Re: [PATCH V4 19/20] xfsprogs: Add support for upgrading to NREXT64 feature

[Chandan Babu R <chandan.babu@oracle.com>]

On 08 Jan 2022 at 00:33, Darrick J. Wong wrote:
> On Fri, Jan 07, 2022 at 09:47:10PM +0530, Chandan Babu R wrote:
>> On 05 Jan 2022 at 06:47, Darrick J. Wong wrote:
>> > On Tue, Dec 14, 2021 at 02:18:10PM +0530, Chandan Babu R wrote:
>> >> This commit adds support to xfs_repair to allow upgrading an existing
>> >> filesystem to support per-inode large extent counters.
>> >> 
>> >> Signed-off-by: Chandan Babu R <chandan.babu@oracle.com>
>> >> ---
>> >>  repair/globals.c    |  1 +
>> >>  repair/globals.h    |  1 +
>> >>  repair/phase2.c     | 35 ++++++++++++++++++++++++++++++++++-
>> >>  repair/xfs_repair.c | 11 +++++++++++
>> >>  4 files changed, 47 insertions(+), 1 deletion(-)
>> >> 
>> >> diff --git a/repair/globals.c b/repair/globals.c
>> >> index d89507b1..2f29391a 100644
>> >> --- a/repair/globals.c
>> >> +++ b/repair/globals.c
>> >> @@ -53,6 +53,7 @@ bool	add_bigtime;		/* add support for timestamps up to 2486 */
>> >>  bool	add_finobt;		/* add free inode btrees */
>> >>  bool	add_reflink;		/* add reference count btrees */
>> >>  bool	add_rmapbt;		/* add reverse mapping btrees */
>> >> +bool	add_nrext64;
>> >>  
>> >>  /* misc status variables */
>> >>  
>> >> diff --git a/repair/globals.h b/repair/globals.h
>> >> index 53ff2532..af0bcb6b 100644
>> >> --- a/repair/globals.h
>> >> +++ b/repair/globals.h
>> >> @@ -94,6 +94,7 @@ extern bool	add_bigtime;		/* add support for timestamps up to 2486 */
>> >>  extern bool	add_finobt;		/* add free inode btrees */
>> >>  extern bool	add_reflink;		/* add reference count btrees */
>> >>  extern bool	add_rmapbt;		/* add reverse mapping btrees */
>> >> +extern bool	add_nrext64;
>> >>  
>> >>  /* misc status variables */
>> >>  
>> >> diff --git a/repair/phase2.c b/repair/phase2.c
>> >> index c811ed5d..c9db3281 100644
>> >> --- a/repair/phase2.c
>> >> +++ b/repair/phase2.c
>> >> @@ -191,6 +191,7 @@ check_new_v5_geometry(
>> >>  	struct xfs_perag	*pag;
>> >>  	xfs_agnumber_t		agno;
>> >>  	xfs_ino_t		rootino;
>> >> +	uint			old_bm_maxlevels[2];
>> >>  	int			min_logblocks;
>> >>  	int			error;
>> >>  
>> >> @@ -201,6 +202,12 @@ check_new_v5_geometry(
>> >>  	memcpy(&old_sb, &mp->m_sb, sizeof(struct xfs_sb));
>> >>  	memcpy(&mp->m_sb, new_sb, sizeof(struct xfs_sb));
>> >>  
>> >> +	old_bm_maxlevels[0] = mp->m_bm_maxlevels[0];
>> >> +	old_bm_maxlevels[1] = mp->m_bm_maxlevels[1];
>> >> +
>> >> +	xfs_bmap_compute_maxlevels(mp, XFS_DATA_FORK);
>> >> +	xfs_bmap_compute_maxlevels(mp, XFS_ATTR_FORK);
>> >
>> > Ahh... I see why you added my (evil) patch that allows upgrading a
>> > filesystem to reflink -- you need the check_new_v5_geometry function so
>> > that you can check if the log size is big enough to handle larger bmbt
>> > trees.
>> >
>> > Hmm, I guess I should work on separating this from the actual
>> > rmap/reflink/finobt upgrade code, since I have no idea if we /ever/ want
>> > to support that.
>> >
>> 
>> I can do that. I will include the trimmed down version of the patch before
>> posting the patchset once again.
>
> I separated that megapatch into smaller pieces yesterday, so I'll point
> you to it once it all goes through QA.
>

Ok. I will wait.

-- 
chandan