[PATCH] scsi: lpfc: Fix a possible data race in lpfc_unregister_fcf_rescan()

[PATCH] scsi: lpfc: Fix a possible data race in lpfc_unregister_fcf_rescan()

[Tuo Li]

The variable phba->fcf.fcf_flag is often protected by the lock 
phba->hbalock() when is accessed. Here is an example in 
lpfc_unregister_fcf_rescan():

  spin_lock_irq(&phba->hbalock);
  phba->fcf.fcf_flag |= FCF_INIT_DISC;
  spin_unlock_irq(&phba->hbalock);

However, in the same function, phba->fcf.fcf_flag is assigned with 0 
without holding the lock, and thus can cause a data race:

  phba->fcf.fcf_flag = 0;

To fix this possible data race, a lock and unlock pair is added when 
accessing the variable phba->fcf.fcf_flag.

Reported-by: BassCheck <bass@buaa.edu.cn>
Signed-off-by: Tuo Li <islituo@gmail.com>
---
 drivers/scsi/lpfc/lpfc_hbadisc.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/scsi/lpfc/lpfc_hbadisc.c b/drivers/scsi/lpfc/lpfc_hbadisc.c
index 5ba3a9ad9501..9d2feb69cae7 100644
--- a/drivers/scsi/lpfc/lpfc_hbadisc.c
+++ b/drivers/scsi/lpfc/lpfc_hbadisc.c
@@ -6961,7 +6961,9 @@ lpfc_unregister_fcf_rescan(struct lpfc_hba *phba)
 	if (rc)
 		return;
 	/* Reset HBA FCF states after successful unregister FCF */
+	spin_lock_irq(&phba->hbalock);
 	phba->fcf.fcf_flag = 0;
+	spin_unlock_irq(&phba->hbalock);
 	phba->fcf.current_rec.flag = 0;
 
 	/*
-- 
2.34.1

Re: [PATCH] scsi: lpfc: Fix a possible data race in lpfc_unregister_fcf_rescan()

[Justin Tee]

Hi Tuo,

Looks good.

Reviewed-by: Justin Tee <justin.tee@broadcom.com>

Thanks,
Justin

On Thu, Jun 29, 2023 at 8:03 PM Tuo Li <islituo@gmail.com> wrote:
>
> The variable phba->fcf.fcf_flag is often protected by the lock
> phba->hbalock() when is accessed. Here is an example in
> lpfc_unregister_fcf_rescan():
>
>   spin_lock_irq(&phba->hbalock);
>   phba->fcf.fcf_flag |= FCF_INIT_DISC;
>   spin_unlock_irq(&phba->hbalock);
>
> However, in the same function, phba->fcf.fcf_flag is assigned with 0
> without holding the lock, and thus can cause a data race:
>
>   phba->fcf.fcf_flag = 0;
>
> To fix this possible data race, a lock and unlock pair is added when
> accessing the variable phba->fcf.fcf_flag.
>
> Reported-by: BassCheck <bass@buaa.edu.cn>
> Signed-off-by: Tuo Li <islituo@gmail.com>
> ---
>  drivers/scsi/lpfc/lpfc_hbadisc.c | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/drivers/scsi/lpfc/lpfc_hbadisc.c b/drivers/scsi/lpfc/lpfc_hbadisc.c
> index 5ba3a9ad9501..9d2feb69cae7 100644
> --- a/drivers/scsi/lpfc/lpfc_hbadisc.c
> +++ b/drivers/scsi/lpfc/lpfc_hbadisc.c
> @@ -6961,7 +6961,9 @@ lpfc_unregister_fcf_rescan(struct lpfc_hba *phba)
>         if (rc)
>                 return;
>         /* Reset HBA FCF states after successful unregister FCF */
> +       spin_lock_irq(&phba->hbalock);
>         phba->fcf.fcf_flag = 0;
> +       spin_unlock_irq(&phba->hbalock);
>         phba->fcf.current_rec.flag = 0;
>
>         /*
> --
> 2.34.1
>

Re: [PATCH] scsi: lpfc: Fix a possible data race in lpfc_unregister_fcf_rescan()

[Laurence Oberman]

On Fri, 2023-06-30 at 10:47 +0800, Tuo Li wrote:
> The variable phba->fcf.fcf_flag is often protected by the lock 
> phba->hbalock() when is accessed. Here is an example in 
> lpfc_unregister_fcf_rescan():
> 
>   spin_lock_irq(&phba->hbalock);
>   phba->fcf.fcf_flag |= FCF_INIT_DISC;
>   spin_unlock_irq(&phba->hbalock);
> 
> However, in the same function, phba->fcf.fcf_flag is assigned with 0 
> without holding the lock, and thus can cause a data race:
> 
>   phba->fcf.fcf_flag = 0;
> 
> To fix this possible data race, a lock and unlock pair is added when 
> accessing the variable phba->fcf.fcf_flag.
> 
> Reported-by: BassCheck <bass@buaa.edu.cn>
> Signed-off-by: Tuo Li <islituo@gmail.com>
> ---
>  drivers/scsi/lpfc/lpfc_hbadisc.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/scsi/lpfc/lpfc_hbadisc.c
> b/drivers/scsi/lpfc/lpfc_hbadisc.c
> index 5ba3a9ad9501..9d2feb69cae7 100644
> --- a/drivers/scsi/lpfc/lpfc_hbadisc.c
> +++ b/drivers/scsi/lpfc/lpfc_hbadisc.c
> @@ -6961,7 +6961,9 @@ lpfc_unregister_fcf_rescan(struct lpfc_hba
> *phba)
>         if (rc)
>                 return;
>         /* Reset HBA FCF states after successful unregister FCF */
> +       spin_lock_irq(&phba->hbalock);
>         phba->fcf.fcf_flag = 0;
> +       spin_unlock_irq(&phba->hbalock);
>         phba->fcf.current_rec.flag = 0;
>  
>         /*

This makes sense and looks good to me
Reviewed-by: Laurence Oberman <loberman@redhat.com>