linux-kernel - Re: [PATCH 2/2] arm: dts: mt6589: Add device tree for Fairphone 1

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: <5755444.lOV4Wx5bFT@g550jk>
Date:   Tue, 12 Oct 2021 19:54:38 +0200
From:   Luca Weiss <luca@...tu.xyz>
To:     linux-mediatek@...ts.infradead.org,
        Matthias Brugger <matthias.bgg@...il.com>
Cc:     ~postmarketos/upstreaming@...ts.sr.ht,
        Arnd Bergmann <arnd@...db.de>,
        Enric Balletbo i Serra <enric.balletbo@...labora.com>,
        Fabien Parent <fparent@...libre.com>,
        Hsin-Yi Wang <hsinyi@...omium.org>,
        Olof Johansson <olof@...om.net>,
        Rob Herring <robh+dt@...nel.org>,
        Sean Wang <sean.wang@...iatek.com>,
        Seiya Wang <seiya.wang@...iatek.com>,
        devicetree@...r.kernel.org, linux-arm-kernel@...ts.infradead.org,
        linux-kernel@...r.kernel.org, soc@...nel.org
Subject: Re: [PATCH 2/2] arm: dts: mt6589: Add device tree for Fairphone 1

Hi Matthias,

On Freitag, 8. Oktober 2021 13:49:25 CEST Matthias Brugger wrote:
> On 05/10/2021 22:28, Luca Weiss wrote:
> > Add rudimentary support for the Fairphone 1, based on MT6589 to boot to
> > UART console.
> > 
> > The recently added SMP support needs to be disabled for this board as
> > the kernel panics executing /init with it, even though the CPUs seem to
> > start up fine - maybe a stability issue.
> > 
> > [    0.072010] smp: Bringing up secondary CPUs ...
> > [    0.131888] CPU1: thread -1, cpu 1, socket 0, mpidr 80000001
> > [    0.191889] CPU2: thread -1, cpu 2, socket 0, mpidr 80000002
> > [    0.251890] CPU3: thread -1, cpu 3, socket 0, mpidr 80000003
> > [    0.251982] smp: Brought up 1 node, 4 CPUs
> > [    0.254745] SMP: Total of 4 processors activated (7982.28 BogoMIPS).
> > [    0.255582] CPU: All CPU(s) started in SVC mode.
> > 
> > [    0.472039] Run /init as init process
> > [    0.473317] Kernel panic - not syncing: Attempted to kill init!
> > exitcode=0x00000004
> Would be nice to find out why. Did you tried to boot the system with
> enable-method set but with bringing up just one or two cpus?

Answered further down.

> 
> > Signed-off-by: Luca Weiss <luca@...tu.xyz>
> > ---
> > 
> >   arch/arm/boot/dts/Makefile                 |  1 +
> >   arch/arm/boot/dts/mt6589-fairphone-fp1.dts | 30 ++++++++++++++++++++++
> >   2 files changed, 31 insertions(+)
> >   create mode 100644 arch/arm/boot/dts/mt6589-fairphone-fp1.dts
> > 
> > diff --git a/arch/arm/boot/dts/Makefile b/arch/arm/boot/dts/Makefile
> > index 7e0934180724..24f402db2613 100644
> > --- a/arch/arm/boot/dts/Makefile
> > +++ b/arch/arm/boot/dts/Makefile
> > @@ -1437,6 +1437,7 @@ dtb-$(CONFIG_ARCH_MEDIATEK) += \
> > 
> >   	mt2701-evb.dtb \
> >   	mt6580-evbp1.dtb \
> >   	mt6589-aquaris5.dtb \
> > 
> > +	mt6589-fairphone-fp1.dtb \
> > 
> >   	mt6592-evb.dtb \
> >   	mt7623a-rfb-emmc.dtb \
> >   	mt7623a-rfb-nand.dtb \
> > 
> > diff --git a/arch/arm/boot/dts/mt6589-fairphone-fp1.dts
> > b/arch/arm/boot/dts/mt6589-fairphone-fp1.dts new file mode 100644
> > index 000000000000..32c14ecf2244
> > --- /dev/null
> > +++ b/arch/arm/boot/dts/mt6589-fairphone-fp1.dts
> > @@ -0,0 +1,30 @@
> > +// SPDX-License-Identifier: BSD-3-Clause
> > +/*
> > + * Copyright (c) 2021, Luca Weiss <luca@...tu.xyz>
> > + */
> > +
> > +/dts-v1/;
> > +#include "mt6589.dtsi"
> > +
> > +/ {
> > +	model = "Fairphone 1";
> > +	compatible = "fairphone,fp1", "mediatek,mt6589";
> > +
> > +	chosen {
> > +		stdout-path = &uart3;
> > +	};
> > +
> > +	cpus {
> 
> I'd expected "&cpus" why can we overwrite delete the node property like this
> here?

Both results in the same, dtc just merges everything together, so as long as 
the node name is identical, it works.
Also I cannot use &cpus because cpus in mt6589.dtsi doesn't have a label set.

Regarding SMP:
I have tried setting maxcpus=2 in cmdline and that still makes the kernel 
panic. With maxcpus=1 and leaving the deleting out of the dts the kernel is 
stable and works properly.

So I think it's better to leave this out of the dts and keep maxcpus=1 in 
cmdline (until this gets fixed).

I've also heard from the person adding enable-method to mt6589.dtsi that it 
still works on their board, so something's different, maybe a different SoC 
revision, different bootloader behavior or whatever.

If that's fine with you, I'll send a v2 with that fixed.

> > +		/* SMP is not stable on this board, makes the kernel 
panic */
> > +		/delete-property/ enable-method;
> > +	};
> > +
> > +	memory {

Also I was told off-list that this should be called memory@...00000 because of 
the reg, will fix in v2.

Regards
Luca

> > +		device_type = "memory";
> > +		reg = <0x80000000 0x40000000>;
> > +	};
> > +};
> > +
> > +&uart3 {
> > +	status = "okay";
> > +};