Eric Paris [Mon, 31 Mar 2008 01:17:33 +0000 (12:17 +1100)]
selinux: introduce permissive types
Introduce the concept of a permissive type. A new ebitmap is introduced to
the policy database which indicates if a given type has the permissive bit
set or not. This bit is tested for the scontext of any denial. The bit is
meaningless on types which only appear as the target of a decision and never
the source. A domain running with a permissive type will be allowed to
perform any action similarly to when the system is globally set permissive.
Signed-off-by: Eric Paris <eparis@redhat.com> Acked-by: Stephen Smalley <sds@tycho.nsa.gov> Signed-off-by: James Morris <jmorris@namei.org>
Roland McGrath [Wed, 26 Mar 2008 22:46:39 +0000 (15:46 -0700)]
selinux: remove ptrace_sid
This changes checks related to ptrace to get rid of the ptrace_sid tracking.
It's good to disentangle the security model from the ptrace implementation
internals. It's sufficient to check against the SID of the ptracer at the
time a tracee attempts a transition.
Signed-off-by: Roland McGrath <roland@redhat.com> Acked-by: Stephen Smalley <sds@tycho.nsa.gov> Signed-off-by: James Morris <jmorris@namei.org>
Eric Paris [Tue, 11 Mar 2008 18:19:34 +0000 (14:19 -0400)]
SELinux: requesting no permissions in avc_has_perm_noaudit is a BUG()
This patch turns the case where we have a call into avc_has_perm with no
requested permissions into a BUG_ON. All callers to this should be in
the kernel and thus should be a function we need to fix if we ever hit
this. The /selinux/access permission checking it done directly in the
security server and not through the avc, so those requests which we
cannot control from userspace should not be able to trigger this BUG_ON.
Signed-off-by: Eric Paris <eparis@redhat.com> Acked-by: Stephen D. Smalley <sds@tycho.nsa.gov> Signed-off-by: James Morris <jmorris@namei.org>
Andrew Morton [Wed, 5 Mar 2008 23:05:08 +0000 (10:05 +1100)]
security: code cleanup
ERROR: "(foo*)" should be "(foo *)"
#168: FILE: security/selinux/hooks.c:2656:
+ "%s, rc=%d\n", __func__, (char*)value, -rc);
total: 1 errors, 0 warnings, 195 lines checked
./patches/security-replace-remaining-__function__-occurences.patch has style problems, please review. If any of these errors
are false positives report them to the maintainer, see
CHECKPATCH in MAINTAINERS.
Please run checkpatch prior to sending patches
Cc: Harvey Harrison <harvey.harrison@gmail.com> Cc: James Morris <jmorris@namei.org> Cc: Stephen Smalley <sds@tycho.nsa.gov> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: James Morris <jmorris@namei.org>
Signed-off-by: Harvey Harrison <harvey.harrison@gmail.com> Cc: James Morris <jmorris@namei.org> Cc: Stephen Smalley <sds@tycho.nsa.gov> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: James Morris <jmorris@namei.org>
Eric Paris [Thu, 28 Feb 2008 17:58:40 +0000 (12:58 -0500)]
SELinux: create new open permission
Adds a new open permission inside SELinux when 'opening' a file. The idea
is that opening a file and reading/writing to that file are not the same
thing. Its different if a program had its stdout redirected to /tmp/output
than if the program tried to directly open /tmp/output. This should allow
policy writers to more liberally give read/write permissions across the
policy while still blocking many design and programing flaws SELinux is so
good at catching today.
Signed-off-by: Eric Paris <eparis@redhat.com> Acked-by: Stephen Smalley <sds@tycho.nsa.gov> Reviewed-by: Paul Moore <paul.moore@hp.com> Signed-off-by: James Morris <jmorris@namei.org>
James Morris [Tue, 26 Feb 2008 09:42:02 +0000 (20:42 +1100)]
SELinux: unify printk messages
Replace "security:" prefixes in printk messages with "SELinux"
to help users identify the source of the messages. Also fix a
couple of minor formatting issues.
