Re: Question about IoMarkIrpPending
From: Jackal Huang (huangjj_at_hotmail.com)
Date: 02/10/04
- Next message: David J. Craig: "Re: SoftIce crashes with Windows 2000 SP4"
- Previous message: Alex: "Re: SoftIce crashes with Windows 2000 SP4"
- In reply to: Doron Holan [MS]: "Re: Question about IoMarkIrpPending"
- Messages sorted by: [ date ] [ thread ]
Date: Tue, 10 Feb 2004 09:49:33 +0800
Sorry! I am confused for your description here.
In the XPDDK usb bulk samplecode, the power dispatch routine always
calls IoMarkIrpPending first , calls HandleDeviceSetPower() and then return
pending status for the device set power case of IRP_MN_SET_POWER.
HandleDeviceSetPower() set an IoComplete routine(FinishDevPoUpIrp)
for power on case. Therefore, it did has a completion routine for the
pending
Irp here. FinishDevPoUpIrp calls IoMarkIrpPending if Irp->PendingReturned
is set. Here is my question.
Since the power dispatch routine has called IoMarkIrpPending, its stack
location should has been set SL_PENDING_RETURNED. Why does the
IoComplete routine needs to call IoMarkIrpPending again?
Best Regards
Jacakal Huang
"Doron Holan [MS]" <doronh@nospam.microsoft.com> ...
> i think there is a bit of confusion here. if you mark an irp pending and
> then queue it up and then compolete it later, there is no completion
> routine. you are the owner of the irp and you have pended it. in the
> sample, the driver is sending the irp down the stack. in this case, you
need
> to propagate the pending status up the stack. the pending status of the
irp
> is set on a per stack location basis and this is a cooperative effort.
all
> drivers in the stack must play by the rules (verifier helps here). the
only
> case where you don't have to propagate the pending bit when you send an
irp
> down the stack is if you are synchronously waiting for the irp to complete
> and then are completing the irp in the dispatch routine.
>
> d
>
> --
> This posting is provided "AS IS" with no warranties, and confers no
rights.
> Please reply to newsgroups only.
>
> "Jackal Huang" <huangjj@hotmail.com> wrote in message
> news:uCCv93v7DHA.632@TK2MSFTNGP12.phx.gbl...
> > The rule of calling IoMarkIrpPending has suffered me for a long time.
> > After studied Walter's WDM book 2nd, I understand it more now, but
> > still have some questions.
> >
> > If the dispatch routine has called IoMarkIrpPending and return pending
> > status, does the IoComplete routine still need to check Irp->
> > PendingReturned and call IoMarkIrpPending if the flag is set?
> >
> > According to my understanding, it isn't necessary to call
IoMarkIrpPending
> > in the IoComplete routine since SL_PENDING_RETURNED has been set
> > in the stack location when the dispatch routine called IoMarkIrpPending.
> >
> > In the power management of XPDDK usb bulk sample code, it always
> > calls IoMarkIrpPending and returns pending status for
> > IRP_MN_SET_POWER in its power dispatch routine. In the complete
> > routine of power on case (FinishDevPoUpIrp), it calls IoMarkIrpPending
> > if Irp->PendingReturned is set. Is it a redundant code segment??
> >
> > Please tell me if my understanding is wrong.
> >
> > Best Regards
> >
> > Jackal Huang
> >
> >
>
>
- Next message: David J. Craig: "Re: SoftIce crashes with Windows 2000 SP4"
- Previous message: Alex: "Re: SoftIce crashes with Windows 2000 SP4"
- In reply to: Doron Holan [MS]: "Re: Question about IoMarkIrpPending"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
