CVS log for src/sys/kern/kern_rwlock_obj.c

Up to [cvs.NetBSD.org] / src / sys / kern

Request diff between arbitrary revisions

Keyword substitution: kv
Default branch: MAIN

Use kmem_intr_*() variants for lock objects since aiodoned was done away
with and we process these I/Os in soft interrupt context now.

Repply this change with a couple of bugs fixed:

- Do away with separate pool_cache for some kernel objects that have no special
  requirements and use the general purpose allocator instead. On one of my
  test systems this makes for a small (~1%) but repeatable reduction in system
  time during builds presumably because it decreases the kernel's cache /
  memory bandwidth footprint a little.
- vfs_lockf: cache a pointer to the uidinfo and put mutex in the data segment.

Back out recent change to replace pool_cache with then general allocator.
Will return to this when I have time again.

- Do away with separate pool_cache for some kernel objects that have no special
  requirements and use the general purpose allocator instead.  On one of my
  test systems this makes for a small (~1%) but repeatable reduction in system
  time during builds presumably because it decreases the kernel's cache /
  memory bandwidth footprint a little.
- vfs_lockf: cache a pointer to the uidinfo and put mutex in the data segment.

kern: Eliminate most __HAVE_ATOMIC_AS_MEMBAR conditionals.

I'm leaving in the conditional around the legacy membar_enters
(store-before-load, store-before-store) in kern_mutex.c and in
kern_lock.c because they may still matter: store-before-load barriers
tend to be the most expensive kind, so eliding them is probably
worthwhile on x86.  (It also may not matter; I just don't care to do
measurements right now, and it's a single valid and potentially
justifiable use case in the whole tree.)

However, membar_release/acquire can be mere instruction barriers on
all TSO platforms including x86, so there's no need to go out of our
way with a bad API to conditionalize them.  If the procedure call
overhead is measurable we just could change them to be macros on x86
that expand into __insn_barrier.

Discussed on tech-kern:
https://mail-index.netbsd.org/tech-kern/2023/02/23/msg028729.html

rwlock(9): Nix extern _rw_init in .c; use sys/rwlock.h.

sys: Use membar_release/acquire around reference drop.

This just goes through my recent reference count membar audit and
changes membar_exit to membar_release and membar_enter to
membar_acquire -- this should make everything cheaper on most CPUs
without hurting correctness, because membar_acquire is generally
cheaper than membar_enter.

sys: Membar audit around reference count releases.

If two threads are using an object that is freed when the reference
count goes to zero, we need to ensure that all memory operations
related to the object happen before freeing the object.

Using an atomic_dec_uint_nv(&refcnt) == 0 ensures that only one
thread takes responsibility for freeing, but it's not enough to
ensure that the other thread's memory operations happen before the
freeing.

Consider:

	  Thread A			  Thread B
	obj->foo = 42;			obj->baz = 73;
	mumble(&obj->bar);		grumble(&obj->quux);
	/* membar_exit(); */		/* membar_exit(); */
	atomic_dec -- not last		atomic_dec -- last
					/* membar_enter(); */
					KASSERT(invariant(obj->foo,
					    obj->bar));
					free_stuff(obj);

The memory barriers ensure that

	obj->foo = 42;
	mumble(&obj->bar);

in thread A happens before

	KASSERT(invariant(obj->foo, obj->bar));
	free_stuff(obj);

in thread B.  Without them, this ordering is not guaranteed.

So in general it is necessary to do

	membar_exit();
	if (atomic_dec_uint_nv(&obj->refcnt) != 0)
		return;
	membar_enter();

to release a reference, for the `last one out hit the lights' style
of reference counting.  (This is in contrast to the style where one
thread blocks new references and then waits under a lock for existing
ones to drain with a condvar -- no membar needed thanks to mutex(9).)

I searched for atomic_dec to find all these.  Obviously we ought to
have a better abstraction for this because there's so much copypasta.
This is a stop-gap measure to fix actual bugs until we have that.  It
would be nice if an abstraction could gracefully handle the different
styles of reference counting in use -- some years ago I drafted an
API for this, but making it cover everything got a little out of hand
(particularly with struct vnode::v_usecount) and I ended up setting
it aside to work on psref/localcount instead for better scalability.

I got bored of adding #ifdef __HAVE_ATOMIC_AS_MEMBAR everywhere, so I
only put it on things that look performance-critical on 5sec review.
We should really adopt membar_enter_preatomic/membar_exit_postatomic
or something (except they are applicable only to atomic r/m/w, not to
atomic_load/store_*, making the naming annoying) and get rid of all
the ifdefs.

Merge changes from current as of 20200406

Add some new functions for lock objects:

mutex_obj_refcnt(), mutex_obj_tryalloc()
rw_obj_refcnt(), rw_obj_tryalloc()

Pull up following revision(s) (requested by ozaki-r in ticket #687):
	sys/kern/kern_rwlock_obj.c: revision 1.4
	sys/rump/librump/rumpkern/locks.c: revision 1.80
	sys/kern/kern_rwlock.c: revision 1.50
	sys/arch/x86/x86/db_memrw.c: revision 1.5,1.6
	sys/ddb/db_command.c: revision 1.150-1.153
	share/man/man4/ddb.4: revision 1.175 (via patch),1.176-1.178
	sys/kern/kern_mutex_obj.c: revision 1.6
	sys/kern/subr_lockdebug.c: revision 1.61-1.64
	sys/sys/lockdebug.h: revision 1.17
	sys/kern/kern_mutex.c: revision 1.71
	sys/sys/lockdebug.h: revision 1.18,1.19
	sys/kern/subr_xcall.c: revision 1.26

Obtain proper initialized addresses of locks allocated by mutex_obj_alloc or rw_obj_alloc

Initialized addresses of locks allocated by mutex_obj_alloc or rw_obj_alloc
were not useful because the addresses were mutex_obj_alloc or rw_obj_alloc
itself. What we want to know are callers of them.

Spinkle ASSERT_SLEEPABLE to xcall functions

Use db_printf instead of printf in ddb

Add a new command, show lockstat, which shows statistics of locks
Currently the command shows the number of allocated locks.
The command is useful only if LOCKDEBUG is enabled.

Add a new command, show all locks, which shows information of active locks

The command shows information of all active (i.e., being held) locks that are
tracked through either of LWPs or CPUs by the LOCKDEBUG facility.  The /t
modifier additionally shows a backtrace for each LWP additionally.  This
feature is useful for debugging especially to analyze deadlocks.
The command is useful only if LOCKDEBUG is enabled.

Don't pass a unset address to lockdebug_lock_print

x86: avoid accessing invalid addresses in ddb like arm32
This avoids that a command stops in the middle of an execution if
a fault occurs due to an access to an invalid address.

Get rid of a redundant output

Improve wording. Fix a Cm argument.

ddb: rename "show lockstat" to "show lockstats" to avoid conflicting with lockstat(8)
Requested by mrg@

Obtain proper initialized addresses of locks allocated by mutex_obj_alloc or rw_obj_alloc

Initialized addresses of locks allocated by mutex_obj_alloc or rw_obj_alloc
were not useful because the addresses were mutex_obj_alloc or rw_obj_alloc
itself. What we want to know are callers of them.

Sync with HEAD.

sync with head

Sprinkle __cacheline_aligned and __read_mostly.

sync with head.

sync with head

Sync with HEAD.

Fix copyrights.

sync with head

file kern_rwlock_obj.c was added on branch yamt-nfs-mp on 2010-03-11 15:04:17 +0000

Heave-ho mutex/rwlock object routines into separate modules -- they
don't have anything to do with the lock internals.