From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <sourceware-bugzilla@sourceware.org>
Received: by sourceware.org (Postfix, from userid 48)
 id 638A33850414; Thu,  3 Dec 2020 12:51:00 +0000 (GMT)
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 638A33850414
From: "jakub at redhat dot com" <sourceware-bugzilla@sourceware.org>
To: glibc-bugs@sourceware.org
Subject: [Bug dynamic-link/27004] ld.so is miscompiled by GCC 11
Date: Thu, 03 Dec 2020 12:51:00 +0000
X-Bugzilla-Reason: CC
X-Bugzilla-Type: changed
X-Bugzilla-Watch-Reason: None
X-Bugzilla-Product: glibc
X-Bugzilla-Component: dynamic-link
X-Bugzilla-Version: 2.33
X-Bugzilla-Keywords: 
X-Bugzilla-Severity: normal
X-Bugzilla-Who: jakub at redhat dot com
X-Bugzilla-Status: NEW
X-Bugzilla-Resolution: 
X-Bugzilla-Priority: P2
X-Bugzilla-Assigned-To: unassigned at sourceware dot org
X-Bugzilla-Target-Milestone: ---
X-Bugzilla-Flags: security-
X-Bugzilla-Changed-Fields: 
Message-ID: <bug-27004-131-8uca3ZOJ3f@http.sourceware.org/bugzilla/>
In-Reply-To: <bug-27004-131@http.sourceware.org/bugzilla/>
References: <bug-27004-131@http.sourceware.org/bugzilla/>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
X-Bugzilla-URL: http://sourceware.org/bugzilla/
Auto-Submitted: auto-generated
MIME-Version: 1.0
X-BeenThere: glibc-bugs@sourceware.org
X-Mailman-Version: 2.1.29
Precedence: list
List-Id: Glibc-bugs mailing list <glibc-bugs.sourceware.org>
List-Unsubscribe: <https://sourceware.org/mailman/options/glibc-bugs>,
 <mailto:glibc-bugs-request@sourceware.org?subject=unsubscribe>
List-Archive: <https://sourceware.org/pipermail/glibc-bugs/>
List-Help: <mailto:glibc-bugs-request@sourceware.org?subject=help>
List-Subscribe: <https://sourceware.org/mailman/listinfo/glibc-bugs>,
 <mailto:glibc-bugs-request@sourceware.org?subject=subscribe>
X-List-Received-Date: Thu, 03 Dec 2020 12:51:00 -0000

https://sourceware.org/bugzilla/show_bug.cgi?id=3D27004

--- Comment #5 from Jakub Jelinek <jakub at redhat dot com> ---
(In reply to Florian Weimer from comment #4)
> __builtin_thread_pointer is potentially trapping (which is the root of the
> problem). Does GCC know about that? It also results in a load followed by=
 an
> add, instead of a single load, I think. Using the __segfs or __seggs
> namespaces looks preferable to me.

GCC makes it const __attribute__((nothrow)), and that is I think ok except =
in
the dynamic linker.

I don't think
+     __self =3D (struct pthread *) (__builtin_thread_pointer ()           =
     \
+                                 + offsetof (struct pthread, header.self))=
;  \
does what the old code did, which was movq %fs:16, __self
but your version is movq %fs:0, %reg; leaq 16(%reg), __self
That would be *(struct pthread **) (__builtin_thread_pointer () + offsetof
(struct pthread, header.self))
if we optimize that back to just movq %fs:16, __self
But, as it is for x86 GCC 11+ only, isn't it better to use __seg_{f,g}s that
should work already since GCC 6?

--=20
You are receiving this mail because:
You are on the CC list for the bug.=