Exofs raid engine was saving on memory space by having a single layout-info,
single pid, and a single device-table, global to the filesystem. Then passing
a credential and object_id info at the io_state level, private for each
inode. It would also devise this contraption of rotating the device table
view for each inode->ino to spread out the device usage.
This is not compatible with the pnfs-objects standard, demanding that
each inode can have it's own layout-info, device-table, and each object
component it's own pid, oid and creds.
So: Bring exofs raid engine to be usable for generic pnfs-objects use by:
* Define an exofs_comp structure that holds obj_id and credential info.
* Break up exofs_layout struct to an exofs_components structure that holds a
possible array of exofs_comp and the array of devices + the size of the
arrays.
* Add a "comps" parameter to get_io_state() that specifies the ids creds
and device array to use for each IO.
This enables to keep the layout global, but the device-table view, creds
and IDs at the inode level. It only adds two 64bit to each inode, since
some of these members already existed in another form.
* ios raid engine now access layout-info and comps-info through the passed
pointers. Everything is pre-prepared by caller for generic access of
these structures and arrays.
At the exofs Level:
* Super block holds an exofs_components struct that holds the device
array, previously in layout. The devices there are in device-table
order. The device-array is twice bigger and repeats the device-table
twice so now each inode's device array can point to a random device
and have a round-robin view of the table, making it compatible to
previous exofs versions.
* Each inode has an exofs_components struct that is initialized at
load time, with it's own view of the device table IDs and creds.
When doing IO this gets passed to the io_state together with the
layout.
While preforming this change. Bugs where found where credentials with the
wrong IDs where used to access the different SB objects (super.c). As well
as some dead code. It was never noticed because the target we use does not
check the credentials.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
ios.c will be moving to an external library, for use by the
objects-layout-driver. Remove from it some exofs specific functions.
Also g_attr_logical_length is used both by inode.c and ios.c
move definition to the later, to keep it independent
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
In future raid code we will need to know the IO offset/length
and if it's a read or write to determine some of the array
sizes we'll need.
So add a new exofs_get_rw_state() API for use when
writeing/reading. All other simple cases are left using the
old way.
The major change to this is that now we need to call
exofs_get_io_state later at inode.c::read_exec and
inode.c::write_exec when we actually know these things. So this
patch is kept separate so I can test things apart from other
changes.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
In the general raid-group case the truncate was wrong in that
it did not also fix the object length of the neighboring groups.
There are two bad cases in the old code:
1. Space that should be freed was not.
2. If a file That was big is truncated small, then made bigger
again, the holes would not contain zeros but could expose old data.
(If the growing of the file expands to more than a full
groups cycle + group size (> S + T))
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Since the beginning we realloced the sbi structure when a bigger
then one device table was specified. (I know that was really stupid).
Then much later when "register bdi" was added (By Jens) it was
registering the pointer to sbi->bdi before the realloc.
We never saw this problem because up till now the realloc did not
do anything since the device table was small enough to fit in the
original allocation. But once we starting testing with large device
tables (Bigger then 28) we noticed the crash of writeback operating
on a deallocated pointer.
* Avoid the all mess by allocating the device-table as a second array
and get rid of the variable-sized structure and the rest of this
mess.
* Take the chance to clean near by structures and comments.
* Add a needed dprint on startup to indicate the loaded layout.
* Also move the bdi registration to the very end because it will
only fail in a low memory, which will probably fail before hand.
There are many more likely causes to not load before that. This
way the error handling is made simpler. (Just doing this would be
enough to fix the BUG)
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Btrfs needs to be able to control how filemap_write_and_wait_range() is called
in fsync to make it less of a painful operation, so push down taking i_mutex and
the calling of filemap_write_and_wait() down into the ->fsync() handlers. Some
file systems can drop taking the i_mutex altogether it seems, like ext3 and
ocfs2. For correctness sake I just pushed everything down in all cases to make
sure that we keep the current behavior the same for everybody, and then each
individual fs maintainer can make up their mind about what to do from there.
Thanks,
Acked-by: Jan Kara <jack@suse.cz>
Signed-off-by: Josef Bacik <josef@redhat.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* 'for-2.6.39/core' of git://git.kernel.dk/linux-2.6-block: (65 commits)
Documentation/iostats.txt: bit-size reference etc.
cfq-iosched: removing unnecessary think time checking
cfq-iosched: Don't clear queue stats when preempt.
blk-throttle: Reset group slice when limits are changed
blk-cgroup: Only give unaccounted_time under debug
cfq-iosched: Don't set active queue in preempt
block: fix non-atomic access to genhd inflight structures
block: attempt to merge with existing requests on plug flush
block: NULL dereference on error path in __blkdev_get()
cfq-iosched: Don't update group weights when on service tree
fs: assign sb->s_bdi to default_backing_dev_info if the bdi is going away
block: Require subsystems to explicitly allocate bio_set integrity mempool
jbd2: finish conversion from WRITE_SYNC_PLUG to WRITE_SYNC and explicit plugging
jbd: finish conversion from WRITE_SYNC_PLUG to WRITE_SYNC and explicit plugging
fs: make fsync_buffers_list() plug
mm: make generic_writepages() use plugging
blk-cgroup: Add unaccounted time to timeslice_used.
block: fixup plugging stubs for !CONFIG_BLOCK
block: remove obsolete comments for blkdev_issue_zeroout.
blktrace: Use rq->cmd_flags directly in blk_add_trace_rq.
