RE: [f2fs-dev] [PATCH] f2fs: modify the procedure of scan free nid

From: Fan Li
Date: Wed Nov 01 2017 - 22:39:54 EST




> -----Original Message-----
> From: Chao Yu [mailto:chao@xxxxxxxxxx]
> Sent: Wednesday, November 01, 2017 8:47 PM
> To: Fan Li; 'Jaegeuk Kim'
> Cc: linux-kernel@xxxxxxxxxxxxxxx; linux-f2fs-devel@xxxxxxxxxxxxxxxxxxxxx
> Subject: Re: [f2fs-dev] [PATCH] f2fs: modify the procedure of scan free nid
>
> On 2017/11/1 18:03, Fan Li wrote:
> >
> >
> >> -----Original Message-----
> >> From: Chao Yu [mailto:chao@xxxxxxxxxx]
> >> Sent: Tuesday, October 31, 2017 10:32 PM
> >> To: Fan Li; 'Jaegeuk Kim'
> >> Cc: linux-kernel@xxxxxxxxxxxxxxx;
> >> linux-f2fs-devel@xxxxxxxxxxxxxxxxxxxxx
> >> Subject: Re: [f2fs-dev] [PATCH] f2fs: modify the procedure of scan
> >> free nid
> >>
> >> On 2017/10/31 21:37, Fan Li wrote:
> >>> In current version, we preserve 8 pages of nat blocks as free nids,
> >>> build bitmaps for it and use them to allocate nids until its number
> >>> drops below NAT_ENTRY_PER_BLOCK.
> >>>
> >>> After that, we have a problem, scan_free_nid_bits will scan the same
> >>> 8 pages trying to find more free nids, but in most cases the free
> >>> nids in these bitmaps are already in free list, scan them won't get
> >>> us any new nids.
> >>> Further more, after scan_free_nid_bits, the search is over if
> >>> nid_cnt[FREE_NID] != 0.
> >>> It causes that we scan the same pages over and over again, yet no
> >>> new free nids are found until nid_cnt[FREE_NID]==0.
> >>>
> >>> This patch mark the range where new free nids could exist and keep
> >>> scan for free nids until nid_cnt[FREE_NID] >= NAT_ENTRY_PER_BLOCK.
> >>> The new vairable first_scan_block marks the start of the range, it's
> >>> initialized with NEW_ADDR, which means all free nids before
> >>> next_scan_nid are already in free list; and use next_scan_nid as the
> >>> end of the range since all free nids which are scanned must be
> >>> smaller next_scan_nid.
> >>>
> >>>
> >>> Signed-off-by: Fan li <fanofcode.li@xxxxxxxxxxx>
> >>> ---
> >>> fs/f2fs/f2fs.h | 1 +
> >>> fs/f2fs/node.c | 30 ++++++++++++++++++++++++++----
> >>> 2 files changed, 27 insertions(+), 4 deletions(-)
> >>>
> >>> diff --git a/fs/f2fs/f2fs.h b/fs/f2fs/f2fs.h index e0ef31c..ae1cf91
> >>> 100644
> >>> --- a/fs/f2fs/f2fs.h
> >>> +++ b/fs/f2fs/f2fs.h
> >>> @@ -705,6 +705,7 @@ struct f2fs_nm_info {
> >>> nid_t max_nid; /* maximum possible node ids */
> >>> nid_t available_nids; /* # of available node ids */
> >>> nid_t next_scan_nid; /* the next nid to be scanned */
> >>> + block_t first_scan_block; /* the first NAT block to be scanned */
> >>
> >> As we are traveling bitmap, so how about using smaller granularity for tracking last-scanned-position. like:
> >>
> >> unsigned next_bitmap_pos; ?
> >>
> > Yes, I think it's a good idea, but original code scans nids by blocks,
> > if I change that, I need to change some other details too, and before that, I want to make sure this idea of patch is right.
> > I also have some ideas about it, if that's OK, I tend to submit other patches to implement them.
> >
> >>> unsigned int ram_thresh; /* control the memory footprint */
> >>> unsigned int ra_nid_pages; /* # of nid pages to be readaheaded */
> >>> unsigned int dirty_nats_ratio; /* control dirty nats ratio threshold */
> >>> diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c index 3d0d1be..7834097
> >>> 100644
> >>> --- a/fs/f2fs/node.c
> >>> +++ b/fs/f2fs/node.c
> >>> @@ -1950,10 +1950,23 @@ static void scan_free_nid_bits(struct f2fs_sb_info *sbi)
> >>> struct curseg_info *curseg = CURSEG_I(sbi, CURSEG_HOT_DATA);
> >>> struct f2fs_journal *journal = curseg->journal;
> >>> unsigned int i, idx;
> >>> + unsigned int max_blocks = NAT_BLOCK_OFFSET(nm_i->next_scan_nid);
> >>>
> >>> - down_read(&nm_i->nat_tree_lock);
> >>> + /* every free nid in blocks scanned previously is in the free list */
> >>> + if (nm_i->first_scan_block == NEW_ADDR)
> >>
> >> How about using nm_i->max_nid as no more free nids in bitmap?
> >>
> > For now, I use the block as the unit of variable first_scan_block, for
> > the same reason above, I tend to change it in another patch.
> >
> >>> + return;
> >>>
> >>> - for (i = 0; i < nm_i->nat_blocks; i++) {
> >>> + /*
> >>> + * TODO: "next_scan_nid == 0" means after searching every nat block,
> >>> + * we still can't find enough free nids, there may not be any
> >>> + * more nid left to be found, we should stop at somewhere
> >>> + * instead of going through these all over again.
> >>> + */
>
> How about trying avoid todo thing in our patch, if our new feature is not so complicate or big.
>
Sure, I will delete this.

