According to the report from Jiro SEKIBA titled "regression in
2.6.37?" (Message-Id: <8739n8vs1f.wl%jir@sekiba.com>), on 2.6.37 and
later kernels, lscp command no longer displays "i" flag on checkpoints
that snapshot operations or garbage collection created.
This is a regression of nilfs2 checkpointing function, and it's
critical since it broke behavior of a part of nilfs2 applications.
For instance, snapshot manager of TimeBrowse gets to create
meaningless snapshots continuously; snapshot creation triggers another
checkpoint, but applications cannot distinguish whether the new
checkpoint contains meaningful changes or not without the i-flag.
This patch fixes the regression and brings that application behavior
back to normal.
Reported-by: Jiro SEKIBA <jir@unicus.jp>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Tested-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Tested-by: Jiro SEKIBA <jir@unicus.jp>
Cc: stable <stable@kernel.org> [2.6.37]
nilfs_dat_inode function was a wrapper to switch between normal dat
inode and gcdat, a clone of the dat inode for garbage collection.
This function got obsolete when the gcdat inode was removed, and now
we can access the dat inode directly from a nilfs object. So, we will
unfold the wrapper and remove it.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Nilfs does not allocate new blocks on disk until they are actually
written to. To implement fiemap, we need to deal with such blocks.
To allow successive fiemap patch to distinguish mapped but unallocated
regions, this marks buffer heads of those new blocks as delayed and
clears the flag after the blocks are written to disk.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Some functions using nilfs bmap routines can wrongly return invalid
argument error (i.e. -EINVAL) that bmap returns as an internal code
for btree corruption.
This fixes the issue by catching and converting the internal EINVAL to
EIO and calling nilfs_error function inside bmap routines.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
To help developers and applications gain visibility into writeback
behaviour this patch adds two counters to /proc/vmstat.
# grep nr_dirtied /proc/vmstat
nr_dirtied 3747
# grep nr_written /proc/vmstat
nr_written 3618
These entries allow user apps to understand writeback behaviour over time
and learn how it is impacting their performance. Currently there is no
way to inspect dirty and writeback speed over time. It's not possible for
nr_dirty/nr_writeback.
These entries are necessary to give visibility into writeback behaviour.
We have /proc/diskstats which lets us understand the io in the block
layer. We have blktrace for more in depth understanding. We have
e2fsprogs and debugsfs to give insight into the file systems behaviour,
but we don't offer our users the ability understand what writeback is
doing. There is no way to know how active it is over the whole system, if
it's falling behind or to quantify it's efforts. With these values
exported users can easily see how much data applications are sending
through writeback and also at what rates writeback is processing this
data. Comparing the rates of change between the two allow developers to
see when writeback is not able to keep up with incoming traffic and the
rate of dirty memory being sent to the IO back end. This allows folks to
understand their io workloads and track kernel issues. Non kernel
engineers at Google often use these counters to solve puzzling performance
problems.
Patch #4 adds a pernode vmstat file with nr_dirtied and nr_written
Patch #5 add writeback thresholds to /proc/vmstat
Currently these values are in debugfs. But they should be promoted to
/proc since they are useful for developers who are writing databases
and file servers and are not debugging the kernel.
The output is as below:
# grep threshold /proc/vmstat
nr_pages_dirty_threshold 409111
nr_pages_dirty_background_threshold 818223
This patch:
This allows code outside of the mm core to safely manipulate page
writeback state and not worry about the other accounting. Not using these
routines means that some code will lose track of the accounting and we get
bugs.
Modify nilfs2 to use interface.
Signed-off-by: Michael Rubin <mrubin@google.com>
Reviewed-by: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Reviewed-by: Wu Fengguang <fengguang.wu@intel.com>
Cc: KONISHI Ryusuke <konishi.ryusuke@lab.ntt.co.jp>
Cc: Jiro SEKIBA <jir@unicus.jp>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Jens Axboe <axboe@kernel.dk>
Cc: KOSAKI Motohiro <kosaki.motohiro@jp.fujitsu.com>
Cc: Nick Piggin <nickpiggin@yahoo.com.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
insert sparse annotations to fix following sparse warning.
fs/nilfs2/segment.c:2681:3: warning: context imbalance in 'nilfs_segctor_kill_thread' - unexpected unlock
nilfs_segctor_kill_thread is only called inside sc_state_lock lock.
sparse doesn't detect the context and warn "unexpected unlock".
