Re: [Nautilus-list] Tiny patch

[Glen Gray <glen antefacto com>]

On Tue, 2002-04-09 at 11:50, Soeren Sonnenburg wrote:
> > I didn't ignore it. It still gives the same figure as the current
> > formula that nautilus uses
> 
> Ok, sorry.
> 
No problems. Its not nice to feel ignored. I should have stated that I
tried some other formulas that had been mentioned on the list. But most
of them offered little advancement. This dirty little patch was
literally a patch while making a cuppa coffee that offered better
results.

> > Here is my bc session to prove
> > 
> > [root cruachann proc]# ll kcore
> > -r--------    1 root     root     401571840 Apr  9 11:06 kcore
> 
> I just logged into some 384MB linux machine and it gave 402644992 as a
> result making the formula work (kernel 2.4.17).
> 
Strange. Though I have seen some variation on the same machine/kernel
setups though I'm not sure why.

> However when doing tests on machines with >=1GB ram they all return
> 939528192 which sound like the size is set to some fixed size.
> 128/256/512 mem sizes did work.
> 
Thats a nice chunk of ram :-)

> So ether the kernel is buggy or the documentation of /proc/kcore.
> 
> > ((((401571840 - 4096) + 1023)/1024) + 1023) / 1024
> > 383
> 
> We should probably ask at lkml.
> 

Probably should. That was always the plan for "correct" solution. I'm
just really busy at the moment.

As I said the patch isn't correct. But it's more accurate than whats out
there at the moment. The hardware view is eventually going to get
re-written to be platform independent so there'll be lots of these
discussions regarding different Unixes etc.

-- 
_______________________________________
Glen Gray             Software Engineer
a n t e f a c t o     t: +353 1 8586006
www.antefacto.com     f: +353 1 8586014
181 Parnell Street - Dublin 1 - Ireland