Re: [PATCH] mmc: part_switch: fixes switch on gp3 partition

From: Dominique Martinet <dominique.martinet@atmark-techno.com>
To: Linus Walleij <linus.walleij@linaro.org>
Cc: Dominique Martinet <asmadeus@codewreck.org>,
	Ulf Hansson <ulf.hansson@linaro.org>,
	Jorge Ramirez-Ortiz <jorge@foundries.io>,
	linux-mmc@vger.kernel.org, linux-kernel@vger.kernel.org,
	stable@vger.kernel.org
Subject: Re: [PATCH] mmc: part_switch: fixes switch on gp3 partition
Date: Wed, 6 Mar 2024 17:15:16 +0900	[thread overview]
Message-ID: <ZegmFOnrLk79Oacf@atmark-techno.com> (raw)
In-Reply-To: <CACRpkda4pVotd9Fc2Qn0Ae=89sZR7-rXDiZ7OdHE3eDvO=049Q@mail.gmail.com>

Thanks for the review!

Linus Walleij wrote on Wed, Mar 06, 2024 at 09:03:45AM +0100:
> > A couple of notes:
> > - this doesn't fail on all eMMCs, I can still access gp3 on some models
> >   but it seems to fail reliably with micron's "G1M15L"
> > - I've encountered this on the 5.10 backport (in 5.10.208), so that'll
> >   need to be backported everywhere the fix was taken...
> 
> Which device is this?

Sorry I gave the 'name' as seen in linux, the actual reference is
MTFC32GAZAQHD-IT
(but we're creating the GP manually)

> I have never seen an eMMC using the GP:s in my life.
> 
> Or did you create the GP manually?

Yes, we're creating them on all our devices:
(japanese only, but you might get something out of google translate or
similar if interested: https://armadillo.atmark-techno.com/about )

I'd be hard pressed to explain why other than "it's always been done"
and I've been replicating that pattern, but as far as I understand the
gp partitions are always created in "write reliable" mode so it might
help a bit with lifetime if the main area is left as default.

We're using them for rare, important logs and things like that (minimal
trace of updates, hard reset etc for diagnostic)

-- 
Dominique