linux-kernel - Re: [PATCH] signal: Restore the stop PTRACE_EVENT

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:   Wed, 13 Feb 2019 15:38:52 +0100
From:   Oleg Nesterov <oleg@...hat.com>
To:     "Eric W. Biederman" <ebiederm@...ssion.com>
Cc:     Dmitry Vyukov <dvyukov@...gle.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        LKML <linux-kernel@...r.kernel.org>,
        Arnaldo Carvalho de Melo <acme@...nel.org>,
        Alexander Shishkin <alexander.shishkin@...ux.intel.com>,
        jolsa@...hat.com, Namhyung Kim <namhyung@...nel.org>,
        luca abeni <luca.abeni@...tannapisa.it>,
        syzkaller <syzkaller@...glegroups.com>,
        Ivan Delalande <colona@...sta.com>
Subject: Re: [PATCH] signal: Restore the stop PTRACE_EVENT_EXIT

sorry for noise, but after I read the changelog I have a minor nit,
feel free to ignore...

On 02/12, Eric W. Biederman wrote:
>
> Skipping past dequeue_signal when we know a fatal signal has already
> been delivered resulted in SIGKILL remaining pending and
> TIF_SIGPENDING remaining set.  This in turn caused the
> scheduler to not sleep in PTACE_EVENT_EXIT as it figured
> a fatal signal was pending.

Yes, but the status of TIF_SIGPENDING doesn't matter. However I agree
with recalc_sigpending() added by this patch, simply because this is what
the "normal" dequeue_signal() paths do.

> This also caused ptrace_freeze_traced
> in ptrace_check_attach to fail because it left a per thread
> SIGKILL pending which is what fatal_signal_pending tests for.

this is possible too, but in the likely case ptrace_check_attach() won't
be even called exactly because the tracee won't stop and thus waitpid()
won't report WIFSTOPPED. And even if waitpid() "wins" the race and debugger
calls ptrace(), most probably ptrace_freeze_traced() will fail because
task_is_traced() will be false.

I think this part of the changelog looks a bit confusing. It doesn't matter
why ptrace_check_attach() fails, it must fail if the tracee didn't stop.

PTACE_EVENT_EXIT won't stop and thus this event won't be reported, that is all.

again, feel free to ignore.

Oleg.