__acquires/__releases pretend to lock/unlock the sc_state_lock for sparse.
Signed-off-by: Jiro SEKIBA <jir@unicus.jp>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Nilfs hasn't supported the freeze/thaw feature because it didn't work
due to the peculiar design that multiple super block instances could
be allocated for a device. This limitation was removed by the patch
"nilfs2: do not allocate multiple super block instances for a device".
So now this adds the freeze/thaw support to nilfs.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Nilfs object holds a back pointer to a writable super block instance
in nilfs->ns_writer, and this became eliminable since sb is now made
per device and all inodes have a valid pointer to it.
This deletes the ns_writer pointer and a reader/writer semaphore
protecting it.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This applies prepared rollback function and redirect function of
metadata file to DAT file, and eliminates GCDAT inode.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
During garbage collection (GC), DAT file, which converts virtual block
number to real block number, may return disk block number that is not
yet written to the device.
To avoid access to unwritten blocks, the current implementation stores
changes to the caches of GCDAT during GC and atomically commit the
changes into the DAT file after they are written to the device.
This patch, instead, adds a function that makes a copy of specified
buffer and stores it in nilfs_shadow_map, and a function to get the
backup copy as needed (nilfs_mdt_freeze_buffer and
nilfs_mdt_get_frozen_buffer respectively).
Before DAT changes block number in an entry block, it makes a copy and
redirect access to the buffer so that address conversion function
(i.e. nilfs_dat_translate) refers to the old address saved in the
copy.
This patch gives requisites for such redirection.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This rewrites functions using ifile so that they get ifile from
nilfs_root object, and will remove sbi->s_ifile. Some functions that
don't know the root object are extended to receive it from caller.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This uses inode hash function that vfs provides instead of the own
hash table for caching gc inodes. This finally removes the own inode
hash from nilfs.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
On-memory inode structures of nilfs have a member "i_cno" which stores
a checkpoint number related to the inode. For gc-inodes, this field
indicates version of data each gc-inode caches for GC. Log writer
temporarily uses "i_cno" to transfer the latest checkpoint number.
This stops the latter use and lets only gc-inodes use it.
The purpose of this patch is to allow the successive change use
"i_cno" for inode lookup.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Super blocks of nilfs are periodically overwritten in order to record
the recent log position. This shortens recovery time after unclean
unmount, but the current implementation performs the update even for a
few blocks of change. If the filesystem gets small changes slowly and
continually, super blocks may be updated excessively.
This moderates the issue by skipping update of log cursor if it does
not cross a segment boundary.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This will sync super blocks in turns instead of syncing duplicate
super blocks at the time. This will help searching valid super root
when super block is written into disk before log is written, which is
happen when barrier-less block devices are unmounted uncleanly. In
the situation, old super block likely points to valid log.
This patch introduces ns_sbwcount member to the nilfs object and adds
nilfs_sb_will_flip() function; ns_sbwcount counts how many times super
blocks write back to the disk. And, nilfs_sb_will_flip() decides
whether flipping required or not based on the count of ns_sbwcount to
sync super blocks asymmetrically.
The following functions are also changed:
- nilfs_prepare_super(): flips super blocks according to the
argument. The argument is calculated by nilfs_sb_will_flip()
function.
- nilfs_cleanup_super(): sets "clean" flag to both super blocks if
they point to the same checkpoint.
To update both of super block information, caller of
nilfs_commit_super must set the information on both super blocks.
Signed-off-by: Jiro SEKIBA <jir@unicus.jp>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This function checks validity of super block pointers.
If first super block is invalid, it will swap the super blocks.
The function should be called before any super block information updates.
Caller must obtain nilfs->ns_sem.
