Commit Graph

99 Commits

Author SHA1 Message Date
joerg f3c5663df0 Require the actual namecache_look around cache_lookup_entry.
Add one last case of missing stat locking.
2014-06-16 12:28:10 +00:00
joerg 7913639b38 Make the stat mutex a leaf. XXX Use atomic counters. 2014-06-14 16:12:34 +00:00
joerg 94adc2671a Provide sysctl for namecache statistics. 2014-06-03 21:16:15 +00:00
joerg 0ae69ecb1a Don't play loop games, just enumerate the 10 fields explicitly. 2014-06-03 19:42:17 +00:00
joerg 11581dcbbb Introduce two helper functions to centralise the namecache statistics
in vfs_cache.c. Use consistent locking around the per-cpu data.
2014-06-03 19:30:29 +00:00
hannken 97834f7ba0 Change vnode operation lookup to return the resulting vnode *vpp unlocked.
Change cache_lookup() to return an unlocked vnode.

Discussed on tech-kern@

Welcome to 6.99.31
2014-02-07 15:29:20 +00:00
hannken 0fa0d339bd Change cache_prune() to test for end-of-list before testing for an
invalid entry.  Prevents a lifelock when the end-of-list marker
gets invalid while scanning the list and all entries are recent.
2014-01-20 07:47:22 +00:00
hannken 65b1f85ab6 Vnode API cleanup pass 1.
- Make these defines and functions private to vfs_vnode.c:

  VC_MASK, VC_LOCK, DOCLOSE, VI_IANCTREDO and VI_INACTNOW
  vclean() and vrelel()

- Remove the long time unused lwp argument from vrecycle().

- Remove vtryget(), it is responsible for ugly hacks and doesn't
  look that effective.

Presented on tech-kern.

Welcome to 6.99.25
2013-10-29 09:53:51 +00:00
dholland 35ed690545 Excise struct componentname from the namecache.
This uglifies the interface, because several operations need to be
passed the namei flags and cache_lookup also needs for the time being
to be passed cnp->cn_nameiop. Nonetheless, it's a net benefit.

The glop should be able to go away eventually but requires structural
cleanup elsewhere first.

This change requires a kernel bump.
2012-11-05 17:27:37 +00:00
dholland 1617a81dd1 Disentangle the namecache from the internals of namei.
- Move the namecache's hash computation to inside the namecache code,
instead of being spread out all over the place. Remove cn_hash from
struct componentname and delete all uses of it.

 - It is no longer necessary (if it ever was) for cache_lookup and
cache_lookup_raw to clear MAKEENTRY from cnp->cn_flags for the cases
that cache_enter already checks for.

 - Rearrange the interface of cache_lookup (and cache_lookup_raw) to
make it somewhat simpler, to exclude certain nonexistent error
conditions, and (most importantly) to make it not require write access
to cnp->cn_flags.

This change requires a kernel bump.
2012-11-05 17:24:09 +00:00
rmind d65753d972 Move some the test for MAKEENTRY into the cache_enter(9). Make some
variables in vfs_cache.c static, __read_mostly, etc.

No objection on tech-kern@.
2012-07-22 00:53:18 +00:00
rmind e225b7bd09 Welcome to 5.99.53! Merge rmind-uvmplock branch:
- Reorganize locking in UVM and provide extra serialisation for pmap(9).
  New lock order: [vmpage-owner-lock] -> pmap-lock.

- Simplify locking in some pmap(9) modules by removing P->V locking.

- Use lock object on vmobjlock (and thus vnode_t::v_interlock) to share
  the locks amongst UVM objects where necessary (tmpfs, layerfs, unionfs).

- Rewrite and optimise x86 TLB shootdown code, make it simpler and cleaner.
  Add TLBSTATS option for x86 to collect statistics about TLB shootdowns.

- Unify /dev/mem et al in MI code and provide required locking (removes
  kernel-lock on some ports).  Also, avoid cache-aliasing issues.

