Pull exofs update from Boaz Harrosh: "They are all mostly fixes, except the most important patch by Artem Bityutskiy which removes the use of s_dirt. After this patch s_dirt can be completely removed from the tree." * 'for-linus' of git://git.open-osd.org/linux-open-osd: ore: Fix out-of-bounds access in _ios_obj() exofs: Use proper max_IO calculations from ore exofs: Fix __r4w_get_page when offset is beyond i_size exofs: stop using s_dirt exofs: readpage_strip: Add a BUG_ON to check for PageLocked(page)
_ios_obj() is accessed by group_index not device_table index. The oc->comps array is only a group_full of devices at a time it is not like ore_comp_dev() which is indexed by a global device_table index. This did not BUG until now because exofs only uses a single COMP for all devices. But with other FSs like PanFS this is not true. This bug was only in the write_path, all other users were using it correctly [This is a bug since 3.2 Kernel] CC: Stable Tree <stable@kernel.org> Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
exofs_max_io_pages should just use the ORE's calculated layout->max_io_length, And avoid unnecessary BUGs, calculations made here were also a layering violation. Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
It is very common for the end of the file to be unaligned on stripe size. But since we know it's beyond file's end then the XOR should be preformed with all zeros. Old code used to just read zeros out of the OSD devices, which is a great waist. But what scares me more about this situation is that, we now have pages attached to the file's mapping that are beyond i_size. I don't like the kind of bugs this calls for. Fix both birds, by returning a global ZERO_PAGE, if offset is beyond i_size. Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Exofs has the '->write_super()' handler and makes some use of the '->s_dirt' superblock flag, but it really needs neither of them because it never sets 's_dirt' to one which means the VFS never calls its '->write_super()' handler. Thus, remove both. Note, I am trying to remove both 's_dirt' and 'write_super()' from VFS altogether once all users are gone. Signed-off-by: Artem Bityutskiy <artem.bityutskiy@linux.intel.com> Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
readpage_strip can be called from several code paths all of which require that the page be locked before any operations are carried out. Since we export the exofs_readpage callback to the VFS, add a BUG_ON to check for PageLocked(page) to make sure that this understanding is never compromised. Signed-off-by: Kautuk Consul <consul.kautuk@gmail.com> Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Pull the big VFS changes from Al Viro:
"This one is *big* and changes quite a few things around VFS. What's in there:
- the first of two really major architecture changes - death to open
intents.
The former is finally there; it was very long in making, but with
Miklos getting through really hard and messy final push in
fs/namei.c, we finally have it. Unlike his variant, this one
doesn't introduce struct opendata; what we have instead is
->atomic_open() taking preallocated struct file * and passing
everything via its fields.
Instead of returning struct file *, it returns -E... on error, 0
on success and 1 in "deal with it yourself" case (e.g. symlink
found on server, etc.).
See comments before fs/namei.c:atomic_open(). That made a lot of
goodies finally possible and quite a few are in that pile:
->lookup(), ->d_revalidate() and ->create() do not get struct
nameidata * anymore; ->lookup() and ->d_revalidate() get lookup
flags instead, ->create() gets "do we want it exclusive" flag.
With the introduction of new helper (kern_path_locked()) we are rid
of all struct nameidata instances outside of fs/namei.c; it's still
visible in namei.h, but not for long. Come the next cycle,
declaration will move either to fs/internal.h or to fs/namei.c
itself. [me, miklos, hch]
- The second major change: behaviour of final fput(). Now we have
__fput() done without any locks held by caller *and* not from deep
in call stack.
That obviously lifts a lot of constraints on the locking in there.
Moreover, it's legal now to call fput() from atomic contexts (which
has immediately simplified life for aio.c). We also don't need
anti-recursion logics in __scm_destroy() anymore.
There is a price, though - the damn thing has become partially
asynchronous. For fput() from normal process we are guaranteed
that pending __fput() will be done before the caller returns to
userland, exits or gets stopped for ptrace.
For kernel threads and atomic contexts it's done via
schedule_work(), so theoretically we might need a way to make sure
it's finished; so far only one such place had been found, but there
might be more.
There's flush_delayed_fput() (do all pending __fput()) and there's
__fput_sync() (fput() analog doing __fput() immediately). I hope
we won't need them often; see warnings in fs/file_table.c for
details. [me, based on task_work series from Oleg merged last
cycle]
- sync series from Jan
- large part of "death to sync_supers()" work from Artem; the only
bits missing here are exofs and ext4 ones. As far as I understand,
those are going via the exofs and ext4 trees resp.; once they are
in, we can put ->write_super() to the rest, along with the thread
calling it.
