Bug#711135: Further investigation

To: 711135@bugs.debian.org
Subject: Bug#711135: Further investigation
From: Chris Collins <chris@collins.id.au>
Date: Thu, 20 Feb 2014 20:52:03 +1100
Message-id: <[🔎] 37E86A0765E94528AFB27C33EDC0086B@collins.id.au>
Reply-to: Chris Collins <chris@collins.id.au>, 711135@bugs.debian.org

I believe I’ve traced this one down completely now.

I’m currently waiting for the kernel packages to finish rebuilding on my build box so I can verify the fix.

Inspection of my ZX2000’s SAL crash logs show that the machine went down from a recoverable MCA (Cache poisoning). The Wheezy kernel does not have the MCA recovery paths compiled in, but rather, compiled as module. If the system is throwing the MCA early enough, it has no chance of recovery because of this.

All of the functional kernel configs Peter Chubb mentioned have the MCA Recovery code path linked in.

I’m building a kernel package set with MCA recovery linked in and will verify on my zx2000. If this fixes it, I’ll post a patch to this bug.

Regards,

Chris.

Chris Collins

Reply to:

Prev by Date: Bug#739593: unshare -m aka unshare(CLONE_NEWNS) mounts are visible globally
Next by Date: Processed: reassign 739593 to systemd, severity of 739593 is grave
Previous by thread: Bug#739593: unshare -m aka unshare(CLONE_NEWNS) mounts are visible globally
Next by thread: Processed: reassign 739593 to systemd, severity of 739593 is grave
Index(es):
- Date
- Thread