08 Feb, 2020
2 commits
-
Pull nfsd updates from Bruce Fields:
"Highlights:- Server-to-server copy code from Olga.
To use it, client and both servers must have support, the target
server must be able to access the source server over NFSv4.2, and
the target server must have the inter_copy_offload_enable module
parameter set.- Improvements and bugfixes for the new filehandle cache, especially
in the container case, from Trond- Also from Trond, better reporting of write errors.
- Y2038 work from Arnd"
* tag 'nfsd-5.6' of git://linux-nfs.org/~bfields/linux: (55 commits)
sunrpc: expiry_time should be seconds not timeval
nfsd: make nfsd_filecache_wq variable static
nfsd4: fix double free in nfsd4_do_async_copy()
nfsd: convert file cache to use over/underflow safe refcount
nfsd: Define the file access mode enum for tracing
nfsd: Fix a perf warning
nfsd: Ensure sampling of the write verifier is atomic with the write
nfsd: Ensure sampling of the commit verifier is atomic with the commit
sunrpc: clean up cache entry add/remove from hashtable
sunrpc: Fix potential leaks in sunrpc_cache_unhash()
nfsd: Ensure exclusion between CLONE and WRITE errors
nfsd: Pass the nfsd_file as arguments to nfsd4_clone_file_range()
nfsd: Update the boot verifier on stable writes too.
nfsd: Fix stable writes
nfsd: Allow nfsd_vfs_write() to take the nfsd_file as an argument
nfsd: Fix a soft lockup race in nfsd_file_mark_find_or_create()
nfsd: Reduce the number of calls to nfsd_file_gc()
nfsd: Schedule the laundrette regularly irrespective of file errors
nfsd: Remove unused constant NFSD_FILE_LRU_RESCAN
nfsd: Containerise filecache laundrette
... -
Fix sparse warning:
fs/nfsd/filecache.c:55:25: warning:
symbol 'nfsd_filecache_wq' was not declared. Should it be static?Reported-by: Hulk Robot
Signed-off-by: Chen Zhou
Signed-off-by: J. Bruce Fields
07 Feb, 2020
4 commits
-
This frees "copy->nf_src" before and again after the goto.
Fixes: ce0887ac96d3 ("NFSD add nfs4 inter ssc to nfsd4_copy")
Signed-off-by: Dan Carpenter
Reviewed-by: Olga Kornievskaia
Signed-off-by: J. Bruce Fields -
Use the 'refcount_t' type instead of 'atomic_t' for improved
refcounting safety.Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields -
Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields -
perf does not know how to deal with a __builtin_bswap32() call, and
complains. All other functions just store the xid etc in host endian
form, so let's do that in the tracepoint for nfsd_file_acquire too.Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields
04 Feb, 2020
1 commit
-
The most notable change is DEFINE_SHOW_ATTRIBUTE macro split in
seq_file.h.Conversion rule is:
llseek => proc_lseek
unlocked_ioctl => proc_ioctlxxx => proc_xxx
delete ".owner = THIS_MODULE" line
[akpm@linux-foundation.org: fix drivers/isdn/capi/kcapi_proc.c]
[sfr@canb.auug.org.au: fix kernel/sched/psi.c]
Link: http://lkml.kernel.org/r/20200122180545.36222f50@canb.auug.org.au
Link: http://lkml.kernel.org/r/20191225172546.GB13378@avx2
Signed-off-by: Alexey Dobriyan
Signed-off-by: Stephen Rothwell
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds
23 Jan, 2020
14 commits
-
When doing an unstable write, we need to ensure that we sample the
write verifier before releasing the lock, and allowing a commit to
the same file to proceed.Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields -
When we have a successful commit, ensure we sample the commit verifier
before releasing the lock.Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields -
Ensure that we can distinguish between synchronous CLONE and
WRITE errors, and that we can assign them correctly.Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields -
Needed in order to fix exclusion w.r.t. writes.
Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields -
We don't know if the error returned by the fsync() call is
exclusive to the data written by the stable write, so play it
safe.Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields -
Strictly speaking, a stable write error needs to reflect the
write + the commit of that write (and only that write). To
ensure that we don't pick up the write errors from other
writebacks, add a rw_semaphore to provide exclusion.Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields -
Needed in order to fix stable writes.
Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields -
If nfsd_file_mark_find_or_create() keeps winning the race for the
nfsd_file_fsnotify_group->mark_mutex against nfsd_file_mark_put()
then it can soft lock up, since fsnotify_add_inode_mark() ends
up always finding an existing entry.Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields -
Don't call nfsd_file_gc() on every put of the reference in nfsd_file_put().
Instead, do it only when we're expecting the refcount to go to 1.Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields -
Emsure we schedule the laundrette even if the struct file is carrying
file errors.Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields -
Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields -
Ensure that if the filecache laundrette gets stuck, it only affects
the knfsd instances of one container.The notifier callbacks can be called from various contexts so avoid
using synchonous filesystem operations that might deadlock.Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields -
Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields -
If the lookup keeps finding a nfsd_file with an unhashed open file,
then retry once only.Signed-off-by: Trond Myklebust
Cc: stable@vger.kernel.org
Fixes: 65294c1f2c5e "nfsd: add a new struct file caching facility to nfsd"
Signed-off-by: J. Bruce Fields
04 Jan, 2020
3 commits
-
Fixes coccicheck warning:
fs/nfsd/nfssvc.c:394:2-14: WARNING: Assignment of 0/1 to bool variable
fs/nfsd/nfssvc.c:407:2-14: WARNING: Assignment of 0/1 to bool variable
fs/nfsd/nfssvc.c:422:2-14: WARNING: Assignment of 0/1 to bool variableReported-by: Hulk Robot
Signed-off-by: zhengbin
Signed-off-by: J. Bruce Fields -
Fixes coccicheck warning:
fs/nfsd/nfs4proc.c:235:1-18: WARNING: Assignment of 0/1 to bool variable
fs/nfsd/nfs4proc.c:368:1-17: WARNING: Assignment of 0/1 to bool variableReported-by: Hulk Robot
Signed-off-by: zhengbin
Signed-off-by: J. Bruce Fields -
Fixes coccicheck warning:
fs/nfsd/vfs.c:1389:5-13: WARNING: Assignment of 0/1 to bool variable
fs/nfsd/vfs.c:1398:5-13: WARNING: Assignment of 0/1 to bool variable
fs/nfsd/vfs.c:1415:2-10: WARNING: Assignment of 0/1 to bool variableReported-by: Hulk Robot
Signed-off-by: zhengbin
Signed-off-by: J. Bruce Fields
20 Dec, 2019
15 commits
-
The function was removed a long time ago, but the declaration
and a dummy implementation are still there, referencing the
deprecated time_t type.Remove both.
Fixes: f958a1320ff7 ("nfsd4: remove unnecessary lease-setting function")
Signed-off-by: Arnd Bergmann
Signed-off-by: J. Bruce Fields -
gen_confirm() generates a unique identifier based on the current
time. This overflows in year 2038, but that is harmless since it
generally does not lead to duplicates, as long as the time has
been initialized by a real-time clock or NTP.Using ktime_get_boottime_seconds() or ktime_get_seconds() would
avoid the overflow, but it would be more likely to result in
non-unique numbers.Signed-off-by: Arnd Bergmann
Signed-off-by: J. Bruce Fields -
A couple of time_t variables are only used to track the state of the
lease time and its expiration. The code correctly uses the 'time_after()'
macro to make this work on 32-bit architectures even beyond year 2038,
but the get_seconds() function and the time_t type itself are deprecated
as they behave inconsistently between 32-bit and 64-bit architectures
and often lead to code that is not y2038 safe.As a minor issue, using get_seconds() leads to problems with concurrent
settimeofday() or clock_settime() calls, in the worst case timeout never
triggering after the time has been set backwards.Change nfsd to use time64_t and ktime_get_boottime_seconds() here. This
is clearly excessive, as boottime by itself means we never go beyond 32
bits, but it does mean we handle this correctly and consistently without
having to worry about corner cases and should be no more expensive than
the previous implementation on 64-bit architectures.The max_cb_time() function gets changed in order to avoid an expensive
64-bit division operation, but as the lease time is at most one hour,
there is no change in behavior.Also do the same for server-to-server copy expiration time.
Signed-off-by: Arnd Bergmann
[bfields@redhat.com: fix up copy expiration]
Signed-off-by: J. Bruce Fields -
The nfsd4_blocked_lock->nbl_time timestamp is recorded in jiffies,
but then compared to a CLOCK_REALTIME timestamp later on, which makes
no sense.For consistency with the other timestamps, change this to use a time_t.
