Bug 9860 - any hope for a fix?

General support questions
Post Reply
SIGBUS
Posts: 14
Joined: 2007/05/28 15:39:06

Bug 9860 - any hope for a fix?

Post by SIGBUS » 2016/02/12 16:29:19

This problem needs some discussion, and I didn't want to clutter the bug report with it.

I have a system affected by Bug 9860, and thus won't boot any released 3.10.0-327 kernel. It has been almost two months now since this bug has surfaced, and there is still no trace of a fix being available to the public. Upstream has the fix behind a paywall.

As a result, my file server is still stuck on 3.10.0-229.20.1, the last available kernel before the 1511 release. Is there any chance this fix will be made public, or are we out of luck? I'd expect a kernel to break now and then if I were running Fedora, but this isn't Fedora.

I'm just astounded that a) such a showstopper got released upstream, and b) the fix is paywalled.

At this time, my choices seem to be:

1. Keep running an old kernel and let the CVEs pile up. Fortunately, this is a home server that isn't exposed to the public Internet, but I'd still rather keep it up-to-date.
2. Use the ELRepo mainline kernel. Since I don't need to run VMs on this system, it's workable.
3. Move to CentOS 6 and hope that it doesn't get broken in the same manner.
4. Move on to another distro or operating system. I'd rather not do this, but I'm getting unhappy enough that it's becoming more and more likely to happen. I had only recently moved this machine to CentOS 7 after running Illumos/OpenIndiana for a long while. (Yes, I'm a ZFS fan.)
5. Get a new motherboard and/or CPU. At that point, I might as well order a complete refurbished server with Intel hardware.
6. Try to figure out what patch caused the problem and unpatch it, hoping I don't break something else. I haven't hacked on a kernel for hardware support since I ran Slackware with a 1.2 kernel.

Option 2 seems to be the best stopgap solution, but I'd much rather run an official CentOS build.

User avatar
toracat
Forum Moderator
Posts: 7383
Joined: 2006/09/03 16:37:24
Location: California, US
Contact:

Re: Bug 9860 - any hope for a fix?

Post by toracat » 2016/02/12 18:05:58

It is indeed unfortunate that the patch is not publicly available.

If have to choose from the options listed, I'd probably go for Option 2 (kernel-ml) or Option 3 (CentOS-6). Please note that kernel-ml should have no problem with VMs (see comment 25638). CentOS-6 is good through Nov 2020.
CentOS Forum FAQ

User avatar
TrevorH
Forum Moderator
Posts: 26590
Joined: 2009/09/24 10:40:56
Location: Brighton, UK

Re: Bug 9860 - any hope for a fix?

Post by TrevorH » 2016/02/12 18:23:30

I guess it's because the workaround of the kernel parameter works for most people.
CentOS 5 died in March 2017 - migrate NOW!
CentOS 6 goes EOL sooner rather than later, get upgrading!
Full time Geek, part time moderator. Use the FAQ Luke

User avatar
toracat
Forum Moderator
Posts: 7383
Joined: 2006/09/03 16:37:24
Location: California, US
Contact:

Re: Bug 9860 - any hope for a fix?

Post by toracat » 2016/02/12 18:52:25

The workaround is .... a workaround. Is it not without issues. As noted by a RH engineer in his comment,
The workaround is only meant to get you through the initial install. After the installation, upgrade to the latest functioning kernel, and then reboot.
CentOS Forum FAQ

SIGBUS
Posts: 14
Joined: 2007/05/28 15:39:06

Re: Bug 9860 - any hope for a fix?

Post by SIGBUS » 2016/02/17 14:47:36

Followup: the just-released 3.10.0-327.10.1 kernel has a fix for this problem. Sorry if I seemed impatient, but it was getting worrisome.

Post Reply

Return to “CentOS 7 - General Support”