| Message ID | 1711346195-24555-1-git-send-email-zhiguo.niu@unisoc.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | [f2fs-dev,V2] f2fs: fix to adjust appropirate defragment pg_end | expand |
On 2024/3/25 13:56, Zhiguo Niu wrote: > A length that exceeds the real size of the inode may be > specified from user, although these out-of-range areas > are not mapped, but they still need to be check in > while loop, which is unnecessary. > > Signed-off-by: Zhiguo Niu <zhiguo.niu@unisoc.com> > --- > v2: check i_size within inode lock according to Chao's suggestions > --- > --- > fs/f2fs/file.c | 11 +++++++---- > 1 file changed, 7 insertions(+), 4 deletions(-) > > diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c > index 128e53d..cf63db7 100644 > --- a/fs/f2fs/file.c > +++ b/fs/f2fs/file.c > @@ -2608,9 +2608,6 @@ static int f2fs_defragment_range(struct f2fs_sb_info *sbi, > bool fragmented = false; > int err; > > - pg_start = range->start >> PAGE_SHIFT; > - pg_end = (range->start + range->len) >> PAGE_SHIFT; > - > f2fs_balance_fs(sbi, true); > > inode_lock(inode); > @@ -2629,10 +2626,16 @@ static int f2fs_defragment_range(struct f2fs_sb_info *sbi, pg_start = range->start >> PAGE_SHIFT; pg_end = min_t(pgoff_t, (range->start + range->len) >> PAGE_SHIFT, DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE)); > > /* writeback all dirty pages in the range */ > err = filemap_write_and_wait_range(inode->i_mapping, range->start, > - range->start + range->len - 1); > + min_t(loff_t, range->start + range->len - 1, > + i_size_read(inode) - 1)); , pg_start << PAGE_SHIFT - 1, pg_end << PAGE_SHIFT - 1); ? Thanks, > if (err) > goto out; > > + pg_start = range->start >> PAGE_SHIFT; > + pg_end = min_t(pgoff_t, > + (range->start + range->len) >> PAGE_SHIFT, > + DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE)); > + > /* > * lookup mapping info in extent cache, skip defragmenting if physical > * block addresses are continuous.
On Tue, Mar 26, 2024 at 7:11 PM Chao Yu <chao@kernel.org> wrote: > > On 2024/3/25 13:56, Zhiguo Niu wrote: > > A length that exceeds the real size of the inode may be > > specified from user, although these out-of-range areas > > are not mapped, but they still need to be check in > > while loop, which is unnecessary. > > > > Signed-off-by: Zhiguo Niu <zhiguo.niu@unisoc.com> > > --- > > v2: check i_size within inode lock according to Chao's suggestions > > --- > > --- > > fs/f2fs/file.c | 11 +++++++---- > > 1 file changed, 7 insertions(+), 4 deletions(-) > > > > diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c > > index 128e53d..cf63db7 100644 > > --- a/fs/f2fs/file.c > > +++ b/fs/f2fs/file.c > > @@ -2608,9 +2608,6 @@ static int f2fs_defragment_range(struct f2fs_sb_info *sbi, > > bool fragmented = false; > > int err; > > > > - pg_start = range->start >> PAGE_SHIFT; > > - pg_end = (range->start + range->len) >> PAGE_SHIFT; > > - > > f2fs_balance_fs(sbi, true); > > > > inode_lock(inode); > > @@ -2629,10 +2626,16 @@ static int f2fs_defragment_range(struct f2fs_sb_info *sbi, > > pg_start = range->start >> PAGE_SHIFT; > pg_end = min_t(pgoff_t, (range->start + range->len) >> PAGE_SHIFT, > DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE)); > > > > > /* writeback all dirty pages in the range */ > > err = filemap_write_and_wait_range(inode->i_mapping, range->start, > > - range->start + range->len - 1); > > + min_t(loff_t, range->start + range->len - 1, > > + i_size_read(inode) - 1)); > > , pg_start << PAGE_SHIFT - 1, pg_end << PAGE_SHIFT - 1); ? should be pg_start << PAGE_SHIFT , pg_end << PAGE_SHIFT - 1)?? if range.start=0, pg_start is also 0, lstart in filemap_write_and_wait_range is 0, but pg_start << PAGE_SHIFT - 1 will get lstart=-1? thanks！ > > Thanks, > > > if (err) > > goto out; > > > > + pg_start = range->start >> PAGE_SHIFT; > > + pg_end = min_t(pgoff_t, > > + (range->start + range->len) >> PAGE_SHIFT, > > + DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE)); > > + > > /* > > * lookup mapping info in extent cache, skip defragmenting if physical > > * block addresses are continuous.
