2013-04-23 08:39:34

by Gu Zheng

[permalink] [raw]
Subject: [PATCH] mm/filemap.c: fix criteria of calling iov_shorten() in generic_file_direct_write()

>From 35947e6535d92c54cf523470cc8811e8b5fee3e5 Mon Sep 17 00:00:00 2001
From: Gu Zheng <[email protected]>
Date: Tue, 23 Apr 2013 16:09:04 +0800
Subject: [PATCH] mm/filemap.c: fix criteria of calling iov_shorten() in generic_file_direct_write()

generic_file_direct_write() compares 'count'(the max count we actually can write)
with 'ocount'(the count we request to write) to see if there is need to call
iov_shorten() to reduce number of segments and the iovec's length. If the
'count' is equal or greater than 'ocount', there is no need to call iov_shorten()
indeed. So the judgement should be changed:
'if (count != ocount)' --> 'if (count < ocount)'

Signed-off-by: Gu Zheng <[email protected]>
---
mm/filemap.c | 2 +-
1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/mm/filemap.c b/mm/filemap.c
index e1979fd..c566b9c 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -2183,7 +2183,7 @@ generic_file_direct_write(struct kiocb *iocb, const struct iovec *iov,
size_t write_len;
pgoff_t end;

- if (count != ocount)
+ if (count < ocount)
*nr_segs = iov_shorten((struct iovec *)iov, *nr_segs, count);

write_len = iov_length(iov, *nr_segs);
--
1.7.7


2013-04-24 12:21:05

by Jan Kara

[permalink] [raw]
Subject: Re: [PATCH] mm/filemap.c: fix criteria of calling iov_shorten() in generic_file_direct_write()

On Tue 23-04-13 16:37:43, Gu Zheng wrote:
> From 35947e6535d92c54cf523470cc8811e8b5fee3e5 Mon Sep 17 00:00:00 2001
> From: Gu Zheng <[email protected]>
> Date: Tue, 23 Apr 2013 16:09:04 +0800
> Subject: [PATCH] mm/filemap.c: fix criteria of calling iov_shorten() in generic_file_direct_write()
>
> generic_file_direct_write() compares 'count'(the max count we actually can write)
> with 'ocount'(the count we request to write) to see if there is need to call
> iov_shorten() to reduce number of segments and the iovec's length. If the
> 'count' is equal or greater than 'ocount', there is no need to call iov_shorten()
> indeed. So the judgement should be changed:
> 'if (count != ocount)' --> 'if (count < ocount)'
Thanks for the patch but it shouldn't be really possible that count >
ocount, should it? So your patch doesn't really fix anything. Or am I
missing something?

Honza
>
> Signed-off-by: Gu Zheng <[email protected]>
> ---
> mm/filemap.c | 2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/mm/filemap.c b/mm/filemap.c
> index e1979fd..c566b9c 100644
> --- a/mm/filemap.c
> +++ b/mm/filemap.c
> @@ -2183,7 +2183,7 @@ generic_file_direct_write(struct kiocb *iocb, const struct iovec *iov,
> size_t write_len;
> pgoff_t end;
>
> - if (count != ocount)
> + if (count < ocount)
> *nr_segs = iov_shorten((struct iovec *)iov, *nr_segs, count);
>
> write_len = iov_length(iov, *nr_segs);
> --
> 1.7.7
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
--
Jan Kara <[email protected]>
SUSE Labs, CR

2013-04-25 06:56:48

by Gu Zheng

[permalink] [raw]
Subject: Re: [PATCH] mm/filemap.c: fix criteria of calling iov_shorten() in generic_file_direct_write()

On 04/24/2013 08:21 PM, Jan Kara wrote:

> On Tue 23-04-13 16:37:43, Gu Zheng wrote:
>> From 35947e6535d92c54cf523470cc8811e8b5fee3e5 Mon Sep 17 00:00:00 2001
>> From: Gu Zheng <[email protected]>
>> Date: Tue, 23 Apr 2013 16:09:04 +0800
>> Subject: [PATCH] mm/filemap.c: fix criteria of calling iov_shorten() in generic_file_direct_write()
>>
>> generic_file_direct_write() compares 'count'(the max count we actually can write)
>> with 'ocount'(the count we request to write) to see if there is need to call
>> iov_shorten() to reduce number of segments and the iovec's length. If the
>> 'count' is equal or greater than 'ocount', there is no need to call iov_shorten()
>> indeed. So the judgement should be changed:
>> 'if (count != ocount)' --> 'if (count < ocount)'
> Thanks for the patch but it shouldn't be really possible that count >
> ocount, should it? So your patch doesn't really fix anything. Or am I
> missing something?

Hi Jan,
You are right, 'count' should not be greater than 'ocount' actually.
Please ignore this patch.

Thanks,
Gu

>
> Honza
>>
>> Signed-off-by: Gu Zheng <[email protected]>
>> ---
>> mm/filemap.c | 2 +-
>> 1 files changed, 1 insertions(+), 1 deletions(-)
>>
>> diff --git a/mm/filemap.c b/mm/filemap.c
>> index e1979fd..c566b9c 100644
>> --- a/mm/filemap.c
>> +++ b/mm/filemap.c
>> @@ -2183,7 +2183,7 @@ generic_file_direct_write(struct kiocb *iocb, const struct iovec *iov,
>> size_t write_len;
>> pgoff_t end;
>>
>> - if (count != ocount)
>> + if (count < ocount)
>> *nr_segs = iov_shorten((struct iovec *)iov, *nr_segs, count);
>>
>> write_len = iov_length(iov, *nr_segs);
>> --
>> 1.7.7
>>
>> --
>> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
>> the body of a message to [email protected]
>> More majordomo info at http://vger.kernel.org/majordomo-info.html
>> Please read the FAQ at http://www.tux.org/lkml/