* [PATCH 1/3] f2fs: clean up codes in get_ssr_segment @ 2017-02-22 12:50 ` Yunlong Song 0 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-22 12:50 UTC (permalink / raw) To: jaegeuk, cm224.lee, yuchao0, chao, sylinux, yunlong.song, miaoxie Cc: bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Signed-off-by: Yunlong Song <yunlong.song@huawei.com> --- fs/f2fs/segment.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index df2ff5c..80997f5 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1541,7 +1541,7 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) struct curseg_info *curseg = CURSEG_I(sbi, type); const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; - if (IS_NODESEG(type) || !has_not_enough_free_secs(sbi, 0, 0)) + if (IS_NODESEG(type)) return v_ops->get_victim(sbi, &(curseg)->next_segno, BG_GC, type, SSR); -- 1.8.5.2 ^ permalink raw reply related [flat|nested] 28+ messages in thread
* [PATCH 1/3] f2fs: clean up codes in get_ssr_segment @ 2017-02-22 12:50 ` Yunlong Song 0 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-22 12:50 UTC (permalink / raw) To: jaegeuk, cm224.lee, yuchao0, chao, sylinux, yunlong.song, miaoxie Cc: bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Signed-off-by: Yunlong Song <yunlong.song@huawei.com> --- fs/f2fs/segment.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index df2ff5c..80997f5 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1541,7 +1541,7 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) struct curseg_info *curseg = CURSEG_I(sbi, type); const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; - if (IS_NODESEG(type) || !has_not_enough_free_secs(sbi, 0, 0)) + if (IS_NODESEG(type)) return v_ops->get_victim(sbi, &(curseg)->next_segno, BG_GC, type, SSR); -- 1.8.5.2 ^ permalink raw reply related [flat|nested] 28+ messages in thread
* [PATCH 2/3] f2fs: provide more chance to get node ssr 2017-02-22 12:50 ` Yunlong Song @ 2017-02-22 12:50 ` Yunlong Song -1 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-22 12:50 UTC (permalink / raw) To: jaegeuk, cm224.lee, yuchao0, chao, sylinux, yunlong.song, miaoxie Cc: bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Signed-off-by: Yunlong Song <yunlong.song@huawei.com> --- fs/f2fs/segment.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 80997f5..9d13743 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1541,9 +1541,13 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) struct curseg_info *curseg = CURSEG_I(sbi, type); const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; - if (IS_NODESEG(type)) - return v_ops->get_victim(sbi, - &(curseg)->next_segno, BG_GC, type, SSR); + if (IS_NODESEG(type)) { + for (; type >= CURSEG_HOT_NODE; type--) + if (v_ops->get_victim(sbi, &(curseg)->next_segno, + BG_GC, type, SSR)) + return 1; + return 0; + } /* For data segments, let's do SSR more intensively */ for (; type >= CURSEG_HOT_DATA; type--) -- 1.8.5.2 ^ permalink raw reply related [flat|nested] 28+ messages in thread
* [PATCH 2/3] f2fs: provide more chance to get node ssr @ 2017-02-22 12:50 ` Yunlong Song 0 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-22 12:50 UTC (permalink / raw) To: jaegeuk, cm224.lee, yuchao0, chao, sylinux, yunlong.song, miaoxie Cc: bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Signed-off-by: Yunlong Song <yunlong.song@huawei.com> --- fs/f2fs/segment.c | 10 +++++++--- 1 file changed, 7 insertions(+), 3 deletions(-) diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 80997f5..9d13743 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1541,9 +1541,13 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) struct curseg_info *curseg = CURSEG_I(sbi, type); const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; - if (IS_NODESEG(type)) - return v_ops->get_victim(sbi, - &(curseg)->next_segno, BG_GC, type, SSR); + if (IS_NODESEG(type)) { + for (; type >= CURSEG_HOT_NODE; type--) + if (v_ops->get_victim(sbi, &(curseg)->next_segno, + BG_GC, type, SSR)) + return 1; + return 0; + } /* For data segments, let's do SSR more intensively */ for (; type >= CURSEG_HOT_DATA; type--) -- 1.8.5.2 ^ permalink raw reply related [flat|nested] 28+ messages in thread
* [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment 2017-02-22 12:50 ` Yunlong Song @ 2017-02-22 12:50 ` Yunlong Song -1 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-22 12:50 UTC (permalink / raw) To: jaegeuk, cm224.lee, yuchao0, chao, sylinux, yunlong.song, miaoxie Cc: bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Signed-off-by: Yunlong Song <yunlong.song@huawei.com> --- fs/f2fs/segment.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 9d13743..5fe71b9 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) { struct curseg_info *curseg = CURSEG_I(sbi, type); const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; + int old_type = type; if (IS_NODESEG(type)) { for (; type >= CURSEG_HOT_NODE; type--) if (v_ops->get_victim(sbi, &(curseg)->next_segno, BG_GC, type, SSR)) return 1; + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) + if (v_ops->get_victim(sbi, &(curseg)->next_segno, + BG_GC, type, SSR)) + return 1; return 0; } @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) if (v_ops->get_victim(sbi, &(curseg)->next_segno, BG_GC, type, SSR)) return 1; + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) + if (v_ops->get_victim(sbi, &(curseg)->next_segno, + BG_GC, type, SSR)) + return 1; return 0; } -- 1.8.5.2 ^ permalink raw reply related [flat|nested] 28+ messages in thread
* [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment @ 2017-02-22 12:50 ` Yunlong Song 0 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-22 12:50 UTC (permalink / raw) To: jaegeuk, cm224.lee, yuchao0, chao, sylinux, yunlong.song, miaoxie Cc: bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Signed-off-by: Yunlong Song <yunlong.song@huawei.com> --- fs/f2fs/segment.c | 9 +++++++++ 1 file changed, 9 insertions(+) diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 9d13743..5fe71b9 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) { struct curseg_info *curseg = CURSEG_I(sbi, type); const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; + int old_type = type; if (IS_NODESEG(type)) { for (; type >= CURSEG_HOT_NODE; type--) if (v_ops->get_victim(sbi, &(curseg)->next_segno, BG_GC, type, SSR)) return 1; + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) + if (v_ops->get_victim(sbi, &(curseg)->next_segno, + BG_GC, type, SSR)) + return 1; return 0; } @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) if (v_ops->get_victim(sbi, &(curseg)->next_segno, BG_GC, type, SSR)) return 1; + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) + if (v_ops->get_victim(sbi, &(curseg)->next_segno, + BG_GC, type, SSR)) + return 1; return 0; } -- 1.8.5.2 ^ permalink raw reply related [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment 2017-02-22 12:50 ` Yunlong Song (?) @ 2017-02-23 1:17 ` Jaegeuk Kim 2017-02-23 2:13 ` Yunlong Song 2017-02-24 9:05 ` Chao Yu -1 siblings, 2 replies; 28+ messages in thread From: Jaegeuk Kim @ 2017-02-23 1:17 UTC (permalink / raw) To: Yunlong Song Cc: cm224.lee, yuchao0, chao, sylinux, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel [-- Attachment #1: Type: text/plain, Size: 1375 bytes --] Hi Yunlong, I've been testing the similar patches as I attached. Thanks, On 02/22, Yunlong Song wrote: > Signed-off-by: Yunlong Song <yunlong.song@huawei.com> > --- > fs/f2fs/segment.c | 9 +++++++++ > 1 file changed, 9 insertions(+) > > diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c > index 9d13743..5fe71b9 100644 > --- a/fs/f2fs/segment.c > +++ b/fs/f2fs/segment.c > @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) > { > struct curseg_info *curseg = CURSEG_I(sbi, type); > const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; > + int old_type = type; > > if (IS_NODESEG(type)) { > for (; type >= CURSEG_HOT_NODE; type--) > if (v_ops->get_victim(sbi, &(curseg)->next_segno, > BG_GC, type, SSR)) > return 1; > + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) > + if (v_ops->get_victim(sbi, &(curseg)->next_segno, > + BG_GC, type, SSR)) > + return 1; > return 0; > } > > @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) > if (v_ops->get_victim(sbi, &(curseg)->next_segno, > BG_GC, type, SSR)) > return 1; > + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) > + if (v_ops->get_victim(sbi, &(curseg)->next_segno, > + BG_GC, type, SSR)) > + return 1; > return 0; > } > > -- > 1.8.5.2 [-- Attachment #2: 0001-f2fs-find-data-segments-across-all-the-types.patch --] [-- Type: text/plain, Size: 1439 bytes --] >From 92b42757ca01d6c03c4af7bb6e169855605c59ae Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim <jaegeuk@kernel.org> Date: Wed, 22 Feb 2017 17:10:18 -0800 Subject: [PATCH 1/2] f2fs: find data segments across all the types Previously, if type is CURSEG_HOT_DATA, we only check CURSEG_HOT_DATA only. This patch fixes to search all the different types for SSR. