Linux

  1. Bug #1419355
  2. Comment #63

Comment 63 for bug 1419355

Revision history for this message
In , Matti.Kurkela (matti.kurkela-linux-kernel-bugs) wrote :

(In reply to Giorgos aka shad0w from comment #6)
> Haswell LynxPoint and LynxPoint-LP with the recent Intel BIOS show
> mysterious wakeups after shutdown occasionally. After discussing with
> BIOS engineers, they explained that the new BIOS expects that the
> wakeup sources are cleared and set to D3 for all wakeup devices when
> the system is going to sleep or power off, but the current xhci driver
> doesn't do this properly (partly intentionally).

I had a similar spurious wakeup problem on my Asus P8Z77-V Premium motherboard (which is a Panther Point chipset). The problem has existed for several kernel and BIOS versions now (currently using the latest available BIOS, version 2104). I'm running Debian stable, with a vanilla kernel.org kernel version 3.12.7. I'm using UEFI-based boot (rEFInd as the boot menu, and ELILO as the Linux loader), dual-booting with Windows 7.

On Windows 7, the system always powers off without problems; on Linux, it consistently wakes up again a few seconds after powering off.

The discussion in this bug contained the clues I needed to test and apply a workaround which seems to be completely effective for my problem so far.

I added these commands to the system halt scripts:
  modprobe -r xhci-hcd
  setpci -s 00:14.0 CAP_PM+4.b=0b

i.e. the shell-level approximation of the XHCI_SPURIOUS_WAKEUP quirk.

Just unloading the xhci-hcd module had no effect to the problem, and I confirmed that the power state was 08 (i.e. D0) after unloading the module manually.

After adding these commands, my system has powered down successfully on Linux 8 times out of 8 so far.

I guess this means my system might benefit from the XHCI_SPURIOUS_WAKEUP quirk too.

Since the need for this quirk seems to be dependent on BIOS issues, this might be applicable on other systems too. My motherboard uses an UEFI-capable BIOS firmware from American Megatrends Inc. (AMI), and after downloading the manual for Giorgos's Asrock H87 pro4 motherboard, I see his MB has the same BIOS manufacturer. Niklas's and Stephan's dmidecode outputs indicate an UEFI-capable AMI BIOS too.

My first guess would be that systems with AMI UEFI BIOS might need XHCI_SPURIOUS_WAKEUP, but obviously more information would be good.