...
Fix up conflicts in fs/{aio.c,super.c}
One leftover from the days of IBM's original code, is an SB counter
that counts in-flight asynchronous commands. And a piece of code that
waits for the counter to reach zero at unmount. I guess it might have
been needed then, cause of some reference missing or something.
I'm not removing it yet but am putting a warning message if ever this
counter triggers at unmount. If I'll never see it triggers or reported
I'll remove the counter for good.
(I had this print as a debug output for a long time and never had it
trigger)
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Before when creating a new inode, we'd set the sb->s_dirt flag,
and sometime later the system would write out s_nextid as part
of the sb_info. Also on inode sync we would force the sb sync
as well.
Define the s_nextid as a new partition attribute and set it
every time we create a new object.
At mount we read it from it's new place.
We now never set sb->s_dirt anywhere in exofs. write_super
is actually never called. The call to exofs_write_super from
exofs_put_super is also removed because the VFS always calls
->sync_fs before calling ->put_super twice.
To stay backward-and-forward compatible we also write the old
s_nextid in the super_block object at unmount, and support zero
length attribute on mount.
This also fixes a BUG where in layouts when group_width was not
a divisor of EXOFS_SUPER_ID (0x10000) the s_nextid was not read
from the device it was written to. Because of the sliding window
layout trick, and because the read was always done from the 0
device but the write was done via the raid engine that might slide
the device view. Now we read and write through the raid engine.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
If /dev/osd* devices are shuffled because more devices
where added, and/or login order has changed. It is hard to
mount the FS you want.
Add an option to mount by osdname. osdname is any osd-device's
osdname as specified to the mkfs.exofs command when formatting
the osd-devices.
The new mount format is:
OPT="osdname=$UUID0,pid=$PID,_netdev"
mount -t exofs -o $OPT $DEV_OSD0 $MOUNTDIR
if "osdname=" is specified in options above $DEV_OSD0 is
ignored and can be empty.
Also while at it: Removed some old unused Opt_* enums.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
* Set all inode->i_mapping->backing_dev_info to point to
the per super-block sb->s_bdi.
* Calculating a read_ahead that is:
- preferable 2 stripes long
(Future patch will add a mount option to override this)
- Minimum 128K aligned up to stripe-size
- Caped to maximum-IO-sizes round down to stripe_size.
(Max sizes are governed by max bio-size that fits in a page
times number-of-devices)
CC: Marc Dionne <marc.c.dionne@gmail.com>
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
It is incorrect to test inode dirty bits without participating in the inode
writeback protocol. Inode writeback sets I_SYNC and clears I_DIRTY_?, then
writes out the particular bits, then clears I_SYNC when it is done. BTW. it
may not completely write all pages out, so I_DIRTY_PAGES would get set
again.
This is a standard pattern used throughout the kernel's writeback caches
(I_SYNC ~= I_WRITEBACK, if that makes it clearer).
And so it is not possible to determine an inode's dirty status just by
checking I_DIRTY bits. Especially not for the purpose of data integrity
syncs.
Missing the check for these bits means that fsync can complete while
writeback to the inode is underway. Inode writeback functions get this
right, so call into them rather than try to shortcut things by testing
dirty state improperly.
Signed-off-by: Nick Piggin <npiggin@kernel.dk>
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
I stumbled on some of these prints in log files so, might
just submit the fixes.
* All i_ino prints in exofs should be hex
* All OSD_ERR prints should end with a "\n"
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Code has been converted over to the new explicit on-stack plugging,
and delay users have been converted to use the new API for that.
So lets kill off the old plugging along with aops->sync_page().
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
This reverts commit 115e19c535.
Apparently setting inode->bdi to one's own sb->s_bdi stops VFS from
sending *read-aheads*. This problem was bisected to this commit. A
revert fixes it. I'll investigate farther why is this happening for the
next Kernel, but for now a revert.
I'm sending to stable@kernel.org as well, since it exists also in
2.6.37. 2.6.36 is good and does not have this patch.
CC: Stable Tree <stable@kernel.org>
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
RCU free the struct inode. This will allow:
- Subsequent store-free path walking patch. The inode must be consulted for
permissions when walking, so an RCU inode reference is a must.
