spin_unlock_irqrestore(&pm8001_ha->lock, flags);
>  	return ret;
>  }
> 
> @@ -4087,12 +4090,9 @@ static u32 pm8001_chip_is_our_interupt(struct
> pm8001_hba_info *pm8001_ha)
>  static irqreturn_t
>  pm8001_chip_isr(struct pm8001_hba_info *pm8001_ha)
>  {
> -	unsigned long flags;
> -	spin_lock_irqsave(&pm8001_ha->lock, flags);
>  	pm8001_chip_interrupt_disable(pm8001_ha);
>  	process_oq(pm8001_ha);
>  	pm8001_chip_interrupt_enable(pm8001_ha);
> -	spin_unlock_irqrestore(&pm8001_ha->lock, flags);
>  	return IRQ_HANDLED;
>  }
> 
> --
> 1.7.4.4
> 
> --
> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html

Re: [PATCH 2/2] [SCSI] pm8001: Fix possible racing codition.

[santosh prasad nayak]

Thanks Jack for your response.

'process_oq'  is called at two places.


 pm8001_8001_dispatch = {
          .......

         .isr             = pm8001_chip_isr --> process_oq,     //   1st place
         .isr_process_oq  = process_oq,                            //
 2nd  place
         .....
}

regards
santosh

On Mon, Feb 27, 2012 at 10:18 AM, Jack Wang <jack_wang@usish.com> wrote:
> Infact , process_oq is only called in pm8001_chip_isr, but this patch looks
> OK for me. Move down the lock to process_oq is OK.
>
> Thanks for fix.
> Acked-by : Jack Wang <jack_wang@usish.com>
>
>> From: Santosh Nayak <santoshprasadnayak@gmail.com>
>>
>> There is a possble racing scenario.
>>
>> 'process_oq' is called by two routines, as shown below.
>>
>> pm8001_8001_dispatch = {
>>          .......
>>
>>         .isr             = pm8001_chip_isr --> process_oq,// A
>>         .isr_process_oq  = process_oq,                   //  B
>>         .....
>> }
>>
>> process_oq() --> process_one_iomb() --> mpi_sata_completion()
>>
>> In 'mpi_sata_completion', "pm8001_ha->lock" is first released.
>> It means lock is taken before,  which is true for
>> the context A, as 'pm8001_ha->lock' is taken in 'pm8001_chip_isr()'
>>
>> But for context B there is no lock taken before and pm8001_ha->lock
>> is unlocked in 'mpi_sata_completion()'. This may unlock the lock
>> taken in context A. Possible racing ??
>>
>> If 'pm8001_ha->lock' is taken in 'process_oq()' instead of
>> 'pm8001_chip_isr' then the above issue can be avoided.
>>
>> Signed-off-by: Santosh Nayak <santoshprasadnayak@gmail.com>
>> ---
>>  drivers/scsi/pm8001/pm8001_hwi.c |    6 +++---
>>  1 files changed, 3 insertions(+), 3 deletions(-)
>>
>> diff --git a/drivers/scsi/pm8001/pm8001_hwi.c
>> b/drivers/scsi/pm8001/pm8001_hwi.c
>> index 838e3e2..6d9973b 100644
>> --- a/drivers/scsi/pm8001/pm8001_hwi.c
>> +++ b/drivers/scsi/pm8001/pm8001_hwi.c
>> @@ -3663,7 +3663,9 @@ static int process_oq(struct pm8001_hba_info
> *pm8001_ha)
>>       void *pMsg1 = NULL;
>>       u8 uninitialized_var(bc);
>>       u32 ret = MPI_IO_STATUS_FAIL;
>> +     unsigned long flags;
>>
>> +     spin_lock_irqsave(&pm8001_ha->lock, flags);
>>       circularQ = &pm8001_ha->outbnd_q_tbl[0];
>>       do {
>>               ret = mpi_msg_consume(pm8001_ha, circularQ, &pMsg1, &bc);
>> @@ -3682,6 +3684,7 @@ static int process_oq(struct pm8001_hba_info
> *pm8001_ha)
>>                               break;
>>               }
>>       } while (1);
>> +     spin_unlock_irqrestore(&pm8001_ha->lock, flags);
>>       return ret;
>>  }
>>
>> @@ -4087,12 +4090,9 @@ static u32 pm8001_chip_is_our_interupt(struct
>> pm8001_hba_info *pm8001_ha)
>>  static irqreturn_t
>>  pm8001_chip_isr(struct pm8001_hba_info *pm8001_ha)
>>  {
>> -     unsigned long flags;
>> -     spin_lock_irqsave(&pm8001_ha->lock, flags);
>>       pm8001_chip_interrupt_disable(pm8001_ha);
>>       process_oq(pm8001_ha);
>>       pm8001_chip_interrupt_enable(pm8001_ha);
>> -     spin_unlock_irqrestore(&pm8001_ha->lock, flags);
>>       return IRQ_HANDLED;
>>  }
>>
>> --
>> 1.7.4.4
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>

Re: [PATCH 2/2] [SCSI] pm8001: Fix possible racing codition.