- preparatory bits from unionmount series (from dhowells).
- assorted cleanups and fixes all over the place, as usual.
This is not the last pile for this cycle; there's at least jlayton's
ESTALE work and fsfreeze series (the latter - in dire need of fixes,
so I'm not sure it'll make the cut this cycle). I'll probably throw
symlink/hardlink restrictions stuff from Kees into the next pile, too.
Plus there's a lot of misc patches I hadn't thrown into that one -
it's large enough as it is..."
* 'for-linus-2' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: (127 commits)
ext4: switch EXT4_IOC_RESIZE_FS to mnt_want_write_file()
btrfs: switch btrfs_ioctl_balance() to mnt_want_write_file()
switch dentry_open() to struct path, make it grab references itself
spufs: shift dget/mntget towards dentry_open()
zoran: don't bother with struct file * in zoran_map
ecryptfs: don't reinvent the wheels, please - use struct completion
don't expose I_NEW inodes via dentry->d_inode
tidy up namei.c a bit
unobfuscate follow_up() a bit
ext3: pass custom EOF to generic_file_llseek_size()
ext4: use core vfs llseek code for dir seeks
vfs: allow custom EOF in generic_file_llseek code
vfs: Avoid unnecessary WB_SYNC_NONE writeback during sys_sync and reorder sync passes
vfs: Remove unnecessary flushing of block devices
vfs: Make sys_sync writeout also block device inodes
vfs: Create function for iterating over block devices
vfs: Reorder operations during sys_sync
quota: Move quota syncing to ->sync_fs method
quota: Split dquot_quota_sync() to writeback and cache flushing part
vfs: Move noop_backing_dev_info check from sync into writeback
...
The read-4-write pages are locked in address ascending order. But where unlocked in a way easiest for coding. Fix that, locks should be released in opposite order of locking, .i.e descending address order. I have not hit this dead-lock. It was found by inspecting the dbug print-outs. I suspect there is an higher lock at caller that protects us, but fix it regardless. Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Do to OOM situations the ore might fail to allocate all resources needed for IO of the full request. If some progress was possible it would proceed with a partial/short request, for the sake of forward progress. Since this crashes NFS-core and exofs is just fine without it just remove this contraption, and fail. TODO: Support real forward progress with some reserved allocations of resources, such as mem pools and/or bio_sets [Bug since 3.2 Kernel] CC: Stable Tree <stable@kernel.org> CC: Benny Halevy <bhalevy@tonian.com> Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
In RAID_5/6 We used to not permit an IO that it's end byte is not stripe_size aligned and spans more than one stripe. .i.e the caller must check if after submission the actual transferred bytes is shorter, and would need to resubmit a new IO with the remainder. Exofs supports this, and NFS was supposed to support this as well with it's short write mechanism. But late testing has exposed a CRASH when this is used with none-RPC layout-drivers. The change at NFS is deep and risky, in it's place the fix at ORE to lift the limitation is actually clean and simple. So here it is below. The principal here is that in the case of unaligned IO on both ends, beginning and end, we will send two read requests one like old code, before the calculation of the first stripe, and also a new site, before the calculation of the last stripe. If any "boundary" is aligned or the complete IO is within a single stripe. we do a single read like before. The code is clean and simple by splitting the old _read_4_write into 3 even parts: 1._read_4_write_first_stripe 2. _read_4_write_last_stripe 3. _read_4_write_execute And calling 1+3 at the same place as before. 2+3 before last stripe, and in the case of all in a single stripe then 1+2+3 is preformed additively. Why did I not think of it before. Well I had a strike of genius because I have stared at this code for 2 years, and did not find this simple solution, til today. Not that I did not try. This solution is much better for NFS than the previous supposedly solution because the short write was dealt with out-of-band after IO_done, which would cause for a seeky IO pattern where as in here we execute in order. At both solutions we do 2 separate reads, only here we do it within a single IO request. (And actually combine two writes into a single submission) NFS/exofs code need not change since the ORE API communicates the new shorter length on return, what will happen is that this case would not occur anymore. hurray!! [Stable this is an NFS bug since 3.2 Kernel should apply cleanly] CC: Stable Tree <stable@kernel.org> Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>