Currently the fscache code expect the netfs to call fscache_readpages_or_alloc
inside the aops readpages callback. It marks all the pages in the list
provided by readahead with PG_private_2. In the cases that the netfs fails to
read all the pages (which is legal) it ends up returning to the readahead and
triggering a BUG. This happens because the page list still contains marked
pages.
This patch implements a simple fscache_readpages_cancel function that the netfs
should call before returning from readpages. It will revoke the pages from the
underlying cache backend and unmark them.
The problem was originally worked out in the Ceph devel tree, but it also
occurs in CIFS. It appears that NFS, AFS and 9P are okay as read_cache_pages()
will clean up the unprocessed pages in the case of an error.
This can be used to address the following oops:
[12410647.597278] BUG: Bad page state in process petabucket pfn:3d504e
[12410647.597292] page:ffffea000f541380 count:0 mapcount:0 mapping:
(null) index:0x0
[12410647.597298] page flags: 0x200000000001000(private_2)
...
[12410647.597334] Call Trace:
[12410647.597345] [<ffffffff815523f2>] dump_stack+0x19/0x1b
[12410647.597356] [<ffffffff8111def7>] bad_page+0xc7/0x120
[12410647.597359] [<ffffffff8111e49e>] free_pages_prepare+0x10e/0x120
[12410647.597361] [<ffffffff8111fc80>] free_hot_cold_page+0x40/0x170
[12410647.597363] [<ffffffff81123507>] __put_single_page+0x27/0x30
[12410647.597365] [<ffffffff81123df5>] put_page+0x25/0x40
[12410647.597376] [<ffffffffa02bdcf9>] ceph_readpages+0x2e9/0x6e0 [ceph]
[12410647.597379] [<ffffffff81122a8f>] __do_page_cache_readahead+0x1af/0x260
[12410647.597382] [<ffffffff81122ea1>] ra_submit+0x21/0x30
[12410647.597384] [<ffffffff81118f64>] filemap_fault+0x254/0x490
[12410647.597387] [<ffffffff8113a74f>] __do_fault+0x6f/0x4e0
[12410647.597391] [<ffffffff810125bd>] ? __switch_to+0x16d/0x4a0
[12410647.597395] [<ffffffff810865ba>] ? finish_task_switch+0x5a/0xc0
[12410647.597398] [<ffffffff8113d856>] handle_pte_fault+0xf6/0x930
[12410647.597401] [<ffffffff81008c33>] ? pte_mfn_to_pfn+0x93/0x110
[12410647.597403] [<ffffffff81008cce>] ? xen_pmd_val+0xe/0x10
[12410647.597405] [<ffffffff81005469>] ? __raw_callee_save_xen_pmd_val+0x11/0x1e
[12410647.597407] [<ffffffff8113f361>] handle_mm_fault+0x251/0x370
[12410647.597411] [<ffffffff812b0ac4>] ? call_rwsem_down_read_failed+0x14/0x30
[12410647.597414] [<ffffffff8155bffa>] __do_page_fault+0x1aa/0x550
[12410647.597418] [<ffffffff8108011d>] ? up_write+0x1d/0x20
[12410647.597422] [<ffffffff8113141c>] ? vm_mmap_pgoff+0xbc/0xe0
[12410647.597425] [<ffffffff81143bb8>] ? SyS_mmap_pgoff+0xd8/0x240
[12410647.597427] [<ffffffff8155c3ae>] do_page_fault+0xe/0x10
[12410647.597431] [<ffffffff81558818>] page_fault+0x28/0x30
Signed-off-by: Milosz Tanski <milosz@adfin.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Extend the fscache netfs API so that the netfs can ask as to whether a cache
object is up to date with respect to its corresponding netfs object:
int fscache_check_consistency(struct fscache_cookie *cookie)
This will call back to the netfs to check whether the auxiliary data associated
with a cookie is correct. It returns 0 if it is and -ESTALE if it isn't; it
may also return -ENOMEM and -ERESTARTSYS.
The backends now have to implement a mandatory operation pointer:
int (*check_consistency)(struct fscache_object *object)
that corresponds to the above API call. FS-Cache takes care of pinning the
object and the cookie in memory and managing this call with respect to the
object state.
