warnings for the recent change of the type of a module event handler.
Fixed a rotted comment (numeric types of filesystems are not listed here).
Made the function protototype in VFS_SET() more like the corresponding
function definition (don't use extern for prototypes).
Enforce a semicolon after the LKM case of VFS_SET().
MALLOC_DEFINE() and MALLOC_DEFINE() is needed by the recently
reenabled "reallocblks" code, but <sys/kernel.h> was only included
if CLUSTERDEBUG was defined. This was too harmless. gcc only
warns about garbage like `SYSINIT(blech);' at file scope ...
devstart_start_transaction() call is misplaced - it is after the
wdustart() call that queues the transaction on the controller queue.
Normally this doesn't matter because we're running at splbio() so
nothing will look at the controller queue. However, obsolescent
code for syncing labels sometimes slept after the transaction was
started, so the transaction sometimes completed before it was
[recorded as] started. This code was misplaced even for syncing
labels. Move it to the right place. It should go away, but
something may depend on its side effects.
o Move fixups into extraction routine so all consumers don't have to duplicate
the right behavior.
o Make some things more orthogonal (just for asthetics sake)
o Add option to go back and do it again if XF86Setup fails (possibly with
a different setup - this one has always annoyed me).
- Interface wth the new resource manager.
- Allow for multiple drivers implementing a single devclass.
- Remove ordering dependencies between header files.
- Style cleanup.
- Add DEVICE_SUSPEND and DEVICE_RESUME methods.
- Move to a single-phase interrupt setup scheme.
Kernel builds on the Alpha are brken until Doug gets a chance to incorporate
these changes on that side.
Agreed to in principle by: dfr
This avoids the fsck-on-reboot symptoms if you're shutting down with a
hung or unreachable NFS server mounted. Also remove non-local
filesystems from the mount list to prevent the system hanging when it tries
to unmount them (for the same reason).
Drew points out that there's a good argument for forcibly removing all
"non syncable" filesystems from the mount list (eg. NFS mounts, disks
that aren't responding, etc.) as this then allows you to sync and
cleanly unmount their parents. No such change is included in this
patch.
Submitted by: Andrew Gallatin <gallatin@cs.duke.edu>
pid space somewhat more sparse which improves the performance of finding
an unused pid on systems with large numbers of processes. The new value
was chosen so that it doesn't overflow the 5 digit pid fields in various
programs.
This yields startling performance increases for NFS clients for many
access profiles, due to the fact that ACCESS results are persistently
cached in the namecache in many cases.
Note that the code is somewhat conservative in that it requires an
exact credential match for a cache hit. This bloats the nfsnode
structure by sizeof(struct ucred) (96 bytes). Any less conservative
approach opens the possibility for a false veto in eg. setuid
applications. Alternative suggestions would be welcomed.
The cache is normally disabled, to activate set the sysctl variable
vfs.nfs.access_cache_timeout to a nonzero value. This is the time in
seconds that a cached entry will be considered valid; useful values appear
to be 2-10 seconds. Performance of the cache can be monitored with the
vfs.nfs.access_cache_hits and vfs.nfs.access_cache_hits variables.
basically do a on-the-fly defragmentation of the FFS filesystem, changing
file block allocations to make them contiguous. Thanks to Kirk McKusick
for providing hints on what needed to be done to get this working.
