[0/5] UFS: Final folio conversions

Message

Matthew Wilcox Oct. 5, 2024, 6:02 p.m. UTC

This is the last use of struct page I've been able to find in UFS.
All the hard work was done earlier; this is just passing in bh->b_folio
instead of bh->b_page.

Matthew Wilcox (Oracle) (5):
  ufs: Convert ufs_inode_getblock() to take a folio
  ufs: Convert ufs_extend_tail() to take a folio
  ufs: Convert ufs_inode_getfrag() to take a folio
  ufs: Pass a folio to ufs_new_fragments()
  ufs: Convert ufs_change_blocknr() to take a folio

 fs/ufs/balloc.c | 16 ++++++++--------
 fs/ufs/inode.c  | 30 ++++++++++++++----------------
 fs/ufs/ufs.h    |  8 ++++----
 3 files changed, 26 insertions(+), 28 deletions(-)

Comments

Al Viro Oct. 5, 2024, 6:31 p.m. UTC | #1

On Sat, Oct 05, 2024 at 07:02:03PM +0100, Matthew Wilcox (Oracle) wrote:
> This is the last use of struct page I've been able to find in UFS.
> All the hard work was done earlier; this is just passing in bh->b_folio
> instead of bh->b_page.
> 
> Matthew Wilcox (Oracle) (5):
>   ufs: Convert ufs_inode_getblock() to take a folio
>   ufs: Convert ufs_extend_tail() to take a folio
>   ufs: Convert ufs_inode_getfrag() to take a folio
>   ufs: Pass a folio to ufs_new_fragments()
>   ufs: Convert ufs_change_blocknr() to take a folio
> 
>  fs/ufs/balloc.c | 16 ++++++++--------
>  fs/ufs/inode.c  | 30 ++++++++++++++----------------
>  fs/ufs/ufs.h    |  8 ++++----
>  3 files changed, 26 insertions(+), 28 deletions(-)

Looks sane (for now - there's really interesting shite around the block
relocation, hole filling, etc.); I've got a bunch of UFS patches and IMO
it makes sense to keep it in the same branch.  Mine is in #work.ufs
and there's pending work locally.

I'll add yours to the mix and see if xfstests catch any problems there;
will post tonight.

Al Viro Oct. 5, 2024, 9:21 p.m. UTC | #2

On Sat, Oct 05, 2024 at 07:31:08PM +0100, Al Viro wrote:
> On Sat, Oct 05, 2024 at 07:02:03PM +0100, Matthew Wilcox (Oracle) wrote:
> > This is the last use of struct page I've been able to find in UFS.
> > All the hard work was done earlier; this is just passing in bh->b_folio
> > instead of bh->b_page.
> > 
> > Matthew Wilcox (Oracle) (5):
> >   ufs: Convert ufs_inode_getblock() to take a folio
> >   ufs: Convert ufs_extend_tail() to take a folio
> >   ufs: Convert ufs_inode_getfrag() to take a folio
> >   ufs: Pass a folio to ufs_new_fragments()
> >   ufs: Convert ufs_change_blocknr() to take a folio
> > 
> >  fs/ufs/balloc.c | 16 ++++++++--------
> >  fs/ufs/inode.c  | 30 ++++++++++++++----------------
> >  fs/ufs/ufs.h    |  8 ++++----
> >  3 files changed, 26 insertions(+), 28 deletions(-)
> 
> Looks sane (for now - there's really interesting shite around the block
> relocation, hole filling, etc.); I've got a bunch of UFS patches and IMO
> it makes sense to keep it in the same branch.  Mine is in #work.ufs
> and there's pending work locally.
> 
> I'll add yours to the mix and see if xfstests catch any problems there;
> will post tonight.

No regressions.