Skip to contentSkip to conversation

nfsd 2.6.28 merge plans

DISCUSSION J. Bruce Fields 4 participants 6 messages Oct 8, 2008 – Nov 20, 2008

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Linus's release announcement for -rc9 suggested 2.6.27 could be released (and the 2.6.28 merge window opened) any day now.

What I'm currently planning to merge is the contents of

	git://linux-nfs.org/~bfields/linux.git for-2.6.28

summarized below.

Chuck's contributed a ton of small patches preparing for ipv6 support (which we expect to complete in 2.6.29), Tom Tucker has added support for fast memory registration to the server rdma code, and Benny's NFSv4.1 work continues to produce miscellaneous NFSv4.1 bugfixes and cleanups.

There's also one pure locking patch, which makes vfs support for posix locking configurable.

And I've got some grace period patches (which I hope will also pave the way for nfsd lock migration support in 2.6.29), and some minor tweaks to our rpcsec_gss code which (together with recent nfs-utils patches) finally makes rpcsec_gss for v2 and v3 work as well as it does for v4.

Thanks to everyone who's contributed!

--b.

Benny Halevy (9):
nfsd: properly xdr-encode stateid4.seqid as uint32_t for cb_recall
nfsd: properly xdr-encode deleg stateid returned from open
nfsd: fix nfsd4_encode_open buffer space reservation
nfsd: nfs4xdr encode_stateid helper function
nfsd: don't declare p in ENCODE_SEQID_OP_HEAD
nfsd: properly xdr-decode NFS4_OPEN_CLAIM_DELEGATE_CUR stateid
nfsd: nfs4xdr decode_stateid helper function
nfsd: do_probe_callback should not clear rpc stats
nfsd: use nfs client rpc callback program

Chuck Lever (32):
SUNRPC: Add address family field to svc_serv data structure
SUNRPC: Use proper INADDR_ANY when setting up RPC services on IPv6
SUNRPC: Set V6ONLY socket option for RPC listener sockets
SUNRPC: Simplify rpcb_register() API
SUNRPC: Split portmap unregister API into separate function
SUNRPC: Support IPv6 when registering kernel RPC services
NLM: Clean up before introducing new debugging messages
lockd: address-family independent printable addresses
lockd: Specify address family for source address
lockd: Add address family-agnostic helper for zeroing the port number
SUNRPC: Make svc_addr's argument a constant
lockd: Use sockaddr_storage + length for h_addr field
lockd: Use sockaddr_storage for h_saddr field
NSM: Use sockaddr_storage for sm_addr field
lockd: Teach nlm_cmp_addr() to support AF_INET6 addresses
lockd: Support AF_INET6 when hashing addresses in nlm_lookup_host
lockd: Combine __nsm_find() and nsm_find().
lockd: Update nsm_find() to support non-AF_INET addresses
SUNRPC: Register both netids for AF_INET6 servers
SUNRPC: Use short-hand IPv6 ANYADDR for RPCB_SET
SUNRPC: Fix up svc_unregister()
SUNRPC: Clean up debug messages in rpcb_clnt.c
NLM: Convert nlm_lookup_host() to use a single argument
lockd: Support non-AF_INET addresses in nlm_lookup_host()
lockd: Adjust nlmclnt_lookup_host() signature to accomodate non-AF_INET
lockd: Adjust nlmsvc_lookup_host() to accomodate AF_INET6 addresses
lockd: change nlmclnt_grant() to take a "struct sockaddr *"
lockd: Add helper to sanity check incoming NOTIFY requests
lockd: Remove unused fields in the nlm_reboot structure
NLM: Always start both UDP and TCP listeners
NLM: Remove "proto" argument from lockd_up()
NLM: Remove unused argument from svc_addsock() function

