Hi Ira! On Sun 28-03-21 23:54:00, ira.weiny@xxxxxxxxx wrote: > From: Ira Weiny <ira.weiny@xxxxxxxxx> > > kmap is inefficient and can be abused so it is being phased out in favor of > kmap_local_page where possible. > > ext2 uses kmap in ext2_[get|put]_page(). All of the calls to > ext2_[get|put]_page() occur in single threads so it is perfectly safe and > preferable to use kmap_local_page(). > > This series has a clean up which matches ext2_put_page() with ext2_dotdot() and > ext2_find_entry(). Those calls use ext2_get_page() to map the page prior to > returning it to their callers. And they document that ext2_put_page() should > be matched up with them. This was the case but the ext2_put_page() calls were > hidden in other functions. We lift the ext2_put_page() calls to match up to > the functions where ext2_dotdot() and ext2_find_entry() are called. > > After that clean up convert ext2_[get|put]_page() to kmap and adjust for > kunmap_local() requiring the page address. > > Nesting of kmap_local_page() calls is maintained with minor changes. Pulled into my tree. Thanks! Honza > > Ira Weiny (2): > ext2: Match up ext2_put_page() with ext2_dotdot() and > ext2_find_entry() > fs/ext2: Replace kmap() with kmap_local_page() > > fs/ext2/dir.c | 94 +++++++++++++++++++++++++++++++------------------ > fs/ext2/ext2.h | 12 ++++--- > fs/ext2/namei.c | 34 +++++++++++------- > 3 files changed, 89 insertions(+), 51 deletions(-) > > -- > 2.28.0.rc0.12.gb6a658bd00c9 > -- Jan Kara <jack@xxxxxxxx> SUSE Labs, CR
