On a read-only filesystem, we won't invoke the block allocator, so we
don't need to prefetch the block bitmaps.
This avoids starting and running the ext4lazyinit thread at all on a
system with no read-write ext4 filesystems (for instance, a container VM
with read-only filesystems underneath an overlayfs).
Fixes: 21175ca434c5 ("ext4: make prefetch_block_bitmaps default")
Signed-off-by: Josh Triplett <[email protected]>
---
Tested in a VM, with a read-only ext4 root filesystem. Confirmed that
ext4lazyinit starts without this patch, and does not start with this
patch.
(For a future merge window, not the current one. Please let me know if I
need to re-send this at a later, more convenient time.)
fs/ext4/super.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/fs/ext4/super.c b/fs/ext4/super.c
index 845f2f8aee5f..20437acc8865 100644
--- a/fs/ext4/super.c
+++ b/fs/ext4/super.c
@@ -3973,9 +3973,9 @@ int ext4_register_li_request(struct super_block *sb,
goto out;
}
- if (test_opt(sb, NO_PREFETCH_BLOCK_BITMAPS) &&
- (first_not_zeroed == ngroups || sb_rdonly(sb) ||
- !test_opt(sb, INIT_INODE_TABLE)))
+ if (sb_rdonly(sb) ||
+ (test_opt(sb, NO_PREFETCH_BLOCK_BITMAPS) &&
+ (first_not_zeroed == ngroups || !test_opt(sb, INIT_INODE_TABLE))))
goto out;
elr = ext4_li_request_new(sb, first_not_zeroed);
--
2.36.1
On Sun, Jul 31, 2022 at 08:24:53PM -0700, Josh Triplett wrote:
> On a read-only filesystem, we won't invoke the block allocator, so we
> don't need to prefetch the block bitmaps.
>
> This avoids starting and running the ext4lazyinit thread at all on a
> system with no read-write ext4 filesystems (for instance, a container VM
> with read-only filesystems underneath an overlayfs).
>
> Fixes: 21175ca434c5 ("ext4: make prefetch_block_bitmaps default")
> Signed-off-by: Josh Triplett <[email protected]>
> ---
Looks good to me.
Reviewed-by: Lukas Czerner <[email protected]>
>
> Tested in a VM, with a read-only ext4 root filesystem. Confirmed that
> ext4lazyinit starts without this patch, and does not start with this
> patch.
>
> (For a future merge window, not the current one. Please let me know if I
> need to re-send this at a later, more convenient time.)
>
> fs/ext4/super.c | 6 +++---
> 1 file changed, 3 insertions(+), 3 deletions(-)
>
> diff --git a/fs/ext4/super.c b/fs/ext4/super.c
> index 845f2f8aee5f..20437acc8865 100644
> --- a/fs/ext4/super.c
> +++ b/fs/ext4/super.c
> @@ -3973,9 +3973,9 @@ int ext4_register_li_request(struct super_block *sb,
> goto out;
> }
>
> - if (test_opt(sb, NO_PREFETCH_BLOCK_BITMAPS) &&
> - (first_not_zeroed == ngroups || sb_rdonly(sb) ||
> - !test_opt(sb, INIT_INODE_TABLE)))
> + if (sb_rdonly(sb) ||
> + (test_opt(sb, NO_PREFETCH_BLOCK_BITMAPS) &&
> + (first_not_zeroed == ngroups || !test_opt(sb, INIT_INODE_TABLE))))
> goto out;
>
> elr = ext4_li_request_new(sb, first_not_zeroed);
> --
> 2.36.1
>
On Wed, Aug 03, 2022 at 09:00:01AM +0200, Lukas Czerner wrote:
> On Sun, Jul 31, 2022 at 08:24:53PM -0700, Josh Triplett wrote:
> > On a read-only filesystem, we won't invoke the block allocator, so we
> > don't need to prefetch the block bitmaps.
> >
> > This avoids starting and running the ext4lazyinit thread at all on a
> > system with no read-write ext4 filesystems (for instance, a container VM
> > with read-only filesystems underneath an overlayfs).
> >
> > Fixes: 21175ca434c5 ("ext4: make prefetch_block_bitmaps default")
> > Signed-off-by: Josh Triplett <[email protected]>
> > ---
>
> Looks good to me.
>
> Reviewed-by: Lukas Czerner <[email protected]>
>
> > Tested in a VM, with a read-only ext4 root filesystem. Confirmed that
> > ext4lazyinit starts without this patch, and does not start with this
> > patch.
> >
> > (For a future merge window, not the current one. Please let me know if I
> > need to re-send this at a later, more convenient time.)
Now that the merge window has been closed for a while, I wanted to
follow up on this to see if it could get into next.
> > fs/ext4/super.c | 6 +++---
> > 1 file changed, 3 insertions(+), 3 deletions(-)
> >
> > diff --git a/fs/ext4/super.c b/fs/ext4/super.c
> > index 845f2f8aee5f..20437acc8865 100644
> > --- a/fs/ext4/super.c
> > +++ b/fs/ext4/super.c
> > @@ -3973,9 +3973,9 @@ int ext4_register_li_request(struct super_block *sb,
> > goto out;
> > }
> >
> > - if (test_opt(sb, NO_PREFETCH_BLOCK_BITMAPS) &&
> > - (first_not_zeroed == ngroups || sb_rdonly(sb) ||
> > - !test_opt(sb, INIT_INODE_TABLE)))
> > + if (sb_rdonly(sb) ||
> > + (test_opt(sb, NO_PREFETCH_BLOCK_BITMAPS) &&
> > + (first_not_zeroed == ngroups || !test_opt(sb, INIT_INODE_TABLE))))
> > goto out;
> >
> > elr = ext4_li_request_new(sb, first_not_zeroed);
> > --
> > 2.36.1
> >
>
On Sun, 31 Jul 2022 20:24:53 -0700, Josh Triplett wrote:
> On a read-only filesystem, we won't invoke the block allocator, so we
> don't need to prefetch the block bitmaps.
>
> This avoids starting and running the ext4lazyinit thread at all on a
> system with no read-write ext4 filesystems (for instance, a container VM
> with read-only filesystems underneath an overlayfs).
>
> [...]
Applied, thanks!
[1/1] ext4: Don't run ext4lazyinit for read-only filesystems
commit: 3490a40364962a2599bc5a8126003a47150b84d3
Best regards,
--
Theodore Ts'o <[email protected]>