GNU bug report logs - #21309
libtool should not use the dangerous, undocumented AC_TRY_EVAL macro, currently broken

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 21309 in the body.
You can then email your comments to 21309 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Vincent Lefevre <vincent <at> vinc17.net>
To: bug-libtool <at> gnu.org
Subject: libtool generates binary data in config.log due to buggy escaping
Date: Thu, 20 Aug 2015 03:56:02 +0200

I've reported the following bug in the Debian BTS:

  https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=796180

but I can reproduce it with the official libtool 2.4.6.

When I run "./configure" with MPFR, the generated config.log file
contains binary data:

configure:9458: /usr/bin/nm -B conftest.o \| sed -n -e 's/^.*[ ]\([ABCDGIRSTW][ABCDGIRSTW]*\)[ ][ ]*\([_A-Za-z][_A-Za-z0-9]*\)$/^A ^B ^B/p' | sed '/ __gnu_lto/d' \> conftest.nm

where ^A and ^B are the ASCII characters of code 1 and 2 respectively.
Thus, the "file" utility thinks that this is a binary file, so that
"less" + "lesspipe" mangles the file. Rather annoying!

The generated "libtool" script contains:

global_symbol_pipe="sed -n -e 's/^.*[    ]\\([ABCDGIRSTW][ABCDGIRSTW]*\\)[       ][      ]*\\([_A-Za-z][_A-Za-z0-9]*\\)\$/\\1 \\2 \\2/p' | sed '/ __gnu_lto/d'"

So, I suppose that the binary data come from the \\1 and \\2.

-- 
Vincent Lefèvre <vincent <at> vinc17.net> - Web: <https://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <https://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)

Message #8 received at 21309 <at> debbugs.gnu.org (full text, mbox):

From: Vincent Lefevre <vincent <at> vinc17.net>
To: 21309 <at> debbugs.gnu.org
Subject: Re: libtool generates binary data in config.log due to buggy escaping
Date: Mon, 8 Feb 2016 17:44:39 +0100

I actually get these binary data with dash (when setting
CONFIG_SHELL=/bin/sh in my environment, where /bin/sh is dash).

On 2015-08-20 03:56:02 +0200, Vincent Lefevre wrote:
> The generated "libtool" script contains:
> 
> global_symbol_pipe="sed -n -e 's/^.*[    ]\\([ABCDGIRSTW][ABCDGIRSTW]*\\)[       ][      ]*\\([_A-Za-z][_A-Za-z0-9]*\\)\$/\\1 \\2 \\2/p' | sed '/ __gnu_lto/d'"
> 
> So, I suppose that the binary data come from the \\1 and \\2.

The problem is that it yields, for instance:

+ echo configure:10030: /usr/bin/nm -B conftest.o \| sed -n -e 's/^.*[ ]\([ABCDGIRSTW][ABCDGIRSTW]*\)[ ][ ]*\([_A-Za-z][_A-Za-z0-9]*\)$/\1 \2 \2/p' | sed '/ __gnu_lto/d' \> conftest.nm

With bash:

$ echo '\1' | hd
00000000  5c 31 0a                                          |\1.|
00000003

With dash:

$ echo '\1' | hd
00000000  01 0a                                             |..|
00000002

POSIX[*] says:

  if any of the operands contain a <backslash> character, the results
  are implementation-defined.

So, the above code is not portable.

[*] http://pubs.opengroup.org/onlinepubs/9699919799/utilities/echo.html

-- 
Vincent Lefèvre <vincent <at> vinc17.net> - Web: <https://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <https://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)

Message #11 received at 21309 <at> debbugs.gnu.org (full text, mbox):

From: Vincent Lefevre <vincent <at> vinc17.net>
To: 21309 <at> debbugs.gnu.org
Subject: Re: libtool generates binary data in config.log due to buggy escaping
Date: Tue, 2 Oct 2018 11:12:52 +0200

On 2016-02-08 17:44:39 +0100, Vincent Lefevre wrote:
> I actually get these binary data with dash (when setting
> CONFIG_SHELL=/bin/sh in my environment, where /bin/sh is dash).
> 
> On 2015-08-20 03:56:02 +0200, Vincent Lefevre wrote:
> > The generated "libtool" script contains:
> > 
> > global_symbol_pipe="sed -n -e 's/^.*[    ]\\([ABCDGIRSTW][ABCDGIRSTW]*\\)[       ][      ]*\\([_A-Za-z][_A-Za-z0-9]*\\)\$/\\1 \\2 \\2/p' | sed '/ __gnu_lto/d'"
> > 
> > So, I suppose that the binary data come from the \\1 and \\2.
> 
> The problem is that it yields, for instance:
> 
> + echo configure:10030: /usr/bin/nm -B conftest.o \| sed -n -e 's/^.*[ ]\([ABCDGIRSTW][ABCDGIRSTW]*\)[ ][ ]*\([_A-Za-z][_A-Za-z0-9]*\)$/\1 \2 \2/p' | sed '/ __gnu_lto/d' \> conftest.nm

More details (with the current MPFR trunk, thus this has changed
a bit):

[...]
+ nlist=conftest.nm
+ eval echo "$as_me":7267: "/usr/bin/nm -B conftest.o \| sed -n -e 's/^.*[ ]\([A
BCDGIRSTW][ABCDGIRSTW]*\)[ ][ ]*\([_A-Za-z][_A-Za-z0-9]*\)$/\1 \2 \2/p' | sed '/ __gnu_lto/d' \> conftest.nm"
+ echo configure:7267: /usr/bin/nm -B conftest.o \| sed -n -e 's/^.*[ ]\([ABCDGIRSTW][ABCDGIRSTW]*\)[ ][ ]*\([_A-Za-z][_A-Za-z0-9]*\)$/\1 \2 \2/p' | sed '/ __gnu_lto/d' \> conftest.nm
[...]

