Suspend issues with ehci-hcd
Matthew Garrett wrote:
> On Fri, May 09, 2008 at 01:27:33PM -0600, Tim Gardner wrote:
>> I've encountered several suspend issues lately involving ehci-hcd. I'm
>> considering what might be the side effects of adding it to pm-utils as a
>> normal operation, e.g., add ehci-hcd to the list of modules that are
>> always removed before suspend.
> We used to do that, but stopped when USB gained reasonable
> suspend/resume support.
>> What happens to a USB disk mounted as root? Will it fix corruption
>> issues where non-root USB drives don't get flushed?
> If you have a USB disk mounted as root, you'll lose your root
> filesystem. I'm not aware of any situations where data should hit USB
> devices on removal but not on suspend, but it's possible.
>> Removing ehci-hcd prior to suspend does fix a whole class of bluetooth
> Is this not because it ends up destroying and recreating the bluetooth
Right - removing ehci-hcd also removes all dependent modules, among
which are bluetooth. I attempted removing just the root bluetooth module
on suspend, but found I couldn't get the Dell laptop I was using to even
finish the suspend, much less resume.
So, if I've read this correctly, removing ehci-hcd on suspend is
generally OK except when your root device is USB. Which means its
probably not a good idea to add ehci-hcd to pm-utils default removal
Tim Gardner firstname.lastname@example.org
kernel-team mailing list