Glibc-2.2.5 compile error in chapter 6 of LFS 3.2

Bill Maltby LFS Related lfsbill at wlmlx1.wlmcs.com
Mon Mar 11 11:01:55 PST 2002


Don,

Based on your earlier post, I began an introspective on swap on the
little machine. And (great minds ...) reached a similar conclusion
that it's swap might be insufficient. So I rolled over there and
made 256MB swap on my 10K RPM 20GB (gee! Is this overkill on a 166MHz
AMD?) HD that I have laying on top of the case.

Pop to top (I love alliteration :)  ) and watch for a few minutes.
Not yet into localization stuff yet, but I saw that make accumulates
gobs of memory until it gets to "..leaving directory..". So, we can
now predict that 2^32 nationalizations will require 38,327 light
years to complete because make will accumulate 74,000TB of memory
(or appx. .00174% of a new IBM NAS appliance minimal config).

Watcha think - in the ballpark?

Bill
billm at wlmcs.com

On Mon, 11 Mar 2002, Don Smith wrote:

> "Bill Maltby LFS Related" <lfsbill at wlmlx1.wlmcs.com> wrote in message
> news:Pine.LNX.4.10.10203111226320.892-100000 at wlmlx1.wlmcs.com...
> > Don,
> >
> > Both machines have "adequate" swap. The "biggie" has two swaps -
> > 1 of 128MB and the secondary of 132MB. But, I never saw "excessive"
> > use of these. I began to wonder if it was being effectivey used.
> > What I saw, in retrospect, was signs of thrashing. _Very_ slow
> > (1 day install time before "unoptimizing") on the "biggie" machine.
> > After my umm... corrections, install was about 10 hours. Since I
> > saw performance improvement in conjunction with reliable operation
> > when I reduced memory requirements, I deduced kernel 2.2 swap
> (actually
> > paging) was not very good and it probably was thrashing. With all
> > i18n being done, it would sit for hours on the last two or three
> > files in glibc install.
> 
> Yes, your install time is limited by you disk speed. I have
> approximately the same times on a somewhat slower processor. 92% idle
> during heavy swapping, er.. paging.
> 
> >
> > The little machine has 64MB of swap. I still have'nt got this one
> > completely through. The gcc install is dying with interrupt 11.
> >
> > But I am tenacious.
> >
> > Bill Maltby
> > billm at wlmcs.com
> >
> 
> 32+64MB may not be enough. I was compiling a package that uses templates
> extensively and the compiles would take ~90MB (and *way* too many hours)
> all by themselves. With everything else thrown in, memory usage
> approached 128MB. Does glibc have any templates in it? (I doubt it but
> you never know...)
> 
> Don
> 
> 
> -- 
> Unsubscribe: send email to listar at linuxfromscratch.org
> and put 'unsubscribe lfs-support' in the subject header of the message
> 

-- 
Unsubscribe: send email to listar at linuxfromscratch.org
and put 'unsubscribe lfs-support' in the subject header of the message



More information about the lfs-support mailing list