Original-author: Hongyi Jia <jiayisuse@gmail.com>
Signed-off-by: David Howells <dhowells@redhat.com>
cc: Hongyi Jia <jiayisuse@gmail.com>
cc: Milosz Tanski <milosz@adfin.com>
Provide a proper invalidation method rather than relying on the netfs retiring
the cookie it has and getting a new one. The problem with this is that isn't
easy for the netfs to make sure that it has completed/cancelled all its
outstanding storage and retrieval operations on the cookie it is retiring.
Instead, have the cache provide an invalidation method that will cancel or wait
for all currently outstanding operations before invalidating the cache, and
will cause new operations to queue up behind that. Whilst invalidation is in
progress, some requests will be rejected until the cache can stack a barrier on
the operation queue to cause new operations to be deferred behind it.
Signed-off-by: David Howells <dhowells@redhat.com>
Add an FS-Cache helper to bulk uncache pages on an inode. This will
only work for the circumstance where the pages in the cache correspond
1:1 with the pages attached to an inode's page cache.
This is required for CIFS and NFS: When disabling inode cookie, we were
returning the cookie and setting cifsi->fscache to NULL but failed to
invalidate any previously mapped pages. This resulted in "Bad page
state" errors and manifested in other kind of errors when running
fsstress. Fix it by uncaching mapped pages when we disable the inode
cookie.
This patch should fix the following oops and "Bad page state" errors
seen during fsstress testing.
------------[ cut here ]------------
kernel BUG at fs/cachefiles/namei.c:201!
invalid opcode: 0000 [#1] SMP
Pid: 5, comm: kworker/u:0 Not tainted 2.6.38.7-30.fc15.x86_64 #1 Bochs Bochs
RIP: 0010: cachefiles_walk_to_object+0x436/0x745 [cachefiles]
RSP: 0018:ffff88002ce6dd00 EFLAGS: 00010282
RAX: ffff88002ef165f0 RBX: ffff88001811f500 RCX: 0000000000000000
RDX: 0000000000000000 RSI: 0000000000000100 RDI: 0000000000000282
RBP: ffff88002ce6dda0 R08: 0000000000000100 R09: ffffffff81b3a300
R10: 0000ffff00066c0a R11: 0000000000000003 R12: ffff88002ae54840
R13: ffff88002ae54840 R14: ffff880029c29c00 R15: ffff88001811f4b0
FS: 00007f394dd32720(0000) GS:ffff88002ef00000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 00007fffcb62ddf8 CR3: 000000001825f000 CR4: 00000000000006e0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process kworker/u:0 (pid: 5, threadinfo ffff88002ce6c000, task ffff88002ce55cc0)
Stack:
0000000000000246 ffff88002ce55cc0 ffff88002ce6dd58 ffff88001815dc00
ffff8800185246c0 ffff88001811f618 ffff880029c29d18 ffff88001811f380
ffff88002ce6dd50 ffffffff814757e4 ffff88002ce6dda0 ffffffff8106ac56
Call Trace:
cachefiles_lookup_object+0x78/0xd4 [cachefiles]
fscache_lookup_object+0x131/0x16d [fscache]
fscache_object_work_func+0x1bc/0x669 [fscache]
process_one_work+0x186/0x298
worker_thread+0xda/0x15d
kthread+0x84/0x8c
kernel_thread_helper+0x4/0x10
RIP cachefiles_walk_to_object+0x436/0x745 [cachefiles]
---[ end trace 1d481c9af1804caa ]---
I tested the uncaching by the following means:
(1) Create a big file on my NFS server (104857600 bytes).
(2) Read the file into the cache with md5sum on the NFS client. Look in
/proc/fs/fscache/stats:
Pages : mrk=25601 unc=0
(3) Open the file for read/write ("bash 5<>/warthog/bigfile"). Look in proc
again:
Pages : mrk=25601 unc=25601
Reported-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: David Howells <dhowells@redhat.com>
Reviewed-and-Tested-by: Suresh Jayaraman <sjayaraman@suse.de>
cc: stable@kernel.org
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Handle netfs pages that the vmscan algorithm wants to evict from the pagecache
under OOM conditions, but that are waiting for write to the cache. Under these
conditions, vmscan calls the releasepage() function of the netfs, asking if a
page can be discarded.
