[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <CAL_Jsq+L6tKw+OzYVHSyou3eW5s7OChEbwqPe=qtQkRW=B+y-w@mail.gmail.com>
Date: Mon, 24 Jun 2024 10:49:34 -0600
From: Rob Herring <robh@...nel.org>
To: Krzysztof Kozlowski <krzk@...nel.org>
Cc: Joel Stanley <joel@....id.au>, Krzysztof Kozlowski <krzk+dt@...nel.org>,
Conor Dooley <conor+dt@...nel.org>, Andrew Jeffery <andrew@...econstruct.com.au>,
devicetree@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
linux-aspeed@...ts.ozlabs.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH] arm: dts: aspeed: Use standard 'i2c' bus node name
On Sun, Jun 23, 2024 at 10:47 PM Krzysztof Kozlowski <krzk@...nel.org> wrote:
>
> On 24/06/2024 03:55, Joel Stanley wrote:
> > On Sat, 1 Jun 2024 at 05:01, Rob Herring (Arm) <robh@...nel.org> wrote:
> >>
> >> The standard node name for I2C buses is 'i2c'.
> >
> > Unfortunately this can't be merged, as it will break userspace. There
> > is a lot of code out there that looks up devices based on the device
> > tree node path:
> >
> > https://github.com/openbmc/phosphor-state-manager/blob/3c1351cc2b63178876ef68f4107c9804d2e17dcc/meson.options#L140
>
> Eh, thanks for sharing.
>
> That's totally stupid user-space. The sysfs path to /sys/devices, or
> some grep by compatible, would be fine, but iterating over firmware/DT
> is just wrong.
Agreed, but Greg (and maybe it is actually documented somewhere) will
tell you that /sys/devices or /sys/bus/ paths are not ABI. /sys/class
is the ABI path.
Rob
Powered by blists - more mailing lists