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org> --- fs/f2fs/segment.c | 15 +++++++++++---- 1 file changed, 11 insertions(+), 4 deletions(-) diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index f3d80ccd556f..605cda3e17e3 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1529,16 +1529,23 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) { struct curseg_info *curseg = CURSEG_I(sbi, type); const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; + int i; + + /* need_SSR() already forces to do this */ + if (v_ops->get_victim(sbi, &(curseg)->next_segno, BG_GC, type, SSR)) + return 1; if (IS_NODESEG(type)) - return v_ops->get_victim(sbi, - &(curseg)->next_segno, BG_GC, type, SSR); + return 0; /* For data segments, let's do SSR more intensively */ - for (; type >= CURSEG_HOT_DATA; type--) + for (i = CURSEG_HOT_DATA; i <= CURSEG_COLD_DATA; i++) { + if (i == type) + continue; if (v_ops->get_victim(sbi, &(curseg)->next_segno, - BG_GC, type, SSR)) + BG_GC, i, SSR)) return 1; + } return 0; } -- 2.11.0 [-- Attachment #3: 0002-f2fs-do-SSR-for-node-segments-more-aggresively.patch --] [-- Type: text/plain, Size: 1382 bytes --] >From cf25da1cb6af3a516f96f6ea3edf98307b089397 Mon Sep 17 00:00:00 2001 From: Jaegeuk Kim <jaegeuk@kernel.org> Date: Wed, 22 Feb 2017 17:02:32 -0800 Subject: [PATCH 2/2] f2fs: do SSR for node segments more aggresively This patch gives more SSR chances for node blocks. Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org> --- fs/f2fs/segment.c | 15 ++++++++++----- 1 file changed, 10 insertions(+), 5 deletions(-) diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index 605cda3e17e3..9f819eb4749e 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1529,17 +1529,22 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) { struct curseg_info *curseg = CURSEG_I(sbi, type); const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; - int i; + int i, n; /* need_SSR() already forces to do this */ if (v_ops->get_victim(sbi, &(curseg)->next_segno, BG_GC, type, SSR)) return 1; - if (IS_NODESEG(type)) - return 0; + /* For node segments, let's do SSR more intensively */ + if (IS_NODESEG(type)) { + i = CURSEG_HOT_NODE; + n = CURSEG_COLD_NODE; + } else { + i = CURSEG_HOT_DATA; + n = CURSEG_COLD_DATA; + } - /* For data segments, let's do SSR more intensively */ - for (i = CURSEG_HOT_DATA; i <= CURSEG_COLD_DATA; i++) { + for (; i <= n; i++) { if (i == type) continue; if (v_ops->get_victim(sbi, &(curseg)->next_segno, -- 2.11.0 ^ permalink raw reply related [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment 2017-02-23 1:17 ` Jaegeuk Kim @ 2017-02-23 2:13 ` Yunlong Song 2017-02-24 9:05 ` Chao Yu 1 sibling, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-23 2:13 UTC (permalink / raw) To: Jaegeuk Kim Cc: cm224.lee, yuchao0, chao, sylinux, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Hi Jay, But there are some differences: 1 Why not take "neighboring temperature" for ssr? For example, if type == CURSEG_COLD_DATA, your new patch selects CURSEG_HOT_DATA first, how about select CURSEG_WARM_DATA first? The patch I sent ensure this "neighboring temperature" for ssr. 2 Your new patch returns 0 for node ssr, does get_ssr_segment work only for data? 3 I see your new patch just "continue" when i == type, why not consider the current type? On 2017/2/23 9:17, Jaegeuk Kim wrote: > Hi Yunlong, > > I've been testing the similar patches as I attached. > > Thanks, > > On 02/22, Yunlong Song wrote: >> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> >> --- >> fs/f2fs/segment.c | 9 +++++++++ >> 1 file changed, 9 insertions(+) >> >> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c >> index 9d13743..5fe71b9 100644 >> --- a/fs/f2fs/segment.c >> +++ b/fs/f2fs/segment.c >> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >> { >> struct curseg_info *curseg = CURSEG_I(sbi, type); >> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; >> + int old_type = type; >> >> if (IS_NODESEG(type)) { >> for (; type >= CURSEG_HOT_NODE; type--) >> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >> BG_GC, type, SSR)) >> return 1; >> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) >> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >> + BG_GC, type, SSR)) >> + return 1; >> return 0; >> } >> >> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >> BG_GC, type, SSR)) >> return 1; >> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) >> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >> + BG_GC, type, SSR)) >> + return 1; >> return 0; >> } >> >> -- >> 1.8.5.2 -- Thanks, Yunlong Song ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment @ 2017-02-23 2:13 ` Yunlong Song 0 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-23 2:13 UTC (permalink / raw) To: Jaegeuk Kim Cc: cm224.lee, yuchao0, chao, sylinux, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Hi Jay, But there are some differences: 1 Why not take "neighboring temperature" for ssr? For example, if type == CURSEG_COLD_DATA, your new patch selects CURSEG_HOT_DATA first, how about select CURSEG_WARM_DATA first? The patch I sent ensure this "neighboring temperature" for ssr. 2 Your new patch returns 0 for node ssr, does get_ssr_segment work only for data? 3 I see your new patch just "continue" when i == type, why not consider the current type? On 2017/2/23 9:17, Jaegeuk Kim wrote: > Hi Yunlong, > > I've been testing the similar patches as I attached. > > Thanks, > > On 02/22, Yunlong Song wrote: >> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> >> --- >> fs/f2fs/segment.c | 9 +++++++++ >> 1 file changed, 9 insertions(+) >> >> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c >> index 9d13743..5fe71b9 100644 >> --- a/fs/f2fs/segment.c >> +++ b/fs/f2fs/segment.c >> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >> { >> struct curseg_info *curseg = CURSEG_I(sbi, type); >> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; >> + int old_type = type; >> >> if (IS_NODESEG(type)) { >> for (; type >= CURSEG_HOT_NODE; type--) >> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >> BG_GC, type, SSR)) >> return 1; >> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) >> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >> + BG_GC, type, SSR)) >> + return 1; >> return 0; >> } >> >> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >> BG_GC, type, SSR)) >> return 1; >> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) >> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >> + BG_GC, type, SSR)) >> + return 1; >> return 0; >> } >> >> -- >> 1.8.5.2 -- Thanks, Yunlong Song ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment 2017-02-23 2:13 ` Yunlong Song @ 2017-02-23 2:41 ` Yunlong Song -1 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-23 2:41 UTC (permalink / raw) To: Jaegeuk Kim Cc: cm224.lee, yuchao0, chao, sylinux, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Hi Jay, I miss some codes from your mail, now I check all your patch codes and make clear with question 2 and 3, it's fine. But still, how about question 1? On 2017/2/23 10:13, Yunlong Song wrote: > Hi Jay, > But there are some differences: > 1 Why not take "neighboring temperature" for ssr? For example, if type == CURSEG_COLD_DATA, > your new patch selects CURSEG_HOT_DATA first, how about select CURSEG_WARM_DATA first? > The patch I sent ensure this "neighboring temperature" for ssr. > 2 Your new patch returns 0 for node ssr, does get_ssr_segment work only for data? > 3 I see your new patch just "continue" when i == type, why