[Openib-windows] RE: Anything new about the IPOIB arp check-in?

Tzachi Dar tzachid at mellanox.co.il
Wed Dec 7 13:26:22 PST 2005


Hi, fab

One minor mistake that was made to this check-in:
(checking for the wrong size: IOCTL_IBAT_PORTS_IN instead of
IOCTL_IBAT_MAC_TO_GID_IN).

Thanks
Tzachi

Index: ipoib_ibat.c
===================================================================
--- ipoib_ibat.c	(revision 806)
+++ ipoib_ibat.c	(working copy)
@@ -253,14 +253,14 @@
 	IPOIB_ENTER(IPOIB_DBG_IOCTL);
 
 	if( pIoStack->Parameters.DeviceIoControl.InputBufferLength !=
-		sizeof(IOCTL_IBAT_PORTS_IN) )
+		sizeof(IOCTL_IBAT_MAC_TO_GID_IN) )
 	{
 		IPOIB_TRACE_EXIT( IPOIB_DBG_ERROR, ("Invalid input
buffer size.\n") );
 		return STATUS_INVALID_PARAMETER;
 	}
 	
 	if( pIoStack->Parameters.DeviceIoControl.OutputBufferLength !=
-		sizeof(IOCTL_IBAT_PORTS_OUT) )
+		sizeof(IOCTL_IBAT_MAC_TO_GID_OUT) )
 	{
 		IPOIB_TRACE_EXIT( IPOIB_DBG_ERROR, ("Invalid output
buffer size.\n") );
 		return STATUS_INVALID_PARAMETER;

>-----Original Message-----
>From: Tzachi Dar 
>Sent: Wednesday, December 07, 2005 12:58 AM
>To: 'Fab Tillier'; openib-windows at openib.org
>Cc: Gilad Shainer
>Subject: RE: [Openib-windows] RE: Anything new about the IPOIB 
>arp check-in?
>
>
>
>
>>-----Original Message-----
>>From: Fab Tillier [mailto:ftillier at silverstorm.com]
>>Sent: Tuesday, December 06, 2005 11:47 PM
>>To: Tzachi Dar; openib-windows at openib.org
>>Cc: Gilad Shainer
>>Subject: RE: [Openib-windows] RE: Anything new about the IPOIB 
>>arp check-in?
>>
>>
>>> From: Tzachi Dar [mailto:tzachid at mellanox.co.il]
>>> Sent: Monday, December 05, 2005 1:59 PM
>>> 
>>> I have looked at the code that you have sent and things look great.
>>> 
>>> Thanks for the good job.
>>
>>Thanks Tzachi!  Sorry it took so long to get done, and thanks
>>for putting together the initial implementation - it helped greatly.
>>
>>> Please check-in the patch. (Please also check-in the test program).
>>
>>Done, as revision 195.
>>
>Thanks a lot.
>
>
>>I'll be converting WSD to use this instead of DAPL ATS, and
>>will be making DAPL ATS support a parameter option in IPoIB.  
>>The default will be "disabled".  Do you think it should be 
>>enabled for a few releases to allow interop (I don't think 
>>anything else in WSD changed)?  I think I'd rather see a 
>>consistent driver level throughout a cluster myself, so 
>>interop with prior versions isn't a high priority.
>>
>>- Fab
>>
>I believe that there is no need to keep interoperability at 
>this stage, Let's wait for one more day to see if there are 
>other opinions.
>
>Thanks
>Tzachi
>



More information about the ofw mailing list