The problem is typified by the following trace of a stuck process:
kslowd005 D 0000000000000000 0 4253 2 0x00000080
ffff88001b14f370 0000000000000046 ffff880020d0d000 0000000000000007
0000000000000006 0000000000000001 ffff88001b14ffd8 ffff880020d0d2a8
000000000000ddf0 00000000000118c0 00000000000118c0 ffff880020d0d2a8
Call Trace:
[<ffffffffa00782d8>] __fscache_wait_on_page_write+0x8b/0xa7 [fscache]
[<ffffffff8104c0f1>] ? autoremove_wake_function+0x0/0x34
[<ffffffffa0078240>] ? __fscache_check_page_write+0x63/0x70 [fscache]
[<ffffffffa00b671d>] nfs_fscache_release_page+0x4e/0xc4 [nfs]
[<ffffffffa00927f0>] nfs_release_page+0x3c/0x41 [nfs]
[<ffffffff810885d3>] try_to_release_page+0x32/0x3b
[<ffffffff81093203>] shrink_page_list+0x316/0x4ac
[<ffffffff8109372b>] shrink_inactive_list+0x392/0x67c
[<ffffffff813532fa>] ? __mutex_unlock_slowpath+0x100/0x10b
[<ffffffff81058df0>] ? trace_hardirqs_on_caller+0x10c/0x130
[<ffffffff8135330e>] ? mutex_unlock+0x9/0xb
[<ffffffff81093aa2>] shrink_list+0x8d/0x8f
[<ffffffff81093d1c>] shrink_zone+0x278/0x33c
[<ffffffff81052d6c>] ? ktime_get_ts+0xad/0xba
[<ffffffff81094b13>] try_to_free_pages+0x22e/0x392
[<ffffffff81091e24>] ? isolate_pages_global+0x0/0x212
[<ffffffff8108e743>] __alloc_pages_nodemask+0x3dc/0x5cf
[<ffffffff81089529>] grab_cache_page_write_begin+0x65/0xaa
[<ffffffff8110f8c0>] ext3_write_begin+0x78/0x1eb
[<ffffffff81089ec5>] generic_file_buffered_write+0x109/0x28c
[<ffffffff8103cb69>] ? current_fs_time+0x22/0x29
[<ffffffff8108a509>] __generic_file_aio_write+0x350/0x385
[<ffffffff8108a588>] ? generic_file_aio_write+0x4a/0xae
[<ffffffff8108a59e>] generic_file_aio_write+0x60/0xae
[<ffffffff810b2e82>] do_sync_write+0xe3/0x120
[<ffffffff8104c0f1>] ? autoremove_wake_function+0x0/0x34
[<ffffffff810b18e1>] ? __dentry_open+0x1a5/0x2b8
[<ffffffff810b1a76>] ? dentry_open+0x82/0x89
[<ffffffffa00e693c>] cachefiles_write_page+0x298/0x335 [cachefiles]
[<ffffffffa0077147>] fscache_write_op+0x178/0x2c2 [fscache]
[<ffffffffa0075656>] fscache_op_execute+0x7a/0xd1 [fscache]
[<ffffffff81082093>] slow_work_execute+0x18f/0x2d1
[<ffffffff8108239a>] slow_work_thread+0x1c5/0x308
[<ffffffff8104c0f1>] ? autoremove_wake_function+0x0/0x34
[<ffffffff810821d5>] ? slow_work_thread+0x0/0x308
[<ffffffff8104be91>] kthread+0x7a/0x82
[<ffffffff8100beda>] child_rip+0xa/0x20
[<ffffffff8100b87c>] ? restore_args+0x0/0x30
[<ffffffff8102ef83>] ? tg_shares_up+0x171/0x227
[<ffffffff8104be17>] ? kthread+0x0/0x82
[<ffffffff8100bed0>] ? child_rip+0x0/0x20
In the above backtrace, the following is happening:
(1) A page storage operation is being executed by a slow-work thread
(fscache_write_op()).
