This is a re-revised patch to shorten freeze period.
This version include a fix of the bug Konishi-san mentioned last time.
When GC is runnning, GC moves live block to difference segments.
Copying live blocks into memory is done in a transaction,
however it is not necessarily to be in the transaction.
This patch will get the nilfs_ioctl_move_blocks() out from
transaction lock and put it before the transaction.
I ran sysbench fileio test against nilfs partition.
I copied some DVD/CD images and created snapshot to create live blocks
before starting the benchmark.
Followings are summary of rc8 and rc8 w/ the patch of per-request
statistics, which is min/max and avg. I ran each test three times and
bellow is average of those numers.
According to this benchmark result, average time is slightly degrated.
However, worstcase (max) result is significantly improved.
This can address a few seconds write freeze.
- random write per-request performance of rc8
min 0.843ms
max 680.406ms
avg 3.050ms
- random write per-request performance of rc8 w/ this patch
min 0.843ms -> 100.00%
max 380.490ms -> 55.90%
avg 3.233ms -> 106.00%
- sequential write per-request performance of rc8
min 0.736ms
max 774.343ms
avg 2.883ms
- sequential write per-request performance of rc8 w/ this patch
min 0.720ms -> 97.80%
max 644.280ms-> 83.20%
avg 3.130ms -> 108.50%
-----8<-----8<-----nilfs_cleanerd.conf-----8<-----8<-----
protection_period 150
selection_policy timestamp # timestamp in ascend order
nsegments_per_clean 2
cleaning_interval 2
retry_interval 60
use_mmap
log_priority info
-----8<-----8<-----nilfs_cleanerd.conf-----8<-----8<-----
Signed-off-by: Jiro SEKIBA <jir@unicus.jp>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Nilfs has some ioctl commands to read out metadata from meta data
files:
- NILFS_IOCTL_GET_CPINFO for checkpoint file,
- NILFS_IOCTL_GET_SUINFO for segment usage file, and
- NILFS_IOCTL_GET_VINFO for Disk Address Transalation (DAT) file,
respectively.
Every routine on these metadata files is implemented so that it allows
future expansion of on-disk format. But, the above ioctl commands do
not support expansion even though nilfs_argv structure can handle
arbitrary size for data exchanged via ioctl.
This allows future expansion of the following structures which give
basic format of the "get information" ioctls:
- struct nilfs_cpinfo
- struct nilfs_suinfo
- struct nilfs_vinfo
So, this introduces forward compatility of such ioctl commands.
In this patch, a sanity check in nilfs_ioctl_get_info() function is
changed to accept larger data structure [1], and metadata read
routines are rewritten so that they become compatible for larger
structures; the routines will just ignore the remaining fields which
the current version of nilfs doesn't know.
[1] The ioctl function already has another upper limit (PAGE_SIZE
against a structure, which appears in nilfs_ioctl_wrap_copy
function), and this will not cause security problem.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
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>
This fixes a new memory leak problem in garbage collection. The
problem was brought by the bugfix patch ("nilfs2: fix lock order
reversal in nilfs_clean_segments ioctl").
Thanks to Kentaro Suzuki for finding this problem.
Reported-by: Kentaro Suzuki <k_suzuki@ms.sylc.co.jp>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Although some ioctls of nilfs2 exchange data in the form of indirectly
referenced array, some of them lack size check on the array elements.
This inserts the missing checks and rejects requests if data of ioctl
does not have a valid format.
We usually don't have to check size of structures that we associated
with ioctl commands because the size is tested implicitly for
identifying ioctl command; the checks this patch adds are for the
cases where the implicit check is not applied.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This is a companion patch to ("nilfs2: fix possible circular locking
for get information ioctls").
