Re: exception following PowerOffSystem() call in power manager PDD

Tech-Archive recommends: Repair Windows Errors & Optimize Windows Performance

From: Andrew Worsley (Andrew.Worsley_at_nospam.4.me)
Date: 12/20/04 

Date: Tue, 21 Dec 2004 09:31:33 +1300

In article <uOVxj5l5EHA.828@TK2MSFTNGP14.phx.gbl>, "David Liao \(MS\)"
<davli@online.microsoft.com> says...
> On Windows CE 4.2, you can't call any API during the power callback(Except
> few like SetInterruptEvent). The message indicate you are calling some API
> during the power callback. Usually, the problem is not the Power Manager
> PDD, it is a driver. Probably, you have a driver which calling some API in
> XXX_PowerDown or XXX_PowerUp handler.
>
> You probably can find out which driver by turn on PNP zone on processor
> device.exe.

Thanks David,

Sorry I replied by email - I meant to reply to the group, but must have
pressed the wrong button.

I was aware of that restriction (drivers can't call APIs during
XXX_PowerDown()), but couldn't find any driver that
breaks the rules. I'll tried turning on the PNP zone, but it didn't give
any more information. In fact, I turned on all the device.exe zones, and
all the pm.dll zones, and this is what I got:

4294844962 PID:ffc0472 TID:6ff998ae 0x8ff968e0: PowerManager: Changing
state from osrunning to osclosing because of Timeout
4294844963 PID:ffc0472 TID:6ff998ae 0x8ff968e0:
+PmSetSystemPowerState_I: name osclosing, hint 0x00000000, options
0x00000000, fInternal 1
4294844966 PID:ffc0472 TID:6ff998ae 0x8ff968e0: ReadSystemPowerState:
state 'osclosing' devices max is D4, flags 0x00200000
4294844967 PID:ffc0472 TID:6ff998ae 0x8ff968e0: PmAlloc: alloc 28 bytes
returned 0x00053010 (55 total objects allocated)
4294844969 PID:ffc0472 TID:6ff998ae 0x8ff968e0: PmAlloc: alloc 36 bytes
returned 0x00053040 (56 total objects allocated)
4294844971 PID:ffc0472 TID:6ff998ae 0x8ff968e0:
RegReadClassDeviceRestrictions: reading information for class {98c5250d-
c29a-4985-ae5f-afe5367e5006}
4294844973 PID:ffc0472 TID:6ff998ae 0x8ff968e0: PmAlloc: alloc 44 bytes
returned 0x00053080 (57 total objects allocated)
4294844975 PID:ffc0472 TID:6ff998ae 0x8ff968e0: PmAlloc: alloc 24 bytes
returned 0x000530c0 (58 total objects allocated)
4294844977 PID:ffc0472 TID:6ff998ae 0x8ff968e0: PmAlloc: alloc 32 bytes
returned 0x000530f0 (59 total objects allocated)
4294844979 PID:ffc0472 TID:6ff998ae 0x8ff968e0:
RegReadClassDeviceRestrictions: device 'Default' restricted to D0
4294844981 PID:ffc0472 TID:6ff998ae 0x8ff968e0:
RegReadClassDeviceRestrictions: reading information for class {a32942b7-
920c-486b-b0e6-92a702a99b35}
4294844983 PID:ffc0472 TID:6ff998ae 0x8ff968e0: PmAlloc: alloc 44 bytes
returned 0x00053130 (60 total objects allocated)
4294844985 PID:ffc0472 TID:6ff998ae 0x8ff968e0: PmAlloc: alloc 24 bytes
returned 0x00053170 (61 total objects allocated)
4294844987 PID:ffc0472 TID:6ff998ae 0x8ff968e0: PmAlloc: alloc 32 bytes
returned 0x000531a0 (62 total objects allocated)
4294844989 PID:ffc0472 TID:6ff998ae 0x8ff968e0:
RegReadClassDeviceRestrictions: device 'Default' restricted to D4
