[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <4FE16E50.9030304@cn.fujitsu.com>
Date: Wed, 20 Jun 2012 14:31:44 +0800
From: Wanlong Gao <gaowanlong@...fujitsu.com>
To: kosaki.motohiro@...il.com
CC: linux-kernel@...r.kernel.org, linux-mm@...ck.org,
KOSAKI Motohiro <kosaki.motohiro@...fujitsu.com>,
Hugh Dickins <hughd@...gle.com>,
Andrew Morton <akpm@...ux-foundation.org>,
Hillf Danton <dhillf@...il.com>,
Eric Wong <normalperson@...t.net>
Subject: Re: [PATCH] mm, fadvise: don't return -EINVAL when filesystem has
no optimization way
On 06/16/2012 04:36 AM, kosaki.motohiro@...il.com wrote:
> From: KOSAKI Motohiro <kosaki.motohiro@...fujitsu.com>
>
> Eric Wong reported his test suite was fail when /tmp is tmpfs.
>
> https://lkml.org/lkml/2012/2/24/479
>
> Current,input check of POSIX_FADV_WILLNEED has two problems.
>
> 1) require a_ops->readpage.
> But in fact, force_page_cache_readahead() only require
> a target filesystem has either ->readpage or ->readpages.
> 2) return -EINVAL when filesystem don't have ->readpage.
> But, posix says, it should be retrieved a hint. Thus fadvise()
> should return 0 if filesystem has no optimization way.
> Especially, userland application don't know a filesystem type
> of TMPDIR directory as Eric pointed out. Then, userland can't
> avoid this error. We shouldn't encourage to ignore syscall
> return value.
>
> Thus, this patch change a return value to 0 when filesytem don't
> support readahead.
>
> Cc: linux-mm@...ck.org
> Cc: Hugh Dickins <hughd@...gle.com>
> Cc: Andrew Morton <akpm@...ux-foundation.org>
> Cc: Hillf Danton <dhillf@...il.com>
> Signed-off-by: Eric Wong <normalperson@...t.net>
> Tested-by: Eric Wong <normalperson@...t.net>
> Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@...fujitsu.com>
> ---
> mm/fadvise.c | 18 +++++++-----------
> 1 files changed, 7 insertions(+), 11 deletions(-)
>
> diff --git a/mm/fadvise.c b/mm/fadvise.c
> index 469491e..33e6baf 100644
> --- a/mm/fadvise.c
> +++ b/mm/fadvise.c
> @@ -93,11 +93,6 @@ SYSCALL_DEFINE(fadvise64_64)(int fd, loff_t offset, loff_t len, int advice)
> spin_unlock(&file->f_lock);
> break;
> case POSIX_FADV_WILLNEED:
> - if (!mapping->a_ops->readpage) {
> - ret = -EINVAL;
> - break;
> - }
Why not check both readpage and readpages, if they are not here,
just beak and no following force_page_cache_readahead ?
Thanks,
Wanlong Gao
> -
> /* First and last PARTIAL page! */
> start_index = offset >> PAGE_CACHE_SHIFT;
> end_index = endbyte >> PAGE_CACHE_SHIFT;
> @@ -106,12 +101,13 @@ SYSCALL_DEFINE(fadvise64_64)(int fd, loff_t offset, loff_t len, int advice)
> nrpages = end_index - start_index + 1;
> if (!nrpages)
> nrpages = ~0UL;
> -
> - ret = force_page_cache_readahead(mapping, file,
> - start_index,
> - nrpages);
> - if (ret > 0)
> - ret = 0;
> +
> + /*
> + * Ignore return value because fadvise() shall return
> + * success even if filesystem can't retrieve a hint,
> + */
> + force_page_cache_readahead(mapping, file, start_index,
> + nrpages);
> break;
> case POSIX_FADV_NOREUSE:
> break;
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists