| Message ID | 20170411001308.12881-4-jaegeuk@kernel.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 2017/4/11 8:13, Jaegeuk Kim wrote: > Now we're doing SSR aggressively more than ever before, so once we reach to > the reserved_segment, f2fs_balance_fs will call f2fs_gc, which triggers > checkpoint everytime. We actually must avoid that. > > Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org> > --- > fs/f2fs/gc.c | 6 +++++- > 1 file changed, 5 insertions(+), 1 deletion(-) > > diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c > index e2f9b2b12b74..4a720f3394d9 100644 > --- a/fs/f2fs/gc.c > +++ b/fs/f2fs/gc.c > @@ -966,7 +966,11 @@ int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background) > * threshold, we can make them free by checkpoint. Then, we > * secure free segments which doesn't need fggc any more. > */ > - ret = write_checkpoint(sbi, &cpc); > + if (prefree_segments(sbi)) { > + ret = write_checkpoint(sbi, &cpc); > + if (ret) > + goto stop; > + } > if (ret) > goto stop; We don't need redundant check here? Otherwise, this patchset looks good to me, anyway please add: Reviewed-by: Chao Yu <yuchao0@huawei.com> Thanks, > if (has_not_enough_free_secs(sbi, 0, 0)) >
On 04/11, Chao Yu wrote: > On 2017/4/11 8:13, Jaegeuk Kim wrote: > > Now we're doing SSR aggressively more than ever before, so once we reach to > > the reserved_segment, f2fs_balance_fs will call f2fs_gc, which triggers > > checkpoint everytime. We actually must avoid that. > > > > Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org> > > --- > > fs/f2fs/gc.c | 6 +++++- > > 1 file changed, 5 insertions(+), 1 deletion(-) > > > > diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c > > index e2f9b2b12b74..4a720f3394d9 100644 > > --- a/fs/f2fs/gc.c > > +++ b/fs/f2fs/gc.c > > @@ -966,7 +966,11 @@ int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background) > > * threshold, we can make them free by checkpoint. Then, we > > * secure free segments which doesn't need fggc any more. > > */ > > - ret = write_checkpoint(sbi, &cpc); > > + if (prefree_segments(sbi)) { > > + ret = write_checkpoint(sbi, &cpc); > > + if (ret) > > + goto stop; > > + } > > if (ret) > > goto stop; > > We don't need redundant check here? Yup, will remove this. ;) > > Otherwise, this patchset looks good to me, anyway please add: > > Reviewed-by: Chao Yu <yuchao0@huawei.com> Thanks, > > Thanks, > > > if (has_not_enough_free_secs(sbi, 0, 0)) > >
diff --git a/fs/f2fs/gc.c b/fs/f2fs/gc.c index e2f9b2b12b74..4a720f3394d9 100644 --- a/fs/f2fs/gc.c +++ b/fs/f2fs/gc.c @@ -966,7 +966,11 @@ int f2fs_gc(struct f2fs_sb_info *sbi, bool sync, bool background) * threshold, we can make them free by checkpoint. Then, we * secure free segments which doesn't need fggc any more. */ - ret = write_checkpoint(sbi, &cpc); + if (prefree_segments(sbi)) { + ret = write_checkpoint(sbi, &cpc); + if (ret) + goto stop; + } if (ret) goto stop; if (has_not_enough_free_secs(sbi, 0, 0))
Now we're doing SSR aggressively more than ever before, so once we reach to the reserved_segment, f2fs_balance_fs will call f2fs_gc, which triggers checkpoint everytime. We actually must avoid that. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org> --- fs/f2fs/gc.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-)