Re: [RFC][PATCH] problem of cont_prepare_write()

[OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>]

Andrew Morton <akpm@osdl.org> writes:

>> Umm... however, if ->i_size is updated before ->commit_write(),
>> doesn't it allow access to those pages, before all write() work is
>> successful?
>
> That's OK.  A thread which is read()ing that page will either
>
> a) decide that the page is outside i_size, and won't read it anyway or
>
> b) decide that the page is inside i_size and will read the page's contents.
>
> Still, I'd be inclined to update i_size after running ->commit_write.  It
> looks like we can simply replace the call to __block_commit_write() with a
> call to generic_commit_write().

If ->prepare_write() failed, I thought we should restore the ->i_size
by vmtruncate() before running ->prepare_write().

But, it's not required... yes?

Anyway, fixed patch is the following.
-- 
OGAWA Hirofumi <hirofumi@mail.parknet.co.jp>



 fs/buffer.c |    3 +--
 1 files changed, 1 insertion(+), 2 deletions(-)

diff -puN fs/buffer.c~cont_prepare_write-fix fs/buffer.c
--- linux-2.6.10-rc2/fs/buffer.c~cont_prepare_write-fix	2004-11-23 11:10:10.000000000 +0900
+++ linux-2.6.10-rc2-hirofumi/fs/buffer.c	2004-11-23 11:10:10.000000000 +0900
@@ -2224,8 +2224,7 @@ int cont_prepare_write(struct page *page
 		memset(kaddr+zerofrom, 0, PAGE_CACHE_SIZE-zerofrom);
 		flush_dcache_page(new_page);
 		kunmap_atomic(kaddr, KM_USER0);
-		__block_commit_write(inode, new_page,
-				zerofrom, PAGE_CACHE_SIZE);
+		generic_commit_write(NULL, new_page, zerofrom, PAGE_CACHE_SIZE);
 		unlock_page(new_page);
 		page_cache_release(new_page);
 	}
_