From: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
To: James.Bottomley@HansenPartnership.com
Cc: jens.axboe@oracle.com, htejun@gmail.com, bharrosh@panasas.com,
fujita.tomonori@lab.ntt.co.jp, efault@gmx.de, tomof@acm.org,
akpm@linux-foundation.org, linux-kernel@vger.kernel.org,
linux-ide@vger.kernel.org, linux-scsi@vger.kernel.org,
jgarzik@pobox.com, bzolnier@gmail.com
Subject: Re: [PATCH] blk: missing add of padded bytes to io completion byte count
Date: Thu, 06 Mar 2008 13:41:46 +0900 [thread overview]
Message-ID: <20080306134146A.fujita.tomonori@lab.ntt.co.jp> (raw)
In-Reply-To: <1204730484.3047.10.camel@localhost.localdomain>
On Wed, 05 Mar 2008 09:21:24 -0600
James Bottomley <James.Bottomley@HansenPartnership.com> wrote:
> On Wed, 2008-03-05 at 14:51 +0100, Jens Axboe wrote:
> > On Wed, Mar 05 2008, Tejun Heo wrote:
> > > This is getting insanely subtle. Let's say there's PIO driver which
> > > transfer certain sized chunks at a time and completes request partially
> > > after completing each chunk and the driver uses draining to eat up
> > > whatever excess data, which seems like a legit use case to me. But it
> > > won't work because __end_that_request_first() will terminate when it
> > > reaches reaches the 'true' transfer size. That's just broken API. FWIW,
> > >
> > > Nacked-by: Tejun Heo <htejun@gmail.com>
> >
> > Yeah, I think I may have gone a bit overboard in applying this so
> > quickly. It's just not a good interface, silently adding the extra
> > length if asked to complete more. It may even happen right now, for a
> > driver that does no padding (it probably wont do any harm here either,
> > but still).
> >
> > I'll try and see if I can come up with something cleaner.
> >
> > My basic design paradigm for this is that the _driver_ (or mid layer, if
> > SCSI wants to handle it) should care about the padding. So make it easy
> > for them to pad, but have it 'unrolled' by completion time. We should
> > NOT need any extra_len checks or additions in the block/ directory,
> > period.
>
> Right, that's why my original proposal was to do nothing for padding
> (other than ensure the driver could adjust the length if it wanted to)
> and to add an extra element always for draining, which the driver could
> ignore. It basically pushed the use paradigm onto the driver.
>
> If we want the use paradigm shared between block and driver, then I
> think the best approach is to keep all the bios the same (so not adjust
> for padding), but do adjust in the blk_rq_map_sg(). That way we have
> the padding and draining unwind information by comparing with the bio.
Adjusting only sg in blk_rq_map_sg (like drain) looks much
better. This works with libata for me.
diff --git a/block/blk-map.c b/block/blk-map.c
index c07d9c8..e949969 100644
--- a/block/blk-map.c
+++ b/block/blk-map.c
@@ -140,26 +140,6 @@ int blk_rq_map_user(struct request_queue *q, struct request *rq,
ubuf += ret;
}
- /*
- * __blk_rq_map_user() copies the buffers if starting address
- * or length isn't aligned to dma_pad_mask. As the copied
- * buffer is always page aligned, we know that there's enough
- * room for padding. Extend the last bio and update
- * rq->data_len accordingly.
- *
- * On unmap, bio_uncopy_user() will use unmodified
- * bio_map_data pointed to by bio->bi_private.