- sb_inode_list_lock to be moved inside i_lock because sb list walkers who want
to take i_lock no longer need to take sb_inode_list_lock to walk the list in
the first place. This will simplify and optimize locking.
- Could remove some nested trylock loops in dcache code
- Could potentially simplify things a bit in VM land. Do not need to take the
page lock to follow page->mapping.
The downsides of this is the performance cost of using RCU. In a simple
creat/unlink microbenchmark, performance drops by about 10% due to inability to
reuse cache-hot slab objects. As iterations increase and RCU freeing starts
kicking over, this increases to about 20%.
In cases where inode lifetimes are longer (ie. many inodes may be allocated
during the average life span of a single inode), a lot of this cache reuse is
not applicable, so the regression caused by this patch is smaller.
The cache-hot regression could largely be avoided by using SLAB_DESTROY_BY_RCU,
however this adds some complexity to list walking and store-free path walking,
so I prefer to implement this at a later date, if it is shown to be a win in
real situations. I haven't found a regression in any non-micro benchmark so I
doubt it will be a problem.
Signed-off-by: Nick Piggin <npiggin@kernel.dk>
Add a new helper to write out the inode using the writeback code,
that is including the correct dirty bit and list manipulation. A few
of filesystems already opencode this, and a lot of others should be
using it instead of using write_inode_now which also writes out the
data.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
exofs_new_inode() was incrementing the inode->i_count and
decrementing it in create_done(), in a bad attempt to make sure
the inode will still be there when the asynchronous create_done()
finally arrives. This was very stupid because iput() was not called,
and if it was actually needed, it would leak the inode.
However all this is not needed, because at exofs_evict_inode()
we already wait for create_done() by waiting for the
object_created event. Therefore remove the superfluous ref counting
and just Thicken the comment at exofs_evict_inode() a bit.
While at it change places that open coded wait_obj_created()
to call the already available wrapper.
CC: Dave Chinner <dchinner@redhat.com>
CC: Christoph Hellwig <hch@lst.de>
CC: Nick Piggin <npiggin@kernel.dk>
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Though it has been promised that inode->i_mapping->backing_dev_info
is not used and the supporting code is fine. Until the pointer
will default to NULL, I'd rather it points to the correct thing
regardless.
At least for future infrastructure coder it is a clear indication
of where are the key points that inodes are initialized.
I know because it took me time to find this out.
Signed-off-by: Boaz Harrosh <Boaz Harrosh bharrosh@panasas.com>
Last BUG fix added a flag to the the page_collect structure
to communicate with readpage_strip. This calls for a clean up
removing that flag's reincarnations in the read functions
parameters.
Signed-off-by: Boaz Harrosh <Boaz Harrosh bharrosh@panasas.com>
This BUG is there since the first submit of the code, but only triggered
in last Kernel. It's timing related do to the asynchronous object-creation
behaviour of exofs. (Which should be investigated farther)
The bug is obvious hence the fixed.
Signed-off-by: Boaz Harrosh <Boaz Harrosh bharrosh@panasas.com>
* 'for-linus' of git://git.open-osd.org/linux-open-osd:
exofs: Fix groups code when num_devices is not divisible by group_width
exofs: Remove useless optimization
exofs: exofs_file_fsync and exofs_file_flush correctness
exofs: Remove superfluous dependency on buffer_head and writeback
* 'for-2.6.36' of git://git.kernel.dk/linux-2.6-block: (149 commits)
block: make sure that REQ_* types are seen even with CONFIG_BLOCK=n
xen-blkfront: fix missing out label
blkdev: fix blkdev_issue_zeroout return value
block: update request stacking methods to support discards
block: fix missing export of blk_types.h
writeback: fix bad _bh spinlock nesting
drbd: revert "delay probes", feature is being re-implemented differently
drbd: Initialize all members of sync_conf to their defaults [Bugz 315]
drbd: Disable delay probes for the upcomming release
writeback: cleanup bdi_register
writeback: add new tracepoints
writeback: remove unnecessary init_timer call
writeback: optimize periodic bdi thread wakeups
writeback: prevent unnecessary bdi threads wakeups
writeback: move bdi threads exiting logic to the forker thread
writeback: restructure bdi forker loop a little
writeback: move last_active to bdi
writeback: do not remove bdi from bdi_list
writeback: simplify bdi code a little
writeback: do not lose wake-ups in bdi threads
...
Fixed up pretty trivial conflicts in drivers/block/virtio_blk.c and
drivers/scsi/scsi_error.c as per Jens.
These changes are crafted based on the similar
conversion done to ext2 by Nick Piggin.
* Remove the deprecated ->truncate vector. Let exofs_setattr
take care of on-disk size updates.
* Call truncate_pagecache on the unused pages if
write_begin/end fails.
* Cleanup exofs_delete_inode that did stupid inode
writes and updates on an inode that will be
removed.
