From: Boaz Harrosh <bharrosh@panasas.com>
To: Weston Andros Adamson <dros@primarydata.com>,
<trond.myklebust@primarydata.com>
Cc: <linux-nfs@vger.kernel.org>
Subject: Re: [PATCH 11/17] nfs: chain calls to pg_test
Date: Wed, 23 Apr 2014 15:20:15 +0300 [thread overview]
Message-ID: <5357AFFF.8030902@panasas.com> (raw)
In-Reply-To: <1398202165-78897-12-git-send-email-dros@primarydata.com>
On 04/23/2014 12:29 AM, Weston Andros Adamson wrote:
> Now that pg_test can change the size of the request (by returning a non-zero
> size smaller than the request), pg_test functions that call other
> pg_test functions must return the minimum of the result - or 0 if any fail.
>
> Also clean up the logic of some pg_test functions so that all checks are
> for contitions where coalescing is not possible.
>
> Signed-off-by: Weston Andros Adamson <dros@primarydata.com>
> ---
> fs/nfs/nfs4filelayout.c | 27 ++++++++++++++-------------
> fs/nfs/objlayout/objio_osd.c | 12 ++++++++----
> fs/nfs/pnfs.c | 15 ++++++++++-----
> 3 files changed, 32 insertions(+), 22 deletions(-)
>
> diff --git a/fs/nfs/nfs4filelayout.c b/fs/nfs/nfs4filelayout.c
> index 3b32c95..cfd76bd 100644
> --- a/fs/nfs/nfs4filelayout.c
> +++ b/fs/nfs/nfs4filelayout.c
> @@ -930,26 +930,27 @@ static size_t
> filelayout_pg_test(struct nfs_pageio_descriptor *pgio, struct nfs_page *prev,
> struct nfs_page *req)
> {
> + unsigned int size;
> u64 p_stripe, r_stripe;
> u32 stripe_unit;
>
> - if (!pnfs_generic_pg_test(pgio, prev, req) ||
> - !nfs_generic_pg_test(pgio, prev, req))
> + /* calls nfs_generic_pg_test */
> + size = pnfs_generic_pg_test(pgio, prev, req);
> + if (!size)
> return 0;
>
> - if (!prev)
> - return req->wb_bytes;
> + if (prev) {
> + p_stripe = (u64)req_offset(prev);
> + r_stripe = (u64)req_offset(req);
> + stripe_unit = FILELAYOUT_LSEG(pgio->pg_lseg)->stripe_unit;
>
> - p_stripe = (u64)req_offset(prev);
> - r_stripe = (u64)req_offset(req);
> - stripe_unit = FILELAYOUT_LSEG(pgio->pg_lseg)->stripe_unit;
> + do_div(p_stripe, stripe_unit);
> + do_div(r_stripe, stripe_unit);
>
> - do_div(p_stripe, stripe_unit);
> - do_div(r_stripe, stripe_unit);
> -
> - if (p_stripe == r_stripe)
> - return req->wb_bytes;
> - return 0;
> + if (p_stripe != r_stripe)
> + return 0;
> + }
> + return min(size, req->wb_bytes);
> }
>
> static void
> diff --git a/fs/nfs/objlayout/objio_osd.c b/fs/nfs/objlayout/objio_osd.c
> index c20352a..31de29e 100644
> --- a/fs/nfs/objlayout/objio_osd.c
> +++ b/fs/nfs/objlayout/objio_osd.c
> @@ -571,13 +571,17 @@ int objio_write_pagelist(struct nfs_write_data *wdata, int how)
> static size_t objio_pg_test(struct nfs_pageio_descriptor *pgio,
> struct nfs_page *prev, struct nfs_page *req)
> {
> - if (!pnfs_generic_pg_test(pgio, prev, req))
> + unsigned int size;
> +
> + size = pnfs_generic_pg_test(pgio, prev, req);
> +
> + if (!size)
> return 0;
>
> - if (pgio->pg_count + req->wb_bytes <=
> + if (pgio->pg_count + req->wb_bytes >
> (unsigned long)pgio->pg_layout_private)
> - return req->wb_bytes;
> - return 0;
> + return 0;
objio_osd can enjoy the new facility by returning the
remainder here:
max_io = (unsigned long)pgio->pg_layout_private);
wb_bytes = min(size, req->wb_bytes);
if (pgio->pg_count + req->wb_bytes > max_io)
wb_bytes = max_io - pgio->pg_count;
return wb_bytes;
Which reminds me that this code sucks and I need to fix it. I will do
so after you send your changes.
[I promise to test these guys soon. Can you please put them on a public tree?]
