Follow @Openwall on Twitter for new release announcements and other news
[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <586ee584-df0d-3c82-a254-16ac1573255a@linux.com>
Date: Mon, 30 Oct 2017 21:06:30 +0300
From: Alexander Popov <alex.popov@...ux.com>
To: Peter Zijlstra <peterz@...radead.org>
Cc: kernel-hardening@...ts.openwall.com, keescook@...omium.org,
 pageexec@...email.hu, spender@...ecurity.net, Ingo Molnar
 <mingo@...nel.org>, Andy Lutomirski <luto@...nel.org>, tycho@...ker.com,
 Laura Abbott <labbott@...hat.com>, Mark Rutland <mark.rutland@....com>,
 Ard Biesheuvel <ard.biesheuvel@...aro.org>, Borislav Petkov <bp@...en8.de>,
 Thomas Gleixner <tglx@...utronix.de>, "H . Peter Anvin" <hpa@...or.com>,
 x86@...nel.org
Subject: Re: [PATCH RFC v5 2/5] gcc-plugins: Add STACKLEAK plugin for tracking
 the kernel stack

Hello Peter,

Thanks for your reply.

On 30.10.2017 20:32, Peter Zijlstra wrote:
> On Mon, Oct 30, 2017 at 07:51:33PM +0300, Alexander Popov wrote:
>> When the thread stack is exhausted, this BUG() is hit. But do_error_trap(),
>> which handles the exception, calls track_stack() itself again (since it is
>> instrumented by the gcc plugin). So this recursion proceeds with exhausting the
>> thread stack.
> 
> Add a __attribute__((nostacktrack)) on it?

Yes, I already tried some blacklisting in the plugin, but it didn't really help,
because:

1. there are other (more than 5) instrumented functions, that are called during
BUG() handling too;

2. decreasing CONFIG_STACKLEAK_TRACK_MIN_SIZE would add more instrumented
functions, which should be manually blacklisted (not good).

I guess handling BUG() in another stack would be a solution. For example, Andy
Lutomirski calls handle_stack_overflow in the DOUBLEFAULT_STACK
(arch/x86/mm/fault.c). Should I do something similar?

Thanks!

Best regards,
Alexander

Powered by blists - more mailing lists

Confused about mailing lists and their use? Read about mailing lists on Wikipedia and check out these guidelines on proper formatting of your messages.