2005-03-06 23:49:33

by Domen Puncer

Subject: [patch 2/8] isdn/capi: replace interruptible_sleep_on() with wait_event_interruptible()

Use wait_event_interruptible() instead of the deprecated
interruptible_sleep_on(). Patch is straight-forward as current sleep is
conditionally looped. Patch is compile-tested.

Signed-off-by: Nishanth Aravamudan <[email protected]>
Signed-off-by: Domen Puncer <[email protected]>
---


kj-domen/drivers/isdn/capi/capi.c | 9 ++-------
1 files changed, 2 insertions(+), 7 deletions(-)

diff -puN drivers/isdn/capi/capi.c~int_sleep_on-drivers_isdn_capi_capi drivers/isdn/capi/capi.c
--- kj/drivers/isdn/capi/capi.c~int_sleep_on-drivers_isdn_capi_capi 2005-03-05 16:11:36.000000000 +0100
+++ kj-domen/drivers/isdn/capi/capi.c 2005-03-05 16:11:36.000000000 +0100
@@ -675,13 +675,8 @@ capi_read(struct file *file, char __user
if (file->f_flags & O_NONBLOCK)
return -EAGAIN;

- for (;;) {
- interruptible_sleep_on(&cdev->recvwait);
- if ((skb = skb_dequeue(&cdev->recvqueue)) != 0)
- break;
- if (signal_pending(current))
- break;
- }
+ wait_event_interruptible(cdev->recvwait,
+ ((skb = skb_dequeue(&cdev->recvqueue)) == 0));
if (skb == 0)
return -ERESTARTNOHAND;
}
_

2005-03-07 03:49:31

by Andrew Morton

Subject: Re: [patch 2/8] isdn/capi: replace interruptible_sleep_on() with wait_event_interruptible()

[email protected] wrote:
>
>
>
> Use wait_event_interruptible() instead of the deprecated
> interruptible_sleep_on(). Patch is straight-forward as current sleep is
> conditionally looped. Patch is compile-tested.
>
> Signed-off-by: Nishanth Aravamudan <[email protected]>
> Signed-off-by: Domen Puncer <[email protected]>
> ---
>
>
> kj-domen/drivers/isdn/capi/capi.c | 9 ++-------
> 1 files changed, 2 insertions(+), 7 deletions(-)
>
> diff -puN drivers/isdn/capi/capi.c~int_sleep_on-drivers_isdn_capi_capi drivers/isdn/capi/capi.c
> --- kj/drivers/isdn/capi/capi.c~int_sleep_on-drivers_isdn_capi_capi 2005-03-05 16:11:36.000000000 +0100
> +++ kj-domen/drivers/isdn/capi/capi.c 2005-03-05 16:11:36.000000000 +0100
> @@ -675,13 +675,8 @@ capi_read(struct file *file, char __user
> if (file->f_flags & O_NONBLOCK)
> return -EAGAIN;
>
> - for (;;) {
> - interruptible_sleep_on(&cdev->recvwait);
> - if ((skb = skb_dequeue(&cdev->recvqueue)) != 0)
> - break;
> - if (signal_pending(current))
> - break;
> - }
> + wait_event_interruptible(cdev->recvwait,
> + ((skb = skb_dequeue(&cdev->recvqueue)) == 0));
> if (skb == 0)
> return -ERESTARTNOHAND;
> }

hmm, OK. Putting an expression with side-effect such as this into a macro
which evaluates the expression multiple times is a bit of a worry, but it
appears that everything will work OK.

That being said, I'd prefer that this come in via the ISDN team, after
having been tested please.

2005-03-07 08:39:42

by Domen Puncer

Subject: Re: [patch 2/8] isdn/capi: replace interruptible_sleep_on() with wait_event_interruptible()

On 06/03/05 19:47 -0800, Andrew Morton wrote:
> [email protected] wrote:
> >
> >
> >
> > Use wait_event_interruptible() instead of the deprecated
> > interruptible_sleep_on(). Patch is straight-forward as current sleep is
> > conditionally looped. Patch is compile-tested.
> >
> > Signed-off-by: Nishanth Aravamudan <[email protected]>
> > Signed-off-by: Domen Puncer <[email protected]>
> > ---
> >
> >
> > kj-domen/drivers/isdn/capi/capi.c | 9 ++-------
> > 1 files changed, 2 insertions(+), 7 deletions(-)
> >
> > diff -puN drivers/isdn/capi/capi.c~int_sleep_on-drivers_isdn_capi_capi drivers/isdn/capi/capi.c
> > --- kj/drivers/isdn/capi/capi.c~int_sleep_on-drivers_isdn_capi_capi 2005-03-05 16:11:36.000000000 +0100
> > +++ kj-domen/drivers/isdn/capi/capi.c 2005-03-05 16:11:36.000000000 +0100
> > @@ -675,13 +675,8 @@ capi_read(struct file *file, char __user
> > if (file->f_flags & O_NONBLOCK)
> > return -EAGAIN;
> >
> > - for (;;) {
> > - interruptible_sleep_on(&cdev->recvwait);
> > - if ((skb = skb_dequeue(&cdev->recvqueue)) != 0)
> > - break;
> > - if (signal_pending(current))
> > - break;
> > - }
> > + wait_event_interruptible(cdev->recvwait,
> > + ((skb = skb_dequeue(&cdev->recvqueue)) == 0));
> > if (skb == 0)
> > return -ERESTARTNOHAND;
> > }
>
> hmm, OK. Putting an expression with side-effect such as this into a macro
> which evaluates the expression multiple times is a bit of a worry, but it
> appears that everything will work OK.
>
> That being said, I'd prefer that this come in via the ISDN team, after
> having been tested please.

I believe last update from ISDN team was 13 months ago. :-(
And I concur, testing would be great.


Domen