Re: [PATCH] aio: aio_complete() will never be called in interrupt context

[Nikanth Karthikesan]

On Friday 27 June 2008 18:41:03 Jeff Moyer wrote:
> Nikanth Karthikesan <knikanth@xxxxxxx> writes:
> > aio_complete() is never called from interrupt context. It is called from
> > user context or worker threads. Remove disabling interrupts and for
> > kmap_atomic use KM_USER slots instead of KM_IRQ slots.
>
> Actually, it can be called from softirq context.  See fs/directio.c.
>

Oh, sorry for my ignorance. Will using KM_SOFTIRQ slots be fine?

Thanks
Nikanth Karthikesan

aio_complete() is never called from hardware interrupt context. Remove 
disabling interrupts and for kmap_atomic use KM_SOFTIRQ slots instead of 
KM_IRQ slots.

Signed-off-by: Nikanth Karthikesan <knikanth@xxxxxxx>

---

Index: linux-2.6/fs/aio.c
===================================================================
--- linux-2.6.orig/fs/aio.c
+++ linux-2.6/fs/aio.c
@@ -901,7 +901,6 @@ int aio_complete(struct kiocb *iocb, lon
 	struct aio_ring_info	*info;
 	struct aio_ring	*ring;
 	struct io_event	*event;
-	unsigned long	flags;
 	unsigned long	tail;
 	int		ret;
 
@@ -924,11 +923,9 @@ int aio_complete(struct kiocb *iocb, lon
 
 	/* add a completion event to the ring buffer.
 	 * must be done holding ctx->ctx_lock to prevent
-	 * other code from messing with the tail
-	 * pointer since we might be called from irq
-	 * context.
+	 * other code from messing with the tail pointer.
 	 */
-	spin_lock_irqsave(&ctx->ctx_lock, flags);
+	spin_lock(&ctx->ctx_lock);
 
 	if (iocb->ki_run_list.prev && !list_empty(&iocb->ki_run_list))
 		list_del_init(&iocb->ki_run_list);
@@ -940,10 +937,10 @@ int aio_complete(struct kiocb *iocb, lon
 	if (kiocbIsCancelled(iocb))
 		goto put_rq;
 
-	ring = kmap_atomic(info->ring_pages[0], KM_IRQ1);
+	ring = kmap_atomic(info->ring_pages[0], KM_SOFTIRQ1);
 
 	tail = info->tail;
-	event = aio_ring_event(info, tail, KM_IRQ0);
+	event = aio_ring_event(info, tail, KM_SOFTIRQ0);
 	if (++tail >= info->nr)
 		tail = 0;
 
@@ -964,8 +961,8 @@ int aio_complete(struct kiocb *iocb, lon
 	info->tail = tail;
 	ring->tail = tail;
 
-	put_aio_ring_event(event, KM_IRQ0);
-	kunmap_atomic(ring, KM_IRQ1);
+	put_aio_ring_event(event, KM_SOFTIRQ0);
+	kunmap_atomic(ring, KM_SOFTIRQ1);
 
 	pr_debug("added to ring %p at [%lu]\n", iocb, tail);
 
@@ -992,7 +989,7 @@ put_rq:
 	if (waitqueue_active(&ctx->wait))
 		wake_up(&ctx->wait);
 
-	spin_unlock_irqrestore(&ctx->ctx_lock, flags);
+	spin_unlock(&ctx->ctx_lock);
 	return ret;
 }
 




--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/