* And finally get rid of exofs_get_block. We never
had any blocks it was all for calling nobh_truncate_page.
nobh_truncate_page is not actually needed in exofs since
the last page is complete and gone, just like all the other
pages. There is no partial blocks in exofs.
I've tested with this patch, and there are no apparent
failures, so far.
CC: Nick Piggin <npiggin@suse.de>
CC: Christoph Hellwig <hch@lst.de>
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Replace inode_setattr with opencoded variants of it in all callers. This
moves the remaining call to vmtruncate into the filesystem methods where it
can be replaced with the proper truncate sequence.
In a few cases it was obvious that we would never end up calling vmtruncate
so it was left out in the opencoded variant:
spufs: explicitly checks for ATTR_SIZE earlier
btrfs,hugetlbfs,logfs,dlmfs: explicitly clears ATTR_SIZE earlier
ufs: contains an opencoded simple_seattr + truncate that sets the filesize just above
In addition to that ncpfs called inode_setattr with handcrafted iattrs,
which allowed to trim down the opencoded variant.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Remove the current bio flags and reuse the request flags for the bio, too.
This allows to more easily trace the type of I/O from the filesystem
down to the block driver. There were two flags in the bio that were
missing in the requests: BIO_RW_UNPLUG and BIO_RW_AHEAD. Also I've
renamed two request flags that had a superflous RW in them.
Note that the flags are in bio.h despite having the REQ_ name - as
blkdev.h includes bio.h that is the only way to go for now.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Jens Axboe <jaxboe@fusionio.com>
There is a bug when num_devices is not divisible by group_width * mirrors.
We would not return to the proper device and offset when looping on to the
next group.
The fix makes code simpler actually.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
We used to compact all used devices in an IO to the beginning
of the device array in an io_state. And keep a last device used
so in later loops we don't iterate on all device slots. This
does not prevent us from checking if slots are empty since in
reads we only read from a single mirror and jump to the next
mirror-set.
This optimization is marginal, and needlessly complicates the
code. Specially when we will later want to support raid/456
with same abstract code. So remove the distinction between
"dev" and "comp". Only "dev" is used both as the device used
and as the index (component) in the device array.
[Note that now the io_state->dev member is redundant but I
keep it because I might want to optimize by only IOing a
single group, though keeping a group_width*mirrors devices
in io_state, we now keep num-devices in each io_state]
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
As per Christoph advise: no need to call filemap_write_and_wait().
In exofs all metadata is at the inode so just writing the inode is
all is needed. ->fsync implies this must be done synchronously.
But now exofs_file_fsync can not be used by exofs_file_flush.
vfs_fsync() should do that job correctly.
FIXME: remove the sb_sync and fix that sb_update better.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
exofs_releasepage && exofs_invalidatepage are never called.
Leave the WARN_ONs but remove any code. Remove the
cleanup other stale #includes.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
* 'for-linus' of git://git.open-osd.org/linux-open-osd:
exofs: confusion between kmap() and kmap_atomic() api
exofs: Add default address_space_operations
For kmap_atomic() we call kunmap_atomic() on the returned pointer.
That's different from kmap() and kunmap() and so it's easy to get them
backwards.
Cc: Stable <stable@kernel.org>
Signed-off-by: Dan Carpenter <error27@gmail.com>
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
All vectors of address_space_operations should be initialized
by the filesystem. Add the missing parts.
This is actually an optimization, by using
__set_page_dirty_nobuffers. The default, in case of NULL,
would be __set_page_dirty_buffers which has these extar if(s).
.releasepage && .invalidatepage should both not be called
because page_private() is NULL in exofs. Put a WARN_ON if
they are called, to indicate the Kernel has changed in this
regard, if when it does.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Commit b3d0ab7e60 ("exofs: add bdi backing
to mount session") has a bug in the placement of the bdi member at
struct exofs_sb_info. The layout member must be kept last.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Acked-by: Jens Axboe <jens.axboe@oracle.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
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>
This gives the filesystem more information about the writeback that
is happening. Trond requested this for the NFS unstable write handling,
and other filesystems might benefit from this too by beeing able to
distinguish between the different callers in more detail.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* _calc_stripe_info() changes to accommodate for grouping
calculations. Returns additional information
* old _prepare_pages() becomes _prepare_one_group()
which stores pages belonging to one device group.
* New _prepare_for_striping iterates on all groups calling
_prepare_one_group().
* Enable mounting of groups data_maps (group_width != 0)
[QUESTION]
what is faster A or B;
A. x += stride;
x = x % width + first_x;
B x += stride
if (x < last_x)
x = first_x;
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
* Rename _offset_dev_unit_off() to _calc_stripe_info()
and recieve a struct for the output params
* In _prepare_for_striping we only need to call
_calc_stripe_info() once. The other componets
are easy to calculate from that. This code
was inspired by what's done in truncate.