not consider the current type? > > On 2017/2/23 9:17, Jaegeuk Kim wrote: >> Hi Yunlong, >> >> I've been testing the similar patches as I attached. >> >> Thanks, >> >> On 02/22, Yunlong Song wrote: >>> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> >>> --- >>> fs/f2fs/segment.c | 9 +++++++++ >>> 1 file changed, 9 insertions(+) >>> >>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c >>> index 9d13743..5fe71b9 100644 >>> --- a/fs/f2fs/segment.c >>> +++ b/fs/f2fs/segment.c >>> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>> { >>> struct curseg_info *curseg = CURSEG_I(sbi, type); >>> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; >>> + int old_type = type; >>> >>> if (IS_NODESEG(type)) { >>> for (; type >= CURSEG_HOT_NODE; type--) >>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>> BG_GC, type, SSR)) >>> return 1; >>> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) >>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>> + BG_GC, type, SSR)) >>> + return 1; >>> return 0; >>> } >>> >>> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>> BG_GC, type, SSR)) >>> return 1; >>> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) >>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>> + BG_GC, type, SSR)) >>> + return 1; >>> return 0; >>> } >>> >>> -- >>> 1.8.5.2 > -- Thanks, Yunlong Song ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment @ 2017-02-23 2:41 ` Yunlong Song 0 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-23 2:41 UTC (permalink / raw) To: Jaegeuk Kim Cc: cm224.lee, yuchao0, chao, sylinux, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Hi Jay, I miss some codes from your mail, now I check all your patch codes and make clear with question 2 and 3, it's fine. But still, how about question 1? On 2017/2/23 10:13, Yunlong Song wrote: > Hi Jay, > But there are some differences: > 1 Why not take "neighboring temperature" for ssr? For example, if type == CURSEG_COLD_DATA, > your new patch selects CURSEG_HOT_DATA first, how about select CURSEG_WARM_DATA first? > The patch I sent ensure this "neighboring temperature" for ssr. > 2 Your new patch returns 0 for node ssr, does get_ssr_segment work only for data? > 3 I see your new patch just "continue" when i == type, why not consider the current type? > > On 2017/2/23 9:17, Jaegeuk Kim wrote: >> Hi Yunlong, >> >> I've been testing the similar patches as I attached. >> >> Thanks, >> >> On 02/22, Yunlong Song wrote: >>> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> >>> --- >>> fs/f2fs/segment.c | 9 +++++++++ >>> 1 file changed, 9 insertions(+) >>> >>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c >>> index 9d13743..5fe71b9 100644 >>> --- a/fs/f2fs/segment.c >>> +++ b/fs/f2fs/segment.c >>> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>> { >>> struct curseg_info *curseg = CURSEG_I(sbi, type); >>> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; >>> + int old_type = type; >>> >>> if (IS_NODESEG(type)) { >>> for (; type >= CURSEG_HOT_NODE; type--) >>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>> BG_GC, type, SSR)) >>> return 1; >>> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) >>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>> + BG_GC, type, SSR)) >>> + return 1; >>> return 0; >>> } >>> >>> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>> BG_GC, type, SSR)) >>> return 1; >>> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) >>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>> + BG_GC, type, SSR)) >>> + return 1; >>> return 0; >>> } >>> >>> -- >>> 1.8.5.2 > -- Thanks, Yunlong Song ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment 2017-02-23 1:17 ` Jaegeuk Kim @ 2017-02-24 9:05 ` Chao Yu 2017-02-24 9:05 ` Chao Yu 1 sibling, 0 replies; 28+ messages in thread From: Chao Yu @ 2017-02-24 9:05 UTC (permalink / raw) To: Jaegeuk Kim, Yunlong Song Cc: chao, sylinux, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Hi Jaegeuk, Reviewed-by: Chao Yu <yuchao0@huawei.com> For your attached two patches. Thanks, On 2017/2/23 9:17, Jaegeuk Kim wrote: > Hi Yunlong, > > I've been testing the similar patches as I attached. > > Thanks, > > On 02/22, Yunlong Song wrote: >> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> >> --- >> fs/f2fs/segment.c | 9 +++++++++ >> 1 file changed, 9 insertions(+) >> >> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c >> index 9d13743..5fe71b9 100644 >> --- a/fs/f2fs/segment.c >> +++ b/fs/f2fs/segment.c >> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >> { >> struct curseg_info *curseg = CURSEG_I(sbi, type); >> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; >> + int old_type = type; >> >> if (IS_NODESEG(type)) { >> for (; type >= CURSEG_HOT_NODE; type--) >> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >> BG_GC, type, SSR)) >> return 1; >> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) >> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >> + BG_GC, type, SSR)) >> + return 1; >> return 0; >> } >> >> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >> BG_GC, type, SSR)) >> return 1; >> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) >> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >> + BG_GC, type, SSR)) >> + return 1; >> return 0; >> } >> >> -- >> 1.8.5.2 ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment @ 2017-02-24 9:05 ` Chao Yu 0 siblings, 0 replies; 28+ messages in thread From: Chao Yu @ 2017-02-24 9:05 UTC (permalink / raw) To: Jaegeuk Kim, Yunlong Song Cc: chao, linux-kernel, linux-f2fs-devel, sylinux, linux-fsdevel Hi Jaegeuk, Reviewed-by: Chao Yu <yuchao0@huawei.com> For your attached two patches. Thanks, On 2017/2/23 9:17, Jaegeuk Kim wrote: > Hi Yunlong, > > I've been testing the similar patches as I attached. > > Thanks, > > On 02/22, Yunlong Song wrote: >> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> >> --- >> fs/f2fs/segment.c | 9 +++++++++ >> 1 file changed, 9 insertions(+) >> >> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c >> index 9d13743..5fe71b9 100644 >> --- a/fs/f2fs/segment.c >> +++ b/fs/f2fs/segment.c >> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >> { >> struct curseg_info *curseg = CURSEG_I(sbi, type); >> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; >> + int old_type = type; >> >> if (IS_NODESEG(type)) { >> for (; type >= CURSEG_HOT_NODE; type--) >> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >> BG_GC, type, SSR)) >> return 1; >> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) >> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >> + BG_GC, type, SSR)) >> + return 1; >> return 0; >> } >> >> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >> BG_GC, type, SSR)) >> return 1; >> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) >> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >> + BG_GC, type, SSR)) >> + return 1; >> return 0; >> } >> >> -- >> 1.8.5.2 ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment 2017-02-24 9:05 ` Chao Yu @ 2017-02-24 9:19 ` Yunlong Song -1 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-24 9:19 UTC (permalink / raw) To: Chao Yu, Jaegeuk Kim Cc: chao, sylinux, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Hi Jaegeuk and Chao, How about the question I pointed out in last mail: Why not take "neighboring temperature" for ssr? For example, if type == CURSEG_COLD_DATA, the new patch selects CURSEG_HOT_DATA first, why not select CURSEG_WARM_DATA first? The patch I sent ensure this "neighboring temperature" for ssr. This is to reduce the influence of mixing different levels of hot/code node types. On 2017/2/24 17:05, Chao Yu wrote: > Hi Jaegeuk, > > Reviewed-by: Chao Yu <yuchao0@huawei.com> > > For your attached two patches. > > Thanks, > > On 2017/2/23 9:17, Jaegeuk Kim wrote: >> Hi Yunlong, >> >> I've been testing the similar patches as I attached. >> >> Thanks, >> >> On 02/22, Yunlong Song wrote: >>> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> >>> --- >>> fs/f2fs/segment.c | 9 +++++++++ >>> 1 file changed, 9 insertions(+) >>> >>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c >>> index 9d13743..5fe71b9 100644 >>> --- a/fs/f2fs/segment.c >>> +++ b/fs/f2fs/segment.c >>> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>> { >>> struct curseg_info *curseg = CURSEG_I(sbi, type); >>> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; >>> + int old_type = type; >>> >>> if (IS_NODESEG(type)) { >>> for (; type >= CURSEG_HOT_NODE; type--) >>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>> BG_GC, type, SSR)) >>> return 1; >>> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) >>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>> + BG_GC, type, SSR)) >>> + return 1; >>> return 0; >>> } >>> >>> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>> BG_GC, type, SSR)) >>> return 1; >>> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) >>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>> + BG_GC, type, SSR)) >>> + return 1; >>> return 0; >>> } >>> >>> -- >>> 1.8.5.2 > > . > -- Thanks, Yunlong Song ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment @ 2017-02-24 9:19 ` Yunlong Song 0 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-24 9:19 UTC (permalink / raw) To: Chao Yu, Jaegeuk Kim Cc: chao, sylinux, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Hi Jaegeuk and Chao, How about the question I pointed out in last mail: Why not take "neighboring temperature" for ssr? For example, if type == CURSEG_COLD_DATA, the new patch selects CURSEG_HOT_DATA first, why not select CURSEG_WARM_DATA first? The patch I sent ensure this "neighboring temperature" for ssr. This is to reduce the influence of mixing different levels of hot/code node types. On 2017/2/24 17:05, Chao Yu wrote: > Hi Jaegeuk, > > Reviewed-by: Chao Yu <yuchao0@huawei.com> > > For your attached two patches. > > Thanks, > > On 2017/2/23 9:17, Jaegeuk Kim wrote: >> Hi Yunlong, >> >> I've been testing the similar patches as I attached. >> >> Thanks, >> >> On 02/22, Yunlong Song wrote: >>> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> >>> --- >>> fs/f2fs/segment.c | 9 +++++++++ >>> 1 file changed, 9 insertions(+) >>> >>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c >>> index 9d13743..5fe71b9 100644 >>> --- a/fs/f2fs/segment.c >>> +++ b/fs/f2fs/segment.c >>> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>> { >>> struct curseg_info *curseg = CURSEG_I(sbi, type); >>> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; >>> + int old_type = type; >>> >>> if (IS_NODESEG(type)) { >>> for (; type >= CURSEG_HOT_NODE; type--) >>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>> BG_GC, type, SSR)) >>> return 1; >>> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) >>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>> + BG_GC, type, SSR)) >>> + return 1; >>> return 0; >>> } >>> >>> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>> BG_GC, type, SSR)) >>> return 1; >>> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) >>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>> + BG_GC, type, SSR)) >>> + return 1; >>> return 0; >>> } >>> >>> -- >>> 1.8.5.2 > > . > -- Thanks, Yunlong Song ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment 2017-02-24 9:19 ` Yunlong Song @ 2017-02-24 10:47 ` Chao Yu -1 siblings, 0 replies; 28+ messages in thread From: Chao Yu @ 2017-02-24 10:47 UTC (permalink / raw) To: Yunlong Song, Jaegeuk Kim Cc: chao, sylinux, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel On 2017/2/24 17:19, Yunlong Song wrote: > Hi Jaegeuk and Chao, > > How about the question I pointed out in last mail: > Why not take "neighboring temperature" for ssr? For example, if type == CURSEG_COLD_DATA, > the new patch selects CURSEG_HOT_DATA first, why not select CURSEG_WARM_DATA first? > The patch I sent ensure this "neighboring temperature" for ssr. This is to reduce the influence of > mixing different levels of hot/code node types. Agreed, I sent one patch for changing the policy of SSR, how do you think of it? Thanks, > > On 2017/2/24 17:05, Chao Yu wrote: >> Hi Jaegeuk, >> >> Reviewed-by: Chao Yu <yuchao0@huawei.com> >> >> For your attached two patches. >> >> Thanks, >> >> On 2017/2/23 9:17, Jaegeuk Kim wrote: >>> Hi Yunlong, >>> >>> I've been testing the similar patches as I attached. >>> >>> Thanks, >>> >>> On 02/22, Yunlong Song wrote: >>>> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> >>>> --- >>>> fs/f2fs/segment.c | 9 +++++++++ >>>> 1 file changed, 9 insertions(+) >>>> >>>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c >>>> index 9d13743..5fe71b9 100644 >>>> --- a/fs/f2fs/segment.c >>>> +++ b/fs/f2fs/segment.c >>>> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>>> { >>>> struct curseg_info *curseg = CURSEG_I(sbi, type); >>>> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; >>>> + int old_type = type; >>>> >>>> if (IS_NODESEG(type)) { >>>> for (; type >= CURSEG_HOT_NODE; type--) >>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>> BG_GC, type, SSR)) >>>> return 1; >>>> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) >>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>> + BG_GC, type, SSR)) >>>> + return 1; >>>> return 0; >>>> } >>>> >>>> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>> BG_GC, type, SSR)) >>>> return 1; >>>> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) >>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>> + BG_GC, type, SSR)) >>>> + return 1; >>>> return 0; >>>> } >>>> >>>> -- >>>> 1.8.5.2 >> >> . >> > > ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment @ 2017-02-24 10:47 ` Chao Yu 0 siblings, 0 replies; 28+ messages in thread From: Chao Yu @ 2017-02-24 10:47 UTC (permalink / raw) To: Yunlong Song, Jaegeuk Kim Cc: chao, sylinux, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel On 2017/2/24 17:19, Yunlong Song wrote: > Hi Jaegeuk and Chao, > > How about the question I pointed out in last mail: > Why not take "neighboring temperature" for ssr? For example, if type == CURSEG_COLD_DATA, > the new patch selects CURSEG_HOT_DATA first, why not select CURSEG_WARM_DATA first? > The patch I sent ensure this "neighboring temperature" for ssr. This is to reduce the influence of > mixing different levels of hot/code node types. Agreed, I sent one patch for changing the policy of SSR, how do you think of it? Thanks, > > On 2017/2/24 17:05, Chao Yu wrote: >> Hi Jaegeuk, >> >> Reviewed-by: Chao Yu <yuchao0@huawei.com> >> >> For your attached two patches. >> >> Thanks, >> >> On 2017/2/23 9:17, Jaegeuk Kim wrote: >>> Hi Yunlong, >>> >>> I've been testing the similar patches as I attached. >>> >>> Thanks, >>> >>> On 02/22, Yunlong Song wrote: >>>> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> >>>> --- >>>> fs/f2fs/segment.c | 9 +++++++++ >>>> 1 file changed, 9 insertions(+) >>>> >>>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c >>>> index 9d13743..5fe71b9 100644 >>>> --- a/fs/f2fs/segment.c >>>> +++ b/fs/f2fs/segment.c >>>> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>>> { >>>> struct curseg_info *curseg = CURSEG_I(sbi, type); >>>> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; >>>> + int old_type = type; >>>> >>>> if (IS_NODESEG(type)) { >>>> for (; type >= CURSEG_HOT_NODE; type--) >>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>> BG_GC, type, SSR)) >>>> return 1; >>>> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) >>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>> + BG_GC, type, SSR)) >>>> + return 1; >>>> return 0; >>>> } >>>> >>>> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>> BG_GC, type, SSR)) >>>> return 1; >>>> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) >>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>> + BG_GC, type, SSR)) >>>> + return 1; >>>> return 0; >>>> } >>>> >>>> -- >>>> 1.8.5.2 >> >> . >> > > ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment 2017-02-24 10:47 ` Chao Yu @ 2017-02-24 11:41 ` Yunlong Song -1 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-24 11:41 UTC (permalink / raw) To: Chao Yu, Jaegeuk Kim Cc: chao, sylinux, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Hi, Chao, Not looks good to me, since there is some case your code does not include: if type is CURSEG_HOT_DATA, and if get_victim also returns 0 for both CURSEG_HOT_DATA and CURSEG_WARM_DATA, then i will be -1 and pass to get_victim in your code. So I still suggest my original patch attached below. On 2017/2/24 18:47, Chao Yu wrote: > On 2017/2/24 17:19, Yunlong Song wrote: >> Hi Jaegeuk and Chao, >> >> How about