- */
- if (len & q->dma_pad_mask) {
- unsigned int pad_len = (q->dma_pad_mask & ~len) + 1;
- struct bio *tail = rq->biotail;
-
- tail->bi_io_vec[tail->bi_vcnt - 1].bv_len += pad_len;
- tail->bi_size += pad_len;
-
- rq->extra_len += pad_len;
- }
-
rq->buffer = rq->data = NULL;
return 0;
unmap_rq:
diff --git a/block/blk-merge.c b/block/blk-merge.c
index 0f58616..2a81c87 100644
--- a/block/blk-merge.c
+++ b/block/blk-merge.c
@@ -220,6 +220,13 @@ new_segment:
bvprv = bvec;
} /* segments in rq */
+ if (sg && (q->dma_pad_mask & rq->data_len)) {
+ unsigned int pad_len = (q->dma_pad_mask & ~rq->data_len) + 1;
+
+ sg->length += pad_len;
+ rq->extra_len += pad_len;
+ }
+
if (q->dma_drain_size && q->dma_drain_needed(rq)) {
if (rq->cmd_flags & REQ_RW)
memset(q->dma_drain_buffer, 0, q->dma_drain_size);
diff --git a/drivers/scsi/scsi.c b/drivers/scsi/scsi.c
index e5c6f6a..fecba05 100644
--- a/drivers/scsi/scsi.c
+++ b/drivers/scsi/scsi.c
@@ -757,7 +757,7 @@ void scsi_finish_command(struct scsi_cmnd *cmd)
"Notifying upper driver of completion "
"(result %x)\n", cmd->result));
- good_bytes = scsi_bufflen(cmd) + cmd->request->extra_len;
+ good_bytes = scsi_bufflen(cmd);
if (cmd->request->cmd_type != REQ_TYPE_BLOCK_PC) {
drv = scsi_cmd_to_driver(cmd);
if (drv->done)
next prev parent reply other threads:[~2008-03-06 4:43 UTC|newest]
Thread overview: 109+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-02-21 8:42 regression: CD burning (k3b) went broke Mike Galbraith
2008-02-22 7:32 ` Jens Axboe
2008-02-23 7:42 ` Mike Galbraith
2008-02-24 7:54 ` Mike Galbraith
2008-02-26 9:48 ` Mike Galbraith
2008-02-26 13:36 ` Mike Galbraith
2008-02-26 23:08 ` Andrew Morton
2008-02-27 0:46 ` Jeff Garzik
2008-02-27 2:58 ` Mike Galbraith
2008-02-27 2:24 ` Mike Galbraith
2008-02-27 6:00 ` Mike Galbraith
2008-02-27 7:07 ` Mike Galbraith
2008-02-28 7:43 ` Tejun Heo
2008-02-28 8:20 ` Mike Galbraith
2008-02-28 8:50 ` [PATCH] block: fix residual byte count handling Tejun Heo
2008-02-28 15:35 ` Jens Axboe
2008-02-28 15:46 ` Tejun Heo
2008-02-29 16:47 ` James Bottomley
2008-02-29 20:11 ` Jens Axboe
2008-03-01 6:17 ` Tejun Heo
2008-03-01 15:19 ` James Bottomley
2008-03-02 14:52 ` FUJITA Tomonori
2008-03-02 18:46 ` Mike Christie
2008-03-03 3:27 ` Mike Galbraith
2008-03-03 2:40 ` Tejun Heo
2008-03-03 3:59 ` FUJITA Tomonori
2008-03-03 4:09 ` Tejun Heo
2008-03-03 6:08 ` [PATCH 1/2] " Tejun Heo
2008-03-03 6:10 ` [PATCH] block: separate out padding from alignment Tejun Heo
2008-03-03 18:27 ` James Bottomley
2008-03-03 8:26 ` [PATCH] block: fix residual byte count handling FUJITA Tomonori
2008-03-03 9:21 ` Tejun Heo
2008-03-03 12:17 ` FUJITA Tomonori
2008-03-03 13:38 ` Tejun Heo
2008-03-03 13:50 ` FUJITA Tomonori
2008-03-03 13:55 ` Tejun Heo
2008-03-03 14:01 ` FUJITA Tomonori
2008-03-03 14:22 ` Tejun Heo
2008-03-03 14:52 ` FUJITA Tomonori
