On 23/12/2007, at 5:34 AM, Alan DuBoff wrote: > On Sun, 23 Dec 2007, David Gwynne wrote: > >> On 22/12/2007, at 5:51 PM, Alan DuBoff wrote: >>> >>> If there is something specific you would like to know, I can try to >>> get some information for you, it is my manager that is responsible >>> for the LSI relationship. >> >> Has OpenSolaris inherited Suns relationship with LSI? > > I can tell you what I know and don't know, don't shoot me I'm just > the messenger...(as they say;-). > > I'm not sure I would say that OpenSolaris is inheriting Sun's > relationship, because I don't know the details of their > relationship. Sun has been working with several vendors to open > source their drivers, and Intel for instance has taken proprietary > drivers such as the e1000g and moved it into OpenSolaris, so the > driver moved recentely from the closed binaries to it's rightful > home in uts.
I read this as "Sun is a very active contributer to OpenSolaris". I know Sun is working hard to get more and more code opened up for inclusion in the OpenSolaris code base. In this situation you explain it as Sun is trying to negotiate with LSI to have the megasas driver supplied. If the source code comes with that, then Sun will pass it on to OpenSolaris. However, isn't that activity separate to the OpenSolaris community choosing to integrate something else if they want to? > I am not sure if the LSI driver will be open sourced, and my gut > tells me that it will not, but that is something I would need to > check on. I *suspect* that the driver is not open, since it's been > playing ping-pong between the lawyers for some time, and normally > the open source drivers don't take so long, it's typically a matter > of getting the contributer agreement signed. > > I'm not clear on the chipsets, but from my a recent staff meeting I > see the LSI 1078 chipset as being in a ping-pong game between both > Sun's and LSI lawyers, but it would appear that it's nearing closure. > > What I am not clear on and will be up front with you, is that I > don't now know if you have a different chipset than the 1078, and > the megasas driver includes multiple chipsets, so my understanding > is that it consolidates more than one chipset into the driver > possibly. I'm not absolutely clear on that though. > However, I would like to point out that if this point is true, it > will be almost impossible to get a driver cleared with partial > support of this driver, in the case that it doesn't support all the > devices that the megasas driver is being developed to support. There's a second generation of the megaraid sas boards that mfi doesn't support, but the changes required to support it are trivial. If someone wants to supply me with hardware I bet I could fix mfi in a few hours. > Here's a page I googled for, on LSI which talks some about MegaRAID > SAS/SATA. There are various documents on this page, not sure if it > will help or confuse the issue. > > http://www.lsi.com/storage_home/products_home/internal_raid/megaraid_sas/megaraid_sas_8888elp/index.html > > Lastly I would like to point out again that Sun has had > relationships ongoing with many of the vendors that pre-date > OpenSolaris. Some of that is business as usual, IOW, things haven't > changed much. In other cases things have changed alot, ala Intel. > Even if LSI does open source this megasas driver, I don't think that > means that Sun is not working with them on other proprietary work, > since Sun does use LSI chipsets in the systems they ship, AFAIK. But > the trend, and more so Sun's intentions, are to open up as much as > they can, and has been encouraging their IHV/OEM partners to do the > same. In general it seems the response is pretty positive from most > vendors, I just do not know where we stand with LSI. > > I'll try to get more details, but most everyone (including my > manager who is responsible for the LSI relationship) is on break and > off until Jan. 7, 2008. > > -- > > Alan DuBoff - Solaris x86 IHV/OEM Group
