linux-kernel - About ftrace+pstore+ramoops

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

Message-ID: <72df55ca-20f3-c3ee-861b-c50c1a55495e@canonical.com>
Date:   Thu, 26 Nov 2020 18:33:09 -0300
From:   "Guilherme G. Piccoli" <gpiccoli@...onical.com>
To:     anton@...msg.org, joel@...lfernandes.org
Cc:     Kees Cook <keescook@...omium.org>, ccross@...roid.com,
        tony.luck@...el.com, Steven Rostedt <rostedt@...dmis.org>,
        Ingo Molnar <mingo@...hat.com>,
        linux-kernel <linux-kernel@...r.kernel.org>,
        u.kleine-koenig@...gutronix.de
Subject: About ftrace+pstore+ramoops

Hi Anton / Joel (and all CCed), sorry for annoyance but I'm not being
able to make pstore ftrace capture to work - I'm not sure if I might be
missing something, or if there's a known limitation.

So, first, my use case: I'd like to be able to collect *ftrace* trace
buffer in a specific point in time, when some issue I'm facing happens;
for that, I enable ftrace through tracing_on() in the correct point in
the code. After this issue, machine is not responsive, so I need to
reboot (hence I'd like to persist the trace buffer in my ramoops
reserved area).

For the basic test purpose though, I'm doing just what is suggested in
the ramoops documentation, which is to write 1 on
<debugfs>/pstore/record_ftrace and reboot - but no ftrace file is
present in the pstore folder after the reboot. I've tested the same
system/config by inducing a kernel oops and the corresponding
dmesg-ramoops file is there after the reboot...

Just for clarification: this ftrace feature in pstore is meant to do
what I want right? It should *always* collect the same information as
seen in the trace buffer, but on the persistent memory region configured
by ramoops, correct?
Any guidance you can provide me is much appreciated.
Thanks in advance,

Guilherme