DevHeads.net

Segfault on /usr/bin/strip

Hello Centos developers,

After upgrading a Centos7 builder, I started to get a segmentation fault
when using `install -s`.

As it turned out, this is related tu /usr/bin/strip, which indeed goes
segfault.

I wasn't able to find this in the redhat bugzilla. Is it a known problem or
should I report it?

Follows stack trace:

Program received signal SIGSEGV, Segmentation fault.
0x00007ffff7afcc55 in section_match (a=0x0, b=b@entry=0x639f18) at elf.c:1236
1236 && a->sh_entsize == b->sh_entsize
(gdb) bt
#0 0x00007ffff7afcc55 in section_match (a=0x0, b=b@entry=0x639f18) at elf.c:1236
#1 0x00007ffff7afd1ae in find_link (iheader=0x639f18, hint=36, obfd=0x63c030) at elf.c:1252
#2 0x00007ffff7afd2ca in copy_special_section_fields (ibfd=ibfd@entry=0x639c90, obfd=obfd@entry=0x63c030, iheader=0x63c1f0,
oheader=oheader@entry=0x642c70, secnum=secnum@entry=1) at elf.c:1321
#3 0x00007ffff7afefa6 in _bfd_elf_copy_private_bfd_data (ibfd=0x639c90, obfd=0x63c030) at elf.c:1441
#4 0x0000000000409d55 in copy_object (ibfd=<optimized out>, obfd=<optimized out>, input_arch=<optimized out>) at objcopy.c:2495
#5 0x000000000040af6a in copy_file (
input_filename=0x7fffffffe51a "esb3008-http-request-router/lib/python2.7/site-packages/_cffi_backend.so",
output_filename=output_filename@entry=0x639c40 "esb3008-http-request-router/lib/python2.7/site-packages/stnLASGM",
input_target=input_target@entry=0x0, output_target=<optimized out>, output_target@entry=0x0, input_arch=input_arch@entry=0x0)
at objcopy.c:2857
#6 0x0000000000406730 in strip_main (argv=<optimized out>, argc=<optimized out>) at objcopy.c:3719
#7 main (argc=2, argv=0x7fffffffe288) at objcopy.c:4816
(gdb)

Comments

Re: Segfault on /usr/bin/strip

By Trevor Hemsley at 05/14/2018 - 09:01

On 14/05/18 13:43, dacav wrote:
It's not a problem that I have seen reported before. If you have a
Redhat Developer Subscription then you can use the included RHEL license
to download a RHEL iso and test there to see if it an upstream bug or a
CentOS one.

For more information on the free Red Hat Developer subscription please
see
<a href="http://developers.redhat.com/blog/2016/03/31/no-cost-rhel-developer-subscription-now-available/" title="http://developers.redhat.com/blog/2016/03/31/no-cost-rhel-developer-subscription-now-available/">http://developers.redhat.com/blog/2016/03/31/no-cost-rhel-developer-subs...</a>

Trevor

Re: Segfault on /usr/bin/strip

By Giovanni at 05/14/2018 - 09:34

On 2018-05-14, Trevor Hemsley
<trevor. ... at ntlworld dot com> wrote:
Without a subscription, I can tell it's probably upstream: I
compiled from scratch the same version (2.27) and it crashes in
the same way.

I'm trying to see if upstream has some bug report open for this.

I will send updates to this thread if I find out :)

Cheers

Re: Segfault on /usr/bin/strip

By Giovanni at 05/14/2018 - 11:25

On 2018-05-14, dacav < ... at teknik dot io> wrote:
So here I am,

I figured out that <a href="https://sourceware.org/bugzilla/show_bug.cgi?id=20922" title="https://sourceware.org/bugzilla/show_bug.cgi?id=20922">https://sourceware.org/bugzilla/show_bug.cgi?id=20922</a>
is probably the upstream bug we are talking about.

The bug was fixed in upstream, commit
a55c9876bb111fd301b4762cf501de0040b8f9db, and indeed I was able
to compile that version under Centos7, and to get `strip` to work
fine.

So version 2.28 should do fine, I guess:

$ git tag --contains a55c9876bb111fd301b4762cf501de0040b8f9db | grep binutils
binutils-2_28
binutils-2_29
binutils-2_29_1
binutils-2_29_1.1
binutils-2_30
users/ARM/embedded-binutils-2_28-branch-2017q1
users/ARM/embedded-binutils-2_28-branch-2017q2
users/ARM/embedded-binutils-master-2017q4

How do I continue from here? Can I be helpful in some way?

Re: Segfault on /usr/bin/strip

By Anssi Johansson at 05/14/2018 - 11:38

dacav kirjoitti 14.5.2018 klo 18.25:
Hmm.. They say it has been assigned a CVE, CVE-2017-7303.

<a href="https://access.redhat.com/security/cve/cve-2017-7303" title="https://access.redhat.com/security/cve/cve-2017-7303">https://access.redhat.com/security/cve/cve-2017-7303</a> and
<a href="https://bugzilla.redhat.com/show_bug.cgi?id=CVE-2017-7303" title="https://bugzilla.redhat.com/show_bug.cgi?id=CVE-2017-7303">https://bugzilla.redhat.com/show_bug.cgi?id=CVE-2017-7303</a> say the
resolution is "wontfix", which is unfortunate.