Olga Kornievskaia says: "I ran into this oops in the nfsd (below)
(4.10-rc3 kernel). To trigger this I had a client (unsuccessfully) try
to mount the server with krb5 where the server doesn't have the
rpcsec_gss_krb5 module built."
The problem is that rsci.cred is copied from a svc_cred structure that
gss_proxy didn't properly initialize. Fix that.
[120408.542387] general protection fault: 0000 [#1] SMP
...
[120408.565724] CPU: 0 PID: 3601 Comm: nfsd Not tainted 4.10.0-rc3+ #16
[120408.567037] Hardware name: VMware, Inc. VMware Virtual =
Platform/440BX Desktop Reference Platform, BIOS 6.00 07/02/2015
[120408.569225] task: ffff8800776f95c0 task.stack: ffffc90003d58000
[120408.570483] RIP: 0010:gss_mech_put+0xb/0x20 [auth_rpcgss]
...
[120408.584946] ? rsc_free+0x55/0x90 [auth_rpcgss]
[120408.585901] gss_proxy_save_rsc+0xb2/0x2a0 [auth_rpcgss]
[120408.587017] svcauth_gss_proxy_init+0x3cc/0x520 [auth_rpcgss]
[120408.588257] ? __enqueue_entity+0x6c/0x70
[120408.589101] svcauth_gss_accept+0x391/0xb90 [auth_rpcgss]
[120408.590212] ? try_to_wake_up+0x4a/0x360
[120408.591036] ? wake_up_process+0x15/0x20
[120408.592093] ? svc_xprt_do_enqueue+0x12e/0x2d0 [sunrpc]
[120408.593177] svc_authenticate+0xe1/0x100 [sunrpc]
[120408.594168] svc_process_common+0x203/0x710 [sunrpc]
[120408.595220] svc_process+0x105/0x1c0 [sunrpc]
[120408.596278] nfsd+0xe9/0x160 [nfsd]
[120408.597060] kthread+0x101/0x140
[120408.597734] ? nfsd_destroy+0x60/0x60 [nfsd]
[120408.598626] ? kthread_park+0x90/0x90
[120408.599448] ret_from_fork+0x22/0x30
Fixes: 1d658336b0 "SUNRPC: Add RPC based upcall mechanism for RPCGSS auth"
Cc: stable@vger.kernel.org
Cc: Simo Sorce <simo@redhat.com>
Reported-by: Olga Kornievskaia <kolga@netapp.com>
Tested-by: Olga Kornievskaia <kolga@netapp.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
In rdma_read_chunk_frmr() when ib_post_send() fails, the error code path
invokes ib_dma_unmap_sg() to unmap the sg list. It then invokes
svc_rdma_put_frmr() which in turn tries to unmap the same sg list through
ib_dma_unmap_sg() again. This second unmap is invalid and could lead to
problems when the iova being unmapped is subsequently reused. Remove
the call to unmap in rdma_read_chunk_frmr() and let svc_rdma_put_frmr()
handle it.
Fixes: 412a15c0fe ("svcrdma: Port to new memory registration API")
Cc: stable@vger.kernel.org
Signed-off-by: Sriharsha Basavapatna <sriharsha.basavapatna@broadcom.com>
Reviewed-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Yuval Shaia <yuval.shaia@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
The inet6addr_chain is an atomic notifier chain, so we can't call
anything that might sleep (like lock_sock)... instead of closing the
socket from svc_age_temp_xprts_now (which is called by the notifier
function), just have the rpc service threads do it instead.
Cc: stable@vger.kernel.org
Fixes: c3d4879e01 "sunrpc: Add a function to close..."
Signed-off-by: Scott Mayhew <smayhew@redhat.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Context expiry times are in units of seconds since boot, not unix time.
The use of get_seconds() here therefore sets the expiry time decades in
the future. This prevents timely freeing of contexts destroyed by
client RPC_GSS_PROC_DESTROY requests. We'd still free them eventually
(when the module is unloaded or the container shut down), but a lot of
contexts could pile up before then.
Cc: stable@vger.kernel.org
Fixes: c5b29f885a "sunrpc: use seconds since boot in expiry cache"
Reported-by: Andy Adamson <andros@netapp.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
ktime is a union because the initial implementation stored the time in
scalar nanoseconds on 64 bit machine and in a endianess optimized timespec
variant for 32bit machines. The Y2038 cleanup removed the timespec variant
and switched everything to scalar nanoseconds. The union remained, but
become completely pointless.