* Some code shifts that make sense now but will make
more sense when group support is added.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
If an object is referenced by a directory but does not
exist on a target, it is a very serious corruption that
means:
1. Either a power failure with very slim chance of it
happening. Because the directory update is always submitted
much after object creation, but if a directory is written
to one device and the object creation to another it might
theoretically happen.
2. It only ever happened to me while developing with BUGs
causing file corruption. Crashes could also cause it but
they are more like case 1.
In any way the object does not exist, so data is surely lost.
If there is a mix-up in the obj-id or data-map, then lost objects
can be salvaged by off-line fsck. The only recoverable information
is the directory name. By letting it appear as a regular empty file,
with date==0 (1970 Jan 1st) ownership to root, we enable recovery
of the only useful information. And also enable deletion or over-write.
I can see how this can hurt.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
* inode.c operations are full-pages based, and not actually
true scatter-gather
* Lets us use more pages at once upto 512 (from 249) in 64 bit
* Brings us much much closer to be able to use exofs's io_state engine
from objlayout driver. (Once I decide where to put the common code)
After RAID0 patch the outer (input) bio was never used as a bio, but
was simply a page carrier into the raid engine. Even in the simple
mirror/single-dev arrangement pages info was copied into a second bio.
It is now easer to just pass a pages array into the io_state and prepare
bio(s) once.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
We now support striping over mirror devices. Including variable sized
stripe_unit.
Some limits:
* stripe_unit must be a multiple of PAGE_SIZE
* stripe_unit * stripe_count is maximum upto 32-bit (4Gb)
Tested RAID0 over mirrors, RAID0 only, mirrors only. All check.
Design notes:
* I'm not using a vectored raid-engine mechanism yet. Following the
pnfs-objects-layout data-map structure, "Mirror" is just a private
case of "group_width" == 1, and RAID0 is a private case of
"Mirrors" == 1. The performance lose of the general case over the
particular special case optimization is totally negligible, also
considering the extra code size.
* In general I added a prepare_stripes() stage that divides the
to-be-io pages to the participating devices, the previous
exofs_ios_write/read, now becomes _write/read_mirrors and a new
write/read upper layer loops on all devices calling
_write/read_mirrors. Effectively the prepare_stripes stage is the all
secret.
Also truncate need fixing to accommodate for striping.
* In a RAID0 arrangement, in a regular usage scenario, if all inode
layouts will start at the same device, the small files fill up the
first device and the later devices stay empty, the farther the device
the emptier it is.
To fix that, each inode will start at a different stripe_unit,
according to it's obj_id modulus number-of-stripe-units. And
will then span all stripe-units in the same incrementing order
wrapping back to the beginning of the device table. We call it
a stripe-units moving window.
Special consideration was taken to keep all devices in a mirror
arrangement identical. So a broken osd-device could just be cloned
from one of the mirrors and no FS scrubbing is needed. (We do that
by rotating stripe-unit at a time and not a single device at a time.)
TODO:
We no longer verify object_length == inode->i_size in exofs_iget.
(since i_size is stripped on multiple objects now).
I should introduce a multiple-device attribute reading, and use
it in exofs_iget.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
* Layouts describe the way a file is spread on multiple devices.
The layout information is stored in the objects attribute introduced
in this patch.
* There can be multiple generating function for the layout.
Currently defined:
- No attribute present - use below moving-window on global
device table, all devices.
(This is the only one currently used in exofs)
- an obj_id generated moving window - the obj_id is a randomizing
factor in the otherwise global map layout.
- An explicit layout stored, including a data_map and a device
index list.
- More might be defined in future ...
* There are two attributes defined of the same structure:
A-data-files-layout - This layout is used by data-files. If present
at a directory, all files of that directory will
be created with this layout.
A-meta-data-layout - This layout is used by a directory and other
meta-data information. Also inherited at creation
of subdirectories.
* At creation time inodes are created with the layout specified above.
A usermode utility may change the creation layout on a give directory
or file. Which in the case of directories, will also apply to newly
created files/subdirectories, children of that directory.
In the simple unaltered case of a newly created exofs, no layout
attributes are present, and all layouts adhere to the layout specified
at the device-table.
* In case of a future file system loaded in an old exofs-driver.
At iget(), the generating_function is inspected and if not supported
will return an IO error to the application and the inode will not
be loaded. So not to damage any data.
Note: After this patch we do not yet support any type of layout
only the RAID0 patch that enables striping at the super-block
level will add support for RAID0 layouts above. This way we
are past and future compatible and fully bisectable.
* Access to the device table is done by an accessor since
it will change according to above information.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
The original idea was that a mirror read can be sub-divided
to multiple devices. But this has very little gain and only
at very large IOes so it's not going to be implemented soon.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
* Abstract away those members in exofs_sb_info that are related/needed
by a layout into a new exofs_layout structure. Embed it in exofs_sb_info.
