Below is a summary of the nfs server patches which I expect to submit
(any day now) for 2.6.25; please let me know if you notice anything
wrong or missing. I'll follow up to [email protected] with the
full list of patches.
The biggest changes come from Tom Tucker's rdma and server transport
switch patches, which Chuck Lever, Greg Banks, Neil Brown, and others
helped review (thanks!). The rest is mostly bug fixes and cleanup.
As a reminder, you can see the patches I've got queued up any time from
the nfs-server-stable branch at:
git://linux-nfs.org/~bfields/linux.git nfs-server-stable
--b.
Chuck Lever (9):
SUNRPC: Prevent length underflow in read_flush()
SUNRPC: Use unsigned string lengths in xdr_decode_string_inplace
NLM: Fix sign of length of NLM variable length strings
NFSD: Use unsigned length argument for decode_filename
NFSD: File name length signage in nfsd request argument structures
NFSD: Adjust filename length argument of nfsd_lookup
NFSD: Use unsigned length argument for decode_pathname
NFSD: Fix mixed sign comparison in nfs3svc_decode_symlinkargs
NFSD: Path name length signage in nfsd request argument structures
Frank Filz (1):
nfsd: Allow AIX client to read dir containing mountpoints
J. Bruce Fields (34):
nfsd4: probe callback channel only once
nfsd: move callback rpc_client creation into separate thread
knfsd: fix broken length check in nfs4idmap.c
knfsd: fix cache.c comment
nfsd: Fix handling of negative lengths in read_buf()
knfsd: cleanup nfsd4 properly on module init failure
nfsd: cleanup nfsd module initialization cleanup
nfsd: fail module init on reply cache init failure
knfsd: cache unregistration needn't return error
nfsd: select CONFIG_PROC_FS in nfsv4 and gss server cases
nfsd: fail init on /proc/fs/nfs/exports creation failure
nfsd: move cache proc (un)registration to separate function
knfsd: allow cache_register to return error on failure
nfsd: move nfsd/auth.h into fs/nfsd
nfsd: minor fs/nfsd/auth.h cleanup
nfsd4: kill some unneeded setclientid comments
nfsd: eliminate final bogus case from setclientid logic
nfsd: uniquify cl_confirm values
nfsd4: kill unnecessary same_name() in setclientid_confirm
nfsd4: remove unnecessary cl_verifier check from setclientid_confirm
nfsd4: kill unneeded cl_confirm check
nfsd: fix encode_entryplus_baggage() indentation
nfsd4: make current_clientid local
nfsd4: miscellaneous nfs4state.c style fixes
nfsd4: recognize callback channel failure earlier
nfsd4: fix bad seqid on lock request incompatible with open mode
nfsd: allow root to set uid and gid on create
nfsd: fix rsi_cache reference count leak
sunrpc: gss: simplify rsi_parse logic
nfsd4: clean up access_valid, deny_valid checks.
svcrpc: ensure gss DESTROY tokens free contexts from cache
knfsd: don't bother mapping putrootfh enoent to eperm
lockd: minor log message fix
nfsd: more careful input validation in nfsctl write methods
Jeff Layton (4):
SUNRPC: spin svc_rqst initialization to its own function
SUNRPC: export svc_sock_update_bufs
NLM: Convert lockd to use kthreads
NLM: have nlm_shutdown_hosts kill off all NLM RPC tasks
Neil Brown (1):
knfsd: change mailing list for nfsd in MAINTAINERS
Oleg Drokin (3):
lockd: fix reference count leaks in async locking case
lockd: fix a leak in nlmsvc_testlock asynchronous request handling
Leak in nlmsvc_testlock for async GETFL case
Prasad P (1):
Fix incorrect assignment
Tom Tucker (47):
svc: Add an svc transport class
svc: Make svc_sock the tcp/udp transport
svc: Change the svc_sock in the rqstp structure to a transport
svc: Add a max payload value to the transport
svc: Move sk_sendto and sk_recvfrom to svc_xprt_class
svc: Add transport specific xpo_release function
svc: Add per-transport delete functions
svc: Add xpo_prep_reply_hdr
svc: Add a transport function that checks for write space
svc: Move close processing to a single place
svc: Add xpo_accept transport function
svc: Remove unnecessary call to svc_sock_enqueue
svc: Move connection limit checking to its own function
svc: Add a generic transport svc_create_xprt function
svc: Change services to use new svc_create_xprt service
svc: Change sk_inuse to a kref
svc: Move sk_flags to the svc_xprt structure
svc: Move sk_server and sk_pool to svc_xprt
svc: Make close transport independent
svc: Move sk_reserved to svc_xprt
svc: Make the enqueue service transport neutral and export it.