Paul Moore [Mon, 25 Feb 2008 16:40:33 +0000 (11:40 -0500)]
SELinux: Correct the NetLabel locking for the sk_security_struct
The RCU/spinlock locking approach for the nlbl_state in the sk_security_struct
was almost certainly overkill. This patch removes both the RCU and spinlock
locking, relying on the existing socket locks to handle the case of multiple
writers. This change also makes several code reductions possible.
Less locking, less code - it's a Good Thing.
Signed-off-by: Paul Moore <paul.moore@hp.com> Signed-off-by: James Morris <jmorris@namei.org>
[XFS] The forward declarations for the xfs_ioctl() helpers and the
associated comment about gcc behavior really aren't needed; all of these
functions are marked STATIC which includes noinline, and the stack usage
won't be a problem.
This effectively just removes the forward declarations and moves
xfs_ioctl() back to the end of the file.
Josef Sipek [Fri, 11 Apr 2008 07:11:02 +0000 (17:11 +1000)]
[XFS] Update XFS documentation for noikeep/ikeep.
Mention how DMAPI affects default for noikeep.
Slightly modified since Josef's patch was based on
an old xfs.txt prior to Dave's (dgc) checkin which
missed going to oss.
Signed-off-by: Josef Sipek <jeffpc@josefsipek.net> Signed-off-by: Tim Shimmin <tes@sgi.com>
Eric Sandeen [Thu, 17 Apr 2008 06:50:22 +0000 (16:50 +1000)]
[XFS] Remove CONFIG_XFS_SECURITY.
There is no point to the CONFIG_XFS_SECURITY option; it disables the
ability to set security attributes at runtime, but it does not actually
slim down or remove any code for runtime. Just remove it and always allow
security attributes to be set.
Tim Shimmin [Thu, 17 Apr 2008 06:50:16 +0000 (16:50 +1000)]
[XFS] xfs_bmap_compute_maxlevels should be based on di_forkoff
Fix up xfs_bmap_compute_maxlevels() to account for the case when we go
from using attr2 to using attr1. In that case attr1 will no longer
necessarily be at m_attr_offset>>3, but could be at a different value for
di_forkoff. Therefore, we return the worst case scenario using MINDBTPTRS
and MINABTPTRS, as this function is used for determining the maximum log
space.
Eric Sandeen [Thu, 17 Apr 2008 06:50:09 +0000 (16:50 +1000)]
[XFS] Always use di_forkoff when checking for attr space.
In the case where we mount a filesystem which was previously using the
attr2 format as attr1, returning the default mp->m_attroffset instead of
the per-inode di_forkoff for inline attribute fit calculations, may result
in corruption, if for example, the data fork is already taking more space
than the default fork offset and we try to add an extended attribute. Fix
tested by xfstests/186.
David Chinner [Thu, 17 Apr 2008 06:49:55 +0000 (16:49 +1000)]
[XFS] Remove periodic logging of in-core superblock counters.
xfssyncd triggers the logging of superblock counters every 30s if the
filesystem is made with lazy-count=1. This will prevent disks from idling
and spinning down as there will be a log write every 30s. With the way
counter recovery works for lazy-count=1, this code is unnecessary and
provides no real benefit, so just remove it.
David Chinner [Thu, 10 Apr 2008 02:24:30 +0000 (12:24 +1000)]
[XFS] Sanitise xfs_log_force error checking.
xfs_log_force() is declared to return an error, but we almost never check
it. We don't need to check it in most cases; if there's a log I/O error
then we'll be shutting down the filesystem anyway and that means we'll
catch the error somewhere else.
However, on certain calls we should be returning an error - sync
transactions, fsync, sync writes, etc. so this isn't a pure black and
white distinction. Hence make xfs_log_force() a void function that issues
a warning to the syslog on error, and call _xfs_log_force() in all the
places where we actually care about the error status returned.