In libtool.m4, this corresponds to:

    nlist=conftest.nm
    if AC_TRY_EVAL(NM conftest.$ac_objext \| "$lt_cv_sys_global_symbol_pipe" \> $nlist) && test -s "$nlist"; then

Thus the problem seems to come from AC_TRY_EVAL, i.e. an autoconf bug.

According to https://debbugs.gnu.org/Packages.html, autoconf doesn't
use this bug tracker, so that I'm going to report a separate bug for
autoconf.

-- 
Vincent Lefèvre <vincent <at> vinc17.net> - Web: <https://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <https://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)

Message #14 received at 21309 <at> debbugs.gnu.org (full text, mbox):

From: Vincent Lefevre <vincent <at> vinc17.net>
To: 21309 <at> debbugs.gnu.org
Subject: Re: libtool generates binary data in config.log due to buggy escaping
Date: Tue, 2 Oct 2018 12:58:04 +0200

retitle 21309 libtool should not use the dangerous, undocumented AC_TRY_EVAL macro, currently broken
thanks

On 2018-10-02 11:12:52 +0200, Vincent Lefevre wrote:
> In libtool.m4, this corresponds to:
> 
>     nlist=conftest.nm
>     if AC_TRY_EVAL(NM conftest.$ac_objext \| "$lt_cv_sys_global_symbol_pipe" \> $nlist) && test -s "$nlist"; then
> 
> Thus the problem seems to come from AC_TRY_EVAL, i.e. an autoconf bug.

I've seen in the autoconf source that AC_TRY_EVAL should not be used:

# AC_TRY_EVAL(VARIABLE)
# ---------------------
# Evaluate $VARIABLE, which should be a valid shell command.
# The purpose of this macro is to write "configure:123: command line"
# into config.log for every test run.
#
# The AC_TRY_EVAL and AC_TRY_COMMAND macros are dangerous and
# undocumented, and should not be used.
# They may be removed or their API changed in a future release.
# Autoconf itself no longer uses these two macros; they are present
# only for backward compatibility with previous versions of Autoconf.
# Not every shell command will work due to problems with eval
# and quoting, and the rules for exactly what does work are tricky.
# Worse, due to double-expansion during evaluation, arbitrary unintended
# shell commands could be executed in some situations.

-- 
Vincent Lefèvre <vincent <at> vinc17.net> - Web: <https://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <https://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)

Message #19 received at 21309 <at> debbugs.gnu.org (full text, mbox):

From: Vincent Lefevre <vincent <at> vinc17.net>
To: 21309 <at> debbugs.gnu.org
Subject: Re: libtool generates binary data in config.log due to buggy escaping
Date: Tue, 2 Oct 2018 15:46:21 +0200

[Message part 1 (text/plain, inline)]

On 2018-10-02 12:58:04 +0200, Vincent Lefevre wrote:
> I've seen in the autoconf source that AC_TRY_EVAL should not be used:
> 
> # AC_TRY_EVAL(VARIABLE)
> # ---------------------
> # Evaluate $VARIABLE, which should be a valid shell command.
> # The purpose of this macro is to write "configure:123: command line"
> # into config.log for every test run.
> #
> # The AC_TRY_EVAL and AC_TRY_COMMAND macros are dangerous and
> # undocumented, and should not be used.
> # They may be removed or their API changed in a future release.
> # Autoconf itself no longer uses these two macros; they are present
> # only for backward compatibility with previous versions of Autoconf.
> # Not every shell command will work due to problems with eval
> # and quoting, and the rules for exactly what does work are tricky.
> # Worse, due to double-expansion during evaluation, arbitrary unintended
> # shell commands could be executed in some situations.

I've attached a patch that replaces the AC_TRY_EVAL occurrence that
generated the binary data. This is a patch against the Debian package
(libtool 2.4.6-4). I've tested it with MPFR and it seems to work fine.

The other AC_TRY_EVAL occurrences do not cause any problem here,
but should also be replaced in the future.

For the reference, my Debian bug report (clone of 796180, which is
now about autoconf):

  https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=910076

-- 
Vincent Lefèvre <vincent <at> vinc17.net> - Web: <https://www.vinc17.net/>
100% accessible validated (X)HTML - Blog: <https://www.vinc17.net/blog/>
Work: CR INRIA - computer arithmetic / AriC project (LIP, ENS-Lyon)

[libtool-eval-nm.patch (text/plain, attachment)]

Message #24 received at 21309-done <at> debbugs.gnu.org (full text, mbox):

From: Ileana Dumitrescu <ileanadumitrescu95 <at> gmail.com>
To: 21309-done <at> debbugs.gnu.org
Cc: Vincent Lefevre <vincent <at> vinc17.net>
Subject: Re: bug#21309: libtool generates binary data in config.log due to
 buggy escaping
Date: Fri, 18 Oct 2024 17:50:35 +0300

[Message part 1 (text/plain, inline)]

Thank you (again :) for the bug report and patch!

The fix you provided has been applied to the development branch:
https://git.savannah.gnu.org/cgit/libtool.git/commit/?h=development&id=f7d11c71dbf23b6a7bc8a273782d3aa3dab08b23

-- 
Ileana Dumitrescu

GPG Public Key: FA26 CA78 4BE1 8892 7F22 B99F 6570 EA01 146F 7354

[OpenPGP_0x6570EA01146F7354.asc (application/pgp-keys, attachment)]

[OpenPGP_signature.asc (application/pgp-signature, attachment)]

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.