Re: [PATCH] HPET driver

[Andrew Morton <akpm@osdl.org>]

Jeff Garzik <jgarzik@pobox.com> wrote:
>
> > I eliminated the request_irq brain damage, eliminated procfs support, 
> > made the check for FMODE_WRITE  in hpet_open and responded to a few 
> > other suggestions.
> 
> Thanks!  I'll look over it too.

Here's one fixlet against Robert's fixes:



- Need to set TASK_INTERRUPTIBLE before checking devp->hd_irqdata. 
  Otherwise the wakeup from hpet_interrupt() could be missed.

Signed-off-by: Andrew Morton <akpm@osdl.org>
---

 25-akpm/drivers/char/hpet.c |   12 +++++-------
 1 files changed, 5 insertions(+), 7 deletions(-)

diff -puN drivers/char/hpet.c~hpet-fixes-fix drivers/char/hpet.c
--- 25/drivers/char/hpet.c~hpet-fixes-fix	Wed Jun 23 14:38:16 2004
+++ 25-akpm/drivers/char/hpet.c	Wed Jun 23 14:38:16 2004
@@ -196,7 +196,8 @@ hpet_read(struct file *file, char *buf, 
 
 	add_wait_queue(&devp->hd_waitqueue, &wait);
 
-	do {
+	for ( ; ; ) {
+		set_current_state(TASK_INTERRUPTIBLE);
 		spin_lock_irq(&hpet_lock);
 		data = devp->hd_irqdata;
 		devp->hd_irqdata = 0;
@@ -211,17 +212,14 @@ hpet_read(struct file *file, char *buf, 
 			retval = -ERESTARTSYS;
 			goto out;
 		}
-
-		set_current_state(TASK_INTERRUPTIBLE);
 		schedule();
-
-	} while (1);
+	}
 
 	retval = put_user(data, (unsigned long *)buf);
 	if (!retval)
 		retval = sizeof(unsigned long);
-      out:
-	current->state = TASK_RUNNING;
+out:
+	__set_current_state(TASK_RUNNING)
 	remove_wait_queue(&devp->hd_waitqueue, &wait);
 
 	return retval;
_