Signed-off-by: Jiro SEKIBA <jir@unicus.jp>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This removes macros to test segment summary flags and redefines a few
relevant macros with inline functions.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This kills the following sparse warnings:
fs/nilfs2/segment.c:567:28: warning: symbol 'nilfs_sc_file_ops' was not declared. Should it be static?
fs/nilfs2/segment.c:617:28: warning: symbol 'nilfs_sc_dat_ops' was not declared. Should it be static?
fs/nilfs2/segment.c:625:28: warning: symbol 'nilfs_sc_dsync_ops' was not declared. Should it be static?
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
In nilfs_segctor_thread(), timer is a local variable allocated on stack. Its
address can't be set to sci->sc_timer and passed in several procedures.
It works now by chance, just because other procedures are called by
nilfs_segctor_thread() directly or indirectly and the stack hasn't been
deallocated yet.
Signed-off-by: Li Hong <lihong.hi@gmail.com>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
There are only two lines of code in nilfs_segctor_init(). From a logic
design view, the first line 'sci->sc_seq_done = sci->sc_seq_request;'
should be put in nilfs_segctor_new(). Even in nilfs_segctor_new(),
this initialization is needless because sci is kzalloc-ed. So
nilfs_segctor_init() is only a wrap call to
nilfs_segctor_start_thread().
Signed-off-by: Li Hong <lihong.hi@gmail.com>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This adds a field to record the latest checkpoint number in the
nilfs_segment_summary structure. This will help to recover the latest
checkpoint number from logs on disk. This field is intended for
crucial cases in which super blocks have lost pointer to the latest
log.
Even though this will change the disk format, both backward and
forward compatibility is preserved by a size field prepared in the
segment summary header.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This cleanup patch gives several improvements:
- Moving all kmem_cache_{create_destroy} calls into one place, which removes
some small function calls, cleans up error check code and clarify the logic.
- Mark all initial code in __init section.
- Remove some very obvious comments.
- Adjust some declarations.
- Fix some space-tab issues.
Signed-off-by: Li Hong <lihong.hi@gmail.com>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This moves a pointer to buffer storing super root block to each log
buffer from nilfs_sc_info struct for simplicity.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
According to the report from Andreas Beckmann (Message-ID:
<4BA54677.3090902@abeckmann.de>), nilfs in 2.6.33 kernel got stuck
after a disk full error.
This turned out to be a regression by log writer updates merged at
kernel 2.6.33. nilfs_segctor_abort_construction, which is a cleanup
function for erroneous cases, was skipping writeback completion for
some logs.
This fixes the bug and would resolve the hang issue.
Reported-by: Andreas Beckmann <debian@abeckmann.de>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Tested-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Cc: stable <stable@kernel.org> [2.6.33.x]
Andreas Beckmann gave me a report that nilfs logged the following
warnings when it got a disk full:
nilfs_sufile_do_cancel_free: segment 0 must be clean
nilfs_sufile_do_cancel_free: segment 1 must be clean
These arise from a duplicate call to nilfs_segctor_cancel_freev in an
error path of log writer. This will fix the issue.
Reported-by: Andreas Beckmann <debian@abeckmann.de>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This moves iterator to submit write requests for a series of logs into
segbuf.c, and hides nilfs_segbuf_write() and nilfs_segbuf_wait() in
the file.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This replaces s_dirt flag use in nilfs with a new flag added on the
nilfs object. The s_dirt flag was used to indicate if
sop->write_super() should be called, however the current version of
nilfs does not use the callback. Thus, it can be replaced with the
own flag.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Cc: Jiro SEKIBA <jir@unicus.jp>
This will clean up nilfs_segctor_req struct and the obscure request
argument passed among private methods of segment constructor.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
nilfs_error() calls nilfs_detach_segment_constructor() if
errors=remount-ro option is specified, and this may lead to a hang due
to recursive locking of, for instance, nilfs->ns_segctor_sem and
others.
In this case, detaching segment constructor is not necessary because
read-only flag is set to the filesystem and further writes are
blocked.
This fixes the potential hang issue by removing the
nilfs_detach_segment_constructor() call from nilfs_error.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This adds a function to send discard requests for given array of
segment numbers, and calls the function when garbage collection
succeeded.