the question I pointed out in last mail: >> Why not take "neighboring temperature" for ssr? For example, if type == CURSEG_COLD_DATA, >> the new patch selects CURSEG_HOT_DATA first, why not select CURSEG_WARM_DATA first? >> The patch I sent ensure this "neighboring temperature" for ssr. This is to reduce the influence of >> mixing different levels of hot/code node types. > Agreed, I sent one patch for changing the policy of SSR, how do you think of it? > > Thanks, > >> On 2017/2/24 17:05, Chao Yu wrote: >>> Hi Jaegeuk, >>> >>> Reviewed-by: Chao Yu <yuchao0@huawei.com> >>> >>> For your attached two patches. >>> >>> Thanks, >>> >>> On 2017/2/23 9:17, Jaegeuk Kim wrote: >>>> Hi Yunlong, >>>> >>>> I've been testing the similar patches as I attached. >>>> >>>> Thanks, >>>> >>>> On 02/22, Yunlong Song wrote: >>>>> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> >>>>> --- >>>>> fs/f2fs/segment.c | 9 +++++++++ >>>>> 1 file changed, 9 insertions(+) >>>>> >>>>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c >>>>> index 9d13743..5fe71b9 100644 >>>>> --- a/fs/f2fs/segment.c >>>>> +++ b/fs/f2fs/segment.c >>>>> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>>>> { >>>>> struct curseg_info *curseg = CURSEG_I(sbi, type); >>>>> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; >>>>> + int old_type = type; >>>>> >>>>> if (IS_NODESEG(type)) { >>>>> for (; type >= CURSEG_HOT_NODE; type--) >>>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>>> BG_GC, type, SSR)) >>>>> return 1; >>>>> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) >>>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>>> + BG_GC, type, SSR)) >>>>> + return 1; >>>>> return 0; >>>>> } >>>>> >>>>> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>>> BG_GC, type, SSR)) >>>>> return 1; >>>>> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) >>>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>>> + BG_GC, type, SSR)) >>>>> + return 1; >>>>> return 0; >>>>> } >>>>> >>>>> -- >>>>> 1.8.5.2 >>> . >>> >> > > . > -- Thanks, Yunlong Song ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment @ 2017-02-24 11:41 ` Yunlong Song 0 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-24 11:41 UTC (permalink / raw) To: Chao Yu, Jaegeuk Kim Cc: chao, linux-kernel, linux-f2fs-devel, sylinux, linux-fsdevel Hi, Chao, Not looks good to me, since there is some case your code does not include: if type is CURSEG_HOT_DATA, and if get_victim also returns 0 for both CURSEG_HOT_DATA and CURSEG_WARM_DATA, then i will be -1 and pass to get_victim in your code. So I still suggest my original patch attached below. On 2017/2/24 18:47, Chao Yu wrote: > On 2017/2/24 17:19, Yunlong Song wrote: >> Hi Jaegeuk and Chao, >> >> How about the question I pointed out in last mail: >> Why not take "neighboring temperature" for ssr? For example, if type == CURSEG_COLD_DATA, >> the new patch selects CURSEG_HOT_DATA first, why not select CURSEG_WARM_DATA first? >> The patch I sent ensure this "neighboring temperature" for ssr. This is to reduce the influence of >> mixing different levels of hot/code node types. > Agreed, I sent one patch for changing the policy of SSR, how do you think of it? > > Thanks, > >> On 2017/2/24 17:05, Chao Yu wrote: >>> Hi Jaegeuk, >>> >>> Reviewed-by: Chao Yu <yuchao0@huawei.com> >>> >>> For your attached two patches. >>> >>> Thanks, >>> >>> On 2017/2/23 9:17, Jaegeuk Kim wrote: >>>> Hi Yunlong, >>>> >>>> I've been testing the similar patches as I attached. >>>> >>>> Thanks, >>>> >>>> On 02/22, Yunlong Song wrote: >>>>> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> >>>>> --- >>>>> fs/f2fs/segment.c | 9 +++++++++ >>>>> 1 file changed, 9 insertions(+) >>>>> >>>>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c >>>>> index 9d13743..5fe71b9 100644 >>>>> --- a/fs/f2fs/segment.c >>>>> +++ b/fs/f2fs/segment.c >>>>> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>>>> { >>>>> struct curseg_info *curseg = CURSEG_I(sbi, type); >>>>> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; >>>>> + int old_type = type; >>>>> >>>>> if (IS_NODESEG(type)) { >>>>> for (; type >= CURSEG_HOT_NODE; type--) >>>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>>> BG_GC, type, SSR)) >>>>> return 1; >>>>> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) >>>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>>> + BG_GC, type, SSR)) >>>>> + return 1; >>>>> return 0; >>>>> } >>>>> >>>>> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>>> BG_GC, type, SSR)) >>>>> return 1; >>>>> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) >>>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>>> + BG_GC, type, SSR)) >>>>> + return 1; >>>>> return 0; >>>>> } >>>>> >>>>> -- >>>>> 1.8.5.2 >>> . >>> >> > > . > -- Thanks, Yunlong Song ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment 2017-02-24 11:41 ` Yunlong Song (?) @ 2017-02-24 17:58 ` Jaegeuk Kim [not found] ` <b5c6b4d.d7.15a7177d92d.Coremail.sylinux@163.com> -1 siblings, 1 reply; 28+ messages in thread From: Jaegeuk Kim @ 2017-02-24 17:58 UTC (permalink / raw) To: Yunlong Song Cc: Chao Yu, chao, sylinux, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel On 02/24, Yunlong Song wrote: > Hi, Chao, > > Not looks good to me, since there is some case your code does not include: > if type is CURSEG_HOT_DATA, and if get_victim also returns 0 for both CURSEG_HOT_DATA and > CURSEG_WARM_DATA, then i will be -1 and pass to get_victim in your code. So I still suggest > my original patch attached below. Why does i become -1? > > On 2017/2/24 18:47, Chao Yu wrote: > > On 2017/2/24 17:19, Yunlong Song wrote: > >> Hi Jaegeuk and Chao, > >> > >> How about the question I pointed out in last mail: > >> Why not take "neighboring temperature" for ssr? For example, if type == CURSEG_COLD_DATA, > >> the new patch selects CURSEG_HOT_DATA first, why not select CURSEG_WARM_DATA first? > >> The patch I sent ensure this "neighboring temperature" for ssr. This is to reduce the influence of > >> mixing different levels of hot/code node types. > > Agreed, I sent one patch for changing the policy of SSR, how do you think of it? > > > > Thanks, > > > >> On 2017/2/24 17:05, Chao Yu wrote: > >>> Hi Jaegeuk, > >>> > >>> Reviewed-by: Chao Yu <yuchao0@huawei.com> > >>> > >>> For your attached two patches. > >>> > >>> Thanks, > >>> > >>> On 2017/2/23 9:17, Jaegeuk Kim wrote: > >>>> Hi Yunlong, > >>>> > >>>> I've been testing the similar patches as I attached. > >>>> > >>>> Thanks, > >>>> > >>>> On 02/22, Yunlong Song wrote: > >>>>> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> > >>>>> --- > >>>>> fs/f2fs/segment.c | 9 +++++++++ > >>>>> 1 file changed, 9 insertions(+) > >>>>> > >>>>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c > >>>>> index 9d13743..5fe71b9 100644 > >>>>> --- a/fs/f2fs/segment.c > >>>>> +++ b/fs/f2fs/segment.c > >>>>> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) > >>>>> { > >>>>> struct curseg_info *curseg = CURSEG_I(sbi, type); > >>>>> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; > >>>>> + int old_type = type; > >>>>> > >>>>> if (IS_NODESEG(type)) { > >>>>> for (; type >= CURSEG_HOT_NODE; type--) > >>>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, > >>>>> BG_GC, type, SSR)) > >>>>> return 1; > >>>>> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) > >>>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, > >>>>> + BG_GC, type, SSR)) > >>>>> + return 1; > >>>>> return 0; > >>>>> } > >>>>> > >>>>> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) > >>>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, > >>>>> BG_GC, type, SSR)) > >>>>> return 1; > >>>>> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) > >>>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, > >>>>> + BG_GC, type, SSR)) > >>>>> + return 1; > >>>>> return 0; > >>>>> } > >>>>> > >>>>> -- > >>>>> 1.8.5.2 > >>> . > >>> > >> > > > > . > > > > > -- > Thanks, > Yunlong Song > ^ permalink raw reply [flat|nested] 28+ messages in thread
[parent not found: <b5c6b4d.d7.15a7177d92d.Coremail.sylinux@163.com>]
