On Sun, 2009-01-25 at 18:13 +0100, Volker Armin Hemmann wrote: > so without Ingo Bormuth's patch you can't use r4 with 2.6.28.1 and later and > with his patch there is a risk of deadlocks? After Edward's reply actually I changed to use the attached mm patch which is actually the same as what I did with additionally setting the AOP_FLAG_NOFS flag before calling grab_cache_page_write_begin. It applies and works nicely with vanilla 2.6.28.2 too. -- Ingo Bormuth, voicebox & telefax: +49-(0)-12125-10226517 PGP public key 86326EC9 at http://ibormuth.efil.de/contact
From: KOSAKI Motohiro <kosaki.motohiro@xxxxxxxxxxxxxx> build fix. Signed-off-by: KOSAKI Motohiro <kosaki.motohiro@xxxxxxxxxxxxxx> Cc: Nick Piggin <npiggin@xxxxxxx> Signed-off-by: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> --- fs/reiser4/plugin/file/file_conversion.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff -puN fs/reiser4/plugin/file/file_conversion.c~fs-symlink-write_begin-allocation-context-fix-reiser4-fix fs/reiser4/plugin/file/file_conversion.c --- a/fs/reiser4/plugin/file/file_conversion.c~fs-symlink-write_begin-allocation-context-fix-reiser4-fix +++ a/fs/reiser4/plugin/file/file_conversion.c @@ -686,7 +686,8 @@ int reiser4_write_begin_careful(struct f start = pos & (PAGE_CACHE_SIZE - 1); end = start + len; - page = __grab_cache_page(mapping, index); + page = grab_cache_page_write_begin(mapping, index, + flags & AOP_FLAG_NOFS); *pagep = page; if (!page) return -ENOMEM; _