Signed-off-by: Jiro SEKIBA <jir@unicus.jp>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This fixes incorrect usage of nilfs_segctor_confirm() test function in
nilfs_segctor_destroy(); nilfs_segctor_confirm() returns zero if the
filesystem is not clean, so its use in nilfs_segctor_destroy() needs
inversion.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This separates wait function for submitted logs from the write
function nilfs_segctor_write(). A new list of segment buffers
"sc_write_logs" is added to hold logs under writing, and double
buffering is partially applied to hide io latency.
At this point, the double buffering is disabled for blocksize <
pagesize because page dirty flag is turned off during write and dirty
buffers are not properly collected for pages crossing over segments.
To receive full benefit of the double buffering, further refinement is
needed to move the io wait outside the lock section of log writer.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This adds a few iterator functions for segment buffers to make it easy
to handle multiple series of logs.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Hides nilfs_write_info struct and nilfs_segbuf_prepare_write function
in segbuf.c to simplify the interface of nilfs_segbuf_write function.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This moves io status variables in nilfs_write_info struct to
nilfs_segment_buffer struct.
This is a preparation to hide nilfs_write_info in segment buffer code.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This applies list_splice_tail (or list_splice_tail_init) operation
instead of list_splice (or list_splice_init, respectively) to append a
new list to tail of an existing list.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This adds nilfs_sufile_set_segment_usage() function in sufile to
replace direct access to the sufile metadata in log writer code.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This adds nilfs_sufile_mark_dirty() function in sufile to replace
nilfs_touch_segusage() function in log writer code. This is a
preparation for the further cleanup which will move out low level
sufile operations in the log writer.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
When nilfs flushes out dirty data to reduce memory pressure, creation
of checkpoints is wrongly postponed. This bug causes irregular
checkpoint creation especially in small footprint systems.
To correct this issue, a timer for the checkpoint creation has to be
continued if a log writer does not create a checkpoint.
This will do the correction.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This removes nilfs_write_super and commit super block in nilfs
internal thread, instead of periodic write_super callback.
VFS layer calls ->write_super callback periodically. However,
it looks like that calling back is ommited when disk I/O is busy.
And when cleanerd (nilfs GC) is runnig, disk I/O tend to be busy thus
nilfs superblock is not synchronized as nilfs designed.
To avoid it, syncing superblock by nilfs thread instead of pdflush.
Signed-off-by: Jiro SEKIBA <jir@unicus.jp>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Andrea Gelmini gave me a report that a kernel oops hit on a nilfs
filesystem with a 1KB block size when doing rsync.
This turned out to be caused by an inconsistency of dirty state
between a page and its buffers storing b-tree node blocks.
If the page had multiple buffers split over multiple logs, and if the
logs were written at a time, a dirty flag remained in the page even
every dirty flag in the buffers was cleared.
This will fix the failure by dropping the dirty flag properly for
pages with the discrete multiple b-tree nodes.
Reported-by: Andrea Gelmini <andrea.gelmini@gmail.com>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Tested-by: Andrea Gelmini <andrea.gelmini@gmail.com>
Cc: stable@kernel.org
Leandro Lucarella gave me a report that nilfs gets stuck after its
write function fails.
The problem turned out to be caused by bugs which leave writeback flag
on pages. This fixes the problem by ensuring to clear the writeback
flag in error path.
Reported-by: Leandro Lucarella <llucax@gmail.com>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Cc: stable@kernel.org
The following error code handling in nilfs_segctor_write() function
wrongly converted negative error codes to a truth value (i.e. 1):
err = unlikely(err) ? : res;
which originaly meant to be
err = err ? : res;
This mis-conversion caused that write or sync functions receive the
unexpected error code. This fixes the bug by removing the unlikely
directive.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Cc: stable@kernel.org
This will clean up the removal list of segments and the related
functions from segment.c and ioctl.c, which have hurt code
readability.
This elimination is applied by using nilfs_sufile_updatev() previously
introduced in the patch ("nilfs2: add sufile function that can modify
multiple segment usages").
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Ryusuke Konishi