David Chinner [Thu, 10 Apr 2008 02:24:17 +0000 (12:24 +1000)]
[XFS] Don't allow silent errors in xfs_inactive().
xfs_inactive() fails to report errors when committing the inactive
transaction. Hence we can get silent failures either finishing off the
truncation or committing the transaction. Even if we get errors, we need
to continue, so simply warn loudly to the system if we get errors here.
David Chinner [Thu, 10 Apr 2008 02:23:52 +0000 (12:23 +1000)]
[XFS] Catch unwritten extent conversion errors.
On unwritten I/O completion, we fail to propagate an error when converting
the extent to a written extent. This means that the I/O silently fails.
propagate the error onto the ioend so that the inode is marked with an
error appropriately.
David Chinner [Thu, 10 Apr 2008 02:23:46 +0000 (12:23 +1000)]
[XFS] xfs_bdwrite() does not return errors.
xfs_bdwrite() cannot return an error; it only queues buffers to the
delayed write list and as such never encounters anything that can fail.
Mark it void.
David Chinner [Thu, 10 Apr 2008 02:22:24 +0000 (12:22 +1000)]
[XFS] Ensure xfs_bawrite() errors are checked.
xfs_bawrite() can return immediate error status on async writes. Unlike
xfsbdstrat() we don't ever check the error on the buffer after the call,
so we currently do not catch errors at all here. Ensure we catch and
propagate or warn to the syslog about up-front async write errors.
David Chinner [Thu, 10 Apr 2008 02:22:17 +0000 (12:22 +1000)]
[XFS] Ensure errors from xfs_bdstrat() are correctly checked.
xfsbdstrat() is declared to return an error. That is never checked because
the error is propagated by the xfs_buf_t that is passed through the
function.
Mark xfsbdstrat() as returning void and comment the prototype on the
methods needed for error checking.
David Chinner [Thu, 10 Apr 2008 02:21:53 +0000 (12:21 +1000)]
[XFS] Check for xfs_free_extent() failing.
xfs_free_extent() can fail, but log recovery never bothers to check if it
successfully free the extent it was supposed to. This could lead to silent
corruption during log recovery. Abort log recovery if we fail to free an
extent.
David Chinner [Thu, 10 Apr 2008 02:21:46 +0000 (12:21 +1000)]
[XFS] Warn if errors come from block_truncate_page().
block_truncate_page() can return errors that we currently ignore and
silently discard. We should not ever get errors reported here - an error
indicates a bug somewhere else. Hence catch the error and issue a stack
dump to the syslog because we cannot propagate the error any further up
the call chain.
David Chinner [Thu, 10 Apr 2008 02:21:32 +0000 (12:21 +1000)]
[XFS] Make xfs_alloc_compute_aligned() void.
xfs_alloc_compute_aligned() returns a value based on a comparison of the
computed extent length and the minimum length allowed. This is only used
by some callers - the other four return parameters are used more often.
Hence move the comparison to the code that actually needs to do it and
make xfs_alloc_compute_aligned() a void function.
David Chinner [Thu, 10 Apr 2008 02:21:25 +0000 (12:21 +1000)]
[XFS] Clean up xfs_alloc_search_busy() return values.
xfs_alloc_search_busy() returns an index into the busy array if the extent
was found in the array. This is never checked, and the
xfs_alloc_search_busy() does a log force to prevent reuse of the extent
before the free transaction hits the disk. Hence the return value is
useless. Declare the function void and remove the slot number from the
tracing as well.
David Chinner [Thu, 10 Apr 2008 02:21:18 +0000 (12:21 +1000)]
[XFS] Propagate errors from xfs_trans_commit().
xfs_trans_commit() can return errors when there are problems in the
transaction subsystem. They are indicative that the entire transaction may
be incomplete, and hence the error should be propagated as there is a good
possibility that there is something fatally wrong in the filesystem. Catch
and propagate or warn about commit errors in the places where they are
currently ignored.
