[PATCH 1/9] lguest: block device speedup

From: Rusty Russell
Date: Thu Mar 08 2007 - 22:06:43 EST


Jens Axboe pointed out that end_request() does not end the entire
request. Go figure. On the upside, he wrote the replacement for me!
Now we do far less block traffic, and our performance sucks less.

Signed-off-by: Rusty Russell <rusty@xxxxxxxxxxxxxxx>

diff -r fdc8cbc1fd61 drivers/block/lguest_blk.c
--- a/drivers/block/lguest_blk.c Thu Mar 08 13:35:39 2007 +1100
+++ b/drivers/block/lguest_blk.c Thu Mar 08 15:51:55 2007 +1100
@@ -45,6 +45,16 @@ struct blockdev
struct request *req;
};

+/* Jens gave me this nice helper to end all chunks of a request. */
+static void end_entire_request(struct request *req, int uptodate)
+{
+ if (end_that_request_first(req, uptodate, req->hard_nr_sectors))
+ BUG();
+ add_disk_randomness(req->rq_disk);
+ blkdev_dequeue_request(req);
+ end_that_request_last(req, uptodate);
+}
+
static irqreturn_t lgb_irq(int irq, void *_bd)
{
struct blockdev *bd = _bd;
@@ -61,7 +71,7 @@ static irqreturn_t lgb_irq(int irq, void
}

spin_lock_irqsave(&bd->lock, flags);
- end_request(bd->req, bd->lb_page->result == 1);
+ end_entire_request(bd->req, bd->lb_page->result == 1);
bd->req = NULL;
bd->dma.used_len = 0;
blk_start_queue(bd->disk->queue);
@@ -149,7 +159,7 @@ again:
pr_debug("Got non-command 0x%08x\n", req->cmd_type);
error:
req->errors++;
- end_request(req, 0);
+ end_entire_request(req, 0);
goto again;
} else {
if (rq_data_dir(req) == WRITE)


-
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/