[PATCH 0/2] Remove PG_error

Matthew Wilcox (Oracle) posted 2 patches 1 year, 6 months ago
Documentation/filesystems/vfs.rst | 3 +--
include/linux/page-flags.h        | 6 +-----
mm/filemap.c                      | 8 --------
mm/migrate.c                      | 2 --
mm/page_io.c                      | 4 +---
5 files changed, 3 insertions(+), 20 deletions(-)
[PATCH 0/2] Remove PG_error
Posted by Matthew Wilcox (Oracle) 1 year, 6 months ago
Hi Linus,

There were four pulls I had to wait for before I could send this to you,
but PG_error is now unused and can be reclaimed.  It could wait until
next merge window, but I don't see the point.

The patches to do all this have been out for review since mid-April,
https://lore.kernel.org/linux-fsdevel/20240420025029.2166544-1-willy@infradead.org/
Current git with these patches builds with allmodconfig.

Matthew Wilcox (Oracle) (2):
  fs: Remove calls to set and clear the folio error flag
  mm: Remove PG_error

 Documentation/filesystems/vfs.rst | 3 +--
 include/linux/page-flags.h        | 6 +-----
 mm/filemap.c                      | 8 --------
 mm/migrate.c                      | 2 --
 mm/page_io.c                      | 4 +---
 5 files changed, 3 insertions(+), 20 deletions(-)

-- 
2.43.0
Re: [PATCH 0/2] Remove PG_error
Posted by Linus Torvalds 1 year, 6 months ago
On Tue, 23 Jul 2024 at 18:05, Matthew Wilcox (Oracle)
<willy@infradead.org> wrote:
>
> There were four pulls I had to wait for before I could send this to you,
> but PG_error is now unused and can be reclaimed.  It could wait until
> next merge window, but I don't see the point.

Well, I was initially inclined to agree with that, but then I did a
test-compile, and that immediately notices that you never tested this
at all and that it's completely broken.

There's still uses of PG_error in fs/proc/page.c, and in
include/trace/events/mmflags.h, and neither of those two cases are
new. One is unchanged from 2009, and the other one has gone through a
few code movements and re-orgs but also goes back over a decade ago.

So clearly this needs to be in linux-next, and as such this very much
is "next merge window".

                 Linus