David Chinner [Thu, 10 Apr 2008 02:21:11 +0000 (12:21 +1000)]
[XFS] Propagate xfs_trans_reserve() errors.
xfs_trans_reserve() reports errors that should not be ignored. For
example, a shutdown filesystem will report errors through
xfs_trans_reserve() to prevent further changes from being attempted on a
damaged filesystem. Catch and propagate all error conditions from
xfs_trans_reserve().
David Chinner [Thu, 10 Apr 2008 02:20:45 +0000 (12:20 +1000)]
[XFS] Catch errors when turning off quotas.
When turning off quota, we need to write various transactions to the log
to ensure that they are cleanly removed in the case of a crash. We need to
check that the transactions hit the disk correctly. If we fail to write
the final quota off transaction, we are corrupt in memory and so the only
option is to shut the filesystem down at this point.
David Chinner [Thu, 10 Apr 2008 02:20:31 +0000 (12:20 +1000)]
[XFS] Clean up quotamount error handling.
xfs_qm_mount_quotas() returns an error status that is ignored. If we fail
to mount quotas, we continue with quota's turned off, which is all handled
inside xfs_qm_mount_quotas(). Mark it as void to indicate that errors need
not be returned to the callers.
David Chinner [Thu, 10 Apr 2008 02:20:24 +0000 (12:20 +1000)]
[XFS] Check for dquot flush errors
xfs_qm_dqflush() can fail, but the return is not checked anywhere. Hence
we never know if we've failed to flush a dquot to disk. Propagate the
error and warn to the syslog if a flush ever fails.
David Chinner [Thu, 10 Apr 2008 02:20:03 +0000 (12:20 +1000)]
[XFS] Report errors from xfs_reserve_blocks().
xfs_reserve_blocks() can fail in interesting ways. In neither case is it a
fatal error, but the result can lead to sub-optimal behaviour. Warn to the
syslog if the call fails but otherwise continue.
David Chinner [Thu, 10 Apr 2008 02:19:02 +0000 (12:19 +1000)]
[XFS] Fix lock inversion in forced shutdown.
Recent changes to xlog_state_release_iclog() placed the grant_lock inside
the icloglock. forced unmount of the log does this the opposite way
around, but does not depend on the order for correct working. Fix the
inversion by changing the order locks are gained in
xfs_log_force_umount().
David Chinner [Thu, 10 Apr 2008 02:18:54 +0000 (12:18 +1000)]
[XFS] Reorganise xlog_t for better cacheline isolation of contention
To reduce contention on the log in large CPU count, separate out different
parts of the xlog_t structure onto different cachelines. Move each lock
onto a different cacheline along with all the members that are
accessed/modified while that lock is held.
David Chinner [Thu, 10 Apr 2008 02:18:46 +0000 (12:18 +1000)]
[XFS] Remove the xlog_ticket allocator
The ticket allocator is just a simple slab implementation internal to the
log. It requires the icloglock to be held when manipulating it and this
contributes to contention on that lock.
Just kill the entire allocator and use a memory zone instead. While there,
allow us to gracefully fail allocation with ENOMEM.
David Chinner [Thu, 10 Apr 2008 02:18:39 +0000 (12:18 +1000)]
[XFS] Per iclog callback chain lock
Rather than use the icloglock for protecting the iclog completion callback
chain, use a new per-iclog lock so that walking the callback chain doesn't
require holding a global lock.
This reduces contention on the icloglock during transaction commit and log
I/O completion by reducing the number of times we need to hold the global
icloglock during these operations.
While investigating the extent corruption bug I ran into this bug in debug
only code. xfs_bmap_check_leaf_extents() loops through the leaf blocks of
the extent btree checking that every extent is entirely before the next
extent. It also compares the last extent in the previous block to the
first extent in the current block when the previous block has been
released and potentially unmapped. So take a copy of the last extent
instead of a pointer. Also move the last extent check out of the loop
because we only need to do it once.
Most VN_RELE calls either directly contain a XFS_ITOV or have the
corresponding xfs_inode already in scope. Use the IRELE helper instead of
VN_RELE to clarify the code. With a little more work we can kill VN_RELE
altogether and define IRELE in terms of iput directly.
