Since max_nid value was minus three in previous commit:
'b63da15e8b475245026bdf2096853683f189706b', Our last three nids could not be
used ever, but 0/node nid/meta nid is still occupied.
And also our ra_meta_pages() could not readahead the last block of NAT.
Let's fix the wrong calculation of this value, and left the reserved nids there.
Signed-off-by: Chao Yu <[email protected]>
---
fs/f2fs/node.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
index 84f9b7b..3f0bb50 100644
--- a/fs/f2fs/node.c
+++ b/fs/f2fs/node.c
@@ -1860,7 +1860,7 @@ static int init_node_manager(struct f2fs_sb_info *sbi)
nat_blocks = nat_segs << le32_to_cpu(sb_raw->log_blocks_per_seg);
/* not used nids: 0, node, meta, (and root counted as valid node) */
- nm_i->max_nid = NAT_ENTRY_PER_BLOCK * nat_blocks - 3;
+ nm_i->max_nid = NAT_ENTRY_PER_BLOCK * nat_blocks;
nm_i->fcnt = 0;
nm_i->nat_cnt = 0;
nm_i->ram_thresh = DEF_RAM_THRESHOLD;
--
1.7.9.5
Hi,
2014-04-17 (목), 14:27 +0800, Chao Yu:
> Since max_nid value was minus three in previous commit:
> 'b63da15e8b475245026bdf2096853683f189706b', Our last three nids could not be
> used ever, but 0/node nid/meta nid is still occupied.
> And also our ra_meta_pages() could not readahead the last block of NAT.
What is the root problem in there?
IIRC, the max_nid should not be turned back since that can cause an
infinite loop during build_free_nids() under a test using whole the
nids.
> Let's fix the wrong calculation of this value, and left the reserved nids there.
>
> Signed-off-by: Chao Yu <[email protected]>
> ---
> fs/f2fs/node.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
> index 84f9b7b..3f0bb50 100644
> --- a/fs/f2fs/node.c
> +++ b/fs/f2fs/node.c
> @@ -1860,7 +1860,7 @@ static int init_node_manager(struct f2fs_sb_info *sbi)
> nat_blocks = nat_segs << le32_to_cpu(sb_raw->log_blocks_per_seg);
>
> /* not used nids: 0, node, meta, (and root counted as valid node) */
> - nm_i->max_nid = NAT_ENTRY_PER_BLOCK * nat_blocks - 3;
> + nm_i->max_nid = NAT_ENTRY_PER_BLOCK * nat_blocks;
> nm_i->fcnt = 0;
> nm_i->nat_cnt = 0;
> nm_i->ram_thresh = DEF_RAM_THRESHOLD;
--
Jaegeuk Kim
Samsung
Hi,
> -----Original Message-----
> From: Jaegeuk Kim [mailto:[email protected]]
> Sent: Thursday, April 17, 2014 3:17 PM
> To: Chao Yu
> Cc: [email protected]; [email protected];
> [email protected]
> Subject: Re: [f2fs-dev] [PATCH] f2fs: fix wrong number of max_nid when init
>
> Hi,
>
> 2014-04-17 (목), 14:27 +0800, Chao Yu:
> > Since max_nid value was minus three in previous commit:
> > 'b63da15e8b475245026bdf2096853683f189706b', Our last three nids could not be
> > used ever, but 0/node nid/meta nid is still occupied.
> > And also our ra_meta_pages() could not readahead the last block of NAT.
>
> What is the root problem in there?
The last one is the root problem, I will submit another patch to fix it.
> IIRC, the max_nid should not be turned back since that can cause an
> infinite loop during build_free_nids() under a test using whole the
> nids.
Got it, thanks for your explanation.
Regards.
>
> > Let's fix the wrong calculation of this value, and left the reserved nids there.
> >
> > Signed-off-by: Chao Yu <[email protected]>
> > ---
> > fs/f2fs/node.c | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
> > index 84f9b7b..3f0bb50 100644
> > --- a/fs/f2fs/node.c
> > +++ b/fs/f2fs/node.c
> > @@ -1860,7 +1860,7 @@ static int init_node_manager(struct f2fs_sb_info *sbi)
> > nat_blocks = nat_segs << le32_to_cpu(sb_raw->log_blocks_per_seg);
> >
> > /* not used nids: 0, node, meta, (and root counted as valid node) */
> > - nm_i->max_nid = NAT_ENTRY_PER_BLOCK * nat_blocks - 3;
> > + nm_i->max_nid = NAT_ENTRY_PER_BLOCK * nat_blocks;
> > nm_i->fcnt = 0;
> > nm_i->nat_cnt = 0;
> > nm_i->ram_thresh = DEF_RAM_THRESHOLD;
>
> --
> Jaegeuk Kim
> Samsung