Cannot PTRACE subthreads due to Yama

[Scott Kitterman]

It's not clear that the fix for Bug #589841 was complete. I just got my first crash since the updated packages were installed and this is what Dr. Konqi was able to provide:

Application: Plasma Netbook (plasma-netbook), signal: Segmentation fault
[Current thread is 1 (Thread 0xb7710720 (LWP 1526))]

Before the fixes for Bug #589841, it wouldn't have even known the current thread, but it's still not making proper backtraces.

ProblemType: Bug
DistroRelease: Ubuntu 10.10
Package: kdelibs5 4:4.4.90-0ubuntu3
ProcVersionSignature: Ubuntu 2.6.35-7.11-generic 2.6.35-rc4
Uname: Linux 2.6.35-7-generic i686
NonfreeKernelModules: wl
Architecture: i386
Date: Fri Jul 9 14:24:06 2010
EcryptfsInUse: Yes
InstallationMedia: Kubuntu-Netbook 10.04 LTS "Lucid Lynx" - Release i386 (20100427)
ProcEnviron:
 LANGUAGE=
 LANG=en_US.UTF-8
 SHELL=/bin/bash
SourcePackage: kde4libs

[Scott Kitterman (kitterman)]

It wasn't just that one crash report. Here's the next one:

Application: rekonq (rekonq), signal: Segmentation fault

[Kees Cook (kees)]

When I clicked the "Developer Information" tab for crashes, I see a full backtrace. For example:

$ kate &
[1] 6927
$ kill -SEGV 6927

And drkonqi shows the crash. Perhaps I'm missing something? What do you see when an application crashes?

[Scott Kitterman (kitterman)]

What I copied in the bug was all I got.
Scott Kitterman

[Philip Muškovac (yofel)]

Confirmed, I just got a plasma crash for some reason and drkonq didn't want to retrace it, even trying your test:

$ kate &
[1] 6927
$ kill -SEGV 6927

does bring drkonqi up, but the backtrace is useless and doesn't match with the gdb output if I run 'gdb kate'. Or taking dolphin as an example, all I get is:
Application: Dolphin (dolphin), signal: Segmentation fault
[Current thread is 1 (Thread 0x7f6ad9877760 (LWP 27989))]

[Philip Muškovac (yofel)]

$ uname -a
Linux yofel-t510 2.6.35-7-generic #12-Ubuntu SMP Fri Jul 9 21:54:03 UTC 2010 x86_64 GNU/Linux

[Kees Cook (kees)]

The PTRACE exception handling was not correctly detecting subthreads. The fiix has been submitted to the kernel team list.

[Launchpad Janitor (janitor)]

This bug was fixed in the package linux - 2.6.35-8.13

---------------
linux (2.6.35-8.13) maverick; urgency=low

  [ Kees Cook ]

  * SAUCE: Yama: check PTRACE using thread group leader
  * SAUCE: Yama: search for PTRACE exceptions via thread group leader
    - LP: #603716

  [ Leann Ogasawara ]

  * rebase to v2.6.35-rc5
  * [Config] update configs following rebase to v2.6.35-rc5

  [ Nicolas Pitre ]

  * SAUCE: make ndiswrapper available on X86 only

  [ Tim Gardner ]

  * [Config] Added ums-cypress to udeb
    - LP: #576066
  * SAUCE: fix build error with CONFIG_BLK_DEV_INITRD=n
  * [Config] CONFIG_NDISWRAPPER=m across all configs

  [ Upstream Kernel Changes ]

  * HID: magicmouse: report last touch up
  * rebase to 2.6.35-rc5
 -- Leann Ogasawara <email address hidden> Tue, 13 Jul 2010 18:57:59 -0700

[Harald Sitter (apachelogger)]

Application: rekonq (rekonq), signal: Segmentation fault

Got that today for my first crash on a system upgraded from lucid yesterday.

Linux logos 2.6.35-10-generic #15-Ubuntu SMP Thu Jul 22 11:04:54 UTC 2010 i686 GNU/Linux

[Harald Sitter (apachelogger)]

Closing again, since original report is fixed -> opening new bug.