On 02/02, Chao Yu wrote: > On 2019-1-29 7:47, Jaegeuk Kim wrote: > > After quota_off, we'll get some dirty blocks. If put_super don't have a > > chance > > to flush them by checkpoint, it causes NULL pointer exception in end_io > > after > > iput(node_inode). (e.g., by checkpoint=disable) > > > > Signed-off-by: Jaegeuk Kim <[email protected]> > > --- > > fs/f2fs/super.c | 6 ++++++ > > 1 file changed, 6 insertions(+) > > > > diff --git a/fs/f2fs/super.c b/fs/f2fs/super.c > > index 5e1f8573a17f..7694cb350734 100644 > > --- a/fs/f2fs/super.c > > +++ b/fs/f2fs/super.c > > @@ -2026,6 +2026,12 @@ void f2fs_quota_off_umount(struct super_block *sb) > > set_sbi_flag(F2FS_SB(sb), SBI_QUOTA_NEED_REPAIR); > > } > > } > > + /* > > + * In case of checkpoint=disable, we must flush quota blocks. > > checkpoint=disable is one of the cases, right? e.g. IO error can be another > case? > > So here I guess we need to change the comments a bit for that.
I didn't test other cases, but checkpoint=disable could give real errors through some tests. So, basically, I'd like to keep known cases only here. > > Thanks, > > > + * This can cause NULL exception for node_inode in end_io, since > > + * put_super already dropped it. > > + */ > > + sync_filesystem(sb); > > } > > > > static void f2fs_truncate_quota_inode_pages(struct super_block *sb) > > _______________________________________________ Linux-f2fs-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
