[PATCH 06/11] uml ubd driver: ubd_io_lock usage fixup

[PATCH 06/11] uml ubd driver: ubd_io_lock usage fixup

[Paolo 'Blaisorblade' Giarrusso]

From: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>

Add some comments about requirements for ubd_io_lock and expand its use.

When an irq signals that the "controller" (i.e. another thread on the host,
which does the actual requests and is the only one blocked on I/O on the host)
has done some work, we call again the request function ourselves
(do_ubd_request).

We now do that with ubd_io_lock held - that's useful to protect against
concurrent calls to elv_next_request and so on.

XXX: Maybe we shouldn't call at all the request function. Input needed on this.
Are we supposed to plug and unplug the queue? That code "indirectly" does that
by setting a flag, called do_ubd, which makes the request function return (it's
a residual of 2.4 block layer interface).

Meanwhile, however, merge this patch, which improves things.

Cc: Jens Axboe <axboe@suse.de>
Signed-off-by: Paolo 'Blaisorblade' Giarrusso <blaisorblade@yahoo.it>
---

 arch/um/drivers/ubd_kern.c |    8 +++++++-
 1 files changed, 7 insertions(+), 1 deletions(-)

diff --git a/arch/um/drivers/ubd_kern.c b/arch/um/drivers/ubd_kern.c
index e4fd29a..1202592 100644
--- a/arch/um/drivers/ubd_kern.c
+++ b/arch/um/drivers/ubd_kern.c
@@ -106,6 +106,8 @@ static inline void ubd_set_bit(__u64 bit
 
 #define DRIVER_NAME "uml-blkdev"
 
+/* Can be taken in interrupt context, and is passed to the block layer to lock
+ * the request queue. Kernel side code knows that. */
 static DEFINE_SPINLOCK(ubd_io_lock);
 
 static DEFINE_MUTEX(ubd_lock);
@@ -497,6 +499,8 @@ static void __ubd_finish(struct request
 	end_request(req, 1);
 }
 
+/* Callable only from interrupt context - otherwise you need to do
+ * spin_lock_irq()/spin_lock_irqsave() */
 static inline void ubd_finish(struct request *req, int error)
 {
  	spin_lock(&ubd_io_lock);
@@ -504,7 +508,7 @@ static inline void ubd_finish(struct req
 	spin_unlock(&ubd_io_lock);
 }
 
-/* Called without ubd_io_lock held */
+/* Called without ubd_io_lock held, and only in interrupt context. */
 static void ubd_handler(void)
 {
 	struct io_thread_req req;
@@ -525,7 +529,9 @@ static void ubd_handler(void)
 
 	ubd_finish(rq, req.error);
 	reactivate_fd(thread_fd, UBD_IRQ);	
+	spin_lock(&ubd_io_lock);
 	do_ubd_request(ubd_queue);
+	spin_unlock(&ubd_io_lock);
 }
 
 static irqreturn_t ubd_intr(int irq, void *dev)
Chiacchiera con i tuoi amici in tempo reale! 
 http://it.yahoo.com/mail_it/foot/*http://it.messenger.yahoo.com 

Re: [PATCH 06/11] uml ubd driver: ubd_io_lock usage fixup

[Jens Axboe]

On Sun, Oct 29 2006, Blaisorblade wrote:
> Add some comments about requirements for ubd_io_lock and expand its use.
> 
> When an irq signals that the "controller" (i.e. another thread on the host,
> which does the actual requests and is the only one blocked on I/O on the
>  host) has done some work, we call again the request function ourselves
> (do_ubd_request).
> 
> We now do that with ubd_io_lock held - that's useful to protect against
> concurrent calls to elv_next_request and so on.

Not only useful, required, as I think I complained about a year or more
ago :-)

> XXX: Maybe we shouldn't call at all the request function. Input needed on
>  this. Are we supposed to plug and unplug the queue? That code "indirectly"
>  does that by setting a flag, called do_ubd, which makes the request function
>  return (it's a residual of 2.4 block layer interface).

Sometimes you need to. I'd probably just remove the do_ubd check and
always recall the request function when handling completions, it's
easier and safe.

-- 
Jens Axboe