linux-kernel - Re: [PATCH] acpi/ghes: Make ghes_panic

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: <87wmfhjusk.fsf@DESKTOP-5N7EMDA>
Date: Mon, 30 Dec 2024 19:04:59 +0800
From: "Huang, Ying" <ying.huang@...ux.alibaba.com>
To: Borislav Petkov <bp@...en8.de>
Cc: Feng Tang <feng.tang@...ux.alibaba.com>,  rafael@...nel.org,  Len Brown
 <lenb@...nel.org>,  James Morse <james.morse@....com>,  Tony Luck
 <tony.luck@...el.com>,  linux-acpi@...r.kernel.org,
  linux-kernel@...r.kernel.org,  Ira Weiny <ira.weiny@...el.com>,  Dave
 Jiang <dave.jiang@...el.com>,  Dan Williams <dan.j.williams@...el.com>,
  Andi Kleen <ak@...ux.intel.com>
Subject: Re: [PATCH] acpi/ghes: Make ghes_panic_timeout adjustable as a
 parameter

Borislav Petkov <bp@...en8.de> writes:

> On Mon, Dec 30, 2024 at 01:54:36PM +0800, Huang, Ying wrote:
>> For example, it may be OK to wait forever for a software error, but it may
>> be better to reboot the system to contain the influence of the hardware
>> error for some hardware errors.
>
> A default panic timeout of 30 seconds for hw errors?! You do realize that 30
> seconds for a machine is an eternity and by that time your hardware error has
> long propagated and corrupted results, right?
>
> So your timeout is not even trying to do what you want.
>
> So unless I'm missing something, this ghes timeout needs to go - if you want
> to "contain the influence" you need to panic *immediately*! And not even that
> would help in some cases - hw has its own protections there so the OS
> panicking is meh. At least on x86, that is.

OK.  30 seconds isn't good enough for hw errors.

Another possible benefit of ghes_panic_timeout is,

rebooting instead of waiting forever can help us to log/report the
hardware errors earlier.

For example, the hardware errors could be logged in some simple
non-volatile storage (such as EFI variables) during panic or kdump, etc.
Then, after reboot, the new kernel could report the hardware errors in
some way.

>> So, we introduced another knob for that.
>
> No, that another knob is piling more of the silly ontop.

---
Best Regards,
Huang, Ying