Re: bug#13588: Pax hangs in case big UID

[Stefano Lattarini]

tags 13588 + patch
stop

[+cc automake-patches]

Hi Jack, Petr, sorry for the ludicrous delay.

On 03/22/2013 02:39 AM, Jack Kelly wrote:
> Petr Hracek <address@hidden> writes:
> 
>> Hello Jack and Stefano,
>>
>> Bellow is corrected patch for automake.
>> Jack thank you for corrections. Now the patch looks like better.
> 
> Yes, it looks a lot better. I have more thoughts, if that's ok:
> 
> +      # Maximum allowed UID in case ustar format is 2097151
> 
> Put a link to
> http://lists.gnu.org/archive/html/bug-automake/2011-11/msg00014.html so
> it's clear where the number came from.
> 
> I'd do it like this (I've also removed the variable assignments for
> $user_id and $group_id):
> 
> m4_if($1, [ustar],
> [AC_CHECK_PROG([am_prog_have_id], [id], [yes], [no])
> if test x"$am_prog_have_id" = x"yes"; then
>   # POSIX 1988 "ustar" format is defined with *fixed size* fields. There
>   # is notably a 21 bits limit (2097151) for the uid and the gid.
>   # http://lists.gnu.org/archive/html/bug-automake/2011-11/msg00014.html so
>   am_ustar_max_id=2097151
>   ...
>   if test $? -eq 0 -a `id -u` -gt $am_ustar_max_id; then
>     ...
> ])
> 
> What should the user be told? It's not really their fault that the
> developer specified ustar-format archives, so perhaps the error message
> should ask the user to run configure with a lower UID or ask the user to
> ask upstream to stop using the ustar option.
>
I still believe that aborting the configure script just to signal the
user that he won't be able to run "make dist" (something only developers
do, usually) is unacceptable.  And in fact, the existing code in tar.m4,
upon failing to find a working tar/pax/whatever propgram good enough to
create a tarball, didn't cause configure to abort, but merely set some
variables to ensure the error would became apparent at "make dist" time.
I think the new code should keep this behaviour.

So, below is my first attempt of a patch.  It sill passes the Automake
testsuite, but before merging it I need to know it works correctly in
real-world problematic situation it is meant to address.  Petr, could
you give the patch a try and report back the results?

Thanks,
  Stefano

---- >8 ---- >8 ---- >8 ---- >8 ---- >8 ---- >8 ---- >8 ---- >8 ---- >8 ----

>From 12e62106381ead9467ef7fbe14940e32aa849a61 Mon Sep 17 00:00:00 2001
Message-Id: <address@hidden>
From: Stefano Lattarini <address@hidden>
Date: Sun, 17 Feb 2013 16:42:46 +0100
Subject: [PATCH] tar: pax could hang configure when big UID are involved

See automake bug#8343 and bug#13588.

Tom Rini address@hidden says (in bug#8343):

    When the user has a UID or GID that is larger than the ustar format
    supports, pax does not error out gracefully in some cases (FC13).

Marc Herbert <address@hidden> adds (in bug#8343):

    When "configure" is run by a user with an UID bigger than 21 bits,
    BSD pax 3.4 aborts when trying to create the 'conftest.tar' test
    archive and leaves an empty or corrupted conftest.tar file behind.
    In the next step, pax tries to extract this incomplete or corrupted
    archive and this *** hangs the whole ./configure script ***.

    Note: GNU cpio 2.9 pretends to pass the test but it is a LIE: it
    silently truncates any big UID to its lower 21 bits. I don't know
    what can be the consequences of this lie.

Months later, Petr Hracek <address@hidden> reports a similar issue
(in bug#13588) for Fedora 17:

    I am trying to solve problem in case a user is created with big
    UID and during configuration pax hangs with message

        ATTENTION! pax archive volume change required.
        Ready for archive volume: 1
        Input archive name or "." to quit pax.
        Archive name >

    and needs user interaction.

    Reference: <https://bugzilla.redhat.com/show_bug.cgi?id=843376>

Time to fix this issue, on the line of a preliminary patch provided by
Petr Hracek in bug#13588.

* m4/tar.m4 (_AM_PROG_TAR): Don't use 'pax' if the UID or GID of the
current user is too high (> 2097151).
* NEWS: Update.
* THANKS: Likewise.

