linux-kernel - Re: [PATCH V2] ACPI: APEI: Use ERST timeout for slow devices

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: <20231025142245.GHZTkktbRl1wjfNc15@fat_crate.local>
Date:   Wed, 25 Oct 2023 16:22:45 +0200
From:   Borislav Petkov <bp@...en8.de>
To:     Jeshua Smith <jeshuas@...dia.com>
Cc:     "Rafael J. Wysocki" <rafael@...nel.org>,
        "Luck, Tony" <tony.luck@...el.com>,
        "james.morse@....com" <james.morse@....com>,
        "keescook@...omium.org" <keescook@...omium.org>,
        "gpiccoli@...lia.com" <gpiccoli@...lia.com>,
        "lenb@...nel.org" <lenb@...nel.org>,
        "linux-acpi@...r.kernel.org" <linux-acpi@...r.kernel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
        "linux-hardening@...r.kernel.org" <linux-hardening@...r.kernel.org>,
        "linux-tegra@...r.kernel.org" <linux-tegra@...r.kernel.org>,
        Thierry Reding <treding@...dia.com>,
        Jonathan Hunter <jonathanh@...dia.com>
Subject: Re: [PATCH V2] ACPI: APEI: Use ERST timeout for slow devices

Hi,

On Wed, Oct 25, 2023 at 02:09:37PM +0000, Jeshua Smith wrote:
<... snip a very detailed and good explanation... >

> Writing 128 nor-flash pages would then take 120us * 128 = 15ms
> typical, or 1800us * 128 = 230.4ms max.

This is perfectly suitable to be in the commit message - it explains in
exact detail why the change is needed.

> Actual use case:
>
> Kernel panic -> Pstore calls APEI's ERST code to write the ~32KB error
> log to persistent store -> ERST code writes the error log to
> nor-flash, which takes more than 1ms to complete. This is expected, as
> communicated by the platform to the OS via the maximum time field in
> the ERST table.

This is actually very important and it justifies the need for that
change even more - you want to flush out the complete panic message to
pstore and not only the first couple of lines.

> ... and therefore the extended (ERST-defined) timeout is only applied
> for implementations that indicate that they are "slow". I assume that
> platforms which bother to set the "slow" bit will also specify actual
> timings, and platforms which don't are OK with the current 1ms
> timeout.

Yap, makes perfect sense to me.

> Does that answer your questions?

Yes, thanks for taking the time to explain this in such a detail and
precisely. I think you should use the main bits of what you wrote here
and add them to the commit message - after this there are no more
questions why this patch is needed, IMO.

Thx.

-- 
Regards/Gruss,
    Boris.

https://people.kernel.org/tglx/notes-about-netiquette