[XFS] cleanup root inode handling in xfs_fs_fill_super
- rename rootvp to root for clarify
- remove useless vn_to_inode call
- check is_bad_inode before calling d_alloc_root
- use iput instead of VN_RELE in the error case
David Chinner [Thu, 27 Mar 2008 07:00:45 +0000 (18:00 +1100)]
[XFS] Ensure a btree insert returns a valid cursor.
When writing into preallocated regions there is a case where XFS can oops
or hang doing the unwritten extent conversion on I/O completion. It turns
out that the problem is related to the btree cursor being invalid.
When we do an insert into the tree, we may need to split blocks in the
tree. When we only split at the leaf level (i.e. level 0), everything
works just fine. However, if we have a multi-level split in the btreee,
the cursor passed to the insert function is no longer valid once the
insert is complete.
The leaf level split is handled correctly because all the operations at
level 0 are done using the original cursor, hence it is updated correctly.
However, when we need to update the next level up the tree, we don't use
that cursor - we use a cloned cursor that points to the index in the next
level up where we need to do the insert.
Hence if we need to split a second level, the changes to the tree are
reflected in the cloned cursor and not the original cursor. This
clone-and-move-up-a-level-on-split behaviour recurses all the way to the
top of the tree.
The complexity here is that these cloned cursors do not point to the
original index that was inserted - they point to the newly allocated block
(the right block) and the original cursor pointer to that level may still
point to the left block. Hence, without deep examination of the cloned
cursor and buffers, we cannot update the original cursor with the new path
from the cloned cursor.
In these cases the original cursor could be pointing to the wrong block(s)
and hence a subsequent modification to the tree using that cursor will
lead to corruption of the tree.
The crash case occurs when the tree changes height - we insert a new level
in the tree, and the cursor does not have a buffer in it's path for that
level. Hence any attempt to walk back up the cursor to the root block will
result in a null pointer dereference.
To make matters even more complex, the BMAP BT is rooted in an inode, so
we can have a change of height in the btree *without a root split*. That
is, if the root block in the inode is full when we split a leaf node, we
cannot fit the pointer to the new block in the root, so we allocate a new
block, migrate all the ptrs out of the inode into the new block and point
the inode root block at the newly allocated block. This changes the height
of the tree without a root split having occurred and hence invalidates the
path in the original cursor.
The patch below prevents xfs_bmbt_insert() from returning with an invalid
cursor by detecting the cases that invalidate the original cursor and
refresh it by do a lookup into the btree for the original index we were
inserting at.
Note that the INOBT, AGFBNO and AGFCNT btree implementations also have
this bug, but the cursor is currently always destroyed or revalidated
after an insert for those trees. Hence this patch only address the problem
in the BMBT code.
David Chinner [Thu, 27 Mar 2008 07:00:38 +0000 (18:00 +1100)]
[XFS] Account for inode cluster alignment in all allocations
At ENOSPC, we can get a filesystem shutdown due to a cancelling a dirty
transaction in xfs_mkdir or xfs_create. This is due to the initial
allocation attempt not taking into account inode alignment and hence we
can prepare the AGF freelist for allocation when it's not actually
possible to do an allocation. This results in inode allocation returning
ENOSPC with a dirty transaction, and hence we shut down the filesystem.
Because the first allocation is an exact allocation attempt, we must tell
the allocator that the alignment does not affect the allocation attempt.
i.e. we will accept any extent alignment as long as the extent starts at
the block we want. Unfortunately, this means that if the longest free
extent is less than the length + alignment necessary for fallback
allocation attempts but is long enough to attempt a non-aligned
allocation, we will modify the free list.
If we then have the exact allocation fail, all other allocation attempts
will also fail due to the alignment constraint being taken into account.
Hence the initial attempt needs to set the "alignment slop" field so that
alignment, while not required, must be taken into account when determining
if there is enough space left in the AG to do the allocation.
