George White ??: > Michael, > > What is the specific fix required for nv105 and later? I'm currently > running nv104, but will want to upgrade. What is the crossbow putback? > http://opensolaris.org/os/project/crossbow PSARC/2006/357 Crossbow - Network Virtualization and Resource Management CR6498311 Crossbow - Network Virtualization and Resource Management You can refer to the changes be made in other native WiFi driver to modify if_ndis.c, you also need to find out mac_flow.h, mac_provider.h from latest gate to build bcmndis module. - Michael > Gorden, > > What is the other 75%? I've noticed some issues with switching between > networks and power cycling the receiver power switch, but it seems to > be nwam related. If I switch it over to the hard wired connection and > then switch it back, it seems to restart OK. I've also noticed issues > with it asking for a password/security key, even though a valid key was > already present in the secobj file. There's also this mysterious > message about 'switching AP' that pops up from time to time, generally > when I'm witching between networks. > > George > >> Hello George and Gordon, >> >> Gordon Marler ??: >> >>>> I noticed the same thing after I sent the ndis1.2.1 >>>> for posting on the ndis page. I have >>>> since found and fixed several bugs in ndisapi. The >>>> version of ndis-1.2.1 on my >>>> web site contains a new version 1.2.1 of the ndisapi >>>> as well as the new version of ndiswrapper. This >>>> should be much more solid. >>>> >>>> http://www.palisad.com/solaris/ndis/ndis-1.2.1.tar.gz >>>> >>>> George >>>> >>>> >>> As you suspected, I was using ndis-1.2.1 (from the Wireless page) already. >>> > The 1.2.1 version from palisad.com is different, and that improves things a > bit. > >>> PowerTop reports that ndisapi`ntoskrnl_timercall is the source of wakeups 7 >>> - >>> > 20% of the time now, rather than 98% of the time with the "old" 1.2.1 version. > >>> >>> >> even with "old" 1.2.1, the powertop reports ndisapi`ntoskrnl_timercall >> is always less than 10% of total wekups, since my test machine has 4 >> cores inside, dtrace shows the ntoskrnl_timercall is dispatched to 4 >> cores with a load balance manner... that's the reason I didn't notice it >> before. Thanks for the fix. >> >> Went through the codes changes in "new" 1.2.1, I think the fix in >> ndisapi is valid, especially the fix for KeSetTimerEx, somehow, >> FreeBSD/NDIS still take period as in 100ns unit, but I agree with >> george's fix since it follows MSDN very well. :-) >> >> Now, "new" ndis-1.2.1 was synced to OpenSolaris/NDIS page. >> Note: v1.2.1 can ony work within build100 - 104, minor changes need to >> be added in if_ndis.c due to crossbow putback after build105. >> >>> The wireless is still about 25% of what it is under Fedora 10 - but it's >>> > certainly usable enough to get what I need to get done - plus NWAM isn't > going > crazy anymore. > >>> Looking forward to your next batch of improvements. >>> >>> >> me too! >> >>> In case I missed it, is using NDIS just a stopgap measure for Broadcom >>> > wireless chipsets, or can we expect to see "native" OpenSolaris drivers in > the > foreseeable future? > >>> >>> >> The wl driver for BCM4312(Broadcom product name, not PCI device id) in >> Fedora 10 was released by Broadcom since July,2008, it is under GPL >> license for now. It seems to me we can expect a "native" OpenSolaris >> driver for BCM43xx in the future... Thanks for your reminding, Gordon. >> >> - Michael >> > >
-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://mail.opensolaris.org/pipermail/laptop-discuss/attachments/20081223/152ebaf2/attachment.html>