This corrects lock order reversal between mm->mmap_sem and
nilfs->ns_segctor_sem in nilfs_clean_segments() which was detected by
lockdep check:
=======================================================
[ INFO: possible circular locking dependency detected ]
2.6.30-rc3-nilfs-00003-g360bdc1 #7
-------------------------------------------------------
mmap/5294 is trying to acquire lock:
(&nilfs->ns_segctor_sem){++++.+}, at: [<d0d0e846>] nilfs_transaction_begin+0xb6/0x10c [nilfs2]
but task is already holding lock:
(&mm->mmap_sem){++++++}, at: [<c043700a>] do_page_fault+0x1d8/0x30a
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #1 (&mm->mmap_sem){++++++}:
[<c01470a5>] __lock_acquire+0x1066/0x13b0
[<c01474a9>] lock_acquire+0xba/0xdd
[<c01836bc>] might_fault+0x68/0x88
[<c023c61d>] copy_from_user+0x2a/0x111
[<d0d120d0>] nilfs_ioctl_prepare_clean_segments+0x1d/0xf1 [nilfs2]
[<d0d0e2aa>] nilfs_clean_segments+0x6d/0x1b9 [nilfs2]
[<d0d11f68>] nilfs_ioctl+0x2ad/0x318 [nilfs2]
[<c01a3be7>] vfs_ioctl+0x22/0x69
[<c01a408e>] do_vfs_ioctl+0x460/0x499
[<c01a4107>] sys_ioctl+0x40/0x5a
[<c01031a4>] sysenter_do_call+0x12/0x38
[<ffffffff>] 0xffffffff
-> #0 (&nilfs->ns_segctor_sem){++++.+}:
[<c0146e0b>] __lock_acquire+0xdcc/0x13b0
[<c01474a9>] lock_acquire+0xba/0xdd
[<c0433f1d>] down_read+0x2a/0x3e
[<d0d0e846>] nilfs_transaction_begin+0xb6/0x10c [nilfs2]
[<d0cfe0e5>] nilfs_page_mkwrite+0xe7/0x154 [nilfs2]
[<c0183b0b>] __do_fault+0x165/0x376
[<c01855cd>] handle_mm_fault+0x287/0x5d1
[<c043712d>] do_page_fault+0x2fb/0x30a
[<c0435462>] error_code+0x72/0x78
[<ffffffff>] 0xffffffff
where nilfs_clean_segments() holds:
nilfs->ns_segctor_sem -> copy_from_user()
--> page fault -> mm->mmap_sem
And, page fault path may hold:
page fault -> mm->mmap_sem
--> nilfs_page_mkwrite() -> nilfs->ns_segctor_sem
Even though nilfs_clean_segments() does not perform write access on
given user pages, it may cause deadlock because nilfs->ns_segctor_sem
is shared per device and mm->mmap_sem can be shared with other tasks.
To avoid this problem, this patch moves all calls of copy_from_user()
outside the nilfs->ns_segctor_sem lock in the ioctl.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This is one of two patches which are to correct possible circular
locking between mm->mmap_sem and nilfs->ns_segctor_sem.
The problem was detected by lockdep check as follows:
=======================================================
[ INFO: possible circular locking dependency detected ]
2.6.30-rc3-nilfs-00002-g3552613 #6
-------------------------------------------------------
mmap/5418 is trying to acquire lock:
(&nilfs->ns_segctor_sem){++++.+}, at: [<d0d0e852>] nilfs_transaction_begin+0xb6/0x10c [nilfs2]
but task is already holding lock:
(&mm->mmap_sem){++++++}, at: [<c043700a>] do_page_fault+0x1d8/0x30a
which lock already depends on the new lock.
the existing dependency chain (in reverse order) is:
-> #1 (&mm->mmap_sem){++++++}:
[<c01470a5>] __lock_acquire+0x1066/0x13b0
[<c01474a9>] lock_acquire+0xba/0xdd
[<c01836bc>] might_fault+0x68/0x88
[<c023c730>] copy_to_user+0x2c/0xfc
[<d0d11b4f>] nilfs_ioctl_wrap_copy+0x103/0x160 [nilfs2]
[<d0d11fa9>] nilfs_ioctl+0x30a/0x3b0 [nilfs2]
[<c01a3be7>] vfs_ioctl+0x22/0x69
[<c01a408e>] do_vfs_ioctl+0x460/0x499
[<c01a4107>] sys_ioctl+0x40/0x5a
[<c01031a4>] sysenter_do_call+0x12/0x38
[<ffffffff>] 0xffffffff
-> #0 (&nilfs->ns_segctor_sem){++++.+}:
[<c0146e0b>] __lock_acquire+0xdcc/0x13b0
[<c01474a9>] lock_acquire+0xba/0xdd
[<c0433f1d>] down_read+0x2a/0x3e
[<d0d0e852>] nilfs_transaction_begin+0xb6/0x10c [nilfs2]
[<d0cfe0e5>] nilfs_page_mkwrite+0xe7/0x154 [nilfs2]
[<c0183b0b>] __do_fault+0x165/0x376
[<c01855cd>] handle_mm_fault+0x287/0x5d1
[<c043712d>] do_page_fault+0x2fb/0x30a
[<c0435462>] error_code+0x72/0x78
[<ffffffff>] 0xffffffff
other info that might help us debug this:
1 lock held by mmap/5418:
#0: (&mm->mmap_sem){++++++}, at: [<c043700a>] do_page_fault+0x1d8/0x30a
stack backtrace:
Pid: 5418, comm: mmap Not tainted 2.6.30-rc3-nilfs-00002-g3552613 #6
Call Trace:
[<c0432145>] ? printk+0xf/0x12
[<c0145c48>] print_circular_bug_tail+0xaa/0xb5
[<c0146e0b>] __lock_acquire+0xdcc/0x13b0
[<d0d10149>] ? nilfs_sufile_get_stat+0x1e/0x105 [nilfs2]
[<c013b59a>] ? up_read+0x16/0x2c
[<d0d10225>] ? nilfs_sufile_get_stat+0xfa/0x105 [nilfs2]
[<c01474a9>] lock_acquire+0xba/0xdd
[<d0d0e852>] ? nilfs_transaction_begin+0xb6/0x10c [nilfs2]
[<c0433f1d>] down_read+0x2a/0x3e
[<d0d0e852>] ? nilfs_transaction_begin+0xb6/0x10c [nilfs2]
[<d0d0e852>] nilfs_transaction_begin+0xb6/0x10c [nilfs2]
[<d0cfe0e5>] nilfs_page_mkwrite+0xe7/0x154 [nilfs2]
[<c0183b0b>] __do_fault+0x165/0x376
[<c01855cd>] handle_mm_fault+0x287/0x5d1
[<c043700a>] ? do_page_fault+0x1d8/0x30a
[<c013b54f>] ? down_read_trylock+0x39/0x43
[<c043712d>] do_page_fault+0x2fb/0x30a
[<c0436e32>] ? do_page_fault+0x0/0x30a
[<c0435462>] error_code+0x72/0x78
[<c0436e32>] ? do_page_fault+0x0/0x30a
This makes the lock granularity of nilfs->ns_segctor_sem finer than
that of the mmap semaphore for ioctl commands except
nilfs_clean_segments().
The successive patch ("nilfs2: fix lock order reversal in
nilfs_clean_segments ioctl") is required to fully resolve the problem.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
This fixes the following circular locking dependency problem:
=======================================================
[ INFO: possible circular locking dependency detected ]
2.6.30-rc3 #5
-------------------------------------------------------
segctord/3895 is trying to acquire lock:
(&nilfs->ns_writer_mutex){+.+...}, at: [<d0d02172>]
nilfs_mdt_get_block+0x89/0x20f [nilfs2]
but task is already holding lock:
(&bmap->b_sem){++++..}, at: [<d0d02d99>]
nilfs_bmap_propagate+0x14/0x2e [nilfs2]
which lock already depends on the new lock.
The bugfix is done by replacing call sites of nilfs_get_writer() which
are never called from read-only context with direct dereferencing of
pointer to a writable FS-instance.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Pekka Enberg advised me:
> It would be nice if BUG(), BUG_ON(), and panic() calls would be
> converted to proper error handling using WARN_ON() calls. The BUG()
> call in nilfs_cpfile_delete_checkpoints(), for example, looks to be
> triggerable from user-space via the ioctl() system call.
This will follow the comment and keep them to a minimum.
Acked-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pekka Enberg suggested converting ->ioctl operations to use
->unlocked_ioctl to avoid BKL.
The conversion was verified to be safe, so I will take it on this
occasion.
Cc: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This removes compat code from the nilfs ioctls and applies the same
function for both .ioctl and .compat_ioctl file operations.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Nilfs ioctl had structures not having fixed sized types such as:
struct nilfs_argv {
void *v_base;
size_t v_nmembs;
size_t v_size;
int v_index;
int v_flags;
};
Further, some of them are wrongly aligned:
e.g.
struct nilfs_cpmode {
__u64 cm_cno;
int cm_mode;
};
The size of wrongly aligned structures varies depending on
architectures, and it breaks the identity of ioctl commands, which
leads to arch dependent errors.
Previously, these are compensated by using compat_ioctl.
This fixes these problems and allows removal of compat ioctl.