Thanks
Boaz
> + return min(size, req->wb_bytes);
> }
>
> static void objio_init_read(struct nfs_pageio_descriptor *pgio, struct nfs_page *req)
> diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c
> index 6201bf6..7c89385 100644
> --- a/fs/nfs/pnfs.c
> +++ b/fs/nfs/pnfs.c
> @@ -1469,8 +1469,12 @@ size_t
> pnfs_generic_pg_test(struct nfs_pageio_descriptor *pgio, struct nfs_page *prev,
> struct nfs_page *req)
> {
> - if (pgio->pg_lseg == NULL)
> - return nfs_generic_pg_test(pgio, prev, req);
> + unsigned int size;
> +
> + size = nfs_generic_pg_test(pgio, prev, req);
> +
> + if (!size)
> + return 0;
>
> /*
> * Test if a nfs_page is fully contained in the pnfs_layout_range.
> @@ -1486,10 +1490,11 @@ pnfs_generic_pg_test(struct nfs_pageio_descriptor *pgio, struct nfs_page *prev,
> * first byte that lies outside the pnfs_layout_range. FIXME?
> *
> */
> - if (req_offset(req) < end_offset(pgio->pg_lseg->pls_range.offset,
> + if (req_offset(req) >= end_offset(pgio->pg_lseg->pls_range.offset,
> pgio->pg_lseg->pls_range.length))
> - return req->wb_bytes;
> - return 0;
> + return 0;
> +
> + return min(size, req->wb_bytes);
> }
> EXPORT_SYMBOL_GPL(pnfs_generic_pg_test);
>
>
next prev parent reply other threads:[~2014-04-23 12:20 UTC|newest]
Thread overview: 35+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-04-22 21:29 [PATCH 00/17] nfs: support multiple requests per page Weston Andros Adamson
2014-04-22 21:29 ` [PATCH 01/17] nfs: clean up PG_* flags Weston Andros Adamson
2014-04-22 21:29 ` [PATCH 02/17] nfs: remove unused arg from nfs_create_request Weston Andros Adamson
2014-04-22 21:29 ` [PATCH 03/17] nfs: modify pg_test interface to return size_t Weston Andros Adamson
2014-04-23 12:30 ` Boaz Harrosh
2014-04-22 21:29 ` [PATCH 04/17] nfs: call nfs_can_coalesce_requests for every req Weston Andros Adamson
2014-04-22 21:29 ` [PATCH 05/17] nfs: add support for multiple nfs reqs per page Weston Andros Adamson
2014-04-22 21:40 ` Weston Andros Adamson
2014-04-23 14:40 ` Weston Andros Adamson
2014-04-24 14:50 ` Jeff Layton
2014-04-24 15:23 ` Weston Andros Adamson
2014-04-24 15:45 ` Jeff Layton
2014-04-24 16:15 ` Weston Andros Adamson
2014-04-24 16:52 ` Jeff Layton
2014-04-24 17:23 ` Weston Andros Adamson
2014-04-22 21:29 ` [PATCH 06/17] nfs: page group syncing in read path Weston Andros Adamson
2014-04-22 21:29 ` [PATCH 07/17] nfs: page group syncing in write path Weston Andros Adamson
2014-04-22 21:29 ` [PATCH 08/17] nfs: page group support in nfs_mark_uptodate Weston Andros Adamson
2014-04-22 21:29 ` [PATCH 09/17] pnfs: clean up filelayout_alloc_commit_info Weston Andros Adamson
2014-04-22 21:29 ` [PATCH 10/17] nfs: allow coalescing of subpage requests Weston Andros Adamson
2014-04-22 21:29 ` [PATCH 11/17] nfs: chain calls to pg_test Weston Andros Adamson
2014-04-23 12:20 ` Boaz Harrosh [this message]
2014-04-23 13:37 ` Weston Andros Adamson
2014-04-22 21:29 ` [PATCH 12/17] nfs: use > 1 request to handle bsize < PAGE_SIZE Weston Andros Adamson
2014-04-22 21:29 ` [PATCH 13/17] nfs: remove list of [rw]data from pgio header Weston Andros Adamson
2014-04-23 14:16 ` Anna Schumaker
2014-04-23 14:31 ` Weston Andros Adamson
2014-04-23 14:36 ` Anna Schumaker
2014-04-23 17:44 ` Weston Andros Adamson
2014-04-23 17:51 ` Anna Schumaker
2014-04-24 11:55 ` Boaz Harrosh
2014-04-22 21:29 ` [PATCH 14/17] pnfs: support multiple verfs per direct req Weston Andros Adamson
2014-04-22 21:29 ` [PATCH 15/17] pnfs: allow non page aligned pnfs layout segments Weston Andros Adamson
2014-04-22 21:29 ` [PATCH 16/17] pnfs: filelayout: support non page aligned layouts Weston Andros Adamson
2014-04-22 21:29 ` [PATCH 17/17] nfs: support page groups in nfs_read_completion Weston Andros Adamson
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=5357AFFF.8030902@panasas.com \
--to=bharrosh@panasas.com \
--cc=dros@primarydata.com \
--cc=linux-nfs@vger.kernel.org \
--cc=trond.myklebust@primarydata.com \
/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).