Bug 80141

Summary: Computer reboots late in the startup process
Product: Platform Specific/Hardware Reporter: Rod Smith (rodsmith)
Component: x86-64Assignee: platform_x86_64 (platform_x86_64)
Status: NEW ---    
Severity: blocking CC: alan
Priority: P1    
Hardware: x86-64   
OS: Linux   
Kernel Version: 3.13.0 and later Subsystem:
Regression: No Bisected commit-id:
Attachments: Output of "lspci -v" on affected computer

Description Rod Smith 2014-07-13 21:52:02 UTC
Created attachment 142871 [details]
Output of "lspci -v" on affected computer

I recently replaced the motherboard in a Mythbuntu 14.04 computer with a new MSI A88X-G43 motherboard with an AMD A10-6700 APU. Unfortunately, I found that the computer would reboot soon after booting using the stock Ubuntu/Mythbuntu 3.13.0 kernel. Sometimes the computer would stay up long enough to log in, but more frequently, it reboots before any evidence of X running appears on the screen. I found that downgrading to a 3.12.23 or earlier kernel (locally compiled from the kernel.org sources) fixes the problem. More recent kernels (up to at least 3.16.0-rc3, the latest I've tried) from kernel.org also suffer from this problem, so it's not an Ubuntu-specific issue -- at least, not in the kernel. (I can't rule out the possibility that something in Ubuntu's init system is interacting with the kernel version to cause the problem.)

I've verified the problem with both EFI-mode and BIOS-mode boots, using both GRUB and the EFI stub loader, although most of my testing has been done in EFI mode with the EFI stub loader.

I'm attaching the output of "lspci -v" on the affected computer.
Comment 1 Alan 2014-07-29 15:38:54 UTC
Does it run memtest86 reliably - ie is this perhaps just a dodgy board ? The randomness sounds like a hardware bug to be honest.
Comment 2 Rod Smith 2014-07-29 18:01:54 UTC
No, that's not it. I ran memory tests on the system, and they passed. I also tried running with just one of the two DIMMs I bought for it; each one by itself produced the same results. Ditto when I tried memory from an entirely different computer. The only thing random about it is the precise time at which the reboot occurs, and that varies by just a few seconds. My interpretation of that is that it's just some minor variation in when whatever process causes the reboot actually executes.