On Mon, 2012-07-02 at 19:29 +0100, Steve Graham wrote: > After building a kernel with X86_CHECK_BIOS_CORRUPTION enabled, I set it to scan (almost) as > advised. No corruption of the low memory was reported after running some hours either with or > without usbhid loaded. And did you plug in a new HID device? > (Booting with memory_corruption_check_size=640K caused an immediate kernel panic -- early exception > 08 --though. Changing it to 620k allowed the boot to continue normally.) > > I took the opportunity to include kmemleak in the new kernel as well. After some uptime, one orphan > object was always found, but it happened either with or without usbhid ever having been loaded. To > judge from the traceback it was something ACPI-related anyway. > > So, basically, no progress. Can you test Linux 3.4, available from the experimental suite? If that still has the problem, we can pass this on to the upstream developers; if not, we can try to work out where it was fixed. Ben. -- Ben Hutchings When in doubt, use brute force. - Ken Thompson
Attachment:
signature.asc
Description: This is a digitally signed message part