public inbox for libc-stable@sourceware.org
 help / color / mirror / Atom feed
From: Florian Weimer <fweimer@redhat.com>
To: Carlos O'Donell <carlos@redhat.com>
Cc: Aurelien Jarno <aurelien@aurel32.net>,  libc-stable@sourceware.org
Subject: Re: Backporting CVE-2016-10739
Date: Tue, 01 Jan 2019 00:00:00 -0000	[thread overview]
Message-ID: <877eefmk3z.fsf@oldenburg2.str.redhat.com> (raw)
In-Reply-To: <aa264078-6910-3415-5955-03527e8ff7b9@redhat.com> (Carlos	O'Donell's message of "Mon, 4 Feb 2019 12:05:06 -0500")

* Carlos O'Donell:

> On 2/4/19 11:18 AM, Florian Weimer wrote:
>> * Carlos O'Donell:
>> 
>>> On 2/4/19 10:44 AM, Florian Weimer wrote:
>>>> * Carlos O'Donell:
>>>>
>>>>>> +#include <arpa/inet.h>
>>>>>> +
>>>>>
>>>>> Please add a comment explaining why this is here.
>>>>
>>>> You mean like this?
>>>>
>>>> /* Obtain the prototype for __inet_aton_exact.  */
>>>
>>> It should reference the bug or CVE to document the intent
>>> of the changes.
>>>
>>> Post v3 and I'll sign off?
>> 
>> This approach does not actually work because copying a prototype this
>> way and adding a hidden visibility attribute does not actually make the
>> symbol hidden.  The patch below however has the desired effect, mainly
>> because interposition no longer happens and the __inet_aton_exact_hidden
>> function is not added to the dynamic symbol table of the nscd
>> executable.
>> 
>> I suspect I would have had to use __attribute__ ((visibility
>> ("hidden"))) directly because we define attribute_hidden thusly:
>> 
>> #if defined SHARED || defined LIBC_NONSHARED \
>>   || (BUILD_PIE_DEFAULT && IS_IN (libc))
>> # define attribute_hidden __attribute__ ((visibility ("hidden")))
>> #else
>> # define attribute_hidden
>> #endif
>> 
>> I can post yet another patch which uses real hidden visibility and
>> avoids the symbol redirect.
>
> This version is a little hack-ish, but I don't object to it. It does
> the job.
>
> It is certainly a minimal set of changes, and that makes it quite
> useful for the release branch.
>
> I don't think you need to go any further unless you think the version
> using real hidden visibility is all that much better?

Well, at least it does not have the completely ineffective
attribute_hidden. 8-)

Patch below.  What do you think?

This one gets the symbol visibility correct.

$ eu-readelf -s ../build/nscd/nscd-inet_addr.o | grep inet_aton_exact
   22: 0000000000000150     56 FUNC    GLOBAL HIDDEN         1 __inet_aton_exact
$ eu-readelf -s ../build/nscd/gai.o | grep inet_aton_exact
   99: 0000000000000000      0 NOTYPE  GLOBAL HIDDEN     UNDEF __inet_aton_exact

And there's no __inet_aton_exact symbol in nscd.  I think this is fairly
standard usage of a hidden symbol, so renaming the symbol should not be
necessary.

Thanks,
Florian

nscd: Do not use __inet_aton_exact@GLIBC_PRIVATE [BZ #20018]

This commit avoids referencing the __inet_aton_exact@GLIBC_PRIVATE
symbol from nscd.  In master, the separately-compiled getaddrinfo
implementation in nscd needs it, however such an internal ABI change
is not desirable on a release branch if it can be avoided.