* At exofs_io_state receive/keep a pointer to an exofs_layout. No need for
an exofs_sb_info pointer, all we need is at exofs_layout.
* Change any usage of above exofs_sb_info members to their new name.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
In check_io, implement the case of reading passed end of
file, by clearing the pages and recover with no error. In
a raid arrangement this can become a legitimate situation
in case of holes in the file.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
optimize the exofs_i_info struct usage by moving the embedded
vfs_inode to be first. A compiler might optimize away an "add"
operation with constant zero. (Which it cannot with other constants)
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
* Last debug trimming left in some stupid print, remove them.
Fixup some other prints
* Shift printing from inode.c to ios.c
* Add couple of prints when memory allocation fails.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
exofs uses simple_write_end() for it's .write_end handler. But
it is not enough because simple_write_end() does not call
mark_inode_dirty() when it extends i_size. So even if we do
call mark_inode_dirty at beginning of write out, with a very
long IO and a saturated system we might get the .write_inode()
called while still extend-writing to file and miss out on the last
i_size updates.
So override .write_end, call simple_write_end(), and afterwords if
i_size was changed call mark_inode_dirty().
It stands to logic that since simple_write_end() was the one extending
i_size it should also call mark_inode_dirty(). But it looks like all
users of simple_write_end() are memory-bound pseudo filesystems, who
could careless about mark_inode_dirty(). I might submit a
warning-comment patch to simple_write_end() in future.
CC: Stable <stable@kernel.org>
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Some on disk exofs constants and types are defined in the pnfs_osd_xdr.h
file. Since we needed these types before the pnfs-objects code was
accepted to mainline we duplicated the minimal needed definitions into
an exofs local header. The definitions where conditionally included
depending on !CONFIG_PNFS defined. So if PNFS was present in the tree
definitions are taken from there and if not they are defined locally.
That was all good but, the CONFIG_PNFS is planed to be included upstream
before the pnfs-objects is also included. (The first pnfs batch might be
pnfs-files only)
So condition exofs local definitions on the absence of pnfs_osd_xdr.h
inclusion (__PNFS_OSD_XDR_H__ not defined). User code must make sure
that in future pnfs_osd_xdr.h will be included before fs/exofs/pnfs.h,
which happens to be so in current code.
Once pnfs-objects hits mainline, exofs's local header will be removed.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
This patch changes on-disk format, it is accompanied with a parallel
patch to mkfs.exofs that enables multi-device capabilities.
After this patch, old exofs will refuse to mount a new formatted FS and
new exofs will refuse an old format. This is done by moving the magic
field offset inside the FSCB. A new FSCB *version* field was added. In
the future, exofs will refuse to mount unmatched FSCB version. To
up-grade or down-grade an exofs one must use mkfs.exofs --upgrade option
before mounting.
Introduced, a new object that contains a *device-table*. This object
contains the default *data-map* and a linear array of devices
information, which identifies the devices used in the filesystem. This
object is only written to offline by mkfs.exofs. This is why it is kept
separate from the FSCB, since the later is written to while mounted.
Same partition number, same object number is used on all devices only
the device varies.
* define the new format, then load the device table on mount time make
sure every thing is supported.
* Change I/O engine to now support Mirror IO, .i.e write same data
to multiple devices, read from a random device to spread the
read-load from multiple clients (TODO: stripe read)
Implementation notes:
A few points introduced in previous patch should be mentioned here:
* Special care was made so absolutlly all operation that have any chance
of failing are done before any osd-request is executed. This is to
minimize the need for a data consistency recovery, to only real IO
errors.
* Each IO state has a kref. It starts at 1, any osd-request executed
will increment the kref, finally when all are executed the first ref
is dropped. At IO-done, each request completion decrements the kref,
the last one to return executes the internal _last_io() routine.
_last_io() will call the registered io_state_done. On sync mode a
caller does not supply a done method, indicating a synchronous
request, the caller is put to sleep and a special io_state_done is
registered that will awaken the caller. Though also in sync mode all
operations are executed in parallel.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
In anticipation for multi-device operations, we separate osd operations
into an abstract I/O API. Currently only one device is used but later
when adding more devices, we will drive all devices in parallel according
to a "data_map" that describes how data is arranged on multiple devices.
The file system level operates, like before, as if there is one object
(inode-number) and an i_size. The io engine will split this to the same
object-number but on multiple device.
At first we introduce Mirror (raid 1) layout. But at the final outcome
we intend to fully implement the pNFS-Objects data-map, including
raid 0,4,5,6 over mirrored devices, over multiple device-groups. And
more. See: http://tools.ietf.org/html/draft-ietf-nfsv4-pnfs-obj-12
* Define an io_state based API for accessing osd storage devices
in an abstract way.