[Jack Wang]

The first is not called, just implement a interface, just reserved.

Thanks anyway.
Jack
> 
> Thanks Jack for your response.
> 
> 'process_oq'  is called at two places.
> 
> 
>  pm8001_8001_dispatch = {
>           .......
> 
>          .isr             = pm8001_chip_isr --> process_oq,     //   1st
> place
>          .isr_process_oq  = process_oq,                            //
>  2nd  place
>          .....
> }
> 
> regards
> santosh
> 
> On Mon, Feb 27, 2012 at 10:18 AM, Jack Wang <jack_wang@usish.com> wrote:
> > Infact , process_oq is only called in pm8001_chip_isr, but this patch
looks
> > OK for me. Move down the lock to process_oq is OK.
> >
> > Thanks for fix.
> > Acked-by : Jack Wang <jack_wang@usish.com>
> >
> >> From: Santosh Nayak <santoshprasadnayak@gmail.com>
> >>
> >> There is a possble racing scenario.
> >>
> >> 'process_oq' is called by two routines, as shown below.
> >>
> >> pm8001_8001_dispatch = {
> >>          .......
> >>
> >>         .isr             = pm8001_chip_isr --> process_oq,// A
> >>         .isr_process_oq  = process_oq,                   //  B
> >>         .....
> >> }
> >>
> >> process_oq() --> process_one_iomb() --> mpi_sata_completion()
> >>
> >> In 'mpi_sata_completion', "pm8001_ha->lock" is first released.
> >> It means lock is taken before,  which is true for
> >> the context A, as 'pm8001_ha->lock' is taken in 'pm8001_chip_isr()'
> >>
> >> But for context B there is no lock taken before and pm8001_ha->lock
> >> is unlocked in 'mpi_sata_completion()'. This may unlock the lock
> >> taken in context A. Possible racing ??
> >>
> >> If 'pm8001_ha->lock' is taken in 'process_oq()' instead of
> >> 'pm8001_chip_isr' then the above issue can be avoided.
> >>
> >> Signed-off-by: Santosh Nayak <santoshprasadnayak@gmail.com>
> >> ---
> >>  drivers/scsi/pm8001/pm8001_hwi.c |    6 +++---
> >>  1 files changed, 3 insertions(+), 3 deletions(-)
> >>
> >> diff --git a/drivers/scsi/pm8001/pm8001_hwi.c
> >> b/drivers/scsi/pm8001/pm8001_hwi.c
> >> index 838e3e2..6d9973b 100644
> >> --- a/drivers/scsi/pm8001/pm8001_hwi.c
> >> +++ b/drivers/scsi/pm8001/pm8001_hwi.c
> >> @@ -3663,7 +3663,9 @@ static int process_oq(struct pm8001_hba_info
> > *pm8001_ha)
> >>       void *pMsg1 = NULL;
> >>       u8 uninitialized_var(bc);
> >>       u32 ret = MPI_IO_STATUS_FAIL;
> >> +     unsigned long flags;
> >>
> >> +     spin_lock_irqsave(&pm8001_ha->lock, flags);
> >>       circularQ = &pm8001_ha->outbnd_q_tbl[0];
> >>       do {
> >>               ret = mpi_msg_consume(pm8001_ha, circularQ, &pMsg1, &bc);
> >> @@ -3682,6 +3684,7 @@ static int process_oq(struct pm8001_hba_info
> > *pm8001_ha)
> >>                               break;
> >>               }
> >>       } while (1);
> >> +     spin_unlock_irqrestore(&pm8001_ha->lock, flags);
> >>       return ret;
> >>  }
> >>
> >> @@ -4087,12 +4090,9 @@ static u32 pm8001_chip_is_our_interupt(struct
> >> pm8001_hba_info *pm8001_ha)
> >>  static irqreturn_t
> >>  pm8001_chip_isr(struct pm8001_hba_info *pm8001_ha)
> >>  {
> >> -     unsigned long flags;
> >> -     spin_lock_irqsave(&pm8001_ha->lock, flags);
> >>       pm8001_chip_interrupt_disable(pm8001_ha);
> >>       process_oq(pm8001_ha);
> >>       pm8001_chip_interrupt_enable(pm8001_ha);
> >> -     spin_unlock_irqrestore(&pm8001_ha->lock, flags);
> >>       return IRQ_HANDLED;
> >>  }
> >>
> >> --
> >> 1.7.4.4
> >>
> >> --
> >> To unsubscribe from this list: send the line "unsubscribe linux-scsi"
in
> >> the body of a message to majordomo@vger.kernel.org
> >> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> >
> --
> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html