Re: 2.6.27-rc2: unable to suspend - one task not freezed

[Andrey Borzenkov]

On Sunday 10 August 2008, Rafael J. Wysocki wrote:
> On Saturday, 9 of August 2008, Andrey Borzenkov wrote:
> > 2.6.27-rc2 + unrevert of e872154921a6b5256a3c412dd69158ac0b135176
> > (USB: don't explicitly reenable root-hub status interrupts) + fix
> > of above from Alan Stern (OHCI: fix system hang caused by earlier patch).
> 
> Why do you have to unrevert commit e872154921a6b5256a3c412dd69158ac0b135176
> exactly?
> 

This commit caused my system to hang after ohci_usb was loaded. It was
reverted in 2.6.26 after my report. Later Alan sent the fix for it which I
have been testing. This commit (with fix) is scheduled back in 2.6.27.

This is regression because 2.6.26 + two patches mentioned above suspend and
resume just fine. 

> > System cannot be suspended - some task cannot be freezed. I think it is
> > khubd ...
> 
> Yes, it is.
> 
> It looks like khubd is running in a loop without checking TIF_FREEZE.
> 
> > dmesg and config attached.
> 
> Hm.  Is this 100% reproducible?
> 

So far yes.

Attachment: signature.asc
Description: This is a digitally signed message part.