svc: Make svc_send transport neutral
svc: Change svc_sock_received to svc_xprt_received and export it
svc: Move accept call to svc_xprt_received to common code
svc: Remove sk_lastrecv
svc: Move the authinfo cache to svc_xprt.
svc: Make deferral processing xprt independent
svc: Move the sockaddr information to svc_xprt
svc: Make svc_sock_release svc_xprt_release
svc: Make svc_recv transport neutral
svc: Make svc_age_temp_sockets svc_age_temp_transports
svc: Move create logic to common code
svc: Removing remaining references to rq_sock in rqstp
svc: Make svc_check_conn_limits xprt independent
svc: Move the xprt independent code to the svc_xprt.c file
svc: Add transport hdr size for defer/revisit
svc: Add /proc/sys/sunrpc/transport files
svc: Add svc API that queries for a transport instance
knfsd: Support adding transports by writing portlist file
svc: Add svc_xprt_names service to replace svc_sock_names
rdma: SVCRMDA Header File
rdma: SVCRDMA Transport Module
rdma: SVCRDMA Core Transport Services
rdma: SVCRDMA recvfrom
rdma: SVCRDMA sendto
rdma: ONCRPC RDMA protocol marshalling
rdma: makefile
Just some idea what we might be working on for 2.6.26, besides continued
bug-fixing and cleanup:
Work that we already have patches for and that I expect to be included
in whole or in 2.6.26:
- ipv6: Aurélien Charbon's patch to add ipv6 support to the
server's export interface is ready. I'm not clear what else
remains for full ipv6 support.
- failover and migration: Wendy Cheng's patches appear to be in
good shape, so I expect them or something with equivalent
functionality to be in 2.6.26.
- gss callbacks: We have patches to add support for rpcsec_gss
on NFSv4's callback channel (allowing us to support
delegations on kerberos mounts), but they've been put on hold
pending improvements to the client's gssd upcall. I hope to
get back to that work in the next few weeks.
Also in progress:
- spkm3 and future gss mechanisms may generate context
initiation rpc's that are very large. Olga Kornievskaia and I
have been working on fixing the server gssd interfaces to
permit this.
- There are some mismatches between the semantics required for
nfsv4 delegations and what Linux's lease subsystem provides
us. David Richter and I have done a little work on this. We
need to start submitting it.
Three items I identified previously as issues I'd like fixed before we
removed the dependency of CONFIG_NFSD_V4 on CONFIG_EXPERIMENTAL:
http://linux-nfs.org/pipermail/nfsv4/2006-December/005497.html
- export paths consistent between v2/v3/v4: We have some code
that fixes this entirely in userspace. That approach doesn't
provide stable filehandles in the NFSv4 pseudofilesystem, and
there seems to be a general sentiment that it's overly
complicated. It has the one advantage that we don't have to
commit to it, since it uses only existing kernel interfaces.
So I think we're probably going to apply that to nfs-utils as
a stopgap measure and start work on fixing this in the kernel
at the same time....
- reboot recovery: there have been complaints about the
server-side nfsv4 reboot recovery code for a while, we've had
code that tries to fix it for a while, and it just hasn't
happened. I'm hoping we can finally get this ready for
2.6.26.
- export security: this was finished in 2.6.23; we now support
export options like sec=krb5:krb5i:krb5p, which have a few
advantages over the special gss/krb5 client names. This could
be better documented, though.
I've probably left a lot out. Let me know of ongoing projects and
todo's that I've forgotten....
--b.
fre, 25 01 2008 kl. 18:15 -0500, skrev J. Bruce Fields:
> Below is a summary of the nfs server patches which I expect to submit
> (any day now) for 2.6.25; please let me know if you notice anything
> wrong or missing. I'll follow up to [email protected] with the
> full list of patches.
There are at least two patches that don't compile and break
bisectability. The errors should be obvious, just checkout and make
6f310df and 23b4e2e (patch no. 29 and 65 from the posting of the patch
set on linux-nfs).