J. Bruce Fields (7):
nfsd: permit unauthenticated stat of export root
locks: allow lockd to process blocked locks during grace period
lockd: don't depend on lockd main loop to end grace
nfsd: common grace period control
lockd: move grace period checks to common code
lockd: reject reclaims outside the grace period Merge branch 'from-tomtucker' into for-2.6.28

Jeff Layton (1):
knfsd: allocate readahead cache in individual chunks

Thomas Petazzoni (1):
Configure out file locking features

Tom Tucker (10):
svcrdma: Add Fast Reg MR Data Types
svcrdma: Add FRMR get/put services
svcrdma: Query device for Fast Reg support during connection setup
svcrdma: Add a service to register a Fast Reg MR with the device
svcrdma: Modify post recv path to use local dma key
svcrdma: Add support to svc_rdma_send to handle chained WR
svcrdma: Modify the RPC recv path to use FRMR when available
svcrdma: Modify the RPC reply path to use FRMR when available
svcrdma: Update svc_rdma_send_error to use DMA LKEY
svcrdma: Fix IRD/ORD polarity

#1 lore raw
Quoted 16 lines · On Wed, 8 Oct 2008 19:22:37 -0400 "J. Bruce Fields" <[email protected]>…

Linus's release announcement for -rc9 suggested 2.6.27 could be released (and the 2.6.28 merge window opened) any day now.

What I'm currently planning to merge is the contents of

	git://linux-nfs.org/~bfields/linux.git for-2.6.28

summarized below.

Chuck's contributed a ton of small patches preparing for ipv6 support (which we expect to complete in 2.6.29), Tom Tucker has added support for fast memory registration to the server rdma code, and Benny's NFSv4.1 work continues to produce miscellaneous NFSv4.1 bugfixes and cleanups.

Quoted 7 lines

There's also one pure locking patch, which makes vfs support for posix locking configurable.

And I've got some grace period patches (which I hope will also pave the way for nfsd lock migration support in 2.6.29), and some minor tweaks to our rpcsec_gss code which (together with recent nfs-utils patches) finally makes rpcsec_gss for v2 and v3 work as well as it does for v4.

Thanks.

Is anything happening with David Howells's fscache and cachefiles work over there?

#2 lore raw
Quoted 2 lines · Andrew Morton

Is anything happening with David Howells's fscache and cachefiles work over there?

That's more NFS than NFSD. I've tried emailing Trond a few times since KS/Plumbers, but I've had no response, so I'm not sure.

David

#3 lore raw
Quoted 9 lines · David Howells
Quoted 2 lines · Andrew Morton

Is anything happening with David Howells's fscache and cachefiles work over there?

That's more NFS than NFSD. I've tried emailing Trond a few times since KS/Plumbers, but I've had no response, so I'm not sure.

David

I'm happy with the changes that you made after the plumbers conf, so as far as the NFS side of the patches is concerned, we should be close to ready to merge.

Signature · 6 lines
Trond Myklebust
Linux NFS client maintainer

NetApp
[email protected]
www.netapp.com
#4 lore raw
Quoted 3 lines · Trond Myklebust

I'm happy with the changes that you made after the plumbers conf, so as far as the NFS side of the patches is concerned, we should be close to ready to merge.

What I'm hoping is that JamesM will pass my creds patches to Linus in this merge window - that way you don't need to carry them in your tree as the main cachefiles patch requires them to deal with security.

David

#5 lore raw
In reply to Trond Myklebust
Quoted 3 lines · Trond Myklebust

I'm happy with the changes that you made after the plumbers conf, so as far as the NFS side of the patches is concerned, we should be close to ready to merge.

Okay... I've fixed a couple of bugs and now it seems to survive multiple parallel tars being repeatedly thrown at it.

How do you want to do this? I could make a GIT tree for you to pull, but it'd have to be based on James Morris's next branch for the security features required.

Alternatively, I could supply you with patches or a GIT tree, including the security patches based on Linus's tree rather than James's.

Or we could wait until after the next merge window and hope Linus pulls them this time (which he ought to, via James's tree).

David