Return-Path: Sender: Johan Hovold Date: Tue, 10 Oct 2017 18:36:30 +0200 From: Johan Hovold To: Marcel Holtmann Cc: Johan Hovold , =?iso-8859-1?Q?Fr=E9d=E9ric?= Danis , robh@kernel.org, sre@kernel.org, loic.poulain@gmail.com, lukas@wunner.de, hdegoede@redhat.com, linux-bluetooth@vger.kernel.org, linux-serial@vger.kernel.org, linux-acpi@vger.kernel.org Subject: Re: [PATCH 1/2] serdev: Add ACPI support Message-ID: <20171010163630.GO4269@localhost> References: <1507107090-15992-1-git-send-email-frederic.danis.oss@gmail.com> <1507107090-15992-2-git-send-email-frederic.danis.oss@gmail.com> <20171007151255.GH2618@localhost> <20171010081528.GJ4269@localhost> <3F3CFF22-4FCE-4283-8A41-73A9A7944494@holtmann.org> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 In-Reply-To: <3F3CFF22-4FCE-4283-8A41-73A9A7944494@holtmann.org> List-ID: On Tue, Oct 10, 2017 at 10:22:19AM +0200, Marcel Holtmann wrote: > > Some of the above are minor nits that can be addressed by a follow-up > > patch indeed, but the handling of nodes with no child devices needs to > > be correct (or you could end up breaking normal serial-port support). > > that sounds good to me as well. Lets get this feature into the ACPI > tree since I am going to push Hans’ patches into bluetooth-next as > soon as he re-submits the missing ID change. That hopefully gives us > then fully serdev support for ACPI and DT when it comes to Broadcom > Bluetooth controllers. But with Hans's work these devices should continue to function using the platform-device hacks until the ACPI patch eventually lands in Linus's tree (via the acpi tree). > Next step is to convert the Intel driver to also use ACPI based serdev > :) > > And for the brave, I think there are Realtek based systems using ACPI > as well. > > What I was wondering the other day is if we need a lsserdev tool or > some integration in lshw to be able to debug what serdev devices and > ID are present. The lsusb and /sys/kernel/debug/usb/devices is just > super powerful and easy when it comes to figuring out what people have > in their system. Maybe /sys/kernel/debug/serdev/devices could be > helpful as well. Just thinking out loud here. Yeah, maybe. Since you'd typically only have small number of serdev devices (say, max 4), using /sys/bus/serial/devices directly should not be too bad meanwhile. Not that much common information we can expose either, at least not in comparison to USB. But I'll keep it mind. :) Johan