Re: DS-207 network failure
Frank Wille <[email protected]> Thu, 26 May 2011 10:56:11 +0200
| Newsgroups | gmane.os.netbsd.ports.sandpoint |
|---|---|
| Organization | Privates Internet Ostwestfalen/Lippe |
| Message-ID | <[email protected]> |
On Wed, 25 May 2011 23:50:49 +0200 Oscar Molin <[email protected]> wrote: > If you have any time to spare, download synologys sourcecode and look > at the SK98 driver they use for your model. The GPL source can be found here: http://gpl.nas-central.org/SYNOLOGY/DS-101g+/ and here: http://gpl.nas-central.org/SYNOLOGY/x07-series/ > There has to be a difference between the altboot SK driver and > synologys SK98 driver that causes this. Yes, I agree. Your NIC is working under Linux and PPCBoot so it cannot be defective. Unfortunately the PPCBoot SK98 driver is hard to read. They copied the Linux driver source and uncommented all the code which isn't needed. > The xmit failed message is written by the altboot SK driver, you can > see this in the sourcecode which isn't too long but not commented in > detail. As with most network chips you have to set up a descriptor (struct desc) which contains information about the frame to be transfered, as for example the 64-bit buffer address in the third and fourth word (xd2/xd3). Little endian, so we set xd3 to NULL. The first 32-bit word (xd0) contains the frame length and some flags. It is the first and last segment at the same time, so CTL_FS and CTL_LS are set. Most important flag is CTL_OWN, which tells the chip that it "owns" the descriptor now and can do something with it. Now we just wait until the chip is finished with our descriptor and resets CTL_OWN to zero. This does never happen in your case. > I have setup a machine for compiling altboot but I havn't had the time > yet to try any changes myself and it's hard to get into it when you > can't find any detailed datasheets anymore. I had the same problem, so I used the NetBSD driver as reference. AFAIK the datasheets have never been available for public use. > Still, it feels like our problem is just a minor obstacle so maybe we > can get this sorted out. I hope so. Any help is appreciated. -- Frank Wille