That means if the exact allocation fails, we will not dirty the freelist
if there is not enough space available fo a subsequent allocation to
succeed. Hence we get an ENOSPC error back to userspace without shutting
down the filesystem.
[XFS] Replace custom AIL linked-list code with struct list_head
Replace the xfs_ail_entry_t with a struct list_head and clean the
surrounding code up. Also fixes a livelock in xfs_trans_first_push_ail()
by terminating the loop at the head of the list correctly.
[XFS] Remove superflous xfs_readsb call in xfs_mountfs.
When xfs_mountfs is called by xfs_mount xfs_readsb was called 35 lines
above unconditionally, so there is no need to try to read the superblock
if it's not present. If any other port doesn't have the superblock read at
this point it should just call it directly from it's xfs_mount equivalent.
Niv Sardi [Thu, 6 Mar 2008 02:49:26 +0000 (13:49 +1100)]
[XFS] kill t_sema member of struct xfs_trans
It's completely unused so we might aswell kill it. Note that there is
another t_sema in struct xlog_ticket, which is used and actually an sv_t
despite the name. That one is left untouched by this patch.
David Chinner [Thu, 6 Mar 2008 02:45:43 +0000 (13:45 +1100)]
[XFS] Use power-of-2 sized buffers to reduce overhead
Now that the ktrace_enter() code is using atomics, the non-power-of-2
buffer sizes - which require modulus operations to get the index - are
showing up as using substantial CPU in the profiles.
Force the buffer sizes to be rounded up to the nearest power of two and
use masking rather than modulus operations to convert the index counter to
the buffer index. This reduces ktrace_enter overhead to 8% of a CPU time,
and again almost halves the trace intensive test runtime.
David Chinner [Thu, 6 Mar 2008 02:45:35 +0000 (13:45 +1100)]
[XFS] Use atomic counters for ktrace buffer indexes
ktrace_enter() is consuming vast amounts of CPU time due to the use of a
single global lock for protecting buffer index increments. Change it to
use per-buffer atomic counters - this reduces ktrace_enter() overhead
during a trace intensive test on a 4p machine from 58% of all CPU time to
12% and halves test runtime.
David Chinner [Thu, 6 Mar 2008 02:45:29 +0000 (13:45 +1100)]
[XFS] Update c/mtime correctly on truncates
XFS changes the c/mtime of an inode when truncating it to the same size.
The c/mtime is only supposed to change if the size is changed. Not to be
confused with ftruncate, where the c/mtime is supposed to be changed even
if the size is not changed.
The Linux VFS encodes this semantic difference in the flags it sends down
to ->setattr, which XFS currently ignores. We need to make XFS pay
attention to the VFS flags and hence Do The Right Thing.
[XFS] don't encode parent in nfs filehandles unless nessecary
As Dave pointed out after the export ops changes we now always encode the
parent into the filehandle for regular files, but it's not actually needed
when the filesystem is export with no_subtree_check. This one-liner fixes
xfs_fs_encode_fh to skip encoding the parent unless nessecary.
Instead of of xfs_get_dir_entry use a macro to get the xfs_inode from the
dentry in the callers and grab the reference manually.
Only grab the reference once as it's fine to keep it over the dmapi calls.
(And even that reference is actually superflous in Linux but I'll leave
that for another patch)
- use proper goto based unwinding instead of the current mess of
multiple conditionals
- rename ip to inode because that's the normal convention for Linux
inodes while ip is the convention for xfs_inodes
- remove unlikely checks for the default_acl - branches marked unlikely
might lead to extreme branch bredictor slowdons if taken and for some
workloads a default acl is quite common
- properly indent the switch statements
- remove xfs_has_fs_struct as nfsd has a fs_struct in any semi-recent
kernel
David Chinner [Thu, 6 Mar 2008 02:44:14 +0000 (13:44 +1100)]
[XFS] Use atomics for iclog reference counting
Now that we update the log tail LSN less frequently on transaction
completion, we pass the contention straight to the global log state lock
(l_iclog_lock) during transaction completion.
