On (01/31/08 09:15), Garrett D'Amore wrote: > > Out of curiousity, why isn't link state recorded for links other than > DL_ETHER? For WiFi it could be recorded, although it has a more complex > link state mechanism (down, associated, authenticated, up).
As I was mentioning in my mail to Seb, there's a separate effort to be undertaken, to get the wifi drivers to export public props. At that time, we'll be converting more of these to ANYMEDIATYPE. > For DLD_PROP_SPEED, I thought we agreed that speed would be in "bps" not > "Mbps". (linkprop.c:201) (Again, recall that 802.11b can have a speed of > 5.5Mbps.) correct, that's a typo. on that line that I will fix. BTW, the kstat itself is printed in dladm after being scaled down to Mbps for commands like 'dladm show-phys', 'dladm show-dev' etc. so the loss of precision is more wide-spread than just this one command (and I believe that should be addressed as a separate bug, so I'm not going to pull that into this fix). > linkprop.c: Link 1259, the comment would be clearer as "we currently cannot > set the rate on wifi links" (double negative). Actually the correct comment would be "we can only set the rate on wifi links" (On ether links, you have to set both speed and duplex in one shot, so speed is a read-only prop). I'll fix this. > Can we really set the rate > on anything other than DL_ETHER? Maybe an explicit DL_ETHER test would be > better? See Seb's review comments from yesterday [#1]. > > bge_main2.c: You removed the ENOTSUP fix? Why? As I mentioned yesterday [#2] Crisson Fu is going to work on this, so I'll work out the details with him. [#1] ttp://mail.opensolaris.org/pipermail/brussels-dev/2008-January/000666.html [#2] http://mail.opensolaris.org/pipermail/brussels-dev/2008-January/000664.html --Sowmini
