[pnfs] linux-pnfs git status 2008-07-06
Benny Halevy
bhalevy at panasas.com
Sun Jul 6 09:55:29 EDT 2008
I've rebased the tree on top of the recent patches I sent to the
upstream branch. There were some cleanups in fs/nfs{,d}/nfs4xdr.c
and consequently code moved around and status return values
were added to the nfs41/pnfs nfsd xdr encoding functions.
Other than that there should be no functional change.
latest patches (since last status, 2008-06-19):
upstream:
Benny Halevy (14):
nfs: initialize timeout variable in nfs4_proc_setclientid_confirm
uml: fix extern inline errors with gcc 4.3.0
usb: fix uninitialized variable warning in keyspan_pda
mm: fix uninitialized variables for find_vma_prepare callers
dlm: fix uninitialized variable for search_rsb_list callers
nfsd: return nfserr_minor_vers_mismatch when compound minorversion != 0
nfsd: tabulate nfs4 xdr decoding functions
nfsd: unsupported nfs4 ops should fail with nfserr_opnotsupp
nfsd: nfs4 minorversion decoder vectors
nfsd: tabulate nfs4 xdr encoding functions
nfsd: dprint operation names
nfs: return nfs4 compound header status on op header decoding error
nfs: remove incorrect usage of nfs4 compound response hdr.status
nfsd: take file and mnt write in nfs4_upgrade_open
nfs41:
Benny Halevy (6):
[SQUASHME] nfs41: rename nfs40 procedures and version names back to nfs4
nfs41: sunrpc: support minorversion for rpc_clnt version table.
[SQUASHME] nfs41: obliterate nfs4_setup_nfs4_program
nfs41: use ptr to rpc procedures via struct nfs_client
nfs: pass recovery status from reclaimer kthread up to nfs4_wait_clnt_reco
nfs41: fallback to lower minorversion if nfs4_create_server fails
Ricardo Labiaga (3):
nfs41: Allow NFSv4 and NFSv4.1 callback services to coexist
nfs41: callback_svc needs to be set for v4 and v4.1 mounts
nfs41: Proper initialization of nfs_client_initdata.rpc_ops
pnfs:
Andy Adamson (3):
pnfs: intialize pnfs on submounts and referrals
pnfs: bump the opencontext reference for ds commit
pnfs: Don't send getattr with Data Server commit
Benny Halevy (2):
[SQUASHME] pnfs: rename -pnfs_rpc_done to pnfs_call_done
[SQUASHME] pnfs: put_lseg in pnfs_call_done
Marc Eshel (1):
pnfsd: remove unused symbol exports
spnfs:
Dan Muntz (1):
Reduce size of I/O through MDS buffer for spNFS.
Mike Sager (2):
Move from nfsctl() to /proc based user/kernel space communication
Add /proc based mechanism to test layout recall
Benny
On Jun. 19, 2008, 22:00 +0300, Benny Halevy <bhalevy at panasas.com> wrote:
> The recent 2.6.26 fixes were applied and back-ported to the
> 2.6.25 series.
>
> I stacked all 2.6.25 branches into a sequential series as
> depicted below to produce a -all patch against v2.6.25:
> http://www.bhalevy.com/pnfs/pnfs-2.6.25-all-2008-06-19.patch
> It's too big (~ 1 MB) to attach here...
> 110 files changed, 27276 insertions(+), 1228 deletions(-)
>
> I ran it through basic connectathon test but please give it a
> good test if you can so we can safely freeze it forever.
>
> Here's the new tree structure:
>
> Note that the pnfs-2.6.25-all branch is just an alias to
> panlayout-2.6.25-all and it's there as an easy-to-find
> anchor and convention for people to look for.
>
> master-2.6.25
> upstream-2.6.25
> nfs41-2.6.25
> pnfs-2.6.25
> pnfsd-lexp-2.6.25
> pnfs-block-2.6.25
> spnfs-2.6.25
> pnfsd-lexp-2.6.25-all
> pnfs-gfs2-2.6.25-all
> pnfs-block-2.6.25-all
> panlayout-2.6.25-all
> pnfs-2.6.25-all
> pnfs-gfs2-2.6.25
> panlayout-2.6.25
> master
> upstream
> nfs41
> pnfs
> pnfsd-lexp
> pnfs-block
> spnfs
> pnfs-gfs2
> panlayout
>
> And the shortlog of what went in:
>
> nfs41{,-2.6.25}:
>
> Andy Adamson (2):
> [SQUASHME] nfs41: only call destroy session from put session
> [SQUASHME] nfs41: don't set flags in a destroyed session
>
> pnfs{,-2.6.25}:
>
> Benny Halevy (7):
> [SQUASHME] pnfs: insert nfsi onto cl_lo_inodes only once
> [SQUASHME] pnfs: release lo_lock around call to alloc_lseg in pnfs_inject_layout
> [SQUASHME] pnfs: refactor pnfs call data out of nfs_{read,write}_data
> [SQUASHME] pnfs: refactor pnfs read, write, and commit done functions
> [SQUASHME] pnfs: don't call rpc_release from pnfs_call_done on rpc completion
> [SQUASHME] pnfs: set PNFS_NO_RPC also on commit path
> [SQUASHME] pnfs: acquire the BKL around rpc ops called from pnfs_rpc_done
>
> Fred Isaman (5):
> [FIXME] pnfs: BUG: spinlock held during call to setup_layoutcommit
> [SQUASHME] pnfs: make pnfs_layout_state flags consistent
> [SQUASHME] pnfs: always set pnfs_layout_suspend
> [SQUASHME] pnfs: shift nfsi structures protected by pnfs_spinlock to nfsi->lo_lock
> [SQUASHME] pnfs: set stateid for LAYOUTCOMMIT
>
> pnfsd-lexp{,-2.6.25}:
>
> Benny Halevy (12):
> pnfsd-lexp: Add CONFIG_PNFSD_LOCAL_EXPORT Kconfig option
> pnfsd-lexp: fs/nfsd/pnfsd_lexp.c skeleton and Makefile changes
> pnfsd-lexp: layout_type
> [FIXME] pnfsd-lexp: local sockaddr hack
> pnfsd-lexp: get_device_iter
> pnfsd-lexp: sunrpc: __svc_print_netaddr
> pnfsd-lexp: get_device_info
> pnfsd-lexp: layout_get
> pnfsd-lexp: layout_commit
> pnfsd-lexp: layout_return
> pnfsd-lexp: pnfsd export operations init function
> pnfsd-lexp: call pnfsd_lexp_init at check_export time
>
> pnfs-block{,-2.6.25}:
>
> Benny Halevy (1):
> [SQUASHME] pnfsblock: use refactored pnfs_call_data
>
> Fred Isaman (2):
> pnfsblock: accomodate write_begin API changes
> pnfsblock: adapt to scsi changes in 2.6.26 # pnfs-block only
>
> panlayout{,-2.6.25}:
>
> Benny Halevy (3):
> [SQUASHME] panlayout: use refactored pnfs_call_data
> [SQUASHME] panlayout: xdr_sz functions should return int not size_t
> [SQUASHME] panlayout: fix pnfs_osd_object_cred_incore_sz
More information about the pNFS
mailing list