28 Mar, 2006
40 commits
-
Cache_fresh is now only used in cache.c, so unexport it.
Part of cache_fresh (setting CACHE_VALID) should really be done under the
lock, while part (calling cache_revisit_request etc) must be done outside the
lock. So we split it up appropriately.Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
This has been replaced by more traditional code.
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
- in cache_check, h must be non-NULL as it has been de-referenced,
so don't bother checking for NULL.- When a cache-item is updated, we need to call cache_revisit_request to see
if there is a pending request waiting for that item. We were using
a transition to CACHE_VALID to see if that was needed, however that is
wrong as an expired entry will still be marked 'valid' (as the data is valid
and will need to be released). So instead use an off transition for
CACHE_PENDING which is exactly the right thing to test.- Add a little bit more debugging info.
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
The C++-like 'template' approach proves to be too ugly and hard to work with.
The old 'template' won't go away until all users are updated.
Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
These were an unnecessary wart. Also only have one 'DefineSimpleCache..'
instead of two.Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Current svc_expkey holds a pointer to the svc_export structure, so updates to
that structure have to be in-place, which is a wart on the whole cache
infrastruct. So we break that linkage and just do a second lookup.If this became a performance issue, it would be possible to put a direct link
back in which was only used conditionally. i.e. when an object is replaced
in the cache, we set a flag in the old object. When dereferencing the link
from svc_expkey, if the flag is set, we drop the reference and do a fresh
lookup.Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
The 'auth_domain's are simply handles on internal data structures. They do
not cache information from user-space, and forcing them into the mold of a
'cache' misrepresents their true nature and causes confusion.Signed-off-by: Neil Brown
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Fix accidental underflow of the atomic counter.
Signed-off-by: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
This functionality is also need for operation of autofs v5 direct mounts.
Signed-off-by: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
We have to have a valid sbi here, or we'd have oopsed already. (There's a
dereference of sbi->catatonic a few lines above)Coverity #740
Signed-off-by: Dave Jones
Cc: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Signed-off-by: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
This patch define a new autofs packet for autofs v5 and updates the waitq.c
functions to handle the additional packet type.Signed-off-by: Ian Kent
Cc: Al Viro
Cc: Christoph Hellwig
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
This patch adds expire logic for autofs direct mounts.
Signed-off-by: Ian Kent
Cc: Al Viro
Cc: Christoph Hellwig
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
This patch adds a follow_link inode method for the root of an autofs direct
mount trigger. It also adds the corresponding mount options and updates the
show_mount method.Signed-off-by: Ian Kent
Cc: Al Viro
Cc: Christoph Hellwig
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
In order to be able to trigger a mount using the follow_link inode method the
nameidata struct that is passed in needs to have the vfsmount of the autofs
trigger not its parent.During a path walk if an autofs trigger is mounted on a dentry, when the
follow_link method is called, the nameidata struct contains the vfsmount and
mountpoint dentry of the parent mount while the dentry that is passed in is
the root of the autofs trigger mount. I believe it is impossible to get the
vfsmount of the trigger mount, within the follow_link method, when only the
parent vfsmount and the root dentry of the trigger mount are known.This patch updates the nameidata struct on entry to __do_follow_link if it
detects that it is out of date. It moves the path_to_nameidata to above
__do_follow_link to facilitate calling it from there. The dput_path is moved
as well as that seemed sensible. No changes are made to these two functions.Signed-off-by: Ian Kent
Cc: Al Viro
Cc: Christoph Hellwig
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Update autofs4 version.
Signed-off-by: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Change the functions may_umount and may_umount_tree to boolean functions to
aid code readability.Signed-off-by: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Rename the function simple_empty_nolock to __simple_empty in line with kernel
naming conventions.Signed-off-by: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Whitespace and formating changes to waitq code.
Signed-off-by: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Add show_options method to display autofs4 mount options in the proc
filesystem.Signed-off-by: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Remove the update of i_atime from autofs4 in favour of having VFS update it.
i_atime is never used for expire in autofs4.Signed-off-by: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Alter the expire semantics that define how "busyness" is determined.
