linux-kernel - Re: [PATCH v4 1/3] watchdog: aspeed: Update bootstatus handling

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <ZyUcIIb1dtoNhX00@heinlein.vulture-banana.ts.net>
Date: Fri, 1 Nov 2024 14:21:20 -0400
From: Patrick Williams <patrick@...cx.xyz>
To: Chin-Ting Kuo <chin-ting_kuo@...eedtech.com>
Cc: joel@....id.au, andrew@...econstruct.com.au, wim@...ux-watchdog.org,
	linux@...ck-us.net, linux-arm-kernel@...ts.infradead.org,
	linux-aspeed@...ts.ozlabs.org, linux-kernel@...r.kernel.org,
	linux-watchdog@...r.kernel.org, Peter.Yin@...ntatw.com,
	Patrick_NC_Lin@...ynn.com, Bonnie_Lo@...ynn.com,
	DELPHINE_CHIU@...ynn.com, bmc-sw@...eedtech.com,
	chnguyen@...erecomputing.com
Subject: Re: [PATCH v4 1/3] watchdog: aspeed: Update bootstatus handling

On Fri, Nov 01, 2024 at 08:11:59PM +0800, Chin-Ting Kuo wrote:
> The boot status mapping rule follows the latest design guide from
> the OpenBMC shown as below.
> https://github.com/openbmc/docs/blob/master/designs/bmc-reboot-cause-update.md#proposed-design
> - WDIOF_EXTERN1   => system is reset by Software
> - WDIOF_CARDRESET => system is reset by WDT SoC reset
> - Others          => other reset events, e.g., power on reset.

I'm quite surprised that the above is relevant for a kernel driver at
all.  Isn't "EXTERN1" a name of a real watchdog signal from your
hardware (my recollection is that there are 2 external watchdogs).  I
think the point of this referenced design document was that most users
of BMCs have "EXTERN1" used a for software reset conditions.
`CARDRESET` should be representing resets by the watchdog itself.

The purpose of this design proposal was not to require very specific
changes to individual watchdog drivers, but to align the userspace use
with the best practices already from other watchdog drivers.  I don't
think the kernel driver should be bending to match a particular
userspace implementation; you should be exposing the information
available to your hardware.

Having said that, it was known that there would need to be changes to
the driver because some of these conditions were not adequately exposed
at all.  I'm just still surprised that we're needing to reference that
document as part of these changes.

> 
> On ASPEED platform, the boot status is recorded in the SCU registers.
> - AST2400: Only a bit represents for any WDT reset.
> - AST2500: The reset triggered by different WDT controllers can be
>            distinguished by different SCU bits. But, WDIOF_EXTERN1 or
>            WDIOF_CARDRESET still cannot be identified due to
>            HW limitation.
> - AST2600: Different from AST2500, additional HW bits are added for
>            distinguishing WDIOF_EXTERN1 and WDIOF_CARDRESET.

-- 
Patrick Williams

Download attachment "signature.asc" of type "application/pgp-signature" (834 bytes)