linux-kernel - Re: Dealing with custom hardware on x86

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: Windows password security audit tool. GUI, reports in PDF.

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

Message-ID: <20211208214028.GA16475@duo.ucw.cz>
Date:   Wed, 8 Dec 2021 22:40:28 +0100
From:   Pavel Machek <pavel@....cz>
To:     Chris Packham <Chris.Packham@...iedtelesis.co.nz>
Cc:     "x86@...nel.org" <x86@...nel.org>,
        "linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>
Subject: Re: Dealing with custom hardware on x86

On Wed 2021-11-24 20:22:38, Chris Packham wrote:
> Hi Pavel,
> 
> Thanks for responding
> 
> On 24/11/21 11:47 pm, Pavel Machek wrote:
> > Hi!
> >
> >> I'm about to start a project involving custom hardware using an x86 CPU.
> >> Complicating things somewhat we will have a CPU board supplied by a 3rd
> >> party vendor using a COM Express type 7 connector which will plug in to
> >> the board we're designing (actually 2 different boards each with some
> >> modular aspects) .
> >>
> >> Coming from an embedded (mostly ARM) background I'm used to describing
> >> the hardware via a devicetree and dealing with plug-in modules by
> >> loading devicetree overlays. How do people achieve this kind of thing on
> >> x86?
> >>
> >> I gather ACPI is involved somewhere but the knowledge that the BIOS has
> >> will go as far as the COM Express header and somehow we'll have to tell
> >> the kernel about the various non-probeable (mostly i2c) devices on our
> >> board which is kind of where I'm stuck. Any pointers would be greatly
> >> appreciated.
> > In x86 land, we usually use PCI/PCIE and have a add-on board expose
> > unique IDs with driver knowing what the IDs are. ACPI should not be
> > involved for add-on boards.
> 
> So if I had a chain of i2c devices (endpoints behind muxes etc) I'd have 
> the PCI add-on board declare that hierarchy with i2c_new_device()?

That sounds reasonable, yes.
									Pavel

-- 
http://www.livejournal.com/~pavelmachek

Download attachment "signature.asc" of type "application/pgp-signature" (196 bytes)