2008-03-03 22:44 ` Tejun Heo
2008-03-04 2:11 ` FUJITA Tomonori
2008-03-04 2:32 ` Tejun Heo
2008-03-04 8:53 ` FUJITA Tomonori
2008-03-04 8:59 ` Jens Axboe
2008-03-04 9:06 ` FUJITA Tomonori
2008-03-04 9:22 ` FUJITA Tomonori
2008-03-04 9:30 ` Tejun Heo
2008-03-04 9:35 ` Jens Axboe
2008-03-04 9:40 ` Tejun Heo
2008-03-04 9:46 ` Jens Axboe
2008-03-04 12:37 ` Mike Galbraith
2008-03-04 12:39 ` Jens Axboe
2008-03-04 12:43 ` Mike Galbraith
2008-03-04 12:58 ` Mike Galbraith
2008-03-04 13:03 ` Jens Axboe
2008-03-04 14:25 ` Mike Galbraith
2008-03-04 18:17 ` Jens Axboe
2008-03-04 18:29 ` Jens Axboe
2008-03-04 18:35 ` Mike Galbraith
2008-03-04 18:45 ` Jens Axboe
2008-03-04 18:49 ` Mike Galbraith
2008-03-04 18:54 ` Jens Axboe
2008-03-04 19:26 ` Mike Galbraith
2008-03-04 19:28 ` Jens Axboe
2008-03-04 16:04 ` James Bottomley
2008-03-04 18:46 ` Jens Axboe
2008-03-04 17:34 ` walt
2008-03-04 17:59 ` Tejun Heo
2008-03-04 19:42 ` Kiyoshi Ueda
2008-03-04 12:40 ` Tejun Heo
2008-03-04 12:45 ` Mike Galbraith
2008-03-04 13:30 ` FUJITA Tomonori
2008-03-04 13:50 ` Tejun Heo
2008-03-04 16:17 ` Tejun Heo
2008-03-04 16:42 ` Tejun Heo
2008-03-04 18:26 ` Boaz Harrosh
2008-03-04 18:35 ` Tejun Heo
2008-03-04 18:27 ` James Bottomley
2008-03-04 18:33 ` Tejun Heo
2008-03-04 18:45 ` Mike Galbraith
2008-03-04 19:25 ` Jens Axboe
2008-03-04 19:33 ` Mike Galbraith
2008-03-04 19:34 ` Jens Axboe
2008-03-04 19:19 ` FUJITA Tomonori
2008-03-04 23:33 ` Tejun Heo
2008-03-04 23:54 ` Tejun Heo
2008-03-05 0:26 ` FUJITA Tomonori
2008-03-05 0:44 ` Tejun Heo
2008-03-06 4:56 ` FUJITA Tomonori
2008-03-06 5:02 ` Tejun Heo
2008-03-05 10:16 ` [PATCH] blk: missing add of padded bytes to io completion byte count Boaz Harrosh
2008-03-05 12:28 ` Mike Galbraith
2008-03-05 12:33 ` Jens Axboe
2008-03-05 12:46 ` Boaz Harrosh
2008-03-05 12:48 ` Jens Axboe
2008-03-05 13:45 ` Tejun Heo
2008-03-05 13:51 ` Jens Axboe
2008-03-05 14:08 ` Tejun Heo
2008-03-05 15:21 ` James Bottomley
2008-03-06 4:41 ` FUJITA Tomonori [this message]
2008-03-06 13:41 ` Jens Axboe
2008-03-07 0:07 ` Tejun Heo
2008-03-07 15:07 ` FUJITA Tomonori
2008-03-08 1:06 ` Tejun Heo
2008-03-20 12:54 ` FUJITA Tomonori
2008-03-05 14:46 ` Boaz Harrosh
2008-03-05 15:11 ` Tejun Heo
2008-03-06 5:02 ` FUJITA Tomonori
2008-03-04 9:29 ` [PATCH] block: fix residual byte count handling Tejun Heo
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20080306134146A.fujita.tomonori@lab.ntt.co.jp \
--to=fujita.tomonori@lab.ntt.co.jp \
--cc=James.Bottomley@HansenPartnership.com \
--cc=akpm@linux-foundation.org \
--cc=bharrosh@panasas.com \
--cc=bzolnier@gmail.com \
--cc=efault@gmx.de \
--cc=htejun@gmail.com \
--cc=jens.axboe@oracle.com \
--cc=jgarzik@pobox.com \
--cc=linux-ide@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-scsi@vger.kernel.org \
--cc=tomof@acm.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).