Get rid of the union and just keep ktime_t as simple typedef of type s64.
The conversion was done with coccinelle and some manual mopping up.
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Peter Zijlstra <peterz@infradead.org>
This was entirely automated, using the script by Al:
PATT='^[[:blank:]]*#[[:blank:]]*include[[:blank:]]*<asm/uaccess.h>'
sed -i -e "s!$PATT!#include <linux/uaccess.h>!" \
$(git grep -l "$PATT"|grep -v ^include/linux/uaccess.h)
to do the replacement at the end of the merge window.
Requested-by: Al Viro <viro@zeniv.linux.org.uk>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pull nfsd updates from Bruce Fields:
"The one new feature is support for a new NFSv4.2 mode_umask attribute
that makes ACL inheritance a little more useful in environments that
default to restrictive umasks. Requires client-side support, also on
its way for 4.10.
Other than that, miscellaneous smaller fixes and cleanup, especially
to the server rdma code"
[ The client side of the umask attribute was merged yesterday ]
* tag 'nfsd-4.10' of git://linux-nfs.org/~bfields/linux:
nfsd: add support for the umask attribute
sunrpc: use DEFINE_SPINLOCK()
svcrdma: Further clean-up of svc_rdma_get_inv_rkey()
svcrdma: Break up dprintk format in svc_rdma_accept()
svcrdma: Remove unused variable in rdma_copy_tail()
svcrdma: Remove unused variables in xprt_rdma_bc_allocate()
svcrdma: Remove svc_rdma_op_ctxt::wc_status
svcrdma: Remove DMA map accounting
svcrdma: Remove BH-disabled spin locking in svc_rdma_send()
svcrdma: Renovate sendto chunk list parsing
svcauth_gss: Close connection when dropping an incoming message
svcrdma: Clear xpt_bc_xps in xprt_setup_rdma_bc() error exit arm
nfsd: constify reply_cache_stats_operations structure
nfsd: update workqueue creation
sunrpc: GFP_KERNEL should be GFP_NOFS in crypto code
nfsd: catch errors in decode_fattr earlier
nfsd: clean up supported attribute handling
nfsd: fix error handling for clients that fail to return the layout
nfsd: more robust allocation failure handling in nfsd_reply_cache_init
Pull NFS client updates from Trond Myklebust:
"Highlights include:
Stable bugfixes:
- Fix a pnfs deadlock between read resends and layoutreturn
- Don't invalidate the layout stateid while a layout return is
outstanding
- Don't schedule a layoutreturn if the layout stateid is marked as
invalid
- On a pNFS error, do not send LAYOUTGET until the LAYOUTRETURN is
complete
- SUNRPC: fix refcounting problems with auth_gss messages.
Features:
- Add client support for the NFSv4 umask attribute.
- NFSv4: Correct support for flock() stateids.
- Add a LAYOUTRETURN operation to CLOSE and DELEGRETURN when
return-on-close is specified
- Allow the pNFS/flexfiles layoutstat information to piggyback on
LAYOUTRETURN
- Optimise away redundant GETATTR calls when doing state recovery
and/or when not required by cache revalidation rules or
close-to-open cache consistency.
- Attribute cache improvements
- RPC/RDMA support for SG_GAP devices
Bugfixes:
- NFS: Fix performance regressions in readdir
- pNFS/flexfiles: Fix a deadlock on LAYOUTGET
- NFSv4: Add missing nfs_put_lock_context()
- NFSv4.1: Fix regression in callback retry handling
- Fix false positive NFSv4.0 trunking detection.
- pNFS/flexfiles: Only send layoutstats updates for mirrors that were
updated
- Various layout stateid related bugfixes
- RPC/RDMA bugfixes"
* tag 'nfs-for-4.10-1' of git://git.linux-nfs.org/projects/trondmy/linux-nfs: (82 commits)
SUNRPC: fix refcounting problems with auth_gss messages.
