Re: [PATCH 25/30] ACPI / hotplug / PCI: Check for new devices on enabled slots

[Alex Williamson <alex.williamson@redhat.com>]

On Thu, 2013-09-05 at 01:35 +0200, Rafael J. Wysocki wrote:
> On Wednesday, September 04, 2013 05:12:14 PM Alex Williamson wrote:
> > On Thu, 2013-09-05 at 00:54 +0200, Rafael J. Wysocki wrote:
> > > On Wednesday, September 04, 2013 02:36:34 PM Alex Williamson wrote:
> > > > On Thu, 2013-07-18 at 01:32 +0200, Rafael J. Wysocki wrote:
> > > > > From: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> > > > > 
> > > > > The current implementation of acpiphp_check_bridge() is pretty dumb:
> > > > >  - It enables a slot if it's not enabled and the slot status is
> > > > >    ACPI_STA_ALL.
> > > > >  - It disables a slot if it's enabled and the slot status is not
> > > > >    ACPI_STA_ALL.
> > > > > 
> > > > > This behavior is not sufficient to handle the Thunderbolt daisy
> > > > > chaining case properly, however, because in that case the bus
> > > > > behind the already enabled slot needs to be rescanned for new
> > > > > devices.
> > > > > 
> > > > > For this reason, modify acpiphp_check_bridge() so that slots are
> > > > > disabled and stopped if they are not in the ACPI_STA_ALL state.
> > > > > 
> > > > > For slots in the ACPI_STA_ALL state, devices behind them that don't
> > > > > respond are trimmed using a new function, trim_stale_devices(),
> > > > > introduced specifically for this purpose.  That function walks
> > > > > the given bus and checks each device on it.  If the device doesn't
> > > > > respond, it is assumed to be gone and is removed.
> > > > > 
> > > > > Once all of the stale devices directy behind the slot have been
> > > > > removed, acpiphp_check_bridge() will start looking for new devices
> > > > > that might have appeared on the given bus.  It will do that even if
> > > > > the slot is already enabled (SLOT_ENABLED is set for it).
> > > > > 
> > > > > In addition to that, make the bus check notification ignore
> > > > > SLOT_ENABLED and go for enable_device() directly if bridge is NULL,
> > > > > so that devices behind the slot are re-enumerated in that case too.
> > > > > 
> > > > > This change is based on earlier patches from Kirill A Shutemov
> > > > > and Mika Westerberg.
> > > > > 
> > > > > Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
> > > > > Tested-by: Mika Westerberg <mika.westerberg@linux.intel.com>
> > > > > ---
> > > > 
> > > > FYI, git bisect landed on this patch as the cause of my serial console
> > > > dying on current upstream.  Further debugging to come...  Thanks,
> > > 
> > > Well, sorry about that.
> > > 
> > > What exactly do you mean by "dying"?
> > 
> > Sorry, I was hoping to have more details quickly, but it's been a pain
> > to debug.  By dying I mean serial console output suddenly stops during
> > kernel boot and nothing more comes out of it until after the system is
> > rebooted.  The problem happens when acpiphp_check_bridge() calls
> > enable_slot().  The serial console dies somewhere down in
> > acpiphp_bus_trim().  I think this is happening on the 00:1f ISA bridge,
> > so there's a good chance the serial ports are described as somewhere
> > under there.
> 
> Can you please check if that is the acpiphp_bus_trim() called by
> acpiphp_bus_add() or the other one called from trim_stale_devices()?
> 
> Just add a dump_stack() or WARN_ON(1) to trim_stale_devices() next to
> the acpiphp_bus_trim() call and see if that triggers.  I *think* it's the one
> in acpiphp_bus_add(), but it won't hurt to verify that.

Here's the call path:

[   16.120824]  [<ffffffff81627e6c>] dump_stack+0x55/0x76
[   16.125979]  [<ffffffff8162132e>] enable_slot+0x4ee/0x5e0
[   16.131396]  [<ffffffff813418fb>] ? trim_stale_devices+0x5b/0xf0
[   16.137420]  [<ffffffff81341b35>] acpiphp_check_bridge+0xd5/0x110
[   16.143531]  [<ffffffff81342acb>] hotplug_event+0x16b/0x260
[   16.149115]  [<ffffffff81072cd9>] ? process_one_work+0x189/0x540
[   16.155136]  [<ffffffff81342bf0>] hotplug_event_work+0x30/0x70
[   16.160978]  [<ffffffff81072d3b>] process_one_work+0x1eb/0x540
[   16.166819]  [<ffffffff81072cd9>] ? process_one_work+0x189/0x540
[   16.172836]  [<ffffffff8107353c>] worker_thread+0x11c/0x370
[   16.178426]  [<ffffffff81073420>] ? rescuer_thread+0x350/0x350
[   16.184276]  [<ffffffff8107b0ea>] kthread+0xea/0xf0
[   16.189165]  [<ffffffff8107b000>] ? kthread_create_on_node+0x160/0x160
[   16.195700]  [<ffffffff816395dc>] ret_from_fork+0x7c/0xb0
[   16.201109]  [<ffffffff8107b000>] ? kthread_create_on_node+0x160/0x160

The actual death of the serial console occurs in acpi_device_set_power()
called from:

enable_slot()
 acpiphp_bus_add()
  acpiphp_bus_trim()
   acpi_bus_trim()
    acpi_walk_namespace()
     acpi_bus_remove()
      acpi_device_unregister()
       acpi_device_set_power()

I can't seem to get a path from the acpi devices in question there, so I
have no idea what's getting trimmed here.  It worries me quite a bit by
introducing this trimming that apparently wasn't happening before
though.  Thanks,

Alex