Since this will change sizes of those structures, binary compatibility
for the past utilities will once break; new utilities have to be used
instead. However, it would be helpful to avoid platform dependent
problems in the long term.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This removes NILFS_IOCTL_TIMEDWAIT command from ioctl interface along
with the related flags and wait queue.
The command is terrible because it just sleeps in the ioctl. I prefer
to avoid this by devising means of event polling in userland program.
By reconsidering the userland GC daemon, I found this is possible
without changing behaviour of the daemon and sacrificing efficiency.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This cleans up the strange indirect function calling convention used in
nilfs to follow the normal kernel coding style.
Signed-off-by: Pekka Enberg <penberg@cs.helsinki.fi>
Acked-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This resolves the following failure of nilfs2 cleaner daemon:
nilfs_cleanerd[20670]: cannot clean segments: No such file or directory
nilfs_cleanerd[20670]: shutdown
When creating thousands of snapshots, the cleaner daemon had rarely died
as above due to an error returned from the kernel code.
After applying the recent patch which fixed memory allocation problems in
ioctl (Message-Id: <20081215.155840.105124170.ryusuke@osrg.net>), the
problem gets more frequent.
It turned out to be a bug of nilfs_ioctl_wrap_copy function and one of its
callback routines to read out information of snapshots; if the
nilfs_ioctl_wrap_copy function divided a large read request into multiple
requests, the second and later requests have failed since a restart
position on snapshot meta data was not properly set forward.
It's a deficiency of the callback interface that cannot pass the restart
position among multiple requests. This patch fixes the issue by allowing
nilfs_ioctl_wrap_copy and snapshot read functions to exchange a position
argument.
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Pekka Enberg pointed out that double error handlings found after
nilfs_transaction_end() can be avoided by separating abort operation:
OK, I don't understand this. The only way nilfs_transaction_end() can
fail is if we have NILFS_TI_SYNC set and we fail to construct the
segment. But why do we want to construct a segment if we don't commit?
I guess what I'm asking is why don't we have a separate
nilfs_transaction_abort() function that can't fail for the erroneous
case to avoid this double error value tracking thing?
This does the separation and renames nilfs_transaction_end() to
nilfs_transaction_commit() for clarification.
Since, some calls of these functions were used just for exclusion control
against the segment constructor, they are replaced with semaphore
operations.
Acked-by: Pekka Enberg <penberg@cs.helsinki.fi>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This is another patch for fixing the following problems of a memory
copy function in nilfs2 ioctl:
(1) It tries to allocate 128KB size of memory even for small objects.
(2) Though the function repeatedly tries large memory allocations
while reducing the size, GFP_NOWAIT flag is not specified.
This increases the possibility of system memory shortage.
(3) During the retries of (2), verbose warnings are printed
because _GFP_NOWARN flag is not used for the kmalloc calls.
The first patch was still doing large allocations by kmalloc which are
repeatedly tried while reducing the size.
Andi Kleen told me that using copy_from_user for large memory is not
good from the viewpoint of preempt latency:
On Fri, 12 Dec 2008 21:24:11 +0100, Andi Kleen <andi@firstfloor.org> wrote:
> > In the current interface, each data item is copied twice: one is to
> > the allocated memory from user space (via copy_from_user), and another
>
> For such large copies it is better to use multiple smaller (e.g. 4K)
> copy user, that gives better real time preempt latencies. Each cfu has a
> cond_resched(), but only one, not multiple times in the inner loop.
He also advised me that:
On Sun, 14 Dec 2008 16:13:27 +0100, Andi Kleen <andi@firstfloor.org> wrote:
> Better would be if you could go to PAGE_SIZE. order 0 allocations
> are typically the fastest / least likely to stall.
>
> Also in this case it's a good idea to use __get_free_pages()
> directly, kmalloc tends to be become less efficient at larger
> sizes.
For the function in question, the size of buffer memory can be reduced
since the buffer is repeatedly used for a number of small objects. On
the other hand, it may incur large preempt latencies for larger buffer
because a copy_from_user (and a copy_to_user) was applied only once
each cycle.
With that, this revision uses the order 0 allocations with
__get_free_pages() to fix the original problems.
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
This adds userland interface implemented with ioctl.
Signed-off-by: Koji Sato <sato.koji@lab.ntt.co.jp>
Signed-off-by: Ryusuke Konishi <konishi.ryusuke@lab.ntt.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
