CVS log for src/sys/uvm/uvm_pdpolicy_clockpro.c

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

Request diff between arbitrary revisions

Keyword substitution: kv
Default branch: MAIN

s/stablize/stabilize/

Start trying to reduce cache misses on vm_page during fault processing.

- Make PGO_LOCKED getpages imply PGO_NOBUSY and remove the latter.  Mark
  pages busy only when there's actually I/O to do.

- When doing COW on a uvm_object, don't mess with neighbouring pages.  In
  all likelyhood they're already entered.

- Don't mess with neighbouring VAs that have existing mappings as replacing
  those mappings with same can be quite costly.

- Don't enqueue pages for neighbour faults unless not enqueued already, and
  don't activate centre pages unless uvmpdpol says its useful.

Also:

- Make PGO_LOCKED getpages on UAOs work more like vnodes: do gang lookup in
  the radix tree, and don't allocate new pages.

- Fix many assertion failures around faults/loans with tmpfs.

Sync with HEAD

Mostly merge changes from HEAD upto 20200411

Update a link to "CLOCK-Pro" paper.

Merge changes from current as of 20200406

uvm_pdpolicy: Require a write lock on the object only for dequeue.
No sense in requiring that for enqueue/activate/deactivate.

Sync with head.

uvmpdpol_estimatepageable(): Don't take any locks here.  This can be called
from DDB, and in any case the numbers are stale the instant the lock is
dropped, so it just doesn't matter.

- Add and use wrapper functions that take and acquire page interlocks, and pairs
  of page interlocks.  Require that the page interlock be held over calls to
  uvm_pageactivate(), uvm_pagewire() and similar.

- Solve the concurrency problem with page replacement state.  Rather than
  updating the global state synchronously, set an intended state on
  individual pages (active, inactive, enqueued, dequeued) while holding the
  page interlock.  After the interlock is released put the pages on a 128
  entry per-CPU queue for their state changes to be made real in batch.
  This results in in a ~400 fold decrease in contention on my test system.
  Proposed on tech-kern but modified to use the page interlock rather than
  atomics to synchronise as it's much easier to maintain that way, and
  cheaper.

Rename uvm_page_locked_p() -> uvm_page_owner_locked_p()

pagedaemon:

- Use marker pages to keep place in the queue when scanning, rather than
  relying on assumptions.

- In uvmpdpol_balancequeue(), lock the object once instead of twice.

- When draining pools, the situation is getting desperate, but try to avoid
  saturating the system with xcall, lock and interrupt activity by sleeping
  for 1 clock tick if being continually awoken and all pools have been
  cycled through at least once.

- Pause & resume the freelist cache during pool draining.

PR kern/54209: NetBSD 8 large memory performance extremely low
PR kern/54210: NetBSD-8 processes presumably not exiting
PR kern/54727: writing a large file causes unreasonable system behaviour

vm_page: Now that listq is gone, give the pagedaemon its own private
TAILQ_ENTRY, so that update of page replacement state can be made
asynchronous/lazy.  No functional change.

Break the global uvm_pageqlock into a per-page identity lock and a private
lock for use of the pagedaemon policy code.  Discussed on tech-kern.

PR kern/54209: NetBSD 8 large memory performance extremely low
PR kern/54210: NetBSD-8 processes presumably not exiting
PR kern/54727: writing a large file causes unreasonable system behaviour

Major changes to uvm.
Support multiple collections (groups) of free pages and run the page
reclaimation algorithm on each group independently.

Catchup with rmind-uvmplock merge.

band-aid fix after the merge of rmind-uvmplock branch.

Sync with HEAD.

sync with head

Sync with HEAD

pageobj: remove a wrong assertion.

sync with head.

Sync w/ -current. 34 merge conflicts to follow.

sync with head.

Sync with HEAD.

Also fix build.

vm_page: put TAILQ_ENTRY into a union with LIST_ENTRY, so we can use both.

Sync with HEAD.

sync with head.

sync with HEAD

Allow this to compile if LISTQ is undefined:

	- Put '#ifdef LISTQ' ... '#endif' pairs around pageq_insert_head()
	  and clockpro_insert_head().

	- Add missing argument to printf statement.

Sync with HEAD.

sync with head.

nonresident_rotate: avoid too long loops which can happen on some workloads.

sync with head

Sync with HEAD

push pmap_clear_reference calls into pdpolicy code, where reference bits
actually matter.

nonresident_rotate: micro optimization

sync with HEAD

Sync with HEAD

Merge vmlocking2 to head.

Pull the vmlocking changes into a new branch.

sync with head.

Sync with HEAD.

Sync with HEAD.

use separate nreslookup evcnt for obj and anon pages.

file uvm_pdpolicy_clockpro.c was added on branch matt-mips64 on 2007-08-01 14:49:56 +0000

use separate nreslookup evcnt for obj and anon pages.

Pull in the initial set of changes for the vmlocking branch.

- sync with head.
- move sched_changepri back to kern_synch.c as it doesn't know PPQ anymore.

sync with head.

TRUE -> true, FALSE -> false

Replace the Mach-derived boolean_t type with the C99 bool type.  A
future commit will replace use of TRUE and FALSE with true and false.

Sync with head.

sync with head.

sync with head.

file uvm_pdpolicy_clockpro.c was added on branch yamt-lazymbuf on 2006-12-30 20:51:05 +0000

uvmpdpol_pagedequeue: clear PQ_INITIALREF.
otherwise, dequeue/enqueue cycles (eg. page loaning) can cause
an assertion failure in clockpro_pageenqueue.

Sync with head.

__unused removal on arguments; approved by core.

sync with head

remove unnecessary #include of vnode.h.

fix some warnings in the case of PDSIM.

merge yamt-pdpolicy branch.
	- separate page replacement policy from the rest of kernel
	- implement an alternative replacement policy

make UVM_KICK_PDAEMON() a real function and stop including
uvm_pdpolicy.h from uvm.h.  this also fixes build of pmap(1).

get rid of bootstrap code from frequently called path.

separate "nresrecord" statistics for obj and anon.

add a sysctl knob to adjust cold target.

reduce BUCKETSIZE to make sizeof(struct bucket) a power of two.

reduce unnecessary c99'ism.

some comments.

add a statistic.

remove unnecessary ";".

simplify #ifdef a little.

an experimental implementation of CLOCK-Pro.

file uvm_pdpolicy_clockpro.c was initially added on branch yamt-pdpolicy.