linux-kernel - Re: [PATCH v3 4/4] arm64: dts: ti: Add support for J784S4 EVM board

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

Message-ID: <3d5e41f6-16a8-4298-ccd3-6db60f94eb47@ti.com>
Date:   Fri, 18 Nov 2022 15:08:23 -0600
From:   Andrew Davis <afd@...com>
To:     Nishanth Menon <nm@...com>
CC:     Apurva Nandan <a-nandan@...com>,
        Vignesh Raghavendra <vigneshr@...com>,
        Tero Kristo <kristo@...nel.org>,
        Rob Herring <robh+dt@...nel.org>,
        Krzysztof Kozlowski <krzysztof.kozlowski+dt@...aro.org>,
        Linus Walleij <linus.walleij@...aro.org>,
        <linux-arm-kernel@...ts.infradead.org>,
        <devicetree@...r.kernel.org>, <linux-kernel@...r.kernel.org>,
        <linux-gpio@...r.kernel.org>, Hari Nagalla <hnagalla@...com>
Subject: Re: [PATCH v3 4/4] arm64: dts: ti: Add support for J784S4 EVM board

On 11/18/22 1:27 PM, Nishanth Menon wrote:
> On 12:15-20221118, Andrew Davis wrote:
>> I don't see either of those addressed in that thread, only that
>> the aliases should go in the .dts files and be trimmed, nothing
> 
> Key is trimmed to what the system and ecosystem needs.
> 
>> stops us from:
>>
>> chosen {
>> 	stdout-path = "serial10:115200n8";
>> };
>>
>> aliases {
>> 	serial10 = &main_uart8;
>> };
> 
> Do we need 10 serial aliases? There are'nt 10 serial ports exposed in
> j782s2. ok - lets say we do this, then: [1] is needed to boot? but why
> do we need to do that for all armv8 platforms when aliases allows us

Why do we need SERIAL_8250_NR_UARTS at all, might be a better question.
These should be dynamically allocated if the number goes over the
default count imposed by the TTY framework. Maybe folks are still a
bit too afraid to touch the TTY subsystem core, I don't blame them..

> to trim it to just the 3 or 4 serial ports the platform really needs
> That + being able to use the convention that serial2 is always linux
> console, is'nt that a good thing? Hence recommending to just expose the
> serialports as aliases to exactly what we need while keeping serial2 as
> the linux console (which in this case happens to be main_uart8 - example
> as j721s2 does).
> 

"serial2 as the linux console" is *not* a convention, we just don't want to
fix up our bootloader/userspace to actually reason about what serial ports to
put logins on. Why not make ttyS10 the default, or ttyS666, it doesn't solve
your multi-distro issue either way since they usually only start a login on
ttyS0, console=, and/or the first virtual tty. Never on ttyS2. So you are
hacking up DT for a solution that doesn't do what you want in the end.

Andrew

> [1] https://lore.kernel.org/lkml/3ab9addf-7938-fcf3-6147-15a998e37d2d@ti.com/
>