On 2024/3/27 9:26, Zhiguo Niu wrote: > On Tue, Mar 26, 2024 at 7:11 PM Chao Yu <chao@kernel.org> wrote: >> >> On 2024/3/25 13:56, Zhiguo Niu wrote: >>> A length that exceeds the real size of the inode may be >>> specified from user, although these out-of-range areas >>> are not mapped, but they still need to be check in >>> while loop, which is unnecessary. >>> >>> Signed-off-by: Zhiguo Niu <zhiguo.niu@unisoc.com> >>> --- >>> v2: check i_size within inode lock according to Chao's suggestions >>> --- >>> --- >>> fs/f2fs/file.c | 11 +++++++---- >>> 1 file changed, 7 insertions(+), 4 deletions(-) >>> >>> diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c >>> index 128e53d..cf63db7 100644 >>> --- a/fs/f2fs/file.c >>> +++ b/fs/f2fs/file.c >>> @@ -2608,9 +2608,6 @@ static int f2fs_defragment_range(struct f2fs_sb_info *sbi, >>> bool fragmented = false; >>> int err; >>> >>> - pg_start = range->start >> PAGE_SHIFT; >>> - pg_end = (range->start + range->len) >> PAGE_SHIFT; >>> - >>> f2fs_balance_fs(sbi, true); >>> >>> inode_lock(inode); >>> @@ -2629,10 +2626,16 @@ static int f2fs_defragment_range(struct f2fs_sb_info *sbi, >> >> pg_start = range->start >> PAGE_SHIFT; >> pg_end = min_t(pgoff_t, (range->start + range->len) >> PAGE_SHIFT, >> DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE)); >> >>> >>> /* writeback all dirty pages in the range */ >>> err = filemap_write_and_wait_range(inode->i_mapping, range->start, >>> - range->start + range->len - 1); >>> + min_t(loff_t, range->start + range->len - 1, >>> + i_size_read(inode) - 1)); >> >> , pg_start << PAGE_SHIFT - 1, pg_end << PAGE_SHIFT - 1); ? > should be pg_start << PAGE_SHIFT , pg_end << PAGE_SHIFT - 1)?? Oh, yes. :) Thanks, > if range.start=0, pg_start is also 0, lstart in > filemap_write_and_wait_range is 0, > but pg_start << PAGE_SHIFT - 1 will get lstart=-1? > thanks！ >> >> Thanks, >> >>> if (err) >>> goto out; >>> >>> + pg_start = range->start >> PAGE_SHIFT; >>> + pg_end = min_t(pgoff_t, >>> + (range->start + range->len) >> PAGE_SHIFT, >>> + DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE)); >>> + >>> /* >>> * lookup mapping info in extent cache, skip defragmenting if physical >>> * block addresses are continuous.
diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c index 128e53d..cf63db7 100644 --- a/fs/f2fs/file.c +++ b/fs/f2fs/file.c @@ -2608,9 +2608,6 @@ static int f2fs_defragment_range(struct f2fs_sb_info *sbi, bool fragmented = false; int err; - pg_start = range->start >> PAGE_SHIFT; - pg_end = (range->start + range->len) >> PAGE_SHIFT; - f2fs_balance_fs(sbi, true); inode_lock(inode); @@ -2629,10 +2626,16 @@ static int f2fs_defragment_range(struct f2fs_sb_info *sbi, /* writeback all dirty pages in the range */ err = filemap_write_and_wait_range(inode->i_mapping, range->start, - range->start + range->len - 1); + min_t(loff_t, range->start + range->len - 1, + i_size_read(inode) - 1)); if (err) goto out; + pg_start = range->start >> PAGE_SHIFT; + pg_end = min_t(pgoff_t, + (range->start + range->len) >> PAGE_SHIFT, + DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE)); + /* * lookup mapping info in extent cache, skip defragmenting if physical * block addresses are continuous.
A length that exceeds the real size of the inode may be specified from user, although these out-of-range areas are not mapped, but they still need to be check in while loop, which is unnecessary. Signed-off-by: Zhiguo Niu <zhiguo.niu@unisoc.com> --- v2: check i_size within inode lock according to Chao's suggestions --- --- fs/f2fs/file.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)