Re: [PATCH 2/4] scsi: pmcraid: fix lock imbalance in pmcraid_reset_reload()

[Christoph Hellwig]

On Thu, Apr 20, 2017 at 07:54:46PM +0200, Arnd Bergmann wrote:
> sparse found a bug that has always been present since the driver
> was merged:
> 
> drivers/scsi/pmcraid.c:2353:12: warning: context imbalance in 'pmcraid_reset_reload' - different lock contexts for basic block
> 
> This adds the missing unlock at the end of the function. I could
> not figure out if this will happen in practice, but I could not
> prove that it doesn't happen, so to be on the safe side, let's
> backport this fix.
> 
> Cc: stable@xxxxxxxxxxxxxxx
> Fixes: 89a368104150 ("[SCSI] pmcraid: PMC-Sierra MaxRAID driver to support 6Gb/s SAS RAID controller")
> Signed-off-by: Arnd Bergmann <arnd@xxxxxxxx>
> ---
>  drivers/scsi/pmcraid.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/drivers/scsi/pmcraid.c b/drivers/scsi/pmcraid.c
> index 096c704ca39a..35087e94c2ad 100644
> --- a/drivers/scsi/pmcraid.c
> +++ b/drivers/scsi/pmcraid.c
> @@ -2411,8 +2411,11 @@ static int pmcraid_reset_reload(
>  		scsi_unblock_requests(pinstance->host);
>  		if (pinstance->ioa_state == target_state)
>  			reset = 0;
> +		spin_lock_irqsave(pinstance->host->host_lock, lock_flags);
>  	}
>  
> +	spin_unlock_irqrestore(pinstance->host->host_lock, lock_flags);
> +

This looks weird.  Using a proper goto label to unwind seem like the
best way to go, e.g. this patch:

---