Bug 151641

Summary: ohci-pci "controller won't resume" "HC died; cleaning up"
Product: Drivers Reporter: elvis has left the building (icanrealizeum+bugzillakernelorg)
Component: USBAssignee: Greg Kroah-Hartman (greg)
Status: NEW ---    
Severity: normal CC: icanrealizeum+bugzillakernelorg, rui.zhang
Priority: P1    
Hardware: All   
OS: Linux   
Kernel Version: 4.7.0-g0cbbc42 Subsystem:
Regression: No Bisected commit-id:
Attachments: lspci -v and lsusb -v included in attachment.
stacktrace

Description elvis has left the building 2016-08-06 17:30:35 UTC 
Created attachment 227781 [details]
lspci -v and lsusb -v included in attachment.

During reboot/shutdown, is the only time that I can see these messages from the title(that is, they aren't present in dmesg any other time).

ohci-pci 0000:00:13.0: PME# disabled
ohci-pci 0000:00:13.0: enabling bus mastering
ohci-pci 0000:00:12.0: PME# disabled
ohci-pci 0000:00:12.0: enabling bus mastering
ohci-pci 0000:00:13.0: controller won't resume
ohci-pci 0000:00:13.0: HC died; cleaning up
sd 3:0:0:0: [sda] Synchronizing SCSI cache
ohci-pci 0000:00:12.0: controller won't resume

I'm not sure if this is related in any way to the bug with Synchronizing SCSI cache here: https://bugzilla.kernel.org/show_bug.cgi?id=151631
but it seems not, unless something is shutting down most subsystems prior to that.

The screenshot of the above messages in context is this: https://i.imgur.com/3P7ztlj.png

(I tried sending to mailing list but I get no message back when trying to subscribe - maybe some temporary issue)

I can try any patches/suggestions. Thanks!
Comment 1 elvis has left the building 2016-08-08 08:47:43 UTC 
Some more stuff from the linux-usb mailing list:

> if I try to sysrq+o (aka poweroff) manually from a file that systemd
> runs at shutdown such as manually created a+x file:
> /usr/lib/systemd/system-shutdown/debug.sh
>
> then I get no ohci errors,
> screenshot: https://i.imgur.com/VVeAyj7.jpg
> only the "PME# disabled" and "enabling bus mastering" are seen.
> But if I allow the kernel(and/or whatever else happens before that) to
> do it, then the errors are back just like in my original post.

this is the stack when 'controller won't resume happened'
> https://i.imgur.com/9wlbure.jpg
> but it looks like something else called that and I don't know
> what(looks forked?), unless it's something from the following stack
> from the point where "Stop disk" happens:
> https://i.imgur.com/z9uq4ez.jpg
> Maybe I'll find out something from there.
Comment 2 elvis has left the building 2016-08-08 12:33:32 UTC 
Created attachment 227941 [details]
stacktrace

Thanks to Alan Stern (from linux-usb ML), I got us this nice stacktrace of what happens on shutdown/poweroff that causes these messages:

also seen here: https://i.imgur.com/VMak4ED.jpg

I mean, am I the only one who likes this? A LOT! :D