linux-kernel - Re: [PATCH 2/2] efi: implement interruptible runtime services

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] [day] [month] [year] [list]

Message-ID: <20160114162142.GD3602@codeblueprint.co.uk>
Date:	Thu, 14 Jan 2016 16:21:42 +0000
From:	Matt Fleming <matt@...eblueprint.co.uk>
To:	Sylvain Chouleur <sylvain.chouleur@...il.com>
Cc:	linux-efi@...r.kernel.org,
	Sylvain Chouleur <sylvain.chouleur@...el.com>,
	linux-kernel@...r.kernel.org, "H. Peter Anvin" <hpa@...or.com>,
	Ingo Molnar <mingo@...nel.org>,
	Thomas Gleixner <tglx@...utronix.de>
Subject: Re: [PATCH 2/2] efi: implement interruptible runtime services

On Fri, 08 Jan, at 02:57:13PM, Sylvain Chouleur wrote:
> 
> I understand, like I said above I'll modify efi_interruptible handlers to
> call legacy ones in case of panic context.
> I would like to avoid removing the panic part of this patch and take time
> to clean it before merging the whole.
 
OK, well at least split out the panic diddling into a separate patch
so that we can discuss the merits of it separately to the other, less
contentious changes.

> > Kconfig is a last resort because it's a build-time decision and
> > greatly limits the flexibility of the kernel. It becomes no longer
> > possible to run a single kernel image with various CONFIG_* enabled on
> > x86 hardware - you now need a special EFI_INTERRUPTIBLE build.
> >
> > Which apart from being a major headache for distributions in general
> > is generally frowned upon for the x86 architecture.
> >
> > If there's any way at all of making this a runtime decision that would
> > be much better.
> 
> I think the best would be to bind this driver with the one which receives the
> interrupts from CSE to write the variables. Then we would have a consistency
> on the feature.
> Does this seems ok for you?

I think that'd be an improvement, yeah.