(2) FS-Cache farms the operation out to the cache to perform
(cachefiles_write_page()).
(3) CacheFiles is then calling Ext3 to perform the actual write, using Ext3's
standard write (do_sync_write()) under KERNEL_DS directly from the netfs
page.
(4) However, for Ext3 to perform the write, it must allocate some memory, in
particular, it must allocate at least one page cache page into which it
can copy the data from the netfs page.
(5) Under OOM conditions, the memory allocator can't immediately come up with
a page, so it uses vmscan to find something to discard
(try_to_free_pages()).
(6) vmscan finds a clean netfs page it might be able to discard (possibly the
one it's trying to write out).
(7) The netfs is called to throw the page away (nfs_release_page()) - but it's
called with __GFP_WAIT, so the netfs decides to wait for the store to
complete (__fscache_wait_on_page_write()).
(8) This blocks a slow-work processing thread - possibly against itself.
The system ends up stuck because it can't write out any netfs pages to the
cache without allocating more memory.
To avoid this, we make FS-Cache cancel some writes that aren't in the middle of
actually being performed. This means that some data won't make it into the
cache this time. To support this, a new FS-Cache function is added
fscache_maybe_release_page() that replaces what the netfs releasepage()
functions used to do with respect to the cache.
The decisions fscache_maybe_release_page() makes are counted and displayed
through /proc/fs/fscache/stats on a line labelled "VmScan". There are four
counters provided: "nos=N" - pages that weren't pending storage; "gon=N" -
pages that were pending storage when we first looked, but weren't by the time
we got the object lock; "bsy=N" - pages that we ignored as they were actively
being written when we looked; and "can=N" - pages that we cancelled the storage
of.
What I'd really like to do is alter the behaviour of the cancellation
heuristics, depending on how necessary it is to expel pages. If there are
plenty of other pages that aren't waiting to be written to the cache that
could be ejected first, then it would be nice to hold up on immediate
cancellation of cache writes - but I don't see a way of doing that.
Signed-off-by: David Howells <dhowells@redhat.com>
Add the API for a generic facility (FS-Cache) by which filesystems (such as AFS
or NFS) may call on local caching capabilities without having to know anything
about how the cache works, or even if there is a cache:
+---------+
| | +--------------+
| NFS |--+ | |
| | | +-->| CacheFS |
+---------+ | +----------+ | | /dev/hda5 |
| | | | +--------------+
+---------+ +-->| | |
| | | |--+
| AFS |----->| FS-Cache |
| | | |--+
+---------+ +-->| | |
| | | | +--------------+
+---------+ | +----------+ | | |
| | | +-->| CacheFiles |
| ISOFS |--+ | /var/cache |
| | +--------------+
+---------+
General documentation and documentation of the netfs specific API are provided
in addition to the header files.
As this patch stands, it is possible to build a filesystem against the facility
and attempt to use it. All that will happen is that all requests will be
immediately denied as if no cache is present.
Further patches will implement the core of the facility. The facility will
transfer requests from networking filesystems to appropriate caches if
possible, or else gracefully deny them.
If this facility is disabled in the kernel configuration, then all its
operations will trivially reduce to nothing during compilation.
WHY NOT I_MAPPING?
==================
I have added my own API to implement caching rather than using i_mapping to do
this for a number of reasons. These have been discussed a lot on the LKML and
CacheFS mailing lists, but to summarise the basics:
(1) Most filesystems don't do hole reportage. Holes in files are treated as
blocks of zeros and can't be distinguished otherwise, making it difficult
to distinguish blocks that have been read from the network and cached from
those that haven't.
(2) The backing inode must be fully populated before being exposed to
userspace through the main inode because the VM/VFS goes directly to the
backing inode and does not interrogate the front inode's VM ops.
Therefore:
(a) The backing inode must fit entirely within the cache.
(b) All backed files currently open must fit entirely within the cache at
the same time.
(c) A working set of files in total larger than the cache may not be
cached.
(d) A file may not grow larger than the available space in the cache.
(e) A file that's open and cached, and remotely grows larger than the
cache is potentially stuffed.
