On 2/27/19 5:03 PM, Darrick J. Wong wrote:
> On Wed, Feb 27, 2019 at 04:19:03PM -0600, Eric Sandeen wrote:
>> On 2/27/19 3:39 PM, Eric Sandeen wrote:
>>> Hi folks,
>>>
>>> The xfsprogs repository at:
>>>
>>> git://git.kernel.org/pub/scm/fs/xfs/xfsprogs-dev.git
>>>
>>> has just been updated.
>>>
>>> This contains only the sync to the kernel's libxfs updates for v5.0.
>>>
>>> The new head of the for-next branch is commit:
>>
>> Fixed a thinko stray ; caught by coverity and did a forced push on
>> these branches right quick before anyone but djwong pulled it down. ;)
>
> Hmmm, I still some discrepancies...
>
> diff -Naurp xfsprogs/libxfs/xfs_ialloc.c kernel/libxfs/xfs_ialloc.c
> --- xfsprogs/libxfs/xfs_ialloc.c 2019-02-27 14:58:30.286319523 -0800
> +++ kernel/libxfs/xfs_ialloc.c 2019-02-27 14:58:48.512355119 -0800
> @@ -34,7 +34,7 @@ xfs_ialloc_cluster_alignment(
> struct xfs_mount *mp)
> {
> if (xfs_sb_version_hasalign(&mp->m_sb) &&
> - mp->m_sb.sb_inoalignmt >= mp->m_blocks_per_cluster)
> + mp->m_sb.sb_inoalignmt >= xfs_icluster_size_fsb(mp))
> return mp->m_sb.sb_inoalignmt;
> return 1;
> }
>
>
> Over on the kernel side we're still using xfs_icluster_size_fsb in the
> comparison test. TBH I don't remember whether this is actually required
> on the kernel side or not, but in any case this ought to go through the
> kernel.
That was a mistake, I was searching for direct users of xfs_icluster_size_fsb
elsewhere and converting them to the precalculated value, and accidentally
hit libxfs. I can change it back.
> I /think/ the reason is that I didn't want the initializer functions for
> the xfs_mount inode goemetry fields to depend on them being set in any
> particular order.
I'll fix it regardless.
>
>
> diff -Naurp xfsprogs/libxfs/xfs_types.h kernel/libxfs/xfs_types.h
> --- xfsprogs/libxfs/xfs_types.h 2019-02-27 14:58:30.287319525 -0800
> +++ kernel/libxfs/xfs_types.h 2019-02-27 14:58:48.516355127 -0800
> @@ -114,8 +114,8 @@ typedef enum {
> { XFS_LOOKUP_GEi, "ge" }
>
> /*
> - * This enum is used in string mapping in xfs_trace.h; please keep the
> - * TRACE_DEFINE_ENUMs for it up to date.
> + * This enum is used in string mapping in xfs_trace.h and scrub/trace.h;
> + * please keep the TRACE_DEFINE_ENUMs for it up to date.
> */
> typedef enum {
> XFS_BTNUM_BNOi, XFS_BTNUM_CNTi, XFS_BTNUM_RMAPi, XFS_BTNUM_BMAPi,
>
> I think you're missing "xfs: stringify scrub types in ftrace output".
forgot to guilt push it on prior to the re-push. :/
>
> --D
>
>> Now we're at:
>>
>> bb83d20 xfs: stringify btree cursor types in ftrace output
>>
>> New Commits:
>>
>> Darrick J. Wong (11):
>> [5afb131] xfs: idiotproof defer op type configuration
>> [5a23b29] xfs: streamline defer op type handling
>> [c2d5dd0] xfs: const-ify xfs_owner_info arguments
>> [9ef0cb4] xfs: remove xfs_rmap_ag_owner and friends
>> [09c23f4] xfs: add a block to inode count converter
>> [ad5eff9] xfs: precalculate inodes and blocks per inode cluster
>> [6180d69] xfs: precalculate cluster alignment in inodes and blocks
>> [ef1b0a2] xfs: fix symbolic enum printing in ftrace output
>> [a97fc90] xfs: move XFS_AG_BTREE_CMP_FORMAT_STR mappings to libxfs
>> [e01dc40] xfs: move XFS_INODE_FORMAT_STR mappings to libxfs
>> [bb83d20] xfs: stringify btree cursor types in ftrace output
>>
>> Dave Chinner (1):
>> [7ab21d5] xfs: zero length symlinks are not valid
>>
>> Omar Sandoval (1):
>> [6890a4c] xfs: cache minimum realtime summary level
>>
>> Pan Bian (1):
>> [972ca59] xfs: libxfs: move xfs_perag_put late
>>
>>
>
>
>
