linux-kernel - Re: [patch] fix !apic build breakage

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-Id: <200609292344.55363.ak@suse.de>
Date:	Fri, 29 Sep 2006 23:44:55 +0200
From:	Andi Kleen <ak@...e.de>
To:	Ingo Molnar <mingo@...e.hu>
Cc:	Jim Cromie <jim.cromie@...il.com>, Andrew Morton <akpm@...l.org>,
	linux-kernel@...r.kernel.org
Subject: Re: [patch] fix !apic build breakage

> so please do it. The fact that there are /other/ reductions possible 
> doesnt mean we can be lax. 

Well with that argument we would put ifdefs nearly everywhere
because most subsystem have some code that you don't need in some
obscure configuration.

Do we do that? No. Clean and maintainable code is more important.

This particular case of APIC CONFIG is just a historical ward.
I eliminated it on 64bit a long time ago (and it undoubtedly
saved me hours of fixing compilation issues and it made the code
cleaner too) and i386 is definitely ripe for that soon too.

> It's like: "oh, the buddy allocator scales  
> better now, so we can slow down the SLAB allocator". No, kernel size is 
> like scalability: we need a million small steps.

Sure you could do a million steps. Just for each step you need 
to look at the ratio of maintainability impact:usefulness
IMHO microconfig loses there usually badly.

[As terminology i call microconfig anything that requires ifdefs 
inside .c or .h files. CONFIGs that only appear in Makefiles are usually
not a problem]

There are lots of other steps to less bloat that make sense, but please don't
advocate that microCONFIG disease.

> the panic_on_unrecovered_nmi thing is gross anyway: it has no place in 
> kernel.h, it should go into include/[asm-i386|x86_64]/nmi.h and not the 
> generic headers. There the prototype can be made #ifdef APIC, hence 
> eliminating the #ifdefs from traps.c. (that's all we care about anyway)

Yes I fixed it already in a cleaner way (without ugly ifdefs) 

ftp://ftp.firstfloor.org/pub/ak/x86_64/quilt/patches/nmi-sysctl-cleanup

> please dont throw away a perfectly fine config option.

I can't count how many that silly option already got broken by 
changes in the APIC code. I definitely wouldn't describe it as "perfectly fine",
more as "fragile and tends to fall over when you even look at it".

-Andi
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/