Currently a last_used counter is updated on every revalidate from processes
other than the mount owner process group.This patch changes that so that an expire candidate is busy only if it has a
reference count greater than the expected minimum, such as when there is an
open file or working directory in use.This method is the only way that busyness can be established for direct mounts
within the new implementation. For consistency the expire semantic is made
the same for all mounts.A side effect of the patch is that mounts which remain mounted unessessarily
in the presence of some GUI programs that scan the filesystem should now
expire.Signed-off-by: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Fix the case where an expire returns busy on a tree mount when it is in fact
not busy. This case was overlooked when the patch to prevent the expiring
away of "scaffolding" directories for tree mounts was applied.The problem arises when a tree of mounts is a member of a map with other keys.
The current logic will not expire the tree if any other mount in the map is
busy. The solution is to maintain a "minimum" use count for each autofs
dentry and compare this to the actual dentry usage count during expire.Signed-off-by: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Simplify the expire tree traversal code by using a function from namespace.c
to calculate the next entry in the top down tree traversals carried out during
the expire operation.Signed-off-by: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Change the names of the boolean functions autofs4_check_mount and
autofs4_check_tree to autofs4_mount_busy and autofs4_tree_busy respectively
and alters their return codes to suit in order to aid code readabilty.A couple of white space cleanups are included as well.
Signed-off-by: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Addresse a problem where stale dentrys stop mounts from happening.
When a mount point directory is pre-created and a non-existent entry within it
is requested a dentry ends up being created within the mount point directory
which stops future mounts. The problem is solved by ignoring negative,
unhashed dentrys in the mount point d_subdirs list.Additionally the apparent cacheing of -ENOENT returns from requests is
removed. The test on d_time is a tautology and d_time is not initialised and
has an unexpected value. In short it doesn't do what it's meant to.The cacheing of failed requests to the daemon is important and will be
followed up later.Signed-off-by: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Change readdir routines to use the cursor based routines in libfs.c. This
removes reliance on old readdir code from 2.4 and should improve efficiency of
readdir in autofs4.Signed-off-by: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Whitespace and formating changes to lookup code.
Signed-off-by: Ian Kent
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Noted by Oleg Drokin:
We initialized an extra slot of struct kstatfs.spare, sometimes
causing stack corruption.Signed-off-by: Jeff Dike
Cc: Paolo 'Blaisorblade' Giarrusso
Cc: Oleg Drokin
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
This fixes a race in the starting of write_sigio_thread. Previously, some of
the data needed by the thread was initialized after the clone. If the thread
ran immediately, it would see the uninitialized data, including an empty
pollfds, which would cause it to hang.We move the data initialization to before the clone, and adjust the error
paths and cleanup accordingly.Signed-off-by: Jeff Dike
Cc: Paolo 'Blaisorblade' Giarrusso
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
Behavior when booting two UMLs with the same umid was broken. The second one
would steal the umid. This fixes that, making the second UML take a random
umid instead.Signed-off-by: Jeff Dike
Cc: Paolo 'Blaisorblade' Giarrusso
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
This fixes a process segfault where a signal was being delivered such that a
new stack page needed to be allocated to hold the signal frame. This was
tripping some logic in the page fault handler which wouldn't allocate the page
if the faulting address was more that 32 bytes lower than the current stack
pointer. Since a signal frame is greater than 32 bytes, this exercised that
case.It's fixed by updating the SP in the pt_regs before starting to copy the
signal frame. Since those are the registers that will be copied on to the
stack, we have to be careful to put the original SP, not the new one which
points to the signal frame, on the stack.Signed-off-by: Jeff Dike
Cc: Paolo 'Blaisorblade' Giarrusso
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
This adds a 'c' option to the ubd switch which turns off host file locking so
that the device can be shared, as with a cluster. There's also some
whitespace cleanup while I was in this file.Signed-off-by: Jeff Dike
Cc: Paolo 'Blaisorblade' Giarrusso
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds -
This rearranges the OS declarations by moving some declarations into os.h.
Signed-off-by: Jeff Dike
Cc: Paolo 'Blaisorblade' Giarrusso
Signed-off-by: Andrew Morton
Signed-off-by: Linus Torvalds