2019-02-04  Florian Weimer  <fweimer@redhat.com>

	[BZ #20018]
	nscd: Do not rely on new GLIBC_PRIVATE ABI after CVE-2016-10739 fix.
	* nscd/nscd-inet_addr.c: New file.  Build resolv/inet_addr.c for
	nscd, without public symbols.
	* nscd/Makefile (nscd-modules): Add it.
	* nscd/gai.c: Include <arpa/inet.h> and change visibility of
	__inet_aton_exact.

diff --git a/nscd/Makefile b/nscd/Makefile
index b713a84c49..eb23c01a39 100644
--- a/nscd/Makefile
+++ b/nscd/Makefile
@@ -36,7 +36,7 @@ nscd-modules := nscd connections pwdcache getpwnam_r getpwuid_r grpcache \
 		getsrvbynm_r getsrvbypt_r servicescache \
 		dbg_log nscd_conf nscd_stat cache mem nscd_setup_thread \
 		xmalloc xstrdup aicache initgrcache gai res_hconf \
-		netgroupcache
+		netgroupcache nscd-inet_addr
 
 ifeq ($(build-nscd)$(have-thread-library),yesyes)
 
diff --git a/nscd/gai.c b/nscd/gai.c
index f57f396f57..68a4abd30e 100644
--- a/nscd/gai.c
+++ b/nscd/gai.c
@@ -33,6 +33,12 @@
 #define __getifaddrs getifaddrs
 #define __freeifaddrs freeifaddrs
 
+/* We do not want to export __inet_aton_exact.  Get the prototype and
+   change its visibility to hidden.  */
+#include <arpa/inet.h>
+__typeof__ (__inet_aton_exact) __inet_aton_exact
+  __attribute__ ((visibility ("hidden")));
+
 /* We are nscd, so we don't want to be talking to ourselves.  */
 #undef  USE_NSCD
 
diff --git a/nscd/nscd-inet_addr.c b/nscd/nscd-inet_addr.c
new file mode 100644
index 0000000000..f366b9567d
--- /dev/null
+++ b/nscd/nscd-inet_addr.c
@@ -0,0 +1,32 @@
+/* Legacy IPv4 text-to-address functions.  Version for nscd.
+   Copyright (C) 2019 Free Software Foundation, Inc.
+   This file is part of the GNU C Library.
+
+   The GNU C Library is free software; you can redistribute it and/or
+   modify it under the terms of the GNU Lesser General Public
+   License as published by the Free Software Foundation; either
+   version 2.1 of the License, or (at your option) any later version.
+
+   The GNU C Library is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+   Lesser General Public License for more details.
+
+   You should have received a copy of the GNU Lesser General Public
+   License along with the GNU C Library; if not, see
+   <http://www.gnu.org/licenses/>.  */
+
+#include <arpa/inet.h>
+
+/* We do not want to export __inet_aton_exact.  Get the prototype and
+   change the visibility to hidden.  */
+#include <arpa/inet.h>
+__typeof__ (__inet_aton_exact) __inet_aton_exact
+  __attribute__ ((visibility ("hidden")));
+
+/* Do not provide definitions of the public symbols exported from
+   libc.  */
+#undef weak_alias
+#define weak_alias(from, to)
+
+#include <resolv/inet_addr.c>

  reply	other threads:[~2019-02-04 19:25 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-01-01  0:00 Aurelien Jarno
2019-01-01  0:00 ` Florian Weimer
2019-01-01  0:00   ` Florian Weimer
2019-01-01  0:00     ` Carlos O'Donell
2019-01-01  0:00       ` Florian Weimer
2019-01-01  0:00         ` Carlos O'Donell
2019-01-01  0:00           ` Florian Weimer
2019-01-01  0:00             ` Carlos O'Donell
2019-01-01  0:00               ` Florian Weimer [this message]
2019-01-01  0:00                 ` Carlos O'Donell
2019-01-01  0:00                   ` Florian Weimer
2019-01-01  0:00                     ` Aurelien Jarno
2019-01-01  0:00                     ` Carlos O'Donell
2019-01-01  0:00   ` Carlos O'Donell
2019-01-01  0:00 ` Florian Weimer
2019-01-01  0:00   ` Aurelien Jarno
2019-01-01  0:00     ` Florian Weimer
2019-01-01  0:00       ` Aurelien Jarno

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=877eefmk3z.fsf@oldenburg2.str.redhat.com \
    --to=fweimer@redhat.com \
    --cc=aurelien@aurel32.net \
    --cc=carlos@redhat.com \
    --cc=libc-stable@sourceware.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).