linux-kernel - Re: [PATCH] also discard .*init sections after module initialization

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: <alpine.DEB.1.10.0907011454050.18560@vinegar-pot.mit.edu>
Date:	Mon, 6 Jul 2009 11:55:40 -0400 (EDT)
From:	Tim Abbott <tabbott@...lice.com>
To:	Jan Beulich <JBeulich@...ell.com>
cc:	Rusty Russell <rusty@...tcorp.com.au>, tony.luck@...el.com,
	Benjamin Herrenschmidt <benh@...nel.crashing.org>,
	Tim Abbott <tabbott@....edu>, sam@...nborg.org,
	Paul Mackerras <paulus@...ba.org>, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] also discard .*init sections after module	
 initialization

On Wed, 1 Jul 2009, Jan Beulich wrote:

> >>> Rusty Russell <rusty@...tcorp.com.au> 01.07.09 07:22 >>>
> >On Tue, 30 Jun 2009 09:27:26 pm Jan Beulich wrote:
> >> Just like .init, these sections are supposed to be unneeded after init,
> >> and modpost warns about improper section references anyway.
> >
> >Tim Abbot had a section renaming patch, for -ffunction-sections.  Can we kill 
> >two birds with one stone and have a unique string for init sections, and 
> >another for exit sections?
> 
> Not sure, would need to see what that patch does first. And then, the patch
> here doesn't do any renaming.

We don't need to rename the .{dev,cpu,mem}init.* sections for 
-ffunction-sections -fdata-sections; the only things that need to be 
renamed for that are sections with names that start with ".text.", 
".data.", ".bss.", and ".rodata." since those are the section names 
generated by -ffunction-sections -fdata-sections.  So I think that this 
change is orthogonal to my efforts.

> >> Likewise
> >> for .*exit which, other than .exit, aren't needed for the module unload
> >> path.
> >
> >Currently true, but I can't see that being true in general.
> 
> Why? .*exit sections deal with device/memory/CPU hot remove, which is
> orthogonal to module unload.

For the core kernel, the .devinit sections are currently being squashed 
into either the .data section or the .init.data section depending on 
whether hotplug is enabled using the vmlinux linker scripts.  Can we use 
the same approach for modules (i.e. have a module linker script that does 
this)?  It seems that would avoid having two duplicate mechanisms for 
doing the same thing, one in the module loader and one in the core 
kernel's linker scripts.

	-Tim Abbott


--
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/