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] [day] [month] [year] [list]
Message-ID: <CANpmjNMtK53SiZwm0N9VuwGJthY0unZ_1_mZ=gXdMH0_LAFr5A@mail.gmail.com>
Date:   Fri, 2 Jul 2021 09:20:56 +0200
From:   Marco Elver <elver@...gle.com>
To:     "Eric W. Biederman" <ebiederm@...ssion.com>
Cc:     peterz@...radead.org, tglx@...utronix.de, mingo@...nel.org,
        kasan-dev@...glegroups.com, linux-kernel@...r.kernel.org,
        mingo@...hat.com, acme@...nel.org, mark.rutland@....com,
        alexander.shishkin@...ux.intel.com, jolsa@...hat.com,
        namhyung@...nel.org, linux-perf-users@...r.kernel.org,
        omosnace@...hat.com, serge@...lyn.com,
        linux-security-module@...r.kernel.org, stable@...r.kernel.org,
        Dmitry Vyukov <dvyukov@...gle.com>
Subject: Re: [PATCH v2] perf: Require CAP_KILL if sigtrap is requested

On Thu, 1 Jul 2021 at 23:41, Eric W. Biederman <ebiederm@...ssion.com> wrote:
>
> Marco Elver <elver@...gle.com> writes:
>
> > If perf_event_open() is called with another task as target and
> > perf_event_attr::sigtrap is set, and the target task's user does not
> > match the calling user, also require the CAP_KILL capability.
> >
> > Otherwise, with the CAP_PERFMON capability alone it would be possible
> > for a user to send SIGTRAP signals via perf events to another user's
> > tasks. This could potentially result in those tasks being terminated if
> > they cannot handle SIGTRAP signals.
> >
> > Note: The check complements the existing capability check, but is not
> > supposed to supersede the ptrace_may_access() check. At a high level we
> > now have:
> >
> >       capable of CAP_PERFMON and (CAP_KILL if sigtrap)
> >               OR
> >       ptrace_may_access() // also checks for same thread-group and uid
>
> Is there anyway we could have a comment that makes the required
> capability checks clear?
>
> Basically I see an inlined version of kill_ok_by_cred being implemented
> without the comments on why the various pieces make sense.

I'll add more comments. It probably also makes sense to factor the
code here into its own helper.

> Certainly ptrace_may_access(task, PTRACE_MODE_READ_REALCREDS) should not
> be a check to allow writing/changing a task.  It needs to be
> PTRACE_MODE_ATTACH_REALCREDS, like /proc/self/mem uses.

So if attr.sigtrap the checked ptrace mode needs to switch to
PTRACE_MODE_ATTACH_REALCREDS. Otherwise, it is possible to send a
signal if only read-ptrace permissions are granted.

Is my assumption here correct?

> Now in practice I think your patch probably has the proper checks in
> place for sending a signal but it is far from clear.

Thanks,
-- Marco

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