On Wed, 11 Jun 2008, Garrett D'Amore wrote: > So maybe instead of a gazillion different CRs, you could just do a single CR > saying "incorporate changes from upstream supplier". I'd still like to have > some explanation in that CR for what all the changes are for, but you can > coordinate with Masa to get the details.
Essentially it is just that. As you might know, there's about 6 months of work that has been going into the codebase. During that time there have been several problems with the test suites, and either things that broke fixing something else, or similar. > (Of course, listing all the CRs individually is even better, but I do > recognize it is potentially a fair bit more effort.) To be honest, it wasn't that at all...we started out failing on test08 on sparc. This was in part due to some of the networking that was taking place in nevada also, and this work has been going on since putback. While I knew that sparc was a problem, I thought it made sense to putback the fix for the auto-negotiate problem, it effects quite a number of folks. But I was told that it would be best to fix it before doing the putback. In a way I feel guilty of code creep, not intentional, but just through the course of evolution in fixing things, we have not been able to issolate the final problem that was found, until just last week, see CR# 6713612. To add fuel to all of this fire, the nicdrv test suites were going through quite a few changes, some which I myself did not and still do not feel were substantiated. In fact, I believe that the test suites should go through some type of ARC so that we know that the changes being made are solid changes. As it is, there is no guidelines for nicdrv, and it seems to have been a moving target. This is an issue in itself, but just pointing this out as the nicdrv suites were changing but we weren't getting updates on a regular basis and fighting some problems with what we were using. Imagine how I felt when I saw the nicdrv test suites released and there was a fix for a timeout in test08...hmmm...that sounds familiar and what brought us in this mess to begin with... Sorry for the long reply, this has consumed a lot of time on both mine and Masa's part. I also tip my hat to Masa, he is running nicdrv himself, and is able to test drivers on his own, has made his code completely cstyle and line compliant (a huge improvement when it integrated), and along with your afe/mxfe drivers, one of the most substantial contributions from the OpenSolaris community to date. A big thanks to him for all of that. -- Alan DuBoff - Solaris x86 IHV/OEM Group _______________________________________________ driver-discuss mailing list driver-discuss@opensolaris.org http://mail.opensolaris.org/mailman/listinfo/driver-discuss