Usage:
First a caller allocates an io state with:
exofs_get_io_state(struct exofs_sb_info *sbi,
struct exofs_io_state** ios);
Then calles one of:
exofs_sbi_create(struct exofs_io_state *ios);
exofs_sbi_remove(struct exofs_io_state *ios);
exofs_sbi_write(struct exofs_io_state *ios);
exofs_sbi_read(struct exofs_io_state *ios);
exofs_oi_truncate(struct exofs_i_info *oi, u64 new_len);
And when done
exofs_put_io_state(struct exofs_io_state *ios);
* Convert all source files to use this new API
* Convert from bio_alloc to bio_kmalloc
* In io engine we make use of the now fixed osd_req_decode_sense
There are no functional changes or on disk additions after this patch.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
If I do a "git mv" together with a massive code change
and commit in one patch, git looses the rename and
records a delete/new instead. This is bad because I want
a rename recorded so later rebased/cherry-picked patches
to the old name will work. Also the --follow is lost.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Even though exofs has a 4k block size, statfs blocks
is in sectors (512 bytes).
Also if target returns 0 for capacity then make it
ULLONG_MAX. df does not like zero-size filesystems
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
It is important to print in the logs when a filesystem was
mounted and eventually unmounted.
Print the osd-device's osd_name and pid the FS was
mounted/unmounted on.
TODO: How to also print the namespace path the filesystem was
mounted on?
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
There are two places that initialize inodes: exofs_iget() and
exofs_new_inode()
As more members of exofs_i_info that need initialization are
added this code will grow. (soon)
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Iner-loops printing is converted to EXOFS_DBG2 which is #defined
to nothing.
It is now almost bareable to just leave debug-on. Every operation
is printed once, with most relevant info (I hope).
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
the two places inside exofs that where taking the BKL were:
exofs_put_super() - .put_super
and
exofs_sync_fs() - which is .sync_fs and is also called from
.write_super.
Now exofs_sync_fs() is protected from itself by also taking
the sb_lock.
exofs_put_super() directly calls exofs_sync_fs() so there is no
danger between these two either.
In anyway there is absolutely nothing dangerous been done
inside exofs_sync_fs().
Unless there is some subtle race with the actual lifetime of
the super_block in regard to .put_super and some other parts
of the VFS. Which is highly unlikely.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* Remove smp_lock.h from files which don't need it (including some headers!)
* Add smp_lock.h to files which do need it
* Make smp_lock.h include conditional in hardirq.h
It's needed only for one kernel_locked() usage which is under CONFIG_PREEMPT
This will make hardirq.h inclusion cheaper for every PREEMPT=n config
(which includes allmodconfig/allyesconfig, BTW)
Signed-off-by: Alexey Dobriyan <adobriyan@gmail.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
The use of file_fsync() in exofs_file_sync() is not necessary since it
does some extra stuff not used by exofs. Open code just the parts that
are currently needed.
TODO: Farther optimization can be done to sync the sb only on inode
update of new files, Usually the sb update is not needed in exofs.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Boaz,
Congrats on getting all the OSD stuff into 2.6.30!
I just pulled the git, and saw that the IBM copyrights are still there.
Please remove them from all files:
* Copyright (C) 2005, 2006
* International Business Machines
IBM has revoked all rights on the code - they gave it to me.
Thanks!
Avishay
Signed-off-by: Avishay Traeger <avishay@gmail.com>
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
When failing a read request in the sync path, called from
write_begin, I forgot to free the allocated bio, fix it.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Conflicts:
drivers/message/fusion/mptsas.c
fixed up conflict between req->data_len accessors and mptsas driver updates.
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Add a ->sync_fs method for data integrity syncs, and reimplement
->write_super ontop of it.
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Push down lock_super into ->write_super instances and remove it from the
caller.
Following filesystem don't need ->s_lock in ->write_super and are skipped:
* bfs, nilfs2 - no other uses of s_lock and have internal locks in
->write_super
* ext2 - uses BKL in ext2_write_super and has internal calls without s_lock
* reiserfs - no other uses of s_lock as has reiserfs_write_lock (BKL) in
->write_super
* xfs - no other uses of s_lock and uses internal lock (buffer lock on
superblock buffer) to serialize ->write_super. Also xfs_fs_write_super
is superflous and will go away in the next merge window
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Move BKL into ->put_super from the only caller. A couple of
filesystems had trivial enough ->put_super (only kfree and NULLing of
s_fs_info + stuff in there) to not get any locking: coda, cramfs, efs,
hugetlbfs, omfs, qnx4, shmem, all others got the full treatment. Most
of them probably don't need it, but I'd rather sort that out individually.
Preferably after all the other BKL pushdowns in that area.
