Hi Stan First of all I want to apology about the slow response.
Mellanox is fully committed to the community. It is important that we work in cooperation to supply open source code for Infiniband stack. As you know the life isn't ideal and it's a question of resources and priorities. in the last period we are very busy with customer issues. I suggest we have a 1x1 to discuss how we can progress and ensure we release a high quality product, with good testing and supportability. Can you do it tomorrow evening? Uri -----Original Message----- From: Smith, Stan [mailto:[email protected]] Sent: Tuesday, September 20, 2011 7:38 PM To: Uri Habusha; Leonid Keller Cc: [email protected] Subject: RE: status on FDR support SVN commits? Hello Uri, Please see inline comments. >-----Original Message----- >From: Uri Habusha [mailto:[email protected]] >Sent: Monday, September 19, 2011 12:21 PM >To: Smith, Stan; Leonid Keller >Cc: [email protected] >Subject: RE: status on FDR support SVN commits? > >Hi Stan, > >Leonid already completed the changes for FDR in the low level driver. >However since the bus driver completely changed we need to see how we can >propagate the changes to OFW svn. Leo suggested the FDR patches were a small set and independent of ROCE, hence could be committed independent of ROCE. Has something changed? > >Is it okay if we propagate it as part of the changes for RoCE branch? Why is it now you choose to ignore the standard open source policy of working with the OFA-Windows community prior to commits? Because you stated ROCE changes were so wide spread that making patches was too time consuming, the WWG agreed to review a ROCE code drop. Yes you made a ROCE code drop into gen1\branches and it's nearly impossible to determine what is old code and what are ROCE patches; many files are untouched. No written ROCE insights to be located in the drop. Your code drop was based on a SVN snapshot taken many months ago....hence the problem understanding what are ROCE changes and what is old/outdated code. Two separate times I have asked Mellanox to clarify the ROCE source drop; no response. At this late date, you are now requesting you dump non-reviewed ROCE code into SVN, which only just recently it was revealed that the winOFED stack would not load without your binary-only Ethernet driver. Has something changed in the mlx4_bus driver such that there no longer a bus driver dependency on your binary-only Ethernet driver? We have asked you two separate times to participate in engineering a solution which allows the OFED stack to load without your binary only Ethernet driver; received no input. Why is it you choose to ignore OFA-windows community input? You behave as if open source participation has no relevance to you. > >By the way do you know if there is any plan to propagate the SM >FDR/FDR10 changes to windows tree. If yes, when it will be done Opensm 3.3.11 is in branches\OpenSMX; work-in-progress: configures the fabric correctly, DAPL and IPoIB tests run correctly, some osmtests pass, MultiCast tests fail. Should be completed by the end of the week. Stan. > >Thanks Uri > >-----Original Message----- >From: [email protected] >[mailto:[email protected]] On Behalf Of Smith, Stan >Sent: Monday, September 19, 2011 7:13 PM >To: Leonid Keller >Cc: [email protected] >Subject: [ofw] status on FDR support SVN commits? > >Hi Leo, > In the last WWG meeting you stated FDR support could be committed by 9/15. >What's the estimated time of arrival for those patches to be committed? > >Thanks, > >Stan. >_______________________________________________ >ofw mailing list >[email protected] >http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ofw _______________________________________________ ofw mailing list [email protected] http://lists.openfabrics.org/cgi-bin/mailman/listinfo/ofw