> >>> + if (max_blocks == 0)
> >>> + max_blocks = nm_i->nat_blocks;
> >>> +
> >>> + down_read(&nm_i->nat_tree_lock);
> >>> + for (i = nm_i->first_scan_block; i < max_blocks; i++) {
> >>
> >> Free nids could be set free after nodes were truncated & checkpoint,
> >> if we start from first_scan_block, we will miss some free
> > nids.
> >>
> > This is the part I'm not sure. To my understanding, after nodes were
> > truncated, the nats will be cached as dirty nats, the IS_CHECKPOINTED
> > flag will be removed from them, as a result, in original code these nats will not be added to free list in scan, so I also
didn't add these nats
> in this patch, but I don't know why it's designed this way in the first place.
> > Please tell me what's wrong about my understanding or why it's like this.
> > And what do you mean by the free nid which could be set free after checkpoint?
>
> You can check the code in __flush_nat_entries:
>
> if (nat_get_blkaddr(ne) == NULL_ADDR) {
> add_free_nid(sbi, nid, false);
> spin_lock(&NM_I(sbi)->nid_list_lock);
> NM_I(sbi)->available_nids++;
> update_free_nid_bitmap(sbi, nid, true, false);
> spin_unlock(&NM_I(sbi)->nid_list_lock);
> }
>
> I mean that we will try to:
> 1. add_free_nid
> 2. update_free_nid_bitmap
>
> But, you know, there is no guarantee that add_free_nid will success, so nid is been set free just in bitmap, if we do not update
> first_scan_block here, we may lose chance to scan that bitmap, right?
>
> Thanks,
>
Now I see it, thanks.
To be clear, those dirty NULL nats without IS_CHECKPOINTED flag weren't added to the free list in the old codes
and still don't need to be added in this patch, right?
I only need to add those nats which couldn't be added due to system failure, like out of memory or
errors of the insertion to radix tree?
I'm away for quite a while, there are some new development in f2fs I'm still catching up, if there's anything
else in this patch that doesn't fit in, please let me know, thanks.

> >
> >> Thanks,
> >>
> >>> if (!test_bit_le(i, nm_i->nat_block_bitmap))
> >>> continue;
> >>> if (!nm_i->free_nid_count[i])
> >>> @@ -1967,10 +1980,13 @@ static void scan_free_nid_bits(struct f2fs_sb_info *sbi)
> >>> nid = i * NAT_ENTRY_PER_BLOCK + idx;
> >>> add_free_nid(sbi, nid, true);
> >>>
> >>> - if (nm_i->nid_cnt[FREE_NID] >= MAX_FREE_NIDS)
> >>> + if (nm_i->nid_cnt[FREE_NID] >= MAX_FREE_NIDS) {
> >>> + nm_i->first_scan_block = i;
> >>> goto out;
> >>> + }
> >>> }
> >>> }
> >>> + nm_i->first_scan_block = NEW_ADDR;
> >>> out:
> >>> down_read(&curseg->journal_rwsem);
> >>> for (i = 0; i < nats_in_cursum(journal); i++) { @@ -2010,7 +2026,7
> >>> @@ static void __build_free_nids(struct f2fs_sb_info *sbi, bool sync, bool mount)
> >>> /* try to find free nids in free_nid_bitmap */
> >>> scan_free_nid_bits(sbi);
> >>>
> >>> - if (nm_i->nid_cnt[FREE_NID])
> >>> + if (nm_i->nid_cnt[FREE_NID] >= NAT_ENTRY_PER_BLOCK)
> >>> return;
> >>> }
> >>>
> >>> @@ -2163,6 +2179,7 @@ int try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink)
> >>> struct f2fs_nm_info *nm_i = NM_I(sbi);
> >>> struct free_nid *i, *next;
> >>> int nr = nr_shrink;
> >>> + nid_t min_nid = nm_i->max_nid;
> >>>
> >>> if (nm_i->nid_cnt[FREE_NID] <= MAX_FREE_NIDS)
> >>> return 0;
> >>> @@ -2176,11 +2193,15 @@ int try_to_free_nids(struct f2fs_sb_info *sbi, int nr_shrink)
> >>> nm_i->nid_cnt[FREE_NID] <= MAX_FREE_NIDS)
> >>> break;
> >>>
> >>> + if (i->nid < min_nid)
> >>> + min_nid = i->nid;
> >>> __remove_free_nid(sbi, i, FREE_NID);
> >>> kmem_cache_free(free_nid_slab, i);
> >>> nr_shrink--;
> >>> }
> >>> spin_unlock(&nm_i->nid_list_lock);
> >>> + if (min_nid != nm_i->max_nid)
> >>> + nm_i->first_scan_block = NAT_BLOCK_OFFSET(min_nid);
> >>
> >> Need to update nm_i->first_scan_block during __flush_nat_entry_set?
> >>
> > The doubt I have is described in above question.
> >
> >> Thanks,
> >>
> >>> mutex_unlock(&nm_i->build_lock);
> >>>
> >>> return nr - nr_shrink;
> >>> @@ -2674,6 +2695,7 @@ static int init_node_manager(struct f2fs_sb_info *sbi)
> >>> init_rwsem(&nm_i->nat_tree_lock);
> >>>
> >>> nm_i->next_scan_nid = le32_to_cpu(sbi->ckpt->next_free_nid);
> >>> + nm_i->first_scan_block = NEW_ADDR;
> >>> nm_i->bitmap_size = __bitmap_size(sbi, NAT_BITMAP);
> >>> version_bitmap = __bitmap_ptr(sbi, NAT_BITMAP);
> >>> if (!version_bitmap)
> >>>
> >
> >