linux-kernel - Re: [PATCH v3 0/6] allow ramoops to collect all kmesg

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: <202005121146.3B3C1FE0D@keescook>
Date:   Tue, 12 May 2020 11:53:42 -0700
From:   Kees Cook <keescook@...omium.org>
To:     Pavel Tatashin <pasha.tatashin@...een.com>
Cc:     Petr Mladek <pmladek@...e.com>, Anton Vorontsov <anton@...msg.org>,
        Colin Cross <ccross@...roid.com>,
        Tony Luck <tony.luck@...el.com>,
        Jonathan Corbet <corbet@....net>,
        Rob Herring <robh+dt@...nel.org>,
        Benson Leung <bleung@...omium.org>,
        Enric Balletbo i Serra <enric.balletbo@...labora.com>,
        Sergey Senozhatsky <sergey.senozhatsky@...il.com>,
        Steven Rostedt <rostedt@...dmis.org>,
        James Morris <jmorris@...ei.org>,
        Sasha Levin <sashal@...nel.org>,
        Linux Doc Mailing List <linux-doc@...r.kernel.org>,
        LKML <linux-kernel@...r.kernel.org>, devicetree@...r.kernel.org
Subject: Re: [PATCH v3 0/6] allow ramoops to collect all kmesg_dump events

On Tue, May 12, 2020 at 12:49:10PM -0400, Pavel Tatashin wrote:
> On Tue, May 12, 2020 at 11:52 AM Petr Mladek <pmladek@...e.com> wrote:
> > I wonder if anyone is actually using the ramoops.dump_oops parameter
> > in reality. I would personally make it deprecated and change the
> > default behavior to work according to printk.always_kmsg_dump parameter.
> 
> This sounds alright to me with one slight problem. I am doing this
> work for an embedded arm64 SoC, so controlling everything via device
> tree is preferable compared to having some settings via device tree
> and others via kernel parameters, especially because the kernel
> parameters are hardcoded by firmware that we try not to update too
> often for uptime reasons.

I'm entirely convinced that this area of pstore needs to be cleaned up
and I want to have the pstore backends be able to declare their kmsg
dump reason filters in a configurable fashion. So at least on the pstore
end, I intend to have some way to do this.

> > IMHO, ramoops.dump_oops just increases complexity and should not have
> > been introduced at all. I would try hard to avoid introducing even bigger
> > complecity and mess.
> 
> I agree, amoops.dump_oops should be depricated with or without
> max_reason change.

Yup. dump_oops will be deprecated in favor of whatever we settle on here.

> > I know that there is the "do not break existing userspace" rule. The
> > question is if there is any user and if it is worth it.
> >
> > > I agree, the reasons in kmsg_dump_reason do not order well  (I
> > > actually want to add another reason for kexec type reboots, and where
> > > do I put it?), so how about if we change the ordering list to
> > > bitfield/flags, and instead of max_reason provide: "reasons" bitset?
> >
> > It looks too complicated. I would really try hard to avoid the
> > parameter at all.
> 
> OK. Should we remove max_reason from struct kmsg_dumper and also
> remove the misleading comment about kmsg_dump_reason ordering?

I'm also fine with this. I can have pstore infrastructure doing the
filtering if kmsg dump doesn't want to. Given the existence of
printk.always_kmsg_dump, though, it seemed like it was better to have
kmsg dump do this filtering instead.

At this point my preference is to switch to a bit field -- I don't see a
reason for ordering. The only cases that remain "special" appear to be
PANIC and EMERG (which, again, aren't ordered adjacent).

-Kees

-- 
Kees Cook