4294844991 PID:ffc0472 TID:6ff998ae 0x8ff968e0:
RegReadClassDeviceRestrictions: skipping flags 0x00200000
4294844993 PID:ffc0472 TID:6ff998ae 0x8ff968e0: GenerateNotifications:
sending type 1 notifications (32 bytes)
4294844996 PID:ffc0472 TID:6ff998ae 0x8ff968e0: SendNotification:
sending notification to 0x000519c0 (owner 0x2fe444f6)
4294844997 PID:ffc0472 TID:6ff998ae 0x8ff968e0: GenerateNotifications:
done sending type 1 notifications
4294844999 PID:ffc0472 TID:6ff998ae 0x8ff968e0: NotifyAndPowerOff():
suspending - notifying non-block drivers
4294845000 PID:ffc0472 TID:6ff998ae 0x8ff968e0: DeviceStateAddRef:
refcnt for 0x00037430 set to 3
4294845002 PID:ffc0472 TID:6ff998ae 0x8ff968e0: UpdateDeviceState: new
state for 'nds0:' is 4 (current 0, req 0, set -1, floor 4, ceiling 4,
actual 0)
4294845004 PID:ffc0472 TID:6ff998ae 0x8ff968e0: MapDevicePowerState:
mapping D4 to D0
4294845005 PID:ffc0472 TID:6ff998ae 0x8ff968e0: SetDevicePower: setting
'nds0:' to D4 (mapped to D0), fForceSet is 0
4294845007 PID:ffc0472 TID:6ff998ae 0x8ff968e0: SetDevicePower: device
is already at D0
4294845008 PID:ffc0472 TID:6ff998ae 0x8ff968e0: DeviceStateDecRef:
refcnt for 0x00037430 set to 2
4294845009 PID:ffc0472 TID:6ff998ae 0x8ff968e0: DeviceStateAddRef:
refcnt for 0x00037670 set to 2
4294845011 PID:ffc0472 TID:6ff998ae 0x8ff968e0: UpdateDeviceState: new
state for 'vmini1' is -1 (current 0, req 0, set -1, floor 4, ceiling 0,
actual 0)
4294845013 PID:ffc0472 TID:6ff998ae 0x8ff968e0: DeviceStateDecRef:
refcnt for 0x00037670 set to 1
4294845015 PID:ffc0472 TID:6ff998ae 0x8ff968e0: NotifyAndPowerOff():
calling KernelIoControl(IOCTL_HAL_PRESUSPEND)
4294845016 PID:cffdd59e TID:6ff998ae 0x8ff968e0: FS: Got Power notif
0x00000002
4294845018 PID:cffdd59e TID:6ff998ae 0x8ff968e0: FSREG: Registry flush
unnecessary
4294845020 PID:cffdd59e TID:6ff998ae 0x8ff968e0: FSRAM: Flushing cached
file data
4294845021 PID:ffc0472 TID:6ff998ae 0x8ff968e0: NotifyPowerOff():
suspending - notifying block drivers
4294845023 PID:ffc0472 TID:6ff998ae 0x8ff968e0: DeviceStateAddRef:
refcnt for 0x000524d0 set to 2
4294845025 PID:ffc0472 TID:6ff998ae 0x8ff968e0: UpdateDeviceState: new
state for 'dsk1:' is 4 (current 0, req 0, set -1, floor 4, ceiling 4,
actual 0)
4294845026 PID:ffc0472 TID:6ff998ae 0x8ff968e0: MapDevicePowerState:
mapping D4 to D4
4294845028 PID:ffc0472 TID:6ff998ae 0x8ff968e0: SetDevicePower: setting
'dsk1:' to D4 (mapped to D4), fForceSet is 0
4294845029 PID:ffc0472 TID:6ff998ae 0x8ff968e0: RequestStreamDevice:
calling DeviceIoControl(0x4fee6ce6) w/ request 3280904
('IOCTL_POWER_SET')
4294845031 PID:ffc0472 TID:6ff998ae 0x8ff968e0: ATADISK: Powering off
system: D4 Count=1
4294845033 PID:ffc0472 TID:6ff998ae 0x8ff968e0: DeviceStateDecRef:
refcnt for 0x000524d0 set to 1
4294845035 PID:ffc0472 TID:6ff998ae 0x8ff968e0: NotifyPowerOff():
calling PowerOffSystem()
4294845036 PID:ffc0472 TID:6ff998ae Exception 040 Thread=8ff968e0 Proc=
0ffc0472 'device.exe'
4294845037 PID:ffc0472 TID:6ff998ae AKY=00000009 PC=8c9b2f4e RA=8c9b01f0
TEA=00003b83