This is a change in behavior, which may cause regressions, but the
current code is not sensible. On a system with CONFIG_HZ=1000,
the 'time_after((unsigned long)nbl->nbl_time, (unsigned long)cutoff))'
check is false for roughly the first 18 days of uptime and then true
for the next 49 days.Fixes: 7919d0a27f1e ("nfsd: add a LRU list for blocked locks")
Signed-off-by: Arnd Bergmann
Signed-off-by: J. Bruce Fields -
The nfsd4_cb_layout_done() function takes a 'time_t' value,
multiplied by NSEC_PER_SEC*2 to get a nanosecond value.This works fine on 64-bit architectures, but on 32-bit, any
value over 1 second results in a signed integer overflow
with unexpected results.Cast one input to a 64-bit type in order to produce the
same result that we have on 64-bit architectures, regarless
of the type of nfsd4_lease.Fixes: 6b9b21073d3b ("nfsd: give up on CB_LAYOUTRECALLs after two lease periods")
Signed-off-by: Arnd Bergmann
Signed-off-by: J. Bruce Fields -
Change to time64_t and ktime_get_real_seconds() to make the
logic work correctly on 32-bit architectures beyond 2038.Signed-off-by: Arnd Bergmann
Signed-off-by: J. Bruce Fields -
Guardtime handling in nfs3 differs between 32-bit and 64-bit
architectures, and uses the deprecated time_t type.Change it to using time64_t, which behaves the same way on
64-bit and 32-bit architectures, treating the number as an
unsigned 32-bit entity with a range of year 1970 to 2106
consistently, and avoiding the y2038 overflow.Signed-off-by: Arnd Bergmann
Signed-off-by: J. Bruce Fields -
The local boot time variable gets truncated to time_t at the moment,
which can lead to slightly odd behavior on 32-bit architectures.Use ktime_get_real_seconds() instead of get_seconds() to always
get a 64-bit result, and keep it that way wherever possible.It still gets truncated in a few places:
- When assigning to cl_clientid.cl_boot, this is already documented
and is only used as a unique identifier.- In clients_still_reclaiming(), the truncation is to 'unsigned long'
in order to use the 'time_before() helper.Signed-off-by: Arnd Bergmann
Signed-off-by: J. Bruce Fields -
The values in encode_time_delta are always small and don't
overflow the range of 'struct timespec', so changing it has
no effect.Change it to timespec64 as a prerequisite for removing the
timespec definition later.Signed-off-by: Arnd Bergmann
Signed-off-by: J. Bruce Fields -
The decode_time3 function behaves differently on 32-bit
and 64-bit architectures: on the former, a 32-bit timestamp
gets converted into an signed number and then into a timestamp
between 1902 and 2038, while on the latter it is interpreted
as unsigned in the range 1970-2106.Change all the remaining 'timespec' in nfsd to 'timespec64'
to make the behavior the same, and use the current interpretation
of the dominant 64-bit architectures.Signed-off-by: Arnd Bergmann
Signed-off-by: J. Bruce Fields -
The nii_time field gets truncated to 'time_t' on 32-bit architectures
before printing.Remove the use of 'struct timespec' to product the correct output
beyond 2038.Signed-off-by: Arnd Bergmann
Signed-off-by: J. Bruce Fields -
The delegation logic in nfsd uses the somewhat inefficient
seconds_since_boot() function to record time intervals.Signed-off-by: Arnd Bergmann
Signed-off-by: J. Bruce Fields -
The replay variable is set in the only caller of nfsd4_encode_replay.
The assertion is unnecessary and the patch removes this check.Signed-off-by: Aditya Pakki
Signed-off-by: J. Bruce Fields -
vfs_clone_file_range() can modify the metadata on the source file too,
so we need to commit that to stable storage as well.Reported-by: Dave Chinner
Signed-off-by: Trond Myklebust
Acked-by: Dave Chinner
Signed-off-by: J. Bruce Fields -
Fixes coccicheck warning:
fs/nfsd/nfs4state.c:3376:2-3: Unneeded semicolon
Reported-by: Hulk Robot
Signed-off-by: zhengbin
Signed-off-by: J. Bruce Fields
18 Dec, 2019
1 commit
-
We must allow for the fact that iov_iter_write() could have returned
a short write (e.g. if there was an ENOSPC issue).Fixes: d890be159a71 "nfsd: Add I/O trace points in the NFSv4 write path"
Signed-off-by: Trond Myklebust
Signed-off-by: J. Bruce Fields