2012-05-30 01:43:20

by Asai Thambi SP

Subject: [PATCH 06/11] mtip32xx: Use plain spinlock for 'cmd_issue_lock'

'cmd_issue_lock' is for only acquiring a free slot, and it is not used
in interrupt context. So replaced irq version with non-irq version of spinlock.

Signed-off-by: Asai Thambi S P <[email protected]>
---
drivers/block/mtip32xx/mtip32xx.c | 6 ++----
1 files changed, 2 insertions(+), 4 deletions(-)

diff --git a/drivers/block/mtip32xx/mtip32xx.c b/drivers/block/mtip32xx/mtip32xx.c
index 801e70c..a8a5522 100644
--- a/drivers/block/mtip32xx/mtip32xx.c
+++ b/drivers/block/mtip32xx/mtip32xx.c
@@ -294,18 +294,16 @@ static int hba_reset_nosleep(struct driver_data *dd)
*/
static inline void mtip_issue_ncq_command(struct mtip_port *port, int tag)
{
- unsigned long flags = 0;
-
atomic_set(&port->commands[tag].active, 1);

- spin_lock_irqsave(&port->cmd_issue_lock, flags);
+ spin_lock(&port->cmd_issue_lock);

writel((1 << MTIP_TAG_BIT(tag)),
port->s_active[MTIP_TAG_INDEX(tag)]);
writel((1 << MTIP_TAG_BIT(tag)),
port->cmd_issue[MTIP_TAG_INDEX(tag)]);

- spin_unlock_irqrestore(&port->cmd_issue_lock, flags);
+ spin_unlock(&port->cmd_issue_lock);

/* Set the command's timeout value.*/
port->commands[tag].comp_time = jiffies + msecs_to_jiffies(
--
1.7.1

2012-05-31 06:45:16

by Jens Axboe

Subject: Re: [PATCH 06/11] mtip32xx: Use plain spinlock for 'cmd_issue_lock'

On 05/30/2012 03:43 AM, Asai Thambi S P wrote:
>
> 'cmd_issue_lock' is for only acquiring a free slot, and it is not used
> in interrupt context. So replaced irq version with non-irq version of spinlock.
>
> Signed-off-by: Asai Thambi S P <[email protected]>
> ---
> drivers/block/mtip32xx/mtip32xx.c | 6 ++----
> 1 files changed, 2 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/block/mtip32xx/mtip32xx.c b/drivers/block/mtip32xx/mtip32xx.c
> index 801e70c..a8a5522 100644
> --- a/drivers/block/mtip32xx/mtip32xx.c
> +++ b/drivers/block/mtip32xx/mtip32xx.c
> @@ -294,18 +294,16 @@ static int hba_reset_nosleep(struct driver_data *dd)
> */
> static inline void mtip_issue_ncq_command(struct mtip_port *port, int tag)
> {
> - unsigned long flags = 0;
> -
> atomic_set(&port->commands[tag].active, 1);
>
> - spin_lock_irqsave(&port->cmd_issue_lock, flags);
> + spin_lock(&port->cmd_issue_lock);
>
> writel((1 << MTIP_TAG_BIT(tag)),
> port->s_active[MTIP_TAG_INDEX(tag)]);
> writel((1 << MTIP_TAG_BIT(tag)),
> port->cmd_issue[MTIP_TAG_INDEX(tag)]);
>
> - spin_unlock_irqrestore(&port->cmd_issue_lock, flags);
> + spin_unlock(&port->cmd_issue_lock);

Looks fine - wasn't going to comment specifically on this one, but
wanted to say, that in general you don't need to pre-clear the flags
variable before passing it to _irqsave().

--
Jens Axboe