[H-GEN] Reducing kernel memory footprint?

[Martin Pool]

[ Humbug *General* list - semi-serious discussions about Humbug and ]
[ Unix-related topics.  Please observe the list's charter.          ]

Chris Baird wrote:

> Martin,
> > It may be possible to configure gcc to run its various stages (cpp, cc1,
> > cc2, as, ld) in sequence: I think by default it tries to run them at the
> > same time with pipes connecting them.
> 
> That's "-pipe".  It's in Linux makefiles as a default now, as anyone
> who've compiled with recent compiler tools on a 8M machine will
> testify.

I thought that was it.  What I meant was to verify that gcc _without_
-pipe will not use pipes -- IIRC the manual threatens to do it if gcc
thinks it's a good idea even if you don't ask for it.  Is there a
-nopipe option?

> With memory this small it actually doesn't add much bypassing make--
> loading in anything will see other programs paged out completely to
> vm, as if they weren't there to begin with.

But isn't the time to load make significant?

> I've gone through all them, including the undocumented ones. :) For a
> i386, the best combination so far has been "-O2 -malign-loops=0
> -malign-jumps=0 -malign-functions=0".  

Ah, that would mean "put every instruction at address 0?"  That should
make things very small. :-) [0]

> Chris,,
> ...deep inside Torvalds' sacred code, made me wish that cpp abuse was
> a felony. :P

M-x hide-ifdef-mode
M-x hide-ifdefs

-- 
Martin 

[0] It's actually 1<<x IIRC

Which reminds me of Java's >>> (turn right quickly) operator.  Not quite
so cool as the <=> spaceship, but getting there.

--
* This is list (humbug) general handled by majordomo at lists.humbug.org.au .
* Postings to this list are only accepted from subscribed addresses of
* lists 'general' or 'general-post'.