[Ucip-devel] Re: Dequeueing bug in nettcp.c (was: Fixed Bug in UCIP)

[Robert D. <od...@pn...>]

Guy and Alberto,

I've looked at the code in question and found that I have=20
definately made a very stupid mistake in performing the=20
manual macro expansion. I have corrected this and will=20
update the CVS repository shortly.

My apologies to Alberto and his team for creating so much=20
needless work to get the source working and thanks for=20
submitting the bug report.

Thanks and regards,

Robert Dickenson.


At 09:51 AM 10/01/2002 -0800, you wrote:
>Alberto Perdomo wrote:
>
>> Hi Guy.
>>
>> We=B4re been working with the UCIP stack. I=B4m porting it to a
>> DSP platform and made some corrections to code. We had
>> several problems. The hardest was, that the TCP layer
>> didn=B4t work, every time a packet was received there was an
>> exception. We found the error. It=B4s located in the file
>> nettcp.c, function tcpReadJiffy. It seems that you mixed up
>> two lines:
>>
>> this is how it was in the original code:
>> tcb->rcvBuf =3D tcb->rcvq.qHead->nextChain;
>> tcb->rcvq.qHead =3D tcb->rcvBuf;
>>
>> this is how it really works:
>> tcb->rcvBuf =3D tcb->rcvq.qHead;
>> tcb->rcvq.qHead =3D tcb->rcvBuf->nextChain;
>>
>> so, that=B4s it. it really took us a few days to find it,
>> because we didn=B4t know the code good, but we managed...
>>
>> yours sincerely,
>>
>> alberto perdomo.
>
>  Thanks Alberto.
>
>  Robert, can you look in to this?  It looks like you replaced
>nDEQUEUE() probably so you could trace it outside of the macro
>but it's going to drop the head chain of the queue.  Can we
>just go back to using nDEQUEUE()?
>
>  Thanks.
>
>            Guy
>--
>"Embedded and Open Source Solutions"
>Guy Lancaster Computer Consultants
>mailto:gu...@gu...
>http://guylancaster.com
>
>
>
>_______________________________________________
>Ucip-devel mailing list
>Uci...@li...
>https://lists.sourceforge.net/lists/listinfo/ucip-devel
>
>