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment [not found] ` <b5c6b4d.d7.15a7177d92d.Coremail.sylinux@163.com> @ 2017-02-24 20:05 ` Jaegeuk Kim 0 siblings, 0 replies; 28+ messages in thread From: Jaegeuk Kim @ 2017-02-24 20:05 UTC (permalink / raw) To: Yunlong Song Cc: Yunlong Song, Chao Yu, chao, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel On 02/25, Yunlong Song wrote: > Hi，Jaegeuk and Chao, > Sorry，I misunderstood CURSEG_HOT_DATA is greater than CURSEG_WARM_DATA and use "i--" in the codes, so just forget that. > But there is still a issue, when type is CURSEG_WARM_DATA, chao's patch will select cold SSR segment first, but my patch will select hot SSR segment first, since I think the probability of having hot SSR segment is bigger than having cold SSR segment due to the temperature itself. One thing that I'm seeing is, theoretically hot segments will get dirty more frequently. So if we select dirty cold segments for warm data, we can gain fully valid cold segments more, which can mitigate log thrashing problem. Thanks, > > > On 02/25/2017 01:58, Jaegeuk Kim wrote: > On 02/24, Yunlong Song wrote: > > Hi, Chao, > > > > Not looks good to me, since there is some case your code does not include: > > if type is CURSEG_HOT_DATA, and if get_victim also returns 0 for both CURSEG_HOT_DATA and > > CURSEG_WARM_DATA, then i will be -1 and pass to get_victim in your code. So I still suggest > > my original patch attached below. > > Why does i become -1? > > > > > On 2017/2/24 18:47, Chao Yu wrote: > > > On 2017/2/24 17:19, Yunlong Song wrote: > > >> Hi Jaegeuk and Chao, > > >> > > >> How about the question I pointed out in last mail: > > >> Why not take "neighboring temperature" for ssr? For example, if type == CURSEG_COLD_DATA, > > >> the new patch selects CURSEG_HOT_DATA first, why not select CURSEG_WARM_DATA first? > > >> The patch I sent ensure this "neighboring temperature" for ssr. This is to reduce the influence of > > >> mixing different levels of hot/code node types. > > > Agreed, I sent one patch for changing the policy of SSR, how do you think of it? > > > > > > Thanks, > > > > > >> On 2017/2/24 17:05, Chao Yu wrote: > > >>> Hi Jaegeuk, > > >>> > > >>> Reviewed-by: Chao Yu <yuchao0@huawei.com> > > >>> > > >>> For your attached two patches. > > >>> > > >>> Thanks, > > >>> > > >>> On 2017/2/23 9:17, Jaegeuk Kim wrote: > > >>>> Hi Yunlong, > > >>>> > > >>>> I've been testing the similar patches as I attached. > > >>>> > > >>>> Thanks, > > >>>> > > >>>> On 02/22, Yunlong Song wrote: > > >>>>> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> > > >>>>> --- > > >>>>> fs/f2fs/segment.c | 9 +++++++++ > > >>>>> 1 file changed, 9 insertions(+) > > >>>>> > > >>>>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c > > >>>>> index 9d13743..5fe71b9 100644 > > >>>>> --- a/fs/f2fs/segment.c > > >>>>> +++ b/fs/f2fs/segment.c > > >>>>> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) > > >>>>> { > > >>>>> struct curseg_info *curseg = CURSEG_I(sbi, type); > > >>>>> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; > > >>>>> + int old_type = type; > > >>>>> > > >>>>> if (IS_NODESEG(type)) { > > >>>>> for (; type >= CURSEG_HOT_NODE; type--) > > >>>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, > > >>>>> BG_GC, type, SSR)) > > >>>>> return 1; > > >>>>> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) > > >>>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, > > >>>>> + BG_GC, type, SSR)) > > >>>>> + return 1; > > >>>>> return 0; > > >>>>> } > > >>>>> > > >>>>> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) > > >>>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, > > >>>>> BG_GC, type, SSR)) > > >>>>> return 1; > > >>>>> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) > > >>>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, > > >>>>> + BG_GC, type, SSR)) > > >>>>> + return 1; > > >>>>> return 0; > > >>>>> } > > >>>>> > > >>>>> -- > > >>>>> 1.8.5.2 > > >>> . > > >>> > > >> > > > > > > . > > > > > > > > > -- > > Thanks, > > Yunlong Song > > ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment @ 2017-02-24 20:05 ` Jaegeuk Kim 0 siblings, 0 replies; 28+ messages in thread From: Jaegeuk Kim @ 2017-02-24 20:05 UTC (permalink / raw) To: Yunlong Song; +Cc: linux-kernel, linux-f2fs-devel, linux-fsdevel, chao On 02/25, Yunlong Song wrote: > Hi，Jaegeuk and Chao, > Sorry，I misunderstood CURSEG_HOT_DATA is greater than CURSEG_WARM_DATA and use "i--" in the codes, so just forget that. > But there is still a issue, when type is CURSEG_WARM_DATA, chao's patch will select cold SSR segment first, but my patch will select hot SSR segment first, since I think the probability of having hot SSR segment is bigger than having cold SSR segment due to the temperature itself. One thing that I'm seeing is, theoretically hot segments will get dirty more frequently. So if we select dirty cold segments for warm data, we can gain fully valid cold segments more, which can mitigate log thrashing problem. Thanks, > > > On 02/25/2017 01:58, Jaegeuk Kim wrote: > On 02/24, Yunlong Song wrote: > > Hi, Chao, > > > > Not looks good to me, since there is some case your code does not include: > > if type is CURSEG_HOT_DATA, and if get_victim also returns 0 for both CURSEG_HOT_DATA and > > CURSEG_WARM_DATA, then i will be -1 and pass to get_victim in your code. So I still suggest > > my original patch attached below. > > Why does i become -1? > > > > > On 2017/2/24 18:47, Chao Yu wrote: > > > On 2017/2/24 17:19, Yunlong Song wrote: > > >> Hi Jaegeuk and Chao, > > >> > > >> How about the question I pointed out in last mail: > > >> Why not take "neighboring temperature" for ssr? For example, if type == CURSEG_COLD_DATA, > > >> the new patch selects CURSEG_HOT_DATA first, why not select CURSEG_WARM_DATA first? > > >> The patch I sent ensure this "neighboring temperature" for ssr. This is to reduce the influence of > > >> mixing different levels of hot/code node types. > > > Agreed, I sent one patch for changing the policy of SSR, how do you think of it? > > > > > > Thanks, > > > > > >> On 2017/2/24 17:05, Chao Yu wrote: > > >>> Hi Jaegeuk, > > >>> > > >>> Reviewed-by: Chao Yu <yuchao0@huawei.com> > > >>> > > >>> For your attached two patches. > > >>> > > >>> Thanks, > > >>> > > >>> On 2017/2/23 9:17, Jaegeuk Kim wrote: > > >>>> Hi Yunlong, > > >>>> > > >>>> I've been testing the similar patches as I attached. > > >>>> > > >>>> Thanks, > > >>>> > > >>>> On 02/22, Yunlong Song wrote: > > >>>>> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> > > >>>>> --- > > >>>>> fs/f2fs/segment.c | 9 +++++++++ > > >>>>> 1 file changed, 9 insertions(+) > > >>>>> > > >>>>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c > > >>>>> index 9d13743..5fe71b9 100644 > > >>>>> --- a/fs/f2fs/segment.c > > >>>>> +++ b/fs/f2fs/segment.c > > >>>>> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) > > >>>>> { > > >>>>> struct curseg_info *curseg = CURSEG_I(sbi, type); > > >>>>> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; > > >>>>> + int old_type = type; > > >>>>> > > >>>>> if (IS_NODESEG(type)) { > > >>>>> for (; type >= CURSEG_HOT_NODE; type--) > > >>>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, > > >>>>> BG_GC, type, SSR)) > > >>>>> return 1; > > >>>>> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) > > >>>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, > > >>>>> + BG_GC, type, SSR)) > > >>>>> + return 1; > > >>>>> return 0; > > >>>>> } > > >>>>> > > >>>>> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) > > >>>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, > > >>>>> BG_GC, type, SSR)) > > >>>>> return 1; > > >>>>> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) > > >>>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, > > >>>>> + BG_GC, type, SSR)) > > >>>>> + return 1; > > >>>>> return 0; > > >>>>> } > > >>>>> > > >>>>> -- > > >>>>> 1.8.5.2 > > >>> . > > >>> > > >> > > > > > > . > > > > > > > > > -- > > Thanks, > > Yunlong Song > > ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot _______________________________________________ Linux-f2fs-devel mailing list Linux-f2fs-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment 2017-02-24 20:05 ` Jaegeuk Kim @ 2017-02-25 2:16 ` Yunlong Song -1 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-25 2:16 UTC (permalink / raw) To: Jaegeuk Kim, Yunlong Song Cc: Chao Yu, chao, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel The get_ssr_segment happens when need_SSR is true, that is to say we must make sure we can achieve a right available SSR segment at least, so why not select the type with larger probability? And I think warm and hot are more similar with each other, maybe the update frequency is difference, but if time goes by, warm segments will finally update their blocks like hot segments, but cold segments are something like which should and will never update, so I prefer to put hot and warm to one side, and put cold to the other side. On 2017/2/25 4:05, Jaegeuk Kim wrote: > On 02/25, Yunlong Song wrote: >> Hi，Jaegeuk and Chao, >> Sorry，I misunderstood CURSEG_HOT_DATA is greater than CURSEG_WARM_DATA and use "i--" in the codes, so just forget that. >> But there is still a issue, when type is CURSEG_WARM_DATA, chao's patch will select cold SSR segment first, but my patch will select hot SSR segment first, since I think the probability of having hot SSR segment is bigger than having cold SSR segment due to the temperature itself. > One thing that I'm seeing is, theoretically hot segments will get dirty more > frequently. So if we select dirty cold segments for warm data, we can gain > fully valid cold segments more, which can mitigate log thrashing problem. > > Thanks, > >> >> On 02/25/2017 01:58, Jaegeuk Kim wrote: >> On 02/24, Yunlong Song wrote: >>> Hi, Chao, >>> >>> Not looks good to me, since there is some case your code does not include: >>> if type is CURSEG_HOT_DATA, and if get_victim also returns 0 for both CURSEG_HOT_DATA and >>> CURSEG_WARM_DATA, then i will be -1 and pass to get_victim in your code. So I still suggest >>> my original patch attached below. >> Why does i become -1? >> >>> On 2017/2/24 18:47, Chao Yu wrote: >>>> On 2017/2/24 17:19, Yunlong Song wrote: >>>>> Hi Jaegeuk and Chao, >>>>> >>>>> How about the question I pointed out in last mail: >>>>> Why not take "neighboring temperature" for ssr? For example, if type == CURSEG_COLD_DATA, >>>>> the new patch selects CURSEG_HOT_DATA first, why not select CURSEG_WARM_DATA first? >>>>> The patch I sent ensure this "neighboring temperature" for ssr. This is to reduce the influence of >>>>> mixing different levels of hot/code node types. >>>> Agreed, I sent one patch for changing the policy of SSR, how do you think of it? >>>> >>>> Thanks, >>>> >>>>> On 2017/2/24 17:05, Chao Yu wrote: >>>>>> Hi Jaegeuk, >>>>>> >>>>>> Reviewed-by: Chao Yu <yuchao0@huawei.com> >>>>>> >>>>>> For your attached two patches. >>>>>> >>>>>> Thanks, >>>>>> >>>>>> On 2017/2/23 9:17, Jaegeuk Kim wrote: >>>>>>> Hi Yunlong, >>>>>>> >>>>>>> I've been testing the similar patches as I attached. >>>>>>> >>>>>>> Thanks, >>>>>>> >>>>>>> On 02/22, Yunlong Song wrote: >>>>>>>> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> >>>>>>>> --- >>>>>>>> fs/f2fs/segment.c | 9 +++++++++ >>>>>>>> 1 file changed, 9 insertions(+) >>>>>>>> >>>>>>>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c >>>>>>>> index 9d13743..5fe71b9 100644 >>>>>>>> --- a/fs/f2fs/segment.c >>>>>>>> +++ b/fs/f2fs/segment.c >>>>>>>> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>>>>>>> { >>>>>>>> struct curseg_info *curseg = CURSEG_I(sbi, type); >>>>>>>> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; >>>>>>>> + int old_type = type; >>>>>>>> >>>>>>>> if (IS_NODESEG(type)) { >>>>>>>> for (; type >= CURSEG_HOT_NODE; type--) >>>>>>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>>>>>> BG_GC, type, SSR)) >>>>>>>> return 1; >>>>>>>> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) >>>>>>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>>>>>> + BG_GC, type, SSR)) >>>>>>>> + return 1; >>>>>>>> return 0; >>>>>>>> } >>>>>>>> >>>>>>>> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>>>>>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>>>>>> BG_GC, type, SSR)) >>>>>>>> return 1; >>>>>>>> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) >>>>>>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>>>>>> + BG_GC, type, SSR)) >>>>>>>> + return 1; >>>>>>>> return 0; >>>>>>>> } >>>>>>>> >>>>>>>> -- >>>>>>>> 1.8.5.2 >>>>>> . >>>>>> >>>> . >>>> >>> >>> -- >>> Thanks, >>> Yunlong Song >>> > . > -- Thanks, Yunlong Song ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment @ 2017-02-25 2:16 ` Yunlong Song 0 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-25 2:16 UTC (permalink / raw) To: Jaegeuk Kim, Yunlong Song Cc: linux-kernel, linux-f2fs-devel, linux-fsdevel, chao The get_ssr_segment happens when need_SSR is true, that is to say we must make sure we can achieve a right available SSR segment at least, so why not select the type with larger probability? And I think warm and hot are more similar with each other, maybe the update frequency is difference, but if time goes by, warm segments will finally update their blocks like hot segments, but cold segments are something like which should and will never update, so I prefer to put hot and warm to one side, and put cold to the other side. On 2017/2/25 4:05, Jaegeuk Kim wrote: > On 02/25, Yunlong Song wrote: >> Hi，Jaegeuk and Chao, >> Sorry，I misunderstood CURSEG_HOT_DATA is greater than CURSEG_WARM_DATA and use "i--" in the codes, so just forget that. >> But there is still a issue, when type is CURSEG_WARM_DATA, chao's patch will select cold SSR segment first, but my patch will select hot SSR segment first, since I think the probability of having hot SSR segment is bigger than having cold SSR segment due to the temperature itself. > One thing that I'm seeing is, theoretically hot segments will get dirty more > frequently. So if we select dirty cold segments for warm data, we can gain > fully valid cold segments more, which can mitigate log thrashing problem. > > Thanks, > >> >> On 02/25/2017 01:58, Jaegeuk Kim wrote: >> On 02/24, Yunlong Song wrote: >>> Hi, Chao, >>> >>> Not looks good to me, since there is some case your code does not include: >>> if type is CURSEG_HOT_DATA, and if get_victim also returns 0 for both CURSEG_HOT_DATA and >>> CURSEG_WARM_DATA, then i will be -1 and pass to get_victim in your code. So I still suggest >>> my original patch attached below. >> Why does i become -1? >> >>> On 2017/2/24 18:47, Chao Yu wrote: >>>> On 2017/2/24 17:19, Yunlong Song wrote: >>>>> Hi Jaegeuk and Chao, >>>>> >>>>> How about the question I pointed out in last mail: >>>>> Why not take "neighboring temperature" for ssr? For example, if type == CURSEG_COLD_DATA, >>>>> the new patch selects CURSEG_HOT_DATA first, why not select CURSEG_WARM_DATA first? >>>>> The patch I sent ensure this "neighboring temperature" for ssr. This is to reduce the influence of >>>>> mixing different levels of hot/code node types. >>>> Agreed, I sent one patch for changing the policy of SSR, how do you think of it? >>>> >>>> Thanks, >>>> >>>>> On 2017/2/24 17:05, Chao Yu wrote: >>>>>> Hi Jaegeuk, >>>>>> >>>>>> Reviewed-by: Chao Yu <yuchao0@huawei.com> >>>>>> >>>>>> For your attached two patches. >>>>>> >>>>>> Thanks, >>>>>> >>>>>> On 2017/2/23 9:17, Jaegeuk Kim wrote: >>>>>>> Hi Yunlong, >>>>>>> >>>>>>> I've been testing the similar patches as I attached. >>>>>>> >>>>>>> Thanks, >>>>>>> >>>>>>> On 02/22, Yunlong Song wrote: >>>>>>>> Signed-off-by: Yunlong Song <yunlong.song@huawei.com> >>>>>>>> --- >>>>>>>> fs/f2fs/segment.c | 9 +++++++++ >>>>>>>> 1 file changed, 9 insertions(+) >>>>>>>> >>>>>>>> diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c >>>>>>>> index 9d13743..5fe71b9 100644 >>>>>>>> --- a/fs/f2fs/segment.c >>>>>>>> +++ b/fs/f2fs/segment.c >>>>>>>> @@ -1540,12 +1540,17 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>>>>>>> { >>>>>>>> struct curseg_info *curseg = CURSEG_I(sbi, type); >>>>>>>> const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; >>>>>>>> + int old_type = type; >>>>>>>> >>>>>>>> if (IS_NODESEG(type)) { >>>>>>>> for (; type >= CURSEG_HOT_NODE; type--) >>>>>>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>>>>>> BG_GC, type, SSR)) >>>>>>>> return 1; >>>>>>>> + for (type = old_type + 1; type <= CURSEG_COLD_NODE; type++) >>>>>>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>>>>>> + BG_GC, type, SSR)) >>>>>>>> + return 1; >>>>>>>> return 0; >>>>>>>> } >>>>>>>> >>>>>>>> @@ -1554,6 +1559,10 