linux-kernel - Re: [PATCH] x86,mm: print likely CPU at segfault time

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]

Date:   Fri, 23 Jul 2021 21:38:38 -0400
From:   Rik van Riel <riel@...riel.com>
To:     Thomas Gleixner <tglx@...utronix.de>, linux-kernel@...r.kernel.org
Cc:     Dave Hansen <dave.hansen@...ux.intel.com>,
        Andy Lutomirski <luto@...nel.org>, kernel-team@...com,
        Peter Zijlstra <peterz@...radead.org>,
        Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
        x86@...nel.org
Subject: Re: [PATCH] x86,mm: print likely CPU at segfault time

On Wed, 2021-07-21 at 22:36 +0200, Thomas Gleixner wrote:
> Rik,
> 
> On Mon, Jul 19 2021 at 15:00, Rik van Riel wrote:
> > 
> > Adding a printk to show_signal_msg() achieves that purpose. It
> > isn't
> > perfect since the task might get rescheduled on another CPU between
> > when the fault hit and when the message is printed, but it should
> > be
> > good enough to show correlation between userspace and kernel errors
> > when dealing with a bad CPU.
> 
> we could collect the cpu number in do_*_addr_fault() before
> interrupts
> are enabled and just hand it through. There are only a few callchains
> which end up in __bad_area_nosemaphore().

We could, but do we really want to add that to the hot path
for page faults, when segfaults are so rare?

I suspect the simple patch I sent will be good enough to
identify a bad CPU, even if only 3 out of 4 userspace crashes
get attributed to the right CPU...

I would be happy to write a patch that does what you want
though, so you can compare them side by side :)

-- 
All Rights Reversed.

Download attachment "signature.asc" of type "application/pgp-signature" (489 bytes)