nfs: add support for the umask attribute
pNFS/flexfiles: Ensure we have enough buffer for layoutreturn
pNFS/flexfiles: Remove a redundant parameter in ff_layout_encode_ioerr()
pNFS/flexfiles: Fix a deadlock on LAYOUTGET
pNFS: Layoutreturn must free the layout after the layout-private data
pNFS/flexfiles: Fix ff_layout_add_ds_error_locked()
NFSv4: Add missing nfs_put_lock_context()
pNFS: Release NFS_LAYOUT_RETURN when invalidating the layout stateid
NFSv4.1: Don't schedule lease recovery in nfs4_schedule_session_recovery()
NFSv4.1: Handle NFS4ERR_BADSESSION/NFS4ERR_DEADSESSION replies to OP_SEQUENCE
NFS: Only look at the change attribute cache state in nfs_check_verifier
NFS: Fix incorrect size revalidation when holding a delegation
NFS: Fix incorrect mapping revalidation when holding a delegation
pNFS/flexfiles: Support sending layoutstats in layoutreturn
pNFS/flexfiles: Minor refactoring before adding iostats to layoutreturn
NFS: Fix up read of mirror stats
pNFS/flexfiles: Clean up layoutstats
pNFS/flexfiles: Refactor encoding of the layoutreturn payload
pNFS: Add a layoutreturn callback to performa layout-private setup
...
There are two problems with refcounting of auth_gss messages.
First, the reference on the pipe->pipe list (taken by a call
to rpc_queue_upcall()) is not counted. It seems to be
assumed that a message in pipe->pipe will always also be in
pipe->in_downcall, where it is correctly reference counted.
However there is no guaranty of this. I have a report of a
NULL dereferences in rpc_pipe_read() which suggests a msg
that has been freed is still on the pipe->pipe list.
One way I imagine this might happen is:
- message is queued for uid=U and auth->service=S1
- rpc.gssd reads this message and starts processing.
This removes the message from pipe->pipe
- message is queued for uid=U and auth->service=S2
- rpc.gssd replies to the first message. gss_pipe_downcall()
calls __gss_find_upcall(pipe, U, NULL) and it finds the
*second* message, as new messages are placed at the head
of ->in_downcall, and the service type is not checked.
- This second message is removed from ->in_downcall and freed
by gss_release_msg() (even though it is still on pipe->pipe)
- rpc.gssd tries to read another message, and dereferences a pointer
to this message that has just been freed.
I fix this by incrementing the reference count before calling
rpc_queue_upcall(), and decrementing it if that fails, or normally in
gss_pipe_destroy_msg().
It seems strange that the reply doesn't target the message more
precisely, but I don't know all the details. In any case, I think the
reference counting irregularity became a measureable bug when the
extra arg was added to __gss_find_upcall(), hence the Fixes: line
below.
The second problem is that if rpc_queue_upcall() fails, the new
message is not freed. gss_alloc_msg() set the ->count to 1,
gss_add_msg() increments this to 2, gss_unhash_msg() decrements to 1,
then the pointer is discarded so the memory never gets freed.
Fixes: 9130b8dbc6 ("SUNRPC: allow for upcalls for same uid but different gss service")
Cc: stable@vger.kernel.org
Link: https://bugzilla.opensuse.org/show_bug.cgi?id=1011250
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
xs_connect() contains an exponential backoff mechanism so the repeated
connection attempts are delayed by longer and longer amounts.
This is appropriate when the connection failed due to a timeout, but
it not appropriate when a definitive "no" answer is received. In such
cases, call_connect_status() imposes a minimum 3-second back-off, so
not having the exponetial back-off will never result in immediate
retries.
The current situation is a problem when the NFS server tries to
register with rpcbind but rpcbind isn't running. All connection
attempts are made on the same "xprt" and as the connection is never
"closed", the exponential back delays successive attempts to register,
or de-register, different protocols. This results in a multi-minute
delay with no benefit.
So, when call_connect_status() receives a definitive "no", use
xprt_conditional_disconnect() to cancel the previous connection attempt.
This will set XPRT_CLOSE_WAIT so that xprt->ops->close() calls xs_close()
which resets the reestablish_timeout.
To ensure xprt_conditional_disconnect() does the right thing, we
ensure that rq_connect_cookie is set before a connection attempt, and
allow xprt_conditional_disconnect() to complete even when the
transport is not fully connected.