[...]
> J. Bruce Fields (34):
> nfsd4: kill unnecessary same_name() in setclientid_confirm
diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
index 508b256..d026c24 100644
--- a/fs/nfsd/nfs4state.c
+++ b/fs/nfsd/nfs4state.c
@@ -858,8 +857,7 @@ nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
}
} else if ((conf && !unconf) ||
((conf && unconf) &&
- (!same_verf(&conf->cl_verifier, &unconf->cl_verifier) ||
- !same_name(conf->cl_recdir, unconf->cl_recdir)))) {
+ !same_verf(&conf->cl_verifier, &unconf->cl_verifier)) {
missing ending parenthesis ---------------------------------------^
[...]
> Tom Tucker (47):
> svc: Make close transport independent
diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
index f86538e..6dffdda 100644
--- a/net/sunrpc/svcsock.c
+++ b/net/sunrpc/svcsock.c
@@ -1912,66 +1911,63 @@ static void svc_sock_free(struct svc_xprt *xprt)
}
/*
- * Remove a dead socket
+ * Remove a dead transport
*/
-static void
-svc_delete_socket(struct svc_sock *svsk)
+static void svc_delete_xprt(struct svc_xprt *xprt)
{
- struct svc_serv *serv;
- struct sock *sk;
-
- dprintk("svc: svc_delete_socket(%p)\n", svsk);
-
- serv = svsk->sk_xprt.xpt_server;
- sk = svsk->sk_sk;
+ struct svc_serv *serv = xprt->xpt_server;
- svsk->sk_xprt.xpt_ops->xpo_detach(&svsk->sk_xprt);
+ dprintk("svc: svc_delete_xprt(%p)\n", xprt);
+ xprt->xpt_ops->xpo_detach(xprt);
spin_lock_bh(&serv->sv_lock);
-
- if (!test_and_set_bit(XPT_DETACHED, &svsk->sk_xprt.xpt_flags))
- list_del_init(&svsk->sk_list);
+ if (!test_and_set_bit(XPT_DETACHED, &xprt->xpt_flags))
+ list_del_init(&xprt->xpt_list);
/*
- * We used to delete the svc_sock from whichever list
- * it's sk_ready node was on, but we don't actually
+ * We used to delete the transport from whichever list
+ * it's sk_xprt.xpt_ready node was on, but we don't actually
* need to. This is because the only time we're called
* while still attached to a queue, the queue itself
* is about to be destroyed (in svc_destroy).
*/
- if (!test_and_set_bit(XPT_DEAD, &svsk->sk_xprt.xpt_flags)) {
- BUG_ON(atomic_read(&svsk->sk_xprt.xpt_ref.refcount) <
2);
- if (test_bit(XPT_TEMP, &svsk->sk_xprt.xpt_flags))
+ if (!test_and_set_bit(XPT_DEAD, &xprt->xpt_flags)) {
+ BUG_ON(atomic_read(&xprt->xpt_ref.refcount) < 2);
+ if (test_bit(XPT_TEMP, &xprt->xpt_flags))
serv->sv_tmpcnt--;
svc_xprt_put(&svsk->sk_xprt);
should have been
svc_xprt_put(&xprt);
There might be more errors in the next 15 patches or so; I didn't test
them with this error corrected.
Simon Holm Thøgersen
On Sun, Jan 27, 2008 at 09:42:23PM +0100, Simon Holm Thøgersen wrote:
>
> fre, 25 01 2008 kl. 18:15 -0500, skrev J. Bruce Fields:
> > Below is a summary of the nfs server patches which I expect to submit
> > (any day now) for 2.6.25; please let me know if you notice anything
> > wrong or missing. I'll follow up to [email protected] with the
> > full list of patches.
>
> There are at least two patches that don't compile and break
> bisectability. The errors should be obvious, just checkout and make
> 6f310df and 23b4e2e (patch no. 29 and 65 from the posting of the patch
> set on linux-nfs).
Whoops, thanks for finding both of those.
>
> [...]
