linux-kernel - Re: [PATCH 00/89] Major reorganization of <linux/sched.h>

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: <20170206221359.GB7481@gmail.com>
Date:   Mon, 6 Feb 2017 23:13:59 +0100
From:   Ingo Molnar <mingo@...nel.org>
To:     Linus Torvalds <torvalds@...ux-foundation.org>
Cc:     Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Andrew Morton <akpm@...ux-foundation.org>,
        Mike Galbraith <efault@....de>,
        Oleg Nesterov <oleg@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Thomas Gleixner <tglx@...utronix.de>
Subject: Re: [PATCH 00/89] Major reorganization of <linux/sched.h>

* Linus Torvalds <torvalds@...ux-foundation.org> wrote:

> On Mon, Feb 6, 2017 at 5:28 AM, Ingo Molnar <mingo@...nel.org> wrote:
> > So 25+ years ago, in Linux-0.01, include/linux/sched.h was already
> > the biggest core kernel header file: [...]
> 
> Ok, so having tried to look through this series I do like it, but I'd be 
> *really* much happier if more of it was just verifiably a semantic no-op.

Yeah - that was the intention.

It's really hard to verify it in an automated fashion I think, in a cross arch 
way. I did a sizeof(task_struct) before/after check on x86 defconfig, which did 
not uncover the extra pointer on CONFIG_TASK_DELAY_ACCT=n, because defconfig has 
that enabled...

On the latest tree I've done a wider test of sizeof(task_struct):

		allnoconfig	defconfig	allmodconfig
  -----------------------------------------------------------
  before:	0x1400		0x19c0		0x3f00
  after:	0x1400		0x19c0		0x3f00

Which seems to support my intention that the series should be an overall invariant 
on 'struct task_struct' semantics.

> There were all those small things in there (Peter pointed out those cpumask 
> things I wouldn't for the life of me have noticed) that were really subtle, and 
> were really hidden by the fact that there was just a lot of non-semantic 
> changes.

I think the ->cpus_allowed bugs Peter noticed are pre-existing - that patch 
doesn't intend to make any semantic changes.

Thanks,

	Ingo