linux-kernel - Re: [PATCH] of: fdt: Honor CONFIG_CMDLINE* even without /chosen node

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]

Date:   Tue, 13 Dec 2022 23:07:44 +0100
From:   Alexander Sverdlin <alexander.sverdlin@...il.com>
To:     Rob Herring <robh@...nel.org>,
        Linus Walleij <linus.walleij@...aro.org>
Cc:     devicetree@...r.kernel.org, Frank Rowand <frowand.list@...il.com>,
        linux-kernel@...r.kernel.org, Arnd Bergmann <arnd@...nel.org>,
        Nikita Shubin <nikita.shubin@...uefel.me>,
        Hartley Sweeten <hsweeten@...ionengravers.com>,
        Lukasz Majewski <lukma@...x.de>,
        Linus Walleij <linusw@...nel.org>
Subject: Re: [PATCH] of: fdt: Honor CONFIG_CMDLINE* even without /chosen node

Hello Rob,

On Tue, 2022-12-13 at 09:29 -0600, Rob Herring wrote:
> On Tue, Dec 13, 2022 at 09:51:33AM +0100, Linus Walleij wrote:
> > On Mon, Dec 12, 2022 at 7:01 AM Alexander Sverdlin
> > <alexander.sverdlin@...il.com> wrote:
> > 
> > > I do not read a strict requirement on /chosen node in either ePAPR or in
> > > Documentation/devicetree. Help text for CONFIG_CMDLINE and
> > > CONFIG_CMDLINE_EXTEND doesn't make their behavior explicitly dependent on
> > > the presence of /chosen or the presense of /chosen/bootargs.
> > > 
> > > However the early check for /chosen and bailing out in
> > > early_init_dt_scan_chosen() skips CONFIG_CMDLINE handling which is not
> > > really related to /chosen node or the particular method of passing cmdline
> > > from bootloader.
> > > 
> > > This leads to counterintuitive combinations (assuming
> > > CONFIG_CMDLINE_EXTEND=y):
> > > 
> > > a) bootargs="foo", CONFIG_CMDLINE="bar" => cmdline=="foo bar"
> > > b) /chosen missing, CONFIG_CMDLINE="bar" => cmdline==""
> > > c) bootargs="", CONFIG_CMDLINE="bar" => cmdline==" bar"
> > > 
> > > Move CONFIG_CMDLINE handling outside of early_init_dt_scan_chosen() so that
> > > cases b and c above result in the same cmdline.
> > > 
> > > Signed-off-by: Alexander Sverdlin <alexander.sverdlin@...il.com>
> > 
> > Excellent debugging Alexander!
> > Reviewed-by: Linus Walleij <linus.walleij@...aro.org>
> > 
> > I also think this should go to stable.
> 
> We have to be careful there. This could change behavior on a working 
> system. A system taking the cmdline entirely from a built kernel and 
> no initrd is going to be pretty customized already, I think they can 
> carry a patch. What platform is this anyways?

I've stumbled upon this testing first DT conversion patches for EP93xx (ARM).

> This has actually been known for some time[1][2]. My concern in the past 
> (besides wanting all the cmdline manipulation being common) was MIPS. 

This "change of behavior" actually changes one exact corner case:
no /chosen node + CONFIG_CMDLINE!="" + CONFIG_CMDLINE_EXTEND=y

If someone was intentionally hiding something in the config file
under CONFIG_CMDLINE but didn't want it to appear on the kernel command
line in the past, he could just reconfigure new kernel version after
the change and remove the above configs.

> MIPS in particular has lots of sources for cmdline and ways to combine 
> it. However, MIPS has since stopped using this code and does their own 
> parsing (not great either IMO).

I agree, this code screams to be common.

-- 
Alexander Sverdlin.