Pete
Thu Dec 09 18:15:51 CST 2004
I'd like to know the exact cause of the problem. Say I disabled some driver
and the problem went away. This doesn't prove the disabled driver is faulty.
UDF writers are not supposed to service or disable interrupts. They are
expected to run at <=DispatchLevel. So if the problem is in a software
driver, ScrollLock should crash the system. I've made a simple test with a
recursively acquired spinlock - the PC locks up, but ScrollLock works.
I'm going to try the checked kernel, then hook the clock ISR.
"Pavel A." <pavel_a@NOwritemeNO.com> wrote in message
news:6A0F5C4E-E7E4-47BB-AC4C-3708DE775677@microsoft.com...
> Aha... this looks like problem in a UDF driver.
> Do you have 3rd party UDF writers (DLA, Roxio Drag to disc...)?
> --PA
>
> "Pete Johnston" wrote:
>> I can crash the machine when it's running with ScrollLock, and I can
>> break
>> in with a debugger. Yesterday I had WinDbg/1394 attached. The machine
>> locks
>> up after being restarted from hibernation, when I insert DVD+RW into
>> DVD-ROM. Inserting DVD+RW into DVD writers is ok. The machine did lock up
>> in
>> other circumstances, but the DVD+RW case is 100% reproducible. Neither
>> ScrollLock nor Ctrl+Break, SysRq, Alt+SysReq, Ctrl+C, F12 work on both
>> machines. The debugger cannot break in. I tried several times with no
>> success, with Verifier and without.
>>
>> The machine seems to be spinning since the CPU temperature is going up.
>>
>> Can the debugger break in when the target is spinning at DISPATCH_LEVEL?
>> Can
>> Verifier detect deadlocks at higher levels, such as ISR?
>>
>> What is the highest priority interrupt handler that is periodically
>> called
>> on Win2k3, if any? The system clock? Maybe I can hack the handler to call
>> KeBugCheck if, for instance, a global variable has not been updated for
>> awhile?
>>
>> Any other suggestions?
>>
>> Thanks!
>> Pete
>>
>> --
>>
>> "Doron Holan [MS]" <doronh@nospam.microsoft.com> wrote in message
>> news:uoPK40g3EHA.1152@TK2MSFTNGP14.phx.gbl...
>> > if crash on control + scroll lock doesn't work, either you have not set
>> > the setting properly (have you verified this crashes the machine when
>> > it
>> > is running?), or the keyboard interrupt isn't running when the machine
>> > locks up (which means spinning at high irql or a hw lockup). in these
>> > cases, getting the debugger to break in can sometimes be impossible.
>> >
>> > 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.
>> >
>> >
>> > "Pete Johnston" <anonymous@discussions.microsoft.com> wrote in message
>> > news:ujPg8zY3EHA.2540@TK2MSFTNGP09.phx.gbl...
>> >> I've got a dual Xeon W2k3 box that locks up intermittently. I'd like
>> >> to
>> >> debug
>> >> or crash it when it stops responding. Ctrl + double ScrollLock doesn't
>> >> work.
>> >> It's not convenient to have a debugger attached all the time. I wonder
>> >> if
>> >> the following will work.
>> >>
>> >> - I set /debug /debugport=1394 in boot.ini and boot the box
>> >> - Once the box has locked up, I connect a 1394 cable to a laptop with
>> >> WinDbg
>> >> - I use WinDbg to debug or crash the box
>> >>
>> >> What are my chances to have WinDbg connected to the target machine
>> >> when
>> >> the latter is stuck - ie spinning at a high IRQL?
>> >>
>> >> Can anyone suggest hardware-based solutions to crash a PC with memory
>> >> dump? Like a cheap PCI card, NIC or whatever, with two wires and a red
>> >> button soldered to it? Though I am afraid the problem is in video
>> >> drivers, so even if I managed to crash the computer, I won't get the
>> >> memory dump anyway.
>> >>
>> >> Thanks
>> >> Pete
>> >>
>> >> --
>> >>
>> >>
>> >>
>> >>
>> >
>> >
>>
>>
>>