Signed-off-by: NeilBrown <neilb@suse.com>
Signed-off-by: Trond Myklebust <trond.myklebust@primarydata.com>
The current code results in:
Nov 7 14:50:19 klimt kernel: svcrdma: newxprt->sc_cm_id=ffff88085590c800,
newxprt->sc_pd=ffff880852a7ce00#012 cm_id->device=ffff88084dd20000,
sc_pd->device=ffff88084dd20000#012 cap.max_send_wr = 272#012
cap.max_recv_wr = 34#012 cap.max_send_sge = 32#012
cap.max_recv_sge = 32
Nov 7 14:50:19 klimt kernel: svcrdma: new connection ffff880855908000
accepted with the following attributes:#012 local_ip :
10.0.0.5#012 local_port#011 : 20049#012 remote_ip :
10.0.0.2#012 remote_port : 59909#012 max_sge : 32#012
max_sge_rd : 30#012 sq_depth : 272#012 max_requests :
32#012 ord : 16
Split up the output over multiple dprintks and take the opportunity
to fix the display of IPv6 addresses.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Clean up.
linux-2.6/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c: In function
‘rdma_copy_tail’:
linux-2.6/net/sunrpc/xprtrdma/svc_rdma_recvfrom.c:376:6: warning:
variable ‘ret’ set but not used [-Wunused-but-set-variable]
int ret;
^
Fixes: a97c331f9a ("svcrdma: Handle additional inline content")
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Clean up.
/linux-2.6/net/sunrpc/xprtrdma/svc_rdma_backchannel.c: In function
‘xprt_rdma_bc_allocate’:
linux-2.6/net/sunrpc/xprtrdma/svc_rdma_backchannel.c:169:23: warning:
variable ‘rdma’ set but not used [-Wunused-but-set-variable]
struct svcxprt_rdma *rdma;
^
Fixes: 5d252f90a8 ("svcrdma: Add class for RDMA backwards ...")
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Clean up: Completion status is already reported in the individual
completion handlers. Save a few bytes in struct svc_rdma_op_ctxt.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Clean up: sc_dma_used is not required for correct operation. It is
simply a debugging tool to report when svcrdma has leaked DMA maps.
However, manipulating an atomic has a measurable CPU cost, and DMA
map accounting specific to svcrdma will be meaningless once svcrdma
is converted to use the new generic r/w API.
A similar kind of debug accounting can be done simply by enabling
the IOMMU or by using CONFIG_DMA_API_DEBUG, CONFIG_IOMMU_DEBUG, and
CONFIG_IOMMU_LEAK.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
svcrdma's current SQ accounting algorithm takes sc_lock and disables
bottom-halves while posting all RDMA Read, Write, and Send WRs.
This is relatively heavyweight serialization. And note that Write and
Send are already fully serialized by the xpt_mutex.
Using a single atomic_t should be all that is necessary to guarantee
that ib_post_send() is called only when there is enough space on the
send queue. This is what the other RDMA-enabled storage targets do.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
The current sendto code appears to support clients that provide only
one of a Read list, a Write list, or a Reply chunk. My reading of
that code is that it doesn't support the following cases:
- Read list + Write list
- Read list + Reply chunk
- Write list + Reply chunk
- Read list + Write list + Reply chunk
The protocol allows more than one Read or Write chunk in those
lists. Some clients do send a Read list and Reply chunk
simultaneously. NFSv4 WRITE uses a Read list for the data payload,
and a Reply chunk because the GETATTR result in the reply can
contain a large object like an ACL.
Generalize one of the sendto code paths needed to support all of
the above cases, and attempt to ensure that only one pass is done
through the RPC Call's transport header to gather chunk list
information for building the reply.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
S5.3.3.1 of RFC 2203 requires that an incoming GSS-wrapped message
whose sequence number lies outside the current window is dropped.
The rationale is:
The reason for discarding requests silently is that the server
is unable to determine if the duplicate or out of range request
was due to a sequencing problem in the client, network, or the
operating system, or due to some quirk in routing, or a replay
attack by an intruder. Discarding the request allows the client
to recover after timing out, if indeed the duplication was
unintentional or well intended.
However, clients may rely on the server dropping the connection to
indicate that a retransmit is needed. Without a connection reset, a
client can wait forever without retransmitting, and the workload
just stops dead. I've reproduced this behavior by running xfstests
generic/323 on an NFSv4.0 mount with proto=rdma and sec=krb5i.
To address this issue, have the server close the connection when it
silently discards an incoming message due to a GSS sequence number
problem.
There are a few other places where the server will never reply.
Change those spots in a similar fashion.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Logic copied from xs_setup_bc_tcp().
Fixes: 39a9beab5a ('rpc: share one xps between all backchannels')
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: J. Bruce Fields <bfields@redhat.com>
Clean up: offset and handle should be zero-filled, just like in the
chunk encoders.
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
J. Bruce Fields
Kinglong Mee
Sriharsha Basavapatna
Scott Mayhew
Thomas Gleixner
Linus Torvalds
Trond Myklebust
NeilBrown
Fabian Frederick
Chuck Lever