[Xitami] SMP Problem (WAS: Xitami 2.4d9/Xitami 2.5b5)

[Dr. Martin R. Hadam]

On Mon, 11 Jun 2001 00:54:48 +1200, Ewen McNeill wrote:

 I've done some deep thinking <g> and then tested one thing I'd
apparently  forgotten (shame on me!): since the problem [see below] is
occuring on an SMP machine, I've disabled the use of both processors
using execmode and voilą - Xitami is now perfectly stable again. In
case you're not familiar with execmode: this IBM utility patches an
exe/dll to only make use of one processor in an smp system. So it seems
that there is a serious new problem with threading in Xitami versions
>2.4d6 [OS/2]. This has been tested / verified on two different machines running OS/2 Warp Server SMP at FP 42/43 resp. The test version I used with execmode was the 2.5b4 patch #2 (Bernd Hohmann).

 To reiterate the problem - since its description is now some time ago:
Xitami starts and quits after a while without any further error message
(other than the one below).

 Hope that this will allow you to better trace the problem.
 
 Martin


>In message <E158S7R-0000Ee-00@Mail.MH-Hannover.DE>, "Dr. Martin R. Hadam" writes:
>> I've been moving machines and today, Xitami provided some message, for
>>the first time:
>>2001/06/08 21:32:16: smthttp: HTTP service is now running
>>Fatal error in application - aborted
>>Abort at smtsock:Check-For-Expired-Requests: (Getting-Event, Empty-Event)
>>Abnormal program termination
>
>Ah, that sort of problem.  You might want to keep an eye on whether it's
>always at the same point, or whether the problems shift around.  If they
>shift around it's most likely that something is writing to memory it
>shouldn't.  If it's always (or usually) one place, then possibly there's
>something wrong with the way that some requests are handled by smtsock.
>Unfortunately either way they're difficult to diagnose problems, since
>they're not easily reproducable.


 Martin R. Hadam
 Kinderklinik - Medizinische Hochschule
 D-30623 Hannover
 Germany
 Email: Hadam.Martin@MH-Hannover.de