lp:~mgedmin/apport/fix-lp1453011

Branch merges

Propose for merging

No branches dependent on this one.

Related source package recipes

No recipes using this branch. (?)

Create packaging recipe

Related snap packages

Related bugs

Bug #1453011: SegvAnalysis: Failure: invalid literal for int() with base 16: '='

Undecided

Confirmed

Link a bug report

Related blueprints

2957. By Marius Gedminas on 2015-05-08

Don't let gdb's stderr interfere with apport's spliting logic

I noticed that _command_output() was used in only one place in the codebase, so
I took the liberty of removing unused arguments and changing the API to return
stdout and strerr separately (and, also, decoded into Unicode).

I don't know if the truncated core file warning shows up in stdout or stderr,
so I made the parser expect it in either.

I also made apport notice when the splitting logic goes wrong and append an

  UnreportableReason: Couldn't split gdb's output correctly (got N parts instead of M)

I didn't make it raise because then there's no chance to see how the splitting
went wrong, and now we can at least find some interesting bits of information
in unrelated fields.

I don't know if UnreportableReason is the right field to use.

I think maybe it would be good to attach any leftover parts (if there are any,
which would happen if gdb happened to print $n = -99 for some reason in the
middle of one of the parts) into a new field, but I couldn't come up with a
name. Also note that old code would've raised an IndexError: pop from an empty
list in this case, but I changed it to use zip() so now the report will
proceed, ignoring any extra parts.

This ought to fix https://bugs.launchpad.net/ubuntu/+source/apport/+bug/1453011,
but I'm not sure how to test it.

2956. By Martin Pitt on 2015-05-06

test_signal_crashes: fix comments

2955. By Martin Pitt on 2015-05-06

* signal_crashes test: Fix test_crash_setuid_* to look at whether suid_dumpable was enabled.

2954. By Martin Pitt on 2015-05-06

test_signal_crashes(): Drop hardcoded /tmp/ path in do_crash(), test_nonwritable_cwd() uses a different dir

2953. By Launchpad Translations on behalf of apport-hackers on 2015-04-29

Launchpad automatic translations update.

2952. By Launchpad Translations on behalf of apport-hackers on 2015-04-28

Launchpad automatic translations update.

2951. By Martin Pitt on 2015-04-27

* apportcheckresume: Fix "occured" typo, thanks Matthew Paul Thomas. (LP: #1448636)

2950. By Launchpad Translations on behalf of apport-hackers on 2015-04-22

Launchpad automatic translations update.

2949. By Martin Pitt on 2015-04-16

release 2.17.2

2948. By Martin Pitt on 2015-04-16

* SECURITY UPDATE: Disable crash forwarding to containers. The previous fix in 2.17.1 was not sufficient against all attack scenarios. By binding to specially crafted sockes, a normal user program could forge arbitrary entries in /proc/net/unix. We cannot currently rely on a kernel-side solution for this; this feature will be re-enabled once it gets re-done to be secure. (LP: #1444518)