Signed-off-by: Stefano Lattarini <address@hidden>
---
 NEWS      |  6 ++++++
 THANKS    |  3 +++
 m4/tar.m4 | 30 +++++++++++++++++++++++++++++-
 3 files changed, 38 insertions(+), 1 deletion(-)

diff --git a/NEWS b/NEWS
index f9a1fb1..863ffdf 100644
--- a/NEWS
+++ b/NEWS
@@ -78,6 +78,12 @@ New in 1.13.2:

 * Bugs fixed:

+  - When the 'ustar' option is used, the generated configure script no
+    longer risks hanging during the tests for the availability of the
+    'pax' utility, even if the user running configure has a UID or GID
+    that requires more than 21 bits to be represented.
+    See automake bug#8343 and bug#13588.
+
   - The obsolete macros AM_CONFIG_HEADER or AM_PROG_CC_STDC work once
     again, as they did in Automake 1.12.x (albeit printing runtime
     warnings in the 'obsolete' category).  Removing them has turned
diff --git a/THANKS b/THANKS
index 66498d4..a574909 100644
--- a/THANKS
+++ b/THANKS
@@ -224,6 +224,7 @@ Luo Yi                          address@hidden
 Maciej Stachowiak               address@hidden
 Maciej W. Rozycki               address@hidden
 Manu Rouat                      address@hidden
+Marc Herbert                    address@hidden
 Marcus Brinkmann                address@hidden
 Marcus G. Daniels               address@hidden
 Marius Vollmer                  address@hidden
@@ -311,6 +312,7 @@ Peter Muir                      address@hidden
 Peter O'Gorman                  address@hidden
 Peter Rosin                     address@hidden
 Peter Seiderer                  address@hidden
+Petr Hracek                     address@hidden
 Petter Reinholdtsen             address@hidden
 Petteri Räty                    address@hidden
 Phil Edwards                    address@hidden
@@ -391,6 +393,7 @@ Tim Rice                        address@hidden
 Tim Van Holder                  address@hidden
 Toshio Kuratomi                 address@hidden
 Tom Epperly                     address@hidden
+Tom Rini                        address@hidden
 Ulrich Drepper                  address@hidden
 Ulrich Eckhardt                 address@hidden
 Václav Haisman                  address@hidden
diff --git a/m4/tar.m4 b/m4/tar.m4
index ec8c83e..8e85d03 100644
--- a/m4/tar.m4
+++ b/m4/tar.m4
@@ -81,7 +81,35 @@ do
   AM_RUN_LOG([tardir=conftest.dir && eval $am__tar_ >conftest.tar])
   rm -rf conftest.dir
   if test -s conftest.tar; then
-    AM_RUN_LOG([$am__untar <conftest.tar])
+    m4_if([$1], [ustar], [
+      if test "$_am_tool" = pax; then
+        # Automake bugs #8343 and #13588: 'pax' can hang when UIDs
+        # or GIDs are involved that requires more than 21 bits to
+        # be stored.
+        am_max_uid=2097151 # 2^21 - 1
+        am_max_gid=$am_max_uid
+        # The $UID and $GID variables are not portable, so we need to
+        # resort to the POSIX-mandated id(1) utility.  Errors here are
+        # definitely unexpected, so allow the users to see them (i.e.,
+        # no stderr redirection).
+        am_uid=`id -u || echo unknown`
+        am_gid=`id -g || echo unknown`
+        AC_MSG_CHECKING([whether UID '$am_uid' is small enough for ustar])
+        if test $am_uid -le $am_max_uid; then
+          AC_MSG_RESULT([yes])
+        else
+          AC_MSG_RESULT([no])
+          continue
+        fi
+        AC_MSG_CHECKING([whether GID '$am_gid' is small enough for ustar])
+        if test $am_gid -le $am_max_gid; then
+          AC_MSG_RESULT([yes])
+        else
+          AC_MSG_RESULT([no])
+          continue
+        fi
+      fi
+    ]) # $1 == ustar
     grep GrepMe conftest.dir/file >/dev/null 2>&1 && break
   fi
 done
-- 
1.8.2.1.389.gcaa7d79