Messages in this thread | | | Date | Wed, 16 Nov 2016 19:41:38 -0800 | From | Jaegeuk Kim <> | Subject | Re: [PATCH] f2fs: fix fdatasync |
| |
On Thu, Nov 17, 2016 at 11:35:58AM +0800, Chao Yu wrote: > On 2016/11/17 10:59, Jaegeuk Kim wrote: > > On Thu, Nov 17, 2016 at 10:51:37AM +0800, Chao Yu wrote: > >> Hi Jaegeuk, > >> > >> On 2016/11/17 3:13, Jaegeuk Kim wrote: > >>> Hi Chao, > >>> > >>> On Wed, Nov 16, 2016 at 08:12:11PM +0800, Chao Yu wrote: > >>>> For below two cases, we can't guarantee data consistence: > >>>> > >>>> a) > >>>> 1. xfs_io "pwrite 0 4195328" "fsync" > >>>> 2. xfs_io "pwrite 4195328 1024" "fdatasync" > >>>> 3. godown > >>>> 4. umount & mount > >>>> --> isize we updated before fdatasync won't be recovered > >>>> > >>>> b) > >>>> 1. xfs_io "pwrite -S 0xcc 0 4202496" "fsync" > >>>> 2. xfs_io "fpunch 4194304 4096" "fdatasync" > >>>> 3. godown > >>>> 4. umount & mount > >>>> --> dnode we punched before fdatasync won't be recovered > >>>> > >>>> The reason is that normally fdatasync won't be aware of modification > >>>> of metadata in file, e.g. isize changing, dnode updating, so in ->fsync > >>>> we will skip flushing node pages for above cases, result in making > >>>> fdatasynced file being lost during recovery. > >>>> > >>>> Introduce FDATASYNC_INO global ino cache for tracking node changing, > >>>> later fdatasync choose to flush nodes depend on ino cache state. > >>> > >>> We don't need to add this additionally, and would be better to consider other > >>> major metadata as well. > >>> > >>> How about this? > >> > >> Seems it can't track file after evict? > > > > Do we need that? That means inode page was already up-to-date? > > I mean if node/data page of inode were writeback by kworker, after evict, if > user open it again and call fdatasync, after sudden power-off, we will not > recover it.
That will be handled by need_inode_block_update() below?
Thanks,
| |