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) >>>>>>>> if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>>>>>> BG_GC, type, SSR)) >>>>>>>> return 1; >>>>>>>> + for (type = old_type + 1; type <= CURSEG_COLD_DATA; type++) >>>>>>>> + if (v_ops->get_victim(sbi, &(curseg)->next_segno, >>>>>>>> + BG_GC, type, SSR)) >>>>>>>> + return 1; >>>>>>>> return 0; >>>>>>>> } >>>>>>>> >>>>>>>> -- >>>>>>>> 1.8.5.2 >>>>>> . >>>>>> >>>> . >>>> >>> >>> -- >>> Thanks, >>> Yunlong Song >>> > . > -- Thanks, Yunlong Song ------------------------------------------------------------------------------ Check out the vibrant tech community on one of the world's most engaging tech sites, SlashDot.org! http://sdm.link/slashdot _______________________________________________ Linux-f2fs-devel mailing list Linux-f2fs-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 1/3] f2fs: clean up codes in get_ssr_segment 2017-02-22 12:50 ` Yunlong Song ` (2 preceding siblings ...) (?) @ 2017-02-23 0:41 ` Jaegeuk Kim 2017-02-23 0:59 ` Jaegeuk Kim -1 siblings, 1 reply; 28+ messages in thread From: Jaegeuk Kim @ 2017-02-23 0:41 UTC (permalink / raw) To: Yunlong Song Cc: cm224.lee, yuchao0, chao, sylinux, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Hi Yunlong, This is NOT clean-up. What's your point here? Thanks, On 02/22, Yunlong Song wrote: > Signed-off-by: Yunlong Song <yunlong.song@huawei.com> > --- > fs/f2fs/segment.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c > index df2ff5c..80997f5 100644 > --- a/fs/f2fs/segment.c > +++ b/fs/f2fs/segment.c > @@ -1541,7 +1541,7 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) > struct curseg_info *curseg = CURSEG_I(sbi, type); > const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; > > - if (IS_NODESEG(type) || !has_not_enough_free_secs(sbi, 0, 0)) > + if (IS_NODESEG(type)) > return v_ops->get_victim(sbi, > &(curseg)->next_segno, BG_GC, type, SSR); > > -- > 1.8.5.2 ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 1/3] f2fs: clean up codes in get_ssr_segment 2017-02-23 0:41 ` [PATCH 1/3] f2fs: clean up codes in get_ssr_segment Jaegeuk Kim @ 2017-02-23 0:59 ` Jaegeuk Kim 2017-02-23 1:55 ` Yunlong Song 0 siblings, 1 reply; 28+ messages in thread From: Jaegeuk Kim @ 2017-02-23 0:59 UTC (permalink / raw) To: Yunlong Song Cc: cm224.lee, yuchao0, chao, sylinux, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel On 02/22, Jaegeuk Kim wrote: > Hi Yunlong, > > This is NOT clean-up. What's your point here? I guess you wanted to do like this. >From 8d344ac45890dd95c5734fd29a19a7c19364c327 Mon Sep 17 00:00:00 2001 From: Yunlong Song <yunlong.song@huawei.com> Date: Wed, 22 Feb 2017 20:50:49 +0800 Subject: [PATCH] f2fs: do SSR for data when there is enough free space In allocate_segment_by_default(), need_SSR() already detected it's time to do SSR. So, let's try to find victims for data segments more aggressively in time. Signed-off-by: Yunlong Song <yunlong.song@huawei.com> Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org> --- fs/f2fs/segment.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c index d01ee7b94702..c27dab43c42f 100644 --- a/fs/f2fs/segment.c +++ b/fs/f2fs/segment.c @@ -1541,7 +1541,7 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) struct curseg_info *curseg = CURSEG_I(sbi, type); const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; - if (IS_NODESEG(type) || !has_not_enough_free_secs(sbi, 0, 0)) + if (IS_NODESEG(type)) return v_ops->get_victim(sbi, &(curseg)->next_segno, BG_GC, type, SSR); -- 2.11.0 ^ permalink raw reply related [flat|nested] 28+ messages in thread
* Re: [PATCH 1/3] f2fs: clean up codes in get_ssr_segment 2017-02-23 0:59 ` Jaegeuk Kim @ 2017-02-23 1:55 ` Yunlong Song 0 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-23 1:55 UTC (permalink / raw) To: Jaegeuk Kim Cc: cm224.lee, yuchao0, chao, sylinux, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Yes, exactly, I just mean it for short with word "clean up", thanks for correction. On 2017/2/23 8:59, Jaegeuk Kim wrote: > On 02/22, Jaegeuk Kim wrote: >> Hi Yunlong, >> >> This is NOT clean-up. What's your point here? > I guess you wanted to do like this. > > >From 8d344ac45890dd95c5734fd29a19a7c19364c327 Mon Sep 17 00:00:00 2001 > From: Yunlong Song <yunlong.song@huawei.com> > Date: Wed, 22 Feb 2017 20:50:49 +0800 > Subject: [PATCH] f2fs: do SSR for data when there is enough free space > > In allocate_segment_by_default(), need_SSR() already detected it's time to do > SSR. So, let's try to find victims for data segments more aggressively in time. > > Signed-off-by: Yunlong Song <yunlong.song@huawei.com> > Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org> > --- > fs/f2fs/segment.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c > index d01ee7b94702..c27dab43c42f 100644 > --- a/fs/f2fs/segment.c > +++ b/fs/f2fs/segment.c > @@ -1541,7 +1541,7 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) > struct curseg_info *curseg = CURSEG_I(sbi, type); > const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; > > - if (IS_NODESEG(type) || !has_not_enough_free_secs(sbi, 0, 0)) > + if (IS_NODESEG(type)) > return v_ops->get_victim(sbi, > &(curseg)->next_segno, BG_GC, type, SSR); > -- Thanks, Yunlong Song ^ permalink raw reply [flat|nested] 28+ messages in thread
* Re: [PATCH 1/3] f2fs: clean up codes in get_ssr_segment @ 2017-02-23 1:55 ` Yunlong Song 0 siblings, 0 replies; 28+ messages in thread From: Yunlong Song @ 2017-02-23 1:55 UTC (permalink / raw) To: Jaegeuk Kim Cc: cm224.lee, yuchao0, chao, sylinux, miaoxie, bintian.wang, linux-fsdevel, linux-f2fs-devel, linux-kernel Yes, exactly, I just mean it for short with word "clean up", thanks for correction. On 2017/2/23 8:59, Jaegeuk Kim wrote: > On 02/22, Jaegeuk Kim wrote: >> Hi Yunlong, >> >> This is NOT clean-up. What's your point here? > I guess you wanted to do like this. > > >From 8d344ac45890dd95c5734fd29a19a7c19364c327 Mon Sep 17 00:00:00 2001 > From: Yunlong Song <yunlong.song@huawei.com> > Date: Wed, 22 Feb 2017 20:50:49 +0800 > Subject: [PATCH] f2fs: do SSR for data when there is enough free space > > In allocate_segment_by_default(), need_SSR() already detected it's time to do > SSR. So, let's try to find victims for data segments more aggressively in time. > > Signed-off-by: Yunlong Song <yunlong.song@huawei.com> > Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org> > --- > fs/f2fs/segment.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c > index d01ee7b94702..c27dab43c42f 100644 > --- a/fs/f2fs/segment.c > +++ b/fs/f2fs/segment.c > @@ -1541,7 +1541,7 @@ static int get_ssr_segment(struct f2fs_sb_info *sbi, int type) > struct curseg_info *curseg = CURSEG_I(sbi, type); > const struct victim_selection *v_ops = DIRTY_I(sbi)->v_ops; > > - if (IS_NODESEG(type) || !has_not_enough_free_secs(sbi, 0, 0)) > + if (IS_NODESEG(type)) > return v_ops->get_victim(sbi, > &(curseg)->next_segno, BG_GC, type, SSR); > -- Thanks, Yunlong Song ^ permalink raw reply [flat|nested] 28+ messages in thread
end of thread, other threads:[~2017-02-25 2:17 UTC | newest]
Thread overview: 28+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-02-22 12:50 [PATCH 1/3] f2fs: clean up codes in get_ssr_segment Yunlong Song
2017-02-22 12:50 ` Yunlong Song
2017-02-22 12:50 ` [PATCH 2/3] f2fs: provide more chance to get node ssr Yunlong Song
2017-02-22 12:50 ` Yunlong Song
2017-02-22 12:50 ` [PATCH 3/3] f2fs: provide more chance for node and data to get ssr segment Yunlong Song
2017-02-22 12:50 ` Yunlong Song
2017-02-23 1:17 ` Jaegeuk Kim
2017-02-23 2:13 ` Yunlong Song
2017-02-23 2:13 ` Yunlong Song
2017-02-23 2:41 ` Yunlong Song
2017-02-23 2:41 ` Yunlong Song
2017-02-24 9:05 ` Chao Yu
2017-02-24 9:05 ` Chao Yu
2017-02-24 9:19 ` Yunlong Song
2017-02-24 9:19 ` Yunlong Song
2017-02-24 10:47 ` Chao Yu
2017-02-24 10:47 ` Chao Yu
2017-02-24 11:41 ` Yunlong Song
2017-02-24 11:41 ` Yunlong Song
2017-02-24 17:58 ` Jaegeuk Kim
[not found] ` <b5c6b4d.d7.15a7177d92d.Coremail.sylinux@163.com>
2017-02-24 20:05 ` Jaegeuk Kim
2017-02-24 20:05 ` Jaegeuk Kim
2017-02-25 2:16 ` Yunlong Song
2017-02-25 2:16 ` Yunlong Song
2017-02-23 0:41 ` [PATCH 1/3] f2fs: clean up codes in get_ssr_segment Jaegeuk Kim
2017-02-23 0:59 ` Jaegeuk Kim
2017-02-23 1:55 ` Yunlong Song
2017-02-23 1:55 ` Yunlong Song
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.