Commit Graph

275 Commits

Author SHA1 Message Date
dg
3f1b0f5075 Set laundry flag when transitioning an inactive page from clean to dirty.
This fixes a performance bug where pages would sometimes not be paged
out when they could be.

Submitted by:	John Dyson
1994-11-14 02:57:40 +00:00
dg
5cf4605dc7 Added support for starting the experimental "vmdaemon" system process.
Enabled via REL2_1.

Added support for doing object collapses "on the fly". Enabled via REL2_1a.

Improved object collapses so that they can happen in more cases. Improved
sensing of modified pages to fix an apparant race condition and improve
clustered pageout opportunities. Fixed an "oops" with not restarting page
scan after a potential block in vm_pageout_clean() (not doing this can result
in strange behavior in some cases).

Submitted by:	John Dyson & David Greenman
1994-11-06 05:07:53 +00:00
dg
5461dd7c15 #if 0'd out the object cache trimming code - there are multiple ways
that the pageout daemon can deadlock otherwise.

Submitted by:	John Dyson
1994-10-25 05:35:44 +00:00
dg
5727d2fc7e Fixed object cache trimming policy so it actually works.
Submitted by:	John Dyson
1994-10-23 21:03:09 +00:00
dg
ee0022da28 Adjusted reserved levels to fix a deadlock condition.
Submitted by:	John Dyson
1994-10-23 20:53:33 +00:00
dg
e8b2d4b14c Various changes to allow operation without any swapspace configured. Note
that this is intended for use only in floppy situations and is done at
the sacrifice of performance in that case (in ther words, this is not the
best solution, but works okay for this exceptional situation).

Submitted by:	John Dyson
1994-10-22 02:18:03 +00:00
dg
2d52a86c44 Fix the remaining vmmeter counters. They all now work correctly. 1994-10-18 14:59:20 +00:00
dg
eb282b107c 1) Some of the counters in the vmmeter struct don't fit well into the Mach VM
scheme of things, so I've changed them to be more appropriate. page in/ous
   are now associated with the pager that did them. Nuked v_fault as the
   only fault of interest that wouldn't be already counted in v_trap is a VM
   fault, and this is counted seperately.
2) Implemented most of the remaining counters and corrected the counting of
   some that were done wrong. They are all almost correct now...just a few
   minor ones left to fix.
1994-10-15 13:33:09 +00:00
dg
bb709a3eec Fixed an object reference count problem that was caused by a call to
vm_object_lookup() being outside of some parens. The bug was introduced
via some recently added code.

Reviewed by:	John Dyson
1994-10-13 21:01:39 +00:00
phk
09c3293a0f Cosmetics: unused vars, ()'s, #include's &c &c to silence gcc.
Reviewed by:	davidg
1994-10-09 01:52:19 +00:00
phk
633ef15194 Cosmetics. Unused vars and other warnings. 1994-10-07 07:23:04 +00:00
dg
66cba7eaa5 Fixed minor bug caused by some missing parens that can result in slightly
reduced paging performance by missing a clustering opportunity. Found
by Poul-Henning Kamp with gcc -Wall.
1994-10-05 09:02:53 +00:00
dg
467ebd6849 Fixed bug related to proper sensing of page modification that we
inadvertantly introduced in pre-1.1.5. This could cause page modifications
to go unnoticed during certain extreme low memory/high paging rate conditions.

Submitted by:	John Dyson and David Greenman
1994-10-04 03:05:09 +00:00
dg
71d8424950 Fixed a bug I introduced when fixing the rss limit code. Changed swapout
policy to be a bit more selective about what processes get swapped out.

Reviewed by:	John Dyson
1994-09-12 15:06:14 +00:00
dg
4c73f2ffb4 Don't deactivate pages in 0-refcount objects. Added a couple of missing
paging stats. Fixed problem with free_reserved becoming depleted during
certain swap_pager operations.

Submitted by:	John Dyson, with a little help from me
1994-09-12 11:31:36 +00:00
dg
0a37b9feee Simple changes to paging algorithms...but boy do they make a difference.
FreeBSD's paging performance has never been better. Wow.

Submitted by:	John Dyson
1994-09-06 11:28:46 +00:00
dg
16e35aa368 Fixed bug caused by change of rlimit variables to quad_t's. The bug was in
using min() to calculate the minimum of rss_cur,rss_max - since these
are now quad_t's and min() takes u_ints...the comparison later for exceeding
the rss limit was always true - resulting in rather serious page thrashing.
Now using new qmin() function for this purpose.

Fixed another bug where PG_BUSY pages would sometimes be paged out (bad!).
This was caused by the PG_BUSY flag not being included in a comparison.
1994-08-30 18:27:44 +00:00
wollman
f9fc827448 Fix up some sloppy coding practices:
- Delete redundant declarations.
- Add -Wredundant-declarations to Makefile.i386 so they don't come back.
- Delete sloppy COMMON-style declarations of uninitialized data in
  header files.
- Add a few prototypes.
- Clean up warnings resulting from the above.

NB: ioconf.c will still generate a redundant-declaration warning, which
is unavoidable unless somebody volunteers to make `config' smarter.
1994-08-18 22:36:09 +00:00
dg
8b20309268 Incorporated post 1.1.5 work from John Dyson. This includes performance
improvements via the new routines pmap_qenter/pmap_qremove and pmap_kenter/
pmap_kremove. These routine allow fast mapping of pages for those
architectures that have "normal" MMUs. Also included is a fix to the
pageout daemon to properly check a queue end condition.

Submitted by:	John Dyson
1994-08-06 09:15:42 +00:00
dg
0711a9cff6 Integrated VM system improvements/fixes from FreeBSD-1.1.5. 1994-08-04 03:06:48 +00:00
dg
8d205697aa Added $Id$ 1994-08-02 07:55:43 +00:00
dg
0e87163cbf Removed all code related to the pagescan daemon, and changed 'act_count'
adjustments to compensate for a world without the pagescan daemon.
1994-08-01 11:25:45 +00:00
dg
0d458db402 Don't move the page's position in the active queue if it is busy or
held. John has noticed some stability problems when doing this.
1994-06-06 11:56:27 +00:00
rgrimes
2469c867a1 The big 4.4BSD Lite to FreeBSD 2.0.0 (Development) patch.
Reviewed by:	Rodney W. Grimes
Submitted by:	John Dyson and David Greenman
1994-05-25 09:21:21 +00:00
rgrimes
8fb65ce818 BSD 4.4 Lite Kernel Sources 1994-05-24 10:09:53 +00:00