Hi Ivan, Please go to Ted You's mail for more information. Maybe we should involve Intel for help about this.
Thanks Chenlu On 09/18/08 12:02, Ivan Wang wrote: > Hi Chenlu, > > I incline to agree with Aubrey although I am not affected by this, pci > device id 0x10be should be supported as a backward compatibility > workaround. > Other OS, BIOS, drivers do this all the time. for example, many linux > driver has device id specific even revision specific code. > Yes, it is indeed a ugly hack and logically correct way is to follow > Intel's decision. However it's the reality that there are chips with > old device id out there. > > In the past Solaris is blamed for the limited device compatibility, > one of the reason is the driver often support strictly one or two > version(s) of the hardware. > But ecosystem in PC (x86) world is hardware vendor provides Windows > driver so they have the habit to make changes to hardware along the > way. > Other OS have no way but to workaround this habit, otherwise there's a > great chance it won't work on untested systems. > > Since e1000g has capability to drive the chip with old device ID, do > you think it is possible to add 0x10be back and put a note in man page > to address the > device id change? > > Thanks. > Ivan. > > >> On Thu, Sep 18, 2008 at 8:53 AM, chenlu chen - Sun Microsystems - >> Beijing China <[EMAIL PROTECTED]> wrote: >> >>> Hi Aubrey, >>> >>> This is because the device id of 0x10be is changed to 0x10f5 since e1000g >>> Ver 5.2.8 (CR6687947.). >>> So this is not a bug of e1000g. >>> >>> Because of some reasons Intel has decided to retire id 10be entirely and >>> 0x10f5 is the correct one. >>> But some systems did go out into the world with the 10be device id. >>> E1000g driver since 5.2.8 will not work on this chip 0x10be. >>> As Intel's suggestion, the right thing to do is upgrading the NVM image. >>> >>> Milan, >>> Thank you for forwarding this mail. >>> >>> Thanks >>> Chenlu >>> >>> >> That is probably not the right way. >> That means network will be missed on some systems. >> You know, on ACPI related work, we are keeping fix weird >> BIOS bug to make solaris work on the more platforms. I guess it's not >> difficult for e1000 driver to support this device id, right? I >> personally strongly >> suggest e1000 driver support this device id, otherwise I have to buy a new >> network card, :-) You know, updating NVM image is painful for the common >> users. >> >> Thanks, >> -Aubrey >> >> _______________________________________________ indiana-discuss mailing list [email protected] http://mail.opensolaris.org/mailman/listinfo/indiana-discuss