[AV: original used to move lock_super() down as well; these changes are
removed since we don't do lock_super() at all in generic_shutdown_super()
now]
[AV: fuse, btrfs and xfs are known to need no damn BKL, exempt]
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
We just did a full fs writeout using sync_filesystem before, and if
that's not enough for the filesystem it can perform it's own writeout
in ->put_super, which many filesystems already do.
Move a call to foofs_write_super into every foofs_put_super for now to
guarantee identical behaviour until it's cleaned up by the individual
filesystem maintainers.
Exceptions:
- affs already has identical copy & pasted code at the beginning of
affs_put_super so no need to do it twice.
- xfs does the right thing without it and I have changes pending for
the xfs tree touching this are so I don't really need conflicts
here..
Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
libosd users that need to work with bios, must sometime use
the request_queue associated with the osd_dev. Make a wrapper for
that, and convert all in-tree users.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
For supporting of chained-bios we can not inspect the first
bio only, as before. Caller shall pass the total length of the
request, ie. sum_bytes(bio-chain).
Also since the bio might be a chain we don't set it's direction
on behalf of it's callers. The bio direction should be properly
set prior to this call. So fix a couple of write users that now
need to set the bio direction properly
[In this patch I change both library code and user sites at
exofs, to make it easy on integration. It should be submitted
via James's scsi-misc tree.]
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
CC: Jeff Garzik <jeff@garzik.org>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
By popular demand, define usefull wrappers for osd_req_read/write
that recieve kernel pointers. All users had their own.
Also remove these from exofs
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
rq->data_len served two purposes - the length of data buffer on issue
and the residual count on completion. This duality creates some
headaches.
First of all, block layer and low level drivers can't really determine
what rq->data_len contains while a request is executing. It could be
the total request length or it coulde be anything else one of the
lower layers is using to keep track of residual count. This
complicates things because blk_rq_bytes() and thus
[__]blk_end_request_all() relies on rq->data_len for PC commands.
Drivers which want to report residual count should first cache the
total request length, update rq->data_len and then complete the
request with the cached data length.
Secondly, it makes requests default to reporting full residual count,
ie. reporting that no data transfer occurred. The residual count is
an exception not the norm; however, the driver should clear
rq->data_len to zero to signify the normal cases while leaving it
alone means no data transfer occurred at all. This reverse default
behavior complicates code unnecessarily and renders block PC on some
drivers (ide-tape/floppy) unuseable.
This patch adds rq->resid_len which is used only for residual count.
While at it, remove now unnecessasry blk_rq_bytes() caching in
ide_pc_intr() as rq->data_len is not changed anymore.
Boaz : spotted missing conversion in osd
Sergei : spotted too early conversion to blk_rq_bytes() in ide-tape
[ Impact: cleanup residual count handling, report 0 resid by default ]
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Cc: Borislav Petkov <petkovbb@googlemail.com>
Cc: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Cc: Mike Miller <mike.miller@hp.com>
Cc: Eric Moore <Eric.Moore@lsi.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Doug Gilbert <dgilbert@interlog.com>
Cc: Mike Miller <mike.miller@hp.com>
Cc: Eric Moore <Eric.Moore@lsi.com>
Cc: Darrick J. Wong <djwong@us.ibm.com>
Cc: Pete Zaitcev <zaitcev@redhat.com>
Cc: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Added some documentation in exofs.txt, as well as a BUGS file.
For further reading, operation instructions, example scripts
and up to date infomation and code please see:
http://open-osd.org
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
This patch ties all operation vectors into a file system superblock
and registers the exofs file_system_type at module's load time.
* The file system control block (AKA on-disk superblock) resides in
an object with a special ID (defined in common.h).
Information included in the file system control block is used to
fill the in-memory superblock structure at mount time. This object
is created before the file system is used by mkexofs.c It contains
information such as:
- The file system's magic number
- The next inode number to be allocated
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
implementation of directory and inode operations.
* A directory is treated as a file, and essentially contains a list
of <file name, inode #> pairs for files that are found in that
directory. The object IDs correspond to the files' inode numbers
and are allocated using a 64bit incrementing global counter.
* Each file's control block (AKA on-disk inode) is stored in its
object's attributes. This applies to both regular files and other
types (directories, device files, symlinks, etc.).
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
OK Now we start to read and write from osd-objects. We try to
collect at most contiguous pages as possible in a single write/read.
The first page index is the object's offset.
TODO:
In 64-bit a single bio can carry at most 128 pages.
Add support of chaining multiple bios
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
implementation of the file_operations and inode_operations for
regular data files.
Most file_operations are generic vfs implementations except:
- exofs_truncate will truncate the OSD object as well
- Generic file_fsync is not good for none_bd devices so open code it
- The default for .flush in Linux is todo nothing so call exofs_fsync
on the file.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
This patch includes osd infrastructure that will be used later by
the file system.
Also the declarations of constants, on disk structures,
and prototypes.
And the Kbuild+Kconfig files needed to build the exofs module.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>