As you can see, it is not until AFTER the call to PowerOffSystem() that
the exception occurs.

I thought it was Power Manager PDD that called the XXX_PowerDown()
functions, right?

I can step all the way through the Power Manger PDD code, past the calls
to the UpdateClassDeviceStates() (is this what calls the XXX_PowerDown()
functions?), right up to the call to PowerOffSystem(). Its only after
stepping into that function that the exception occurs.

Does the call to PowerOffStstem() cause other driver functions to be
called? I thought all it did was set a flag, so the kernel calls
OEMPowerOff() on the next reschedule.

It is called like this (same as default driver, with some extra logging
so I can see when it happens:

  FileSystemPowerFunction(FSNOTIFY_POWER_OFF);

  // update block device power states
  DEBUGMSG(ZONE_PLATFORM || ZONE_RESUME, (_T("NotifyPowerOff():
suspending - notifying block drivers\r\n")));
  pdl = GetDeviceListFromClass(&idBlockDevices);
  if(pdl != NULL) {
      UpdateClassDeviceStates(pdl);
  }

  DEBUGMSG(ZONE_PLATFORM || ZONE_RESUME, (_T("NotifyPowerOff(): calling
PowerOffSystem()\r\n")));
  PowerOffSystem(); // Set Flag in the system
  //Sleep(INFINITE); // Run the scheduler and never come back
  Sleep(0); // Run the scheduler and never come back
  DEBUGMSG(ZONE_PLATFORM || ZONE_RESUME, (_T("NotifyPowerOff(): returned
from PowerOffSystem() - shouldn't get here!\r\n")));

Relevant Pages

  • Re: [linux-pm] Power Management framework proposal
    ... if they are why should this new API need to worry about that? ... clocks seems like it fits well into the model) becouse some power modes ... the onenand driver needs to adjust its timing parameters ... the software needs to know what the interconnects and dependancies ...
    (Linux-Kernel)
  • Re: exception following PowerOffSystem() call in power manager PDD
    ... the exception is inside PowerOffSystem. ... the driver's power callback is called. ... So, Most likely, it is device driver ... power callback calling API. ...
    (microsoft.public.windowsce.platbuilder)
  • Re: [PATCH v4 3/3] MFD MAX8998/LP3974: Support Charger
    ... current and turn on and off the charger. ... It's probably better to split the power supply driver into a separate ... One thing that jumps out here is that there's no regulator API usage at ...
    (Linux-Kernel)
  • Re: Nvidia MCP55 Machine reboots on ixgb driver load
    ... There are some serious hardware compatibility issues with the ixgb mixing it with other cards on the same PCI-X bus, ... loading the driver with debug does not appear to produce ... 02:00.0 PCI bridge: NEC Corporation uPD720400 PCI Express - PCI/PCI-X Bridge ... ACPI: Power Button ...
    (Linux-Kernel)
  • Re: [RFC] Staging:IIO: New ABI
    ... Hardware chip or device accessed by on communication port. ... An event driven driver of data capture to an in kernel buffer. ... There is already a well-defined api for these things. ... don't have a custom interface just for these device. ...
    (Linux-Kernel)