From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <archer-return-2362-listarch-archer=sourceware.org@sourceware.org>
Received: (qmail 29018 invoked by alias); 19 Feb 2011 19:48:12 -0000
Mailing-List: contact archer-help@sourceware.org; run by ezmlm
Sender: <archer@sourceware.org>
Precedence: bulk
List-Post: <mailto:archer@sourceware.org>
List-Help: <mailto:archer-help@sourceware.org>
List-Subscribe: <mailto:archer-subscribe@sourceware.org>
List-Id: <archer.sourceware.org>
Received: (qmail 29009 invoked by uid 22791); 19 Feb 2011 19:48:11 -0000
X-SWARE-Spam-Status: No, hits=-6.3 required=5.0
	tests=AWL,BAYES_00,RCVD_IN_DNSWL_HI,SPF_HELO_PASS,TW_TG,T_RP_MATCHES_RCVD
X-Spam-Check-By: sourceware.org
Date: Sat, 19 Feb 2011 19:48:00 -0000
From: Jan Kratochvil <jan.kratochvil@redhat.com>
To: Oleg Nesterov <oleg@redhat.com>
Cc: Roland McGrath <roland@redhat.com>, Project Archer <archer@sourceware.org>
Subject: Re: ptrace improvement: PTRACE_O_INHERIT
Message-ID: <20110219194758.GA8994@host1.dyn.jankratochvil.net>
References: <20110212190253.GA31866@redhat.com>
 <20110214193052.3EC8D1814BA@magilla.sf.frob.com>
 <20110214193812.GA20765@redhat.com>
 <20110215003551.BC1EA1802A2@magilla.sf.frob.com>
 <20110215130805.GA30742@redhat.com>
 <20110215214333.GA18086@host1.dyn.jankratochvil.net>
 <20110215215634.2665A1806E0@magilla.sf.frob.com>
 <20110216193356.GB15576@redhat.com>
 <20110216194534.2D39018020E@magilla.sf.frob.com>
 <20110216200110.GA17601@redhat.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <20110216200110.GA17601@redhat.com>
User-Agent: Mutt/1.5.21 (2010-09-15)
X-SW-Source: 2011-q1/txt/msg00080.txt.bz2

On Wed, 16 Feb 2011 21:01:10 +0100, Oleg Nesterov wrote:
> On 02/16, Roland McGrath wrote:
> > If the process-wide PID (tgid) is not reused
> > ...
> > then tgkill does protect completely against TID reuse.
> 
> How? sooner or later the process which creates/destroys a thread
> in a loop will reuse some pid number.

(a) GDB tracks separately  threads (libpthread-managed) and LWPs (TIDs).
    (So the libthread_db TD_DEATH notification is offtopic for this mail.)

(b) For LWPs it does not use PTRACE_O_TRACEEXIT.
    /* Do not enable PTRACE_O_TRACEEXIT until GDB is more prepared to support
       read-only process state.  */
    (I do not know more regarding this comment.)

Therefore it occasionally uses tkill (TID, 0) instead of PTRACE_O_TRACEEXIT.

It uses PTRACE_O_TRACECLONE so it should see a reusal of thread TID in the same
PID, although it seems to me it will just create a duplicate {PID, TID}.

Just currently there are enough bugreports about users hitting threading races
of GDB in real world so we do not have to think about more race possibilities.


Thanks,
Jan