Re: [2.5.75] S3 and S4

[Pavel Machek <pavel@suse.cz>]

Hi!

> > > > > Suspending devices
> > > > > Badness in local_bh_enable at kernel/softirq.c:113
> > > > > Call Trace:
> > > > >  [<c0130078>] local_bh_enable+0x88/0x90
> > > > >  [<f0a44fa4>] e100_do_wol+0x14/0x60 [e100]
> > > > >  [<f0a461ee>] e100_suspend+0x3e/0xa0 [e100]
> > > > >  [<f0a461b0>] e100_suspend+0x0/0xa0 [e100]
> > > > >  [<c0212577>] pci_device_suspend+0x47/0x70
> > > > >  [<c029bc99>] device_suspend+0xd9/0x100
> > > > >  [<c023e047>] acpi_system_save_state+0x42/0x8c
> > > > >  [<c023e153>] acpi_suspend+0x5e/0xb3
> > > > >  [<c023e394>] acpi_system_write_sleep+0xe3/0x132
> > > > >  [<c0177de0>] filp_open+0x60/0x70
> > > > >  [<c017952d>] vfs_write+0xad/0x120
> > > > >  [<c017963f>] sys_write+0x3f/0x60
> > > > >  [<c010b10f>] syscall_call+0x7/0xb
> > > > > 
> > > > 
> > > > If e100. You have the hardware...
> > > 
> > > No, it's acpi_system_save_state() illegally calling device_suspend() under
> > > local_irq_disable().
> > 
> > Oops, I failed to see this is S3.
> > 
> > I can see that device_suspend( ..., SUSPEND_POWER_DOWN) is called with
> > interrupts disabled. But thats okay: (driver.txt) All calls are made
> > with interrupts enabled, except for the SUSPEND_POWER_DOWN level.
> 
> OK, it's an e100 bug then.  Not allowed to sleep or do spin_unloch_bh() in
> device_suspend( ..., SUSPEND_POWER_DOWN).
> 
> That's a fairly awkward restriction.

Actually, its not. Some phase with interrupts off is needed for
devices such as interrupt controller. Fix is very simple: move the
e100 suspend from SUSPEND_POWER_DOWN to some other level.

								Pavel

-- 
When do you have a heart between your knees?
[Johanka's followup: and *two* hearts?]