coredump_filter omits build IDs

[Bruce Dawson]

According to "man core 5" the default value for /proc/PID/coredump_filter is supposed to be 0x33, but on Ubuntu (12.04) it is actually 0x23. This is value the value for /proc/1/coredump_filter and therefore for all other processes on the system.

The consequence of this is that "Dump ELF headers" is zero instead of one which means that build IDs do not get recorded in core files. Since build IDs are the recommended way of identifying which shared objects were present in a core file this is a problem.

We use build IDs extensively and we would like to see them in core files. The additional file size is trivial, gdb should be compatible with these core files, it is the default on Fedora, and it is supposed to be the default everywhere.

[Ubuntu Foundations Team Bug Bot (crichton)]

Thank you for taking the time to report this bug and helping to make Ubuntu better. It seems that your bug report is not filed about a specific source package though, rather it is just filed against Ubuntu in general. It is important that bug reports be filed about source packages so that people interested in the package can find the bugs about it. You can find some hints about determining what package your bug might be about at https://wiki.ubuntu.com/Bugs/FindRightPackage. You might also ask for help in the #ubuntu-bugs irc channel on Freenode.

To change the source package that this bug is filed about visit https://bugs.launchpad.net/ubuntu/+bug/1108082/+editstatus and add the package name in the text box next to the word Package.

[This is an automated message. I apologize if it reached you inappropriately; please just reply to this message indicating so.]

[Brad Figg (brad-figg)]

This bug is missing log files that will aid in diagnosing the problem. From a terminal window please run:

apport-collect 1108082

and then change the status of the bug to 'Confirmed'.

If, due to the nature of the issue you have encountered, you are unable to run this command, please add a comment stating that fact and change the bug status to 'Confirmed'.

This change has been made by an automated script, maintained by the Ubuntu Kernel Team.

[Joseph Salisbury (jsalisbury)]

It looks like this is the value of /proc/PID/coredump_filter is 0x23 on Raring as well. Do you know if releases prior to 12.04 also had this issue?

[Bruce Dawson (bruced)]

I don't have any previous versions of Ubuntu installed so I don't know if this issue is long-standing.

Apparently having 0x33 be the default started around kernel version 2.6.26 (CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS=y).

Fedora defaults to 0x33. I haven't tested any other distributions.

[Bruce Dawson (bruced)]

System information

[Bruce Dawson (bruced)]

lspci-vvnn.log system information

[Bruce Dawson (bruced)]

uname-a system information

[Bruce Dawson (bruced)]

version system information

[Martin Pitt (pitti)]

This doesn't look incomplete -- the description is complete.

I don't think this is a regression -- we (i. e. Apport etc.) doesn't currently rely on build IDs in the core dump. It would be nice to do that eventually, though, and change our debug symbol server to be build ID based instead of package based.

[Bruce Dawson (bruced)]

While apport doesn't rely on build IDs, there are other uses for core files besides apport. This lack of build IDs in core files makes core files less useful on Ubuntu than on Fedora.

We don't use apport.

[Martin Pitt (pitti)]

I wanted to rule out upstart or our procps settings (which incur some security restrictions); even if I boot with init=/bin/bash, all processes (including pid 1) has a coredump_filter of 23. So it seems this is really a kernel configuration issue. I think this is CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS option, which we don't set by default: http://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/tree/fs/Kconfig.binfmt#n46

[Andy Whitcroft (apw)]

Applied to saucy master-next and saucy unstable.

[Launchpad Janitor (janitor)]

This bug was fixed in the package linux - 3.9.0-7.15

---------------
linux (3.9.0-7.15) saucy; urgency=low

  [ Tim Gardner ]

  * [debian] Use dh_strip
    - LP: #1192759
  * [Config] Enable perf for armhf
  * do_tools=false when cross compiling
  * rebase to v3.9.7
  * [Config] CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS=y
    - LP: #1108082
  * Release Tracking Bug
    - LP: #1193189

  [ Upstream Kernel Changes ]

  * rebase to v3.9.7
 -- Tim Gardner <email address hidden> Thu, 20 Jun 2013 22:32:49 -0400

[Launchpad Janitor (janitor)]

This bug was fixed in the package linux-grouper - 3.1.10-6.16

---------------
linux-grouper (3.1.10-6.16) saucy; urgency=low

  [ Andy Whitcroft ]

  * [Config] CONFIG_DEVPTS_MULTIPLE_INSTANCES=y
    - LP: #1194850
  * [Config] CONFIG_ELF_CORE=y
    - LP: #1108082, #1197862
 -- Andy Whitcroft <email address hidden> Wed, 26 Jun 2013 15:28:18 +0100

[Launchpad Janitor (janitor)]

This bug was fixed in the package linux-maguro - 3.0.0-3.11

---------------
linux-maguro (3.0.0-3.11) saucy; urgency=low

  [ Andy Whitcroft ]

  * [Config] CONFIG_CORE_DUMP_DEFAULT_ELF_HEADERS=y
    - LP: #1108082

  [ Tim Gardner ]

  * [Config] Enable more NF options
    - LP: #1199351
 -- Andy Whitcroft <email address hidden> Thu, 04 Jul 2013 17:00:15 +0100

[Launchpad Janitor (janitor)]

This bug was fixed in the package linux-mako - 3.4.0-3.15

---------------
linux-mako (3.4.0-3.15) saucy; urgency=low

  [ Tim Gardner ]

  * [Config] Harmonize NF, NETFILTER, IP_NF, and IP6_NF options
 -- Tim Gardner <email address hidden> Tue, 09 Jul 2013 13:05:23 -0600