linux-kernel - Re: [PATCH 1/7] Make section names compatible with -ffunction-sections -fdata-sections

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: <20090112135127.7a0bb392.akpm@linux-foundation.org>
Date:	Mon, 12 Jan 2009 13:51:27 -0800
From:	Andrew Morton <akpm@...ux-foundation.org>
To:	Tim Abbott <tabbott@....EDU>
Cc:	andi@...stfloor.org, jbarnold@....EDU,
	linux-kernel@...r.kernel.org, vda.linux@...glemail.com,
	andersk@....EDU, wdaher@....EDU, knikanth@...e.de
Subject: Re: [PATCH 1/7] Make section names compatible with
 -ffunction-sections -fdata-sections

On Sun, 4 Jan 2009 14:20:35 -0500 (EST)
Tim Abbott <tabbott@....EDU> wrote:

> On Thu, 1 Jan 2009, Andrew Morton wrote:
> 
> > OK, it hits a lot of files in minor ways.
> >
> > You could ask Linus to merge it late in the -rc1 merge window (please
> > cc me on the email).  Or you could just send it to me (with the usual
> > cc's) and I'll toss it into my 2.6.29-rc1 pile.  I doubt if it'll cause
> > much trouble.  But please do check its applicability to linux-next - people
> > have been moving arch files into different directories lately.
> 
> Andrew,
> 
> I've attached for you a version of the patch the applies to current 
> linux-next (next-20090102).
> 
> It does not apply to Linus' current master because d3794979a in linux-next 
> adds use of the ".data.percpu.first" section name, which this patch 
> updates to ".kernel.data.percpu.first".  I think it'll be easiest to just 
> merge this after d3794979a.
> 
> If a version of d3794979a is not going to be merged first, there will be 
> two merge conflicts that are easy to resolve as they are changes to code 
> added by d3794979a, but we'll then need to fix d3794979a before merging 
> d3794979a.
> 

I don't think anyone will want to carry this in a non-mainline tree for
any amount of time.  it'd be best to squeeze it into mainline pretty
much immediately.  But...

> This patch was originally written by Denys Vlasenko, but it has been
> substantially modified by Anders Kaseorg and Tim Abbott to update it
> to apply to the current kernel and fix bugs.  Any bugs were added by
> Anders Kaseorg and Tim Abbott.
> 
> The purpose of this patch is to make it possible to build the kernel
> with "gcc -ffunction-sections -fdata-sections".
> 
> The problem is that with -ffunction-sections -fdata-sections, gcc
> creates sections like .text.head and .data.nosave whenever someone has
> innocuous code like this:
> 
> static void head(...) {...}
> 
> or this:
> 
> static int nosave = 1;
> 
> somewhere in the kernel.
> 
> The kernel linker scripts are confused by such names, and thus put
> these sections in the wrong places.
> 
> This patch renames all "magic" section names used by the kernel to not
> have this format, eliminating the possibility of such collisions.
> 
> Ksplice's 'run-pre matching' process is much simpler if the original
> kernel was compiled with -ffunction-sections and -fdata-sections.
> 

The changelog gives us no reason to do that.

So we `make it possible to build the kernel > with "gcc
-ffunction-sections -fdata-sections"'.  So what?  Where is the value in
that?

--
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/