Re: lookup on memory shortage

[Manuel Bouyer <bouyer%antioche.eu.org@localhost>]

I've got another pattern, this time on amd64:
db> sh uvm
Current UVM status:
  pagesize=4096 (0x1000), pagemask=0xfff, pageshift=12
  125114 VM pages: 70735 active, 34544 inactive, 1496 wired, 1 free
  pages  85450 anon, 17915 file, 3410 exec
  freemin=256, free-target=341, wired-max=41704
  faults=-1327104546, traps=-1322414536, intrs=176043552, ctxswitch=839178655
  softint=371000870, syscalls=486221471, swapins=11844, swapouts=11895
  fault counts:
    noram=100400, noanon=1, pgwait=532, pgrele=0
    ok relocks(total)=6850555(6850657), anget(retrys)=1148394825(6398108), amapc
opy=765214926
    neighbor anon/obj pg=1137984846/2128720875, gets(lock/unlock)=-1678869668/45
2531
    cases: anon=751769112, anoncow=304543516, obj=2124123134, prcopy=491974410, 
p
rzero=-853554526
  daemon and swap counts:
    woke=241469, revs=157816, scans=59342629, obscans=39042500, anscans=9525274
    busy=311381, freed=46278362, reactivate=5041657, deactivate=81143354
    pageouts=6719439, pending=858856, nswget=6395820
    nswapdev=1, swpgavail=65535
    swpages=65535, swpginuse=65535, swpgonly=58046, paging=0
db> tr
breakpoint() at netbsd:breakpoint+0x5
xencons_tty_input() at netbsd:xencons_tty_input+0xc5
xencons_handler() at netbsd:xencons_handler+0x96
evtchn_do_event() at netbsd:evtchn_do_event+0xfa
call_evtchn_do_event() at netbsd:call_evtchn_do_event+0xd
hypervisor_callback() at netbsd:hypervisor_callback+0xa3
uvmpd_trylockowner() at netbsd:uvmpd_trylockowner+0x3b
uvmpd_trydropswap() at netbsd:uvmpd_trydropswap+0x2c
uvmpdpol_balancequeue() at netbsd:uvmpdpol_balancequeue+0xa8
uvm_pageout() at netbsd:uvm_pageout+0x535

I guess swap is full again, and there's only one page free in RAM ...

-- 
Manuel Bouyer <bouyer%antioche.eu.org@localhost>
     NetBSD: 26 ans d'experience feront toujours la difference
--