| Message ID | 20190115211028.15742-1-hch@lst.de (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | iomap: get/put the page in iomap_page_create/release() | expand |
On Tue, Jan 15, 2019 at 10:10:28PM +0100, Christoph Hellwig wrote: > migrate_page_move_mapping() expects pages with private data set to have > a page_count elevated by 1. This is what used to happen for xfs through > the buffer_heads code before the switch to iomap in commit 82cb14175e7d > ("xfs: add support for sub-pagesize writeback without buffer_heads"). > Not having the count elevated causes move_pages() to fail on memory > mapped files coming from xfs. > > Make iomap compatible with the migrate_page_move_mapping() assumption by > elevating the page count as part of iomap_page_create() and lowering it > in iomap_page_release(). > > It causes the move_pages() syscall to misbehave on memory mapped files > from xfs. It does not not move any pages, which I suppose is "just" a > perf issue, but it also ends up returning a positive number which is out > of spec for the syscall. Talking to Michal Hocko, it sounds like > returning positive numbers might be a necessary update to move_pages() > anyway though. > > Based on an earlier patch and changelog from Piotr Jaroszynski, except > that this version actually puts/gets the page aswell when migrating > pages using a copy, and thus actually survives xfstests. > > Fixes: 82cb14175e7d ("xfs: add support for sub-pagesize writeback without buffer_heads") > Cc: Piotr Jaroszynski <pjaroszynski@nvidia.com> Missing a SOB. But apart from that, I've been running an identical patch in my test tree, so: Reviewed-by: Dave Chinner <dchinner@redhat.com>
On Wed, Jan 16, 2019 at 10:07:45AM +1100, Dave Chinner wrote: > Missing a SOB. > > But apart from that, I've been running an identical patch in my test > tree, so: Piotr: do you still want to claim authorship for this fixed up patch, in which case I'll need your sob. Otherwise I can submit it under my name.
On 1/16/19 11:51 PM, Christoph Hellwig wrote: > On Wed, Jan 16, 2019 at 10:07:45AM +1100, Dave Chinner wrote: >> Missing a SOB. >> >> But apart from that, I've been running an identical patch in my test >> tree, so: > > Piotr: do you still want to claim authorship for this fixed up patch, > in which case I'll need your sob. Otherwise I can submit it under my > name. Sorry for the delay and thank you for handling the revert while I was out on paternity leave. I'm fine with whatever is easiest for you, and thanks for fixing up the patch! Thanks, Piotr
On Thu, Jan 17, 2019 at 09:32:32AM -0800, Piotr Jaroszynski wrote: > On 1/16/19 11:51 PM, Christoph Hellwig wrote: > > On Wed, Jan 16, 2019 at 10:07:45AM +1100, Dave Chinner wrote: > >> Missing a SOB. > >> > >> But apart from that, I've been running an identical patch in my test > >> tree, so: > > > > Piotr: do you still want to claim authorship for this fixed up patch, > > in which case I'll need your sob. Otherwise I can submit it under my > > name. > > Sorry for the delay and thank you for handling the revert while I was > out on paternity leave. I'm fine with whatever is easiest for you, > and thanks for fixing up the patch! In that case I'll happily keep you as the author. Can you re-cofirm your signoff for this modified variant so we pick it up?
On 1/19/19 1:45 AM, Christoph Hellwig wrote: > On Thu, Jan 17, 2019 at 09:32:32AM -0800, Piotr Jaroszynski wrote: >> On 1/16/19 11:51 PM, Christoph Hellwig wrote: >>> On Wed, Jan 16, 2019 at 10:07:45AM +1100, Dave Chinner wrote: >>>> Missing a SOB. >>>> >>>> But apart from that, I've been running an identical patch in my test >>>> tree, so: >>> >>> Piotr: do you still want to claim authorship for this fixed up patch, >>> in which case I'll need your sob. Otherwise I can submit it under my >>> name. >> >> Sorry for the delay and thank you for handling the revert while I was >> out on paternity leave. I'm fine with whatever is easiest for you, >> and thanks for fixing up the patch! > > In that case I'll happily keep you as the author. Can you re-cofirm > your signoff for this modified variant so we pick it up? > Sure: Signed-off-by: Piotr Jaroszynski <pjaroszynski@nvidia.com> Thanks, Piotr
diff --git a/fs/iomap.c b/fs/iomap.c index a3088fae567b..cb184ff68680 100644 --- a/fs/iomap.c +++ b/fs/iomap.c @@ -116,6 +116,12 @@ iomap_page_create(struct inode *inode, struct page *page) atomic_set(&iop->read_count, 0); atomic_set(&iop->write_count, 0); bitmap_zero(iop->uptodate, PAGE_SIZE / SECTOR_SIZE); + + /* + * migrate_page_move_mapping() assumes that pages with private data have + * their count elevated by 1. + */ + get_page(page); set_page_private(page, (unsigned long)iop); SetPagePrivate(page); return iop; @@ -132,6 +138,7 @@ iomap_page_release(struct page *page) WARN_ON_ONCE(atomic_read(&iop->write_count)); ClearPagePrivate(page); set_page_private(page, 0); + put_page(page); kfree(iop); } @@ -569,8 +576,10 @@ iomap_migrate_page(struct address_space *mapping, struct page *newpage, if (page_has_private(page)) { ClearPagePrivate(page); + get_page(newpage); set_page_private(newpage, page_private(page)); set_page_private(page, 0); + put_page(page); SetPagePrivate(newpage); }
migrate_page_move_mapping() expects pages with private data set to have a page_count elevated by 1. This is what used to happen for xfs through the buffer_heads code before the switch to iomap in commit 82cb14175e7d ("xfs: add support for sub-pagesize writeback without buffer_heads"). Not having the count elevated causes move_pages() to fail on memory mapped files coming from xfs. Make iomap compatible with the migrate_page_move_mapping() assumption by elevating the page count as part of iomap_page_create() and lowering it in iomap_page_release(). It causes the move_pages() syscall to misbehave on memory mapped files from xfs. It does not not move any pages, which I suppose is "just" a perf issue, but it also ends up returning a positive number which is out of spec for the syscall. Talking to Michal Hocko, it sounds like returning positive numbers might be a necessary update to move_pages() anyway though. Based on an earlier patch and changelog from Piotr Jaroszynski, except that this version actually puts/gets the page aswell when migrating pages using a copy, and thus actually survives xfstests. Fixes: 82cb14175e7d ("xfs: add support for sub-pagesize writeback without buffer_heads") Cc: Piotr Jaroszynski <pjaroszynski@nvidia.com> --- fs/iomap.c | 9 +++++++++ 1 file changed, 9 insertions(+)