linux-hardening - Re: [REGRESSION] 5.16rc1: SA

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: <CAHk-=wjsyDcohs=i0XGu0GRb6AkXUNyWCE_f6JMy0RY9wdXUXg@mail.gmail.com>
Date:   Wed, 17 Nov 2021 16:15:21 -0800
From:   Linus Torvalds <torvalds@...ux-foundation.org>
To:     Kees Cook <keescook@...omium.org>
Cc:     "Eric W. Biederman" <ebiederm@...ssion.com>,
        Kyle Huey <me@...ehuey.com>,
        Andrea Righi <andrea.righi@...onical.com>,
        Shuah Khan <shuah@...nel.org>,
        Alexei Starovoitov <ast@...nel.org>,
        Andy Lutomirski <luto@...capital.net>,
        Will Drewry <wad@...omium.org>,
        "open list:KERNEL SELFTEST FRAMEWORK" 
        <linux-kselftest@...r.kernel.org>, bpf <bpf@...r.kernel.org>,
        open list <linux-kernel@...r.kernel.org>,
        linux-hardening@...r.kernel.org,
        "Robert O'Callahan" <rocallahan@...il.com>
Subject: Re: [REGRESSION] 5.16rc1: SA_IMMUTABLE breaks debuggers

On Wed, Nov 17, 2021 at 4:05 PM Kees Cook <keescook@...omium.org> wrote:
>
> (nit: should the "sigdfl" argument be renamed "immutable" for clarity
> in this function?)

I don't think that would necessarily clarify anything. Neither
"sigdfl" nor "immutable" makes at least me go "Ahh, that explains
things".

Both "sigdfl" and "immutable" are about random internal implementation
choices ("force SIGDFL" and "set SA_IMMUTABLE" respectively).

I think naming things by random internal implementation things is
questionable in general, but it's particularly questionable when they
aren't even some really fundamental thing.

I think you generally want to name things not by how they do
something, but by *WHAT* they do.

So I think the proper name for it would be "fatal" or something like
that. It's basically saying "This signal is fatal, even if you have a
handler for it or not". That "set it to SIGDFL" just happens to be how
we made it fatal.

And then we should perhaps also make such a signal uncatchable by the
debugger (rather than just "debugger cannot undo or modify it" like
the SA_IMMUTABLE bit does).

Anybody want to take on that renaming / uncatchable part? Please take
my (now at least tested by Kees) patch and make it your own.

              Linus