We currently have to take this lock to decrement the iclog reference
count. there is a reference count on each iclog, so we need to take þhe
global lock for all refcount changes.
When large numbers of processes are all doing small trnasctions, the iclog
reference counts will be quite high, and the state change that absolutely
requires the l_iclog_lock is the except rather than the norm.
Change the reference counting on the iclogs to use atomic_inc/dec so that
we can use atomic_dec_and_lock during transaction completion and avoid the
need for grabbing the l_iclog_lock for every reference count decrement
except the one that matters - the last.
David Chinner [Thu, 6 Mar 2008 02:44:06 +0000 (13:44 +1100)]
[XFS] Prevent AIL lock contention during transaction completion
When hundreds of processors attempt to commit transactions at the same
time, they can contend on the AIL lock when updating the tail LSN held in
the in-core log structure.
At the moment, the tail LSN is only needed when actually writing out an
iclog, so it really does not need to be updated on every single
transaction completion - only those that result in switching iclogs and
flushing them to disk.
The result is that we reduce the number of times we need to grab the AIL
lock and the log grant lock by up to two orders of magnitude on large
processor count machines. The problem has previously been hidden by AIL
lock contention walking the AIL list which was recently solved and
uncovered this issue.
David Chinner [Thu, 6 Mar 2008 02:43:49 +0000 (13:43 +1100)]
[XFS] Remove the xfs_icluster structure
Remove the xfs_icluster structure and replace with a radix tree lookup.
We don't need to keep a list of inodes in each cluster around anymore as
we can look them up quickly when we need to. The only time we need to do
this now is during inode writeback.
Factor the inode cluster writeback code out of xfs_iflush and convert it
to use radix_tree_gang_lookup() instead of walking a list of inodes built
when we first read in the inodes.
This remove 3 pointers from each xfs_inode structure and the xfs_icluster
structure per inode cluster. Hence we reduce the cache footprint of the
xfs_inodes by between 5-10% depending on cluster sparseness.
To be truly efficient we need a radix_tree_gang_lookup_range() call to
stop searching once we are past the end of the cluster instead of trying
to find a full cluster's worth of inodes.
David Chinner [Thu, 6 Mar 2008 02:43:42 +0000 (13:43 +1100)]
[XFS] Don't block pdflush when writing back inodes
When pdflush is writing back inodes, it can get stuck on inode cluster
buffers that are currently under I/O. This occurs when we write data to
multiple inodes in the same inode cluster at the same time.
Effectively, delayed allocation marks the inode dirty during the data
writeback. Hence if the inode cluster was flushed during the writeback of
the first inode, the writeback of the second inode will block waiting for
the inode cluster write to complete before writing it again for the newly
dirtied inode.
Basically, we want to avoid this from happening so we don't block pdflush
and slow down all of writeback. Hence we introduce a non-blocking async
inode flush flag that pdflush uses. If this flag is set, we use
non-blocking operations (e.g. try locks) whereever we can to avoid
blocking or extra I/O being issued.
David Chinner [Thu, 6 Mar 2008 02:43:34 +0000 (13:43 +1100)]
[XFS] Factor xfs_itobp() and xfs_inotobp().
The only difference between the functions is one passes an inode for the
lookup, the other passes an inode number. However, they don't do the same
validity checking or set all the same state on the buffer that is returned
yet they should.
Factor the functions into a common implementation.
Lachlan McIlroy [Thu, 6 Mar 2008 02:43:11 +0000 (13:43 +1100)]
[XFS] make inode reclaim synchronise with xfs_iflush_done()
On a forced shutdown, xfs_finish_reclaim() will skip flushing the inode.
If the inode flush lock is not already held and there is an outstanding
xfs_iflush_done() then we might free the inode prematurely. By acquiring
and releasing the flush lock we will synchronise with xfs_iflush_done().
projects / linux-2.6-omap-h63xx.git / log