In bug 470214 we're inspecting an overflow ocucring in the now-stable 3.8.6-hardened. While no definitive answer has been given so far, it seems that the bug is not present in 3.8.12-hardened. Please stabilize 3.8.12. See 470214 for details (still classified). Reproducible: Always
Or, perhaps better yet, s/3.8.12/3.8.13/.
can you test 3.9.5
A quick test that is known to have triggered the bug at least once in 3.8.6 doesn't ruffle 3.9.5. I'm switching to 3.9.5 as the main kernel and will report here if I hit something. 3.8.12 is better tested on my part.
(In reply to Roman Žilka from comment #3) > A quick test that is known to have triggered the bug at least once in 3.8.6 > doesn't ruffle 3.9.5. I'm switching to 3.9.5 as the main kernel and will > report here if I hit something. 3.8.12 is better tested on my part. Thanks. I'll target 3.9.5 as the next set to stabilize in 2 weeks. There have been some small incremental fixes between 3.8.12 and 3.9.5. The grsec/pax team drop support for their experimental 3.x.y branch as soon as there's a bump to 3.(x+1).z. They have been long term maintaiing 2.6.32.u and 3.2.v. Please reopen this bug if you hit this problem again.