2012-10-31 10:40:31

by Xiaotian Feng

Subject: [PATCH] dma: ioat: use tasklet_kill in device remove/release process

Some driver uses tasklet_disable in device remove/release process,
tasklet_disable will inc tasklet->count and return. If the tasklet
is not handled yet under some softirq pressure, the tasklet will be
placed on the tasklet_vec, never have a chance to be excuted. This might
lead to a heavy loaded ksoftirqd, wakeup with pending_softirq, but
tasklet is disabled. tasklet_kill should be used in this case.

Signed-off-by: Xiaotian Feng <[email protected]>
Cc: Vinod Koul <[email protected]>
Cc: Dan Williams <[email protected]>
---
drivers/dma/ioat/dma.c | 2 +-
drivers/dma/ioat/dma_v2.c | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/dma/ioat/dma.c b/drivers/dma/ioat/dma.c
index 73b2b65..8bec4a2 100644
--- a/drivers/dma/ioat/dma.c
+++ b/drivers/dma/ioat/dma.c
@@ -379,7 +379,7 @@ static void ioat1_dma_free_chan_resources(struct dma_chan *c)
if (ioat->desccount == 0)
return;

- tasklet_disable(&chan->cleanup_task);
+ tasklet_kill(&chan->cleanup_task);
del_timer_sync(&chan->timer);
ioat1_cleanup(ioat);

diff --git a/drivers/dma/ioat/dma_v2.c b/drivers/dma/ioat/dma_v2.c
index b9d6678..df3ccb0 100644
--- a/drivers/dma/ioat/dma_v2.c
+++ b/drivers/dma/ioat/dma_v2.c
@@ -794,7 +794,7 @@ void ioat2_free_chan_resources(struct dma_chan *c)
if (!ioat->ring)
return;

- tasklet_disable(&chan->cleanup_task);
+ tasklet_kill(&chan->cleanup_task);
del_timer_sync(&chan->timer);
device->cleanup_fn((unsigned long) c);
device->reset_hw(chan);
--
1.7.9.5

2012-10-31 11:24:38

by Xiaotian Feng

Subject: Re: [PATCH] dma: ioat: use tasklet_kill in device remove/release process

On Wed, Oct 31, 2012 at 6:41 PM, Xiaotian Feng <[email protected]> wrote:
> Some driver uses tasklet_disable in device remove/release process,
> tasklet_disable will inc tasklet->count and return. If the tasklet
> is not handled yet under some softirq pressure, the tasklet will be
> placed on the tasklet_vec, never have a chance to be excuted. This might
> lead to a heavy loaded ksoftirqd, wakeup with pending_softirq, but
> tasklet is disabled. tasklet_kill should be used in this case.
>
> Signed-off-by: Xiaotian Feng <[email protected]>
> Cc: Vinod Koul <[email protected]>
> Cc: Dan Williams <[email protected]>

Please ignore this one, I made a mistake here.

> ---
> drivers/dma/ioat/dma.c | 2 +-
> drivers/dma/ioat/dma_v2.c | 2 +-
> 2 files changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/dma/ioat/dma.c b/drivers/dma/ioat/dma.c
> index 73b2b65..8bec4a2 100644
> --- a/drivers/dma/ioat/dma.c
> +++ b/drivers/dma/ioat/dma.c
> @@ -379,7 +379,7 @@ static void ioat1_dma_free_chan_resources(struct dma_chan *c)
> if (ioat->desccount == 0)
> return;
>
> - tasklet_disable(&chan->cleanup_task);
> + tasklet_kill(&chan->cleanup_task);
> del_timer_sync(&chan->timer);
> ioat1_cleanup(ioat);
>
> diff --git a/drivers/dma/ioat/dma_v2.c b/drivers/dma/ioat/dma_v2.c
> index b9d6678..df3ccb0 100644
> --- a/drivers/dma/ioat/dma_v2.c
> +++ b/drivers/dma/ioat/dma_v2.c
> @@ -794,7 +794,7 @@ void ioat2_free_chan_resources(struct dma_chan *c)
> if (!ioat->ring)
> return;
>
> - tasklet_disable(&chan->cleanup_task);
> + tasklet_kill(&chan->cleanup_task);
> del_timer_sync(&chan->timer);
> device->cleanup_fn((unsigned long) c);
> device->reset_hw(chan);
> --
> 1.7.9.5
>