Kernel crash upon resume due to memory corruption
|Reported by:||wad||Owned by:||wad|
|Priority:||high||Milestone:||8.2.0 (was Update.2)|
|Deployments affected:||Action Needed:||never set|
When doing a large number of suspend/resume cycles (tens of thousands), we occasionally see a kernel crash upon resume.
After filtering out the kernel crashes due to actual kernel bugs, we are left with a number of crashes which appear due to hardware problems, specifically memory corruption.
This seems to favor certain laptops, and is not seen on all machines. There is a special version of OFW (q2d01a) which implements a comparison of memory before and after the resume. It may be used to test specifically for this case.