Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
MIME-Version: 1.0
References: <20220817202538.21493-1-leoyang.li@nxp.com> <20220817202538.21493-2-leoyang.li@nxp.com>
 <CAOesGMhz8PYNG_bgMX-6gka77k1hJOZUv6xqJRqATaJ6mFbk6A@mail.gmail.com>
In-Reply-To: <CAOesGMhz8PYNG_bgMX-6gka77k1hJOZUv6xqJRqATaJ6mFbk6A@mail.gmail.com>
From:   Olof Johansson <olof@lixom.net>
Date:   Mon, 12 Sep 2022 11:54:06 -0700
Message-ID: <CAOesGMjsegmLUVxbjZJ3kgxnDkVTyxy5BB=is2CM_7Se7TQrWQ@mail.gmail.com>
Subject: Re: [PATCH v4 1/2] arm64: dts: lx2160a: update PCIe nodes to match
 rev2 silicon
To:     Li Yang <leoyang.li@nxp.com>
Cc:     shawnguo@kernel.org, devicetree@vger.kernel.org,
        robh+dt@kernel.org, linux-arm-kernel@lists.infradead.org,
        linux-kernel@vger.kernel.org, Hou Zhiqiang <Zhiqiang.Hou@nxp.com>
Content-Type: text/plain; charset="UTF-8"
Precedence: bulk

On Mon, Sep 12, 2022 at 12:05 AM Olof Johansson <olof@lixom.net> wrote:
>
> Hi,
>
> On Wed, Aug 17, 2022 at 1:26 PM Li Yang <leoyang.li@nxp.com> wrote:
> >
> > The original dts was created based on the non-production rev1 silicon
> > which was only used for evaluation.  Update the PCIe nodes to align with
> > the different controller used in production rev2 silicon.
>
> How can I confirm what version of silicon I have on a system?
>
> My non-evaluation commercially purchased system (HoneyComb LX2K) has:
>
> # cat /sys/bus/soc/devices/soc0/revision
> 1.0
>
> And I will be really grumpy if this system stops working. It's what I
> use to do all my maintainer work, even if that's been fairly dormant
> this year.
>
> It's overall setting off red flags to update an in-place devicetree to
> a "new revision" of silicon instead of adding a new DT for said
> revision. 2160A has been on the market for several years, so it just
> seems odd to all of the sudden retroactively make things
> non-backwards-compatible.

Confirmed that this patch renders my HoneyComb unbootable -- PCIe doesn't probe.

Shawn, please revert, and be on the lookout for similar problematic
approaches in the future. Thanks!


-Olof