Doron
Wed Oct 17 10:38:44 PDT 2007
the handle can be opened on a child device below usbstor.
d
--
Please do not send e-mail directly to this alias. this alias is for
newsgroup purposes only.
This posting is provided "AS IS" with no warranties, and confers no rights.
"Abhijit Mmirajkar" <abhijit.mirajkar@gmail.com> wrote in message
news:1192633263.357864.95700@k35g2000prh.googlegroups.com...
> Thanks Doron, but the !object also reveals that there are no open
> handles ! There is some entry under 'PointerCount', but that number is
> also there when it successfully unloads when no file transfer is going
> on.
>
> The values for my device object are : HandleCount: 0 PointerCount: 3
> and for USBSTOR are : HandleCount: 0 PointerCount: 2
>
> I also experimented with plugging a mass storage device to a physical
> usb port on the same system and repeating the procedure. I noticed
> that there are 2 calls to "IoInvalidateDeviceRelations" when directly
> plugged into the usb port, one from the USBHUB (as expected) and other
> from ClassPnp, disk (as shown in the trace below).
> With my driver there's only one call, made from my virtual bus driver
> itseld.
>
> Does it mean that I need to register some interface or something
> similar by which system (ClassPnP or disk) will come to know of the
> change in topology.
>
> Here's the trace. Thanks for all the suggesstions.
>
> Breakpoint 0 hit
> nt!IoInvalidateDeviceRelations:
> 8050c9e1 8bff mov edi,edi
> 1: kd> kb
> ChildEBP RetAddr Args to Child
> f7a5ad30 f78103c5 853dc030 00000000 00000008 nt!
> IoInvalidateDeviceRelations
> f7a5ad50 f7816eeb f7a5ad74 00000004 8560f800 usbhub!
> USBH_ProcessPortStateChange+0x111
> f7a5ad7c 804e29d6 855d8ca8 00000000 857c63c8 usbhub!
> USBH_ChangeIndicationWorker+0x145
> f7a5adac 80576b24 855d8ca8 00000000 00000000 nt!ExpWorkerThread+0xef
> f7a5addc 804eed86 804e2912 00000001 00000000 nt!PspSystemThreadStartup
> +0x34
> 00000000 00000000 00000000 00000000 00000000 nt!KiThreadStartup+0x16
>
> Breakpoint 0 hit
> nt!IoInvalidateDeviceRelations:
> 8050c9e1 8bff mov edi,edi
> 0: kd> kb
> ChildEBP RetAddr Args to Child
> 80558064 f7630961 8549bea0 00000000 c0000185 nt!
> IoInvalidateDeviceRelations
> 80558080 f7641270 8510d420 855cdc6c 8510d1f0 disk!DiskFdoProcessError
> +0x15b
> 80558184 f7643bde 8510d420 855cdc6c 00000004 CLASSPNP!
> ClassInterpretSenseInfo+0x678
> 805581b4 f76426e0 855cdbc0 8510d2db 8510d1d8 CLASSPNP!
> InterpretTransferPacketError+0x106
> 805581d4 804e1f14 00000000 8510d1d8 855cdbc0 CLASSPNP!
> TransferPktComplete+0x67
> *** ERROR: Module load completed but symbols could not be loaded for
> IrpSys.sys
> 80558204 f63bdf97 850ef740 855cdc6c 00000000 nt!IopfCompleteRequest
> +0xa2
> WARNING: Stack unwind information not available. Following frames may
> be wrong.
> 80558220 f79e1043 850ef688 8510d1d8 f79df564 IrpSys+0x1f97
> 80558240 804e6691 850ef688 0110d1d8 8549bf58 USBSTOR!USBSTOR_StartIo
> +0x11d
> 80558260 f79dfaee 850ef688 8510d1d8 855cdc98 nt!IoStartPacket+0xa1
> 80558284 804e19ee 8549bea0 8510d1d8 00000000 USBSTOR!USBSTOR_Scsi
> +0x108
> 80558294 f63bde48 8510d2d8 855cdbc0 8549bea0 nt!IopfCallDriver+0x31
> 805582c0 f7640061 f7644f9c 80558308 00000000 IrpSys+0x1e48
> 805582d4 f7644fa9 855cdbc0 805583fc 804e31d4 CLASSPNP!
> SubmitTransferPacket+0x82
> 805582e0 804e31d4 855cdc08 855cdbc0 227e27c8 CLASSPNP!
> TransferPacketRetryTimerDpc+0xd
> 805583fc 804e269e 80562f00 ffdff9c0 ffdff000 nt!KiTimerListExpire
> +0x14b
> 80558428 804dd26b 80563300 00000000 00005d0d nt!KiTimerExpiration+0xb1
> 80558450 804dd14b 00000000 0000000e 00000000 nt!KiRetireDpcList+0x61
> 80558454 00000000 0000000e 00000000 00000000 nt!KiIdleLoop+0x28
>
>
> Regards,
> Abhijit
>
> On Oct 16, 9:39 pm, "Doron Holan [MSFT]" <dor...@online.microsoft.com>
> wrote:
>> as long as there is an open handle, you will not get the remove device
>> and
>> will be stuck in the surprise removed state. !object on your device
>> object
>> and usbstor will tell you the handle count
>>
>> d
>>
>> --
>> Please do not send e-mail directly to this alias. this alias is for
>> newsgroup purposes only.
>> This posting is provided "AS IS" with no warranties, and confers no
>> rights.
>>
>> "Abhijit Mmirajkar" <abhijit.miraj...@gmail.com> wrote in message
>>
>> news:1192547397.878393.138270@y27g2000pre.googlegroups.com...
>>
>>
>>
>> > Hi,
>>
>> > I have a typical problem of IRP_MN_REMOVE not getting called. The
>> > scenario
>> > is like this.
>>
>> > We have a virtual USB bus which creates PDOs for each simulated
>> > device. We
>> > could simulate a mass storage device. We can 'unplug' it (from an
>> > application) by means of our bus driver sending
>> > IoInvalidateDeviceRelations
>> > and then reporting the device as missing. We subsequently get
>> > REMOVE_DEVICE
>> > for the PDO and it successfully unloads.
>>
>> > However when we are copying some data to the device, if we 'unplug' it
>> > while
>> > still copy is in progress then we don't get the IRP_MN_REMOVE_DEVICE
>> > from
>> > the PNP manager.
>> > We see that there are no pending requests (IRPs) with us and we have
>> > successfully completed IRP_MN_SURPRISE_REMOVAL, still there is no
>> > REMOVE.
>>
>> > After dumping the devobj I get the following trace :
>>
>> > 0: kd> !devobj 852fe9f0
>> > Device object (852fe9f0) is for:
>> > 00000089
>> > \Driver\VUSB DriverObject 854cc858
>> > Current Irp 00000000 RefCount 0 Type 0000002a Flags 00003040
>> > Dacl e2430954 DevExt 852feaa8 DevObjExt 852fec08 DevNode 85538ee8
>> > ExtensionFlags (0x00000008) DOE_REMOVE_PROCESSED
>> > AttachedDevice (Upper) 8527b030 \Driver\USBSTOR
>> > Device queue is not busy.
>>
>> > This tells that there are no open references as well. I am also
>> > doubtful about "DOE_REMOVE_PROCESSED". Does this mean that system
>> > thinks that it has sent me IRP_MN_REMOVE_DEVICE ??
>>
>> > Any hints please?
>> > Thanks,
>>
>> > Regards,
>> > Abhijit- Hide quoted text -
>>
>> - Show quoted text -
>
>