Thanks to Andrew Doran and Joerg Sonnenberger, as their initial patches
formed the core changes of this branch.
2011-06-12 03:35:36 +00:00
hannken fb62bef947 Make holding v_interlock mandatory for callers of vget().
Announced some time ago on tech-kern.
2010-07-21 17:52:09 +00:00
hannken e698a5d246 Using cache_revlookup() leads to vnode races as it returns an unreferenced
vnode that may disappear before the caller has a chance to reference it.

Reference the vnode while the name cache is locked.

No objections on tech-kern.
2010-07-21 09:01:35 +00:00
hannken 1423e65b26 Clean up vnode lock operations pass 2:
VOP_UNLOCK(vp, flags) -> VOP_UNLOCK(vp): Remove the unneeded flags argument.

Welcome to 5.99.32.

Discussed on tech-kern.
2010-06-24 12:58:48 +00:00
yamt 777ded00ac cache_lookup_entry: add an assertion. 2009-02-18 13:36:11 +00:00
yamt 947efbeab9 cache_purge1: consistently unlock ncp a little earlier. 2009-02-18 13:24:18 +00:00
yamt cea19a4d14 malloc -> kmem_alloc. 2009-01-17 07:02:35 +00:00
yamt ae07703384 cache_prune: use (a - b > 0) rather than (a > b) to compare ticks. 2009-01-16 06:59:21 +00:00
yamt ea8f5a99d5 cache_lookup_raw: update "goodhits" statistics to be less confused.
add an XXX comment.
2008-10-25 14:20:17 +00:00
yamt 2fb0334ca1 add assertions. 2008-10-25 14:18:18 +00:00
pooka 3e7c7228f4 Release vnode in cache_lookup() error branch if locking fails. 2008-08-20 15:34:59 +00:00
ad 1b23b70818 vfs_cache:
- Don't use goto in critical paths, it can confuse the compiler.
- Sprinkle some branch hints.
- Make namecache stats per-CPU and collate once per second.
- Use vtryget().
2008-06-03 15:50:22 +00:00
ad e071d39c84 - Convert hashinit() to use kmem_alloc(). The hash tables can be large
and it's better to not have them in kmem_map.
- Convert a couple of minor items along the way to kmem_alloc().
- Fix some memory leaks.
2008-05-05 17:11:16 +00:00
martin ce099b4099 Remove clause 3 and 4 from TNF licenses 2008-04-28 20:22:51 +00:00
ad a78ad62cfb cache_enter: inline LIST_INSERT_HEAD so that the membar_producer() can be
put in the right spot. The 'next' link in the new entry must become globally
visible before the list head is updated. This could have affected systems
with weak memory ordering like the alpha.
2008-04-12 17:34:26 +00:00
ad 1e11b07bfa Restructure the name cache code to eliminate most lock contention
resulting from forward lookups. Discussed on tech-kern@.
2008-04-11 15:25:24 +00:00
matt 11910619f7 Change some initialization of static queues to compile time.
(xxx_INIT to xxx_HEAD_INITIALIZER).  Drop code which inits
non-auto (global or static) variables to 0 since that's
already implied by being non-auto.  Init some static/global
cpu_simple_locks at compile time.
2007-11-11 23:22:23 +00:00
ad d18c6ca4de Merge from vmlocking:
- pool_cache changes.
- Debugger/procfs locking fixes.
- Other minor changes.
2007-11-07 00:23:13 +00:00
ad 88ab7da936 Merge some of the less invasive changes from the vmlocking branch:
- kthread, callout, devsw API changes
- select()/poll() improvements
- miscellaneous MT safety improvements
2007-07-09 20:51:58 +00:00
ad 59d979c5f1 Pass an ipl argument to pool_init/POOL_INIT to be used when initializing
the pool's lock.
2007-03-12 18:18:22 +00:00
ad b07ec3fc38 Merge newlock2 to head. 2007-02-09 21:55:00 +00:00
chs c398ae9734 a smorgasbord of improvements to vnode locking and path lookup:
- LOCKPARENT is no longer relevant for lookup(), relookup() or VOP_LOOKUP().
   these now always return the parent vnode locked.  namei() works as before.
   lookup() and various other paths no longer acquire vnode locks in the
   wrong order via vrele().  fixes PR 32535.
   as a nice side effect, path lookup is also up to 25% faster.
 - the above allows us to get rid of PDIRUNLOCK.
 - also get rid of WANTPARENT (just use LOCKPARENT and unlock it).
 - remove an assumption in layer_node_find() that all file systems implement
   a recursive VOP_LOCK() (unionfs doesn't).
 - require that all file systems supply vfs_vptofh and vfs_fhtovp routines.
   fill in eopnotsupp() for file systems that don't support being exported
   and remove the checks for NULL.  (layerfs calls these without checking.)
 - in union_lookup1(), don't change refcounts in the ISDOTDOT case, just
   adjust which vnode is locked.  fixes PR 33374.
 - apply fixes for ufs_rename() from ufs_vnops.c rev. 1.61 to ext2fs_rename().
2006-12-09 16:11:50 +00:00
christos 5ce26ddf80 enable namecache enter reverse for linux procfs lookups.
XXX: how expensive is that?
2006-10-25 18:56:38 +00:00
dbj 8bcbd194da make comment match logic in cache_lookup dvp unlock code 2006-09-10 05:51:33 +00:00
christos e073deb555 Coverity CID 602: Remove always true if statement. 2006-04-15 04:33:48 +00:00
perry a2cd732268 Remove leading __ from __(const|inline|signed|volatile) -- it is obsolete. 2005-12-24 19:12:23 +00:00
christos 95e1ffb156 merge ktrace-lwp. 2005-12-11 12:16:03 +00:00
yamt ce02ffbc68 introduce a new function, cache_lookup_raw(), for filesystems which
want more flexible namecache handling.
it just looks up a dnlc entry and vget() the result vnode.
ie. no automatic entry removal, no automatic vnode locking.