(3) Writes go to the backing filesystem, and can only be transferred to the
network when the file is closed.
(4) There's no record of what changes have been made, so the whole file must
be written back.
(5) The pages belong to the backing filesystem, and all metadata associated
with that page are relevant only to the backing filesystem, and not
anything stacked atop it.
OVERVIEW
========
FS-Cache provides (or will provide) the following facilities:
(1) Caches can be added / removed at any time, even whilst in use.
(2) Adds a facility by which tags can be used to refer to caches, even if
they're not available yet.
(3) More than one cache can be used at once. Caches can be selected
explicitly by use of tags.
(4) The netfs is provided with an interface that allows either party to
withdraw caching facilities from a file (required for (1)).
(5) A netfs may annotate cache objects that belongs to it. This permits the
storage of coherency maintenance data.
(6) Cache objects will be pinnable and space reservations will be possible.
(7) The interface to the netfs returns as few errors as possible, preferring
rather to let the netfs remain oblivious.
(8) Cookies are used to represent indices, files and other objects to the
netfs. The simplest cookie is just a NULL pointer - indicating nothing
cached there.
(9) The netfs is allowed to propose - dynamically - any index hierarchy it
desires, though it must be aware that the index search function is
recursive, stack space is limited, and indices can only be children of
indices.
(10) Indices can be used to group files together to reduce key size and to make
group invalidation easier. The use of indices may make lookup quicker,
but that's cache dependent.
(11) Data I/O is effectively done directly to and from the netfs's pages. The
netfs indicates that page A is at index B of the data-file represented by
cookie C, and that it should be read or written. The cache backend may or
may not start I/O on that page, but if it does, a netfs callback will be
invoked to indicate completion. The I/O may be either synchronous or
asynchronous.
(12) Cookies can be "retired" upon release. At this point FS-Cache will mark
them as obsolete and the index hierarchy rooted at that point will get
recycled.
(13) The netfs provides a "match" function for index searches. In addition to
saying whether a match was made or not, this can also specify that an
entry should be updated or deleted.
FS-Cache maintains a virtual index tree in which all indices, files, objects
and pages are kept. Bits of this tree may actually reside in one or more
caches.
FSDEF
|
+------------------------------------+
| |
NFS AFS
| |
+--------------------------+ +-----------+
| | | |
homedir mirror afs.org redhat.com
| | |
+------------+ +---------------+ +----------+
| | | | | |
00001 00002 00007 00125 vol00001 vol00002
| | | | |
+---+---+ +-----+ +---+ +------+------+ +-----+----+
| | | | | | | | | | | | |
PG0 PG1 PG2 PG0 XATTR PG0 PG1 DIRENT DIRENT DIRENT R/W R/O Bak
| |
PG0 +-------+
| |
00001 00003
|
+---+---+
| | |
PG0 PG1 PG2
In the example above, two netfs's can be seen to be backed: NFS and AFS. These
have different index hierarchies:
(*) The NFS primary index will probably contain per-server indices. Each
server index is indexed by NFS file handles to get data file objects.
Each data file objects can have an array of pages, but may also have
further child objects, such as extended attributes and directory entries.
Extended attribute objects themselves have page-array contents.
(*) The AFS primary index contains per-cell indices. Each cell index contains
per-logical-volume indices. Each of volume index contains up to three
indices for the read-write, read-only and backup mirrors of those volumes.
Each of these contains vnode data file objects, each of which contains an
array of pages.
The very top index is the FS-Cache master index in which individual netfs's
have entries.
Any index object may reside in more than one cache, provided it only has index
children. Any index with non-index object children will be assumed to only
reside in one cache.
The FS-Cache overview can be found in:
Documentation/filesystems/caching/fscache.txt
The netfs API to FS-Cache can be found in:
Documentation/filesystems/caching/netfs-api.txt
Signed-off-by: David Howells <dhowells@redhat.com>
Acked-by: Steve Dickson <steved@redhat.com>
Acked-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Acked-by: Al Viro <viro@zeniv.linux.org.uk>
Tested-by: Daire Byrne <Daire.Byrne@framestore.com>