> > J. Bruce Fields (34):
> > nfsd4: kill unnecessary same_name() in setclientid_confirm
> diff --git a/fs/nfsd/nfs4state.c b/fs/nfsd/nfs4state.c
> index 508b256..d026c24 100644
> --- a/fs/nfsd/nfs4state.c
> +++ b/fs/nfsd/nfs4state.c
> @@ -858,8 +857,7 @@ nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
> }
> } else if ((conf && !unconf) ||
> ((conf && unconf) &&
> - (!same_verf(&conf->cl_verifier, &unconf->cl_verifier) ||
> - !same_name(conf->cl_recdir, unconf->cl_recdir)))) {
> + !same_verf(&conf->cl_verifier, &unconf->cl_verifier)) {
> missing ending parenthesis ---------------------------------------^
Yep; fixed.
>
>
> [...]
> > Tom Tucker (47):
> > svc: Make close transport independent
> diff --git a/net/sunrpc/svcsock.c b/net/sunrpc/svcsock.c
> index f86538e..6dffdda 100644
> --- a/net/sunrpc/svcsock.c
> +++ b/net/sunrpc/svcsock.c
> @@ -1912,66 +1911,63 @@ static void svc_sock_free(struct svc_xprt *xprt)
> }
>
> /*
> - * Remove a dead socket
> + * Remove a dead transport
> */
> -static void
> -svc_delete_socket(struct svc_sock *svsk)
> +static void svc_delete_xprt(struct svc_xprt *xprt)
> {
> - struct svc_serv *serv;
> - struct sock *sk;
> -
> - dprintk("svc: svc_delete_socket(%p)\n", svsk);
> -
> - serv = svsk->sk_xprt.xpt_server;
> - sk = svsk->sk_sk;
> + struct svc_serv *serv = xprt->xpt_server;
>
> - svsk->sk_xprt.xpt_ops->xpo_detach(&svsk->sk_xprt);
> + dprintk("svc: svc_delete_xprt(%p)\n", xprt);
> + xprt->xpt_ops->xpo_detach(xprt);
>
> spin_lock_bh(&serv->sv_lock);
> -
> - if (!test_and_set_bit(XPT_DETACHED, &svsk->sk_xprt.xpt_flags))
> - list_del_init(&svsk->sk_list);
> + if (!test_and_set_bit(XPT_DETACHED, &xprt->xpt_flags))
> + list_del_init(&xprt->xpt_list);
> /*
> - * We used to delete the svc_sock from whichever list
> - * it's sk_ready node was on, but we don't actually
> + * We used to delete the transport from whichever list
> + * it's sk_xprt.xpt_ready node was on, but we don't actually
> * need to. This is because the only time we're called
> * while still attached to a queue, the queue itself
> * is about to be destroyed (in svc_destroy).
> */
> - if (!test_and_set_bit(XPT_DEAD, &svsk->sk_xprt.xpt_flags)) {
> - BUG_ON(atomic_read(&svsk->sk_xprt.xpt_ref.refcount) <
> 2);
> - if (test_bit(XPT_TEMP, &svsk->sk_xprt.xpt_flags))
> + if (!test_and_set_bit(XPT_DEAD, &xprt->xpt_flags)) {
> + BUG_ON(atomic_read(&xprt->xpt_ref.refcount) < 2);
> + if (test_bit(XPT_TEMP, &xprt->xpt_flags))
> serv->sv_tmpcnt--;
> svc_xprt_put(&svsk->sk_xprt);
> should have been
> svc_xprt_put(&xprt);
Actually looks like it should be svc_xprt_put(xprt); fixed.
> There might be more errors in the next 15 patches or so; I didn't test
> them with this error corrected.
OK. I've got a machine running
for c in $(git rev-list origin..linux-nfs/nfs-server-stable^); do
git checkout $c
make -s -j4
done
now just to make sure.... Thanks again!
--b.
On Sun, Jan 27, 2008 at 05:10:17PM -0500, bfields wrote:
> OK. I've got a machine running
>
> for c in $(git rev-list origin..linux-nfs/nfs-server-stable^); do
> git checkout $c
> make -s -j4
> done
>
> now just to make sure.... Thanks again!
Pfft, that should have been:
for c in $(git rev-list --reverse origin..linux-nfs/nfs-server-stable^); do
git checkout $c
make -s -j4
done
The results are a lot easier to understand if you do the compiles
in chronological order!
Anyway, that found one more compile failure, now fixed. Hopefully
that's the last. (Probably I did an incomplete job of reordering one of
the patches the other week....)
--b.