Thread (7 messages) 7 messages, 3 authors, 2012-06-07

Re: [PATCH 2/2] umem: fix up unplugging

From: Tao Guo <hidden>
Date: 2012-06-04 14:49:39
Also in: lkml

This patch is to fix a regression introduced by 7eaceaccab5f40
("block: remove per-queueplugging").

-Tao

On Mon, Jun 4, 2012 at 10:41 PM, Tao Guo [off-list ref] wrote:
quoted hunk ↗ jump to hunk
In that patch, Jens removed the whole mm_unplug_device()
function, which used to be the trigger to make umem start to work.

We need to implement unplugging to make umem start to work, or I/O
will never be triggered.

Signed-off-by: Tao Guo <redacted>
Cc: Neil Brown <redacted>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: <redacted>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: <redacted>
---
 drivers/block/umem.c |   11 +++++++++++
 1 files changed, 11 insertions(+), 0 deletions(-)
diff --git a/drivers/block/umem.c b/drivers/block/umem.c
index aa27120..89cc9a6 100644
--- a/drivers/block/umem.c
+++ b/drivers/block/umem.c
@@ -513,6 +513,15 @@ static void process_page(unsigned long data)
       }
 }

+static void mm_unplug(struct blk_plug_cb *cb)
+{
+       struct cardinfo *card = cb->q->queuedata;
+
+       spin_lock_irq(&card->lock);
+       activate(card);
+       spin_unlock_irq(&card->lock);
+}
+
 static void mm_make_request(struct request_queue *q, struct bio *bio)
 {
       struct cardinfo *card = q->queuedata;
@@ -523,6 +532,8 @@ static void mm_make_request(struct request_queue *q, struct bio *bio)
       *card->biotail = bio;
       bio->bi_next = NULL;
       card->biotail = &bio->bi_next;
+       if (bio->bi_rw & REQ_SYNC || !blk_check_plugged(q, mm_unplug))
+               activate(card);
       spin_unlock_irq(&card->lock);

       return;
--
1.7.7.6
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help