I have a SCSI HBA serving as the boot controller in a dual processor
hyper-threaded system running WinXP x64 edition. Upon generating a crash via
Crash-On-Ctrl-Scroll or the HCT crashdumptest, I have seen proof via WinDbg
that the runtime (miniport) driver is still running when the crashdump
driver is being loaded. Specifically, I have seen an interrupt generated
during crashdump initialization get serviced by the runtime driver. I have
also seen the one second timer I have continue to run on the runtime driver.
No need to explain what a driver collision like this can do to the hardware.
If I disable hyper-threading and one of the processors, the problem goes
away. Why isn't crashdump processing disabling the other processors?

Thanks

Scott

Re: Windows XP x64 crashdump problem by James

James
Wed Aug 10 16:04:07 CDT 2005

"Being loaded" isn't the same as having control. Obviously, 2 instances of a
miniport shouldn't have control concurrently. So please clarify that point.

Second, are you saying that the original miniport instance "saw" the timer
expiration (eg, in a DPC) after the second instance got control at least
once? Are you sure that what got control was the first instance, namely, at
an address in the executable of the first instance?

--
James Antognini
Windows DDK and WDK Support


This posting is provided "AS IS" with no warranties, and confers no rights.



"Scott" <sschmitz@nospam.nospam> wrote in message
news:%23C5zXYdmFHA.3300@TK2MSFTNGP15.phx.gbl...
>I have a SCSI HBA serving as the boot controller in a dual processor
>hyper-threaded system running WinXP x64 edition. Upon generating a crash
>via Crash-On-Ctrl-Scroll or the HCT crashdumptest, I have seen proof via
>WinDbg that the runtime (miniport) driver is still running when the
>crashdump driver is being loaded. Specifically, I have seen an interrupt
>generated during crashdump initialization get serviced by the runtime
>driver. I have also seen the one second timer I have continue to run on the
>runtime driver. No need to explain what a driver collision like this can do
>to the hardware. If I disable hyper-threading and one of the processors,
>the problem goes away. Why isn't crashdump processing disabling the other
>processors?
>
> Thanks
>
> Scott
>
>