Re: boot fail with large RAM

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

On Thu, Nov 19, 2009 at 09:59:30AM +0100, Christoph Egger wrote:
> Can you dump the memory usages of the various pools we have
> in the kernel with 1GB, 2GB, 4GB, 8GB, 16GB, 32GB and 48GB RAM
> and plot graphs ?

I did a 'show all pools' from ddb (I added a call to debugger when
bus_dma allocation fails); most of the pools have 0 or 1 page at this
point. Worst case is uarea with 101 pages allocated.
'show uvm' also says:
Current UVM status:
  pagesize=4096 (0x1000), pagemask=0xfff, pageshift=12
  12163158 VM pages: 0 active, 0 inactive, 404 wired, 12089513 free
  pages  404 anon, 0 file, 0 exec
  freemin=0, free-target=0, wired-max=0
  faults=404, traps=0, intrs=0, ctxswitch=0
  softint=0, syscalls=0, swapins=0, swapouts=0
  fault counts:
    noram=0, noanon=0, pgwait=0, pgrele=0
    ok relocks(total)=0(0), anget(retrys)=0(0), amapcopy=0
    neighbor anon/obj pg=0/0, gets(lock/unlock)=404/0
    cases: anon=0, anoncow=0, obj=404, prcopy=0, przero=0
  daemon and swap counts:
    woke=0, revs=0, scans=0, obscans=0, anscans=0
    busy=0, freed=0, reactivate=0, deactivate=0
    pageouts=0, pending=0, nswget=0
    nswapdev=0, swpgavail=0
    swpages=0, swpginuse=0, swpgonly=0, paging=0

init_x86_64() has loaded 12544034 pages to uvm; uvm claims to have
12163158 VM pages. That's a difference of 380876 pages or 1487MB.
Any idea where these 380876 pages have gone ?

-- 
Manuel Bouyer, LIP6, Universite Paris VI.           
Manuel.Bouyer%lip6.fr@localhost
     NetBSD: 26 ans d'experience feront toujours la difference
--