discussed on tech-kern@.
2004-06-27 08:50:44 +00:00
yamt 8a2c13021f cache_lookup: avoid to grab two vnode's v_interlock.
just hold a reference (usecount) to a vnode instead.
2004-06-19 18:49:47 +00:00
yamt 68b4772ef6 redo the previous (rev.1.58; overwrite a duplicate entry rather than leave it)
differently so that entries entered during we're doing pool_get() are
checked as well.  pointed by Paul Kranenburg on source-changes@.
2004-05-07 12:05:41 +00:00
yamt 8d615f3e18 cache_enter: when we found a duplicate entry,
simply overwrite it rather than leaving a stale entry.
2004-05-06 22:02:02 +00:00
pk 5c36071518 cache_enter: concurrent lookups in the same directory may race for a
cache entry. Upon detection, free our tentative entry and return.
2004-05-02 12:00:34 +00:00
simonb b5d0e6bf06 Initialise (most) pools from a link set instead of explicit calls
to pool_init.  Untouched pools are ones that either in arch-specific
code, or aren't initialiased during initial system startup.

 Convert struct session, ucred and lockf to pools.
2004-04-25 16:42:40 +00:00
yamt 4972de4cca make cache_purge more controlable.
namely, allow following operations.
	- purge only an entry specified by a component name.
	- purge only child entries.
	- purge only parent entries.
no objections on tech-kern@.
2004-04-05 10:20:52 +00:00
yamt b3d9c9a976 remove an obsolete comment. pointed by enami@ 2003-09-01 12:13:55 +00:00
yamt a01f6f4ea5 - make this a bit MP friendly.
(although no actual changes under the kernel lock)
- remove a test that isn't meaningful anymore.
2003-08-08 20:19:56 +00:00
yamt 1f08924c29 arrange namecache lru before vget (and releasing namecache_slock)
since our namecache entry can go away during we're sleeping on the vnode.

the bug pointed by enami tsugutomo.
tested by Matthias Scheler.
PR/22363.
2003-08-08 20:18:19 +00:00
agc aad01611e7 Move UCB-licensed code from 4-clause to 3-clause licence.
Patches provided by Joel Baker in PR 22364, verified by myself.
2003-08-07 16:26:28 +00:00
yamt 428f108569 remove remaining v_id. 2003-07-31 15:43:06 +00:00