On Tue, Jul 27, 2010 at 4:42 PM, Andrew Lutomirski <luto@xxxxxxx> wrote: > On Tue, Jul 27, 2010 at 7:26 PM, Luis R. Rodriguez <mcgrof@xxxxxxxxx> wrote: >> On Tue, Jul 27, 2010 at 4:24 PM, Andrew Lutomirski <luto@xxxxxxx> wrote: >>> On Mon, Jul 26, 2010 at 3:48 PM, Andrew Lutomirski <luto@xxxxxxx> wrote: >>>>> In the attached log, you'll see the one successful connection to >>>>> 00:02:6f:6e:d7:6c as well as a failure. The failure is just: >>>>> >>>>> [ 668.569218] wlan0: authenticate with 00:02:6f:6e:d7:6c (try 1) >>>>> [ 668.769269] wlan0: authenticate with 00:02:6f:6e:d7:6c (try 2) >>>>> [ 668.969150] wlan0: authenticate with 00:02:6f:6e:d7:6c (try 3) >>>>> [ 669.169262] wlan0: authentication with 00:02:6f:6e:d7:6c timed out >>>> >>>> This time with actual attachment. >>> >>> Not sure if this is the same issue, but I've got another AP here >>> (Ubiquiti RouterStation Pro + R52N ath9k minipci card running latest >>> OpenWRT "backfire") that works quite nicely on 2.6.34 but requires me >>> to reconnect every minute or so on 2.6.35-rc6+ to keep it working. >> >> Define 2.6.35-rc6+ >> >> If you are using wireless-testing there was a regression there that I >> just sent patches for. > > It's just Linus' kernel (specifically > 6aa033d7efb85830535bb83cf6713d6025ae6e59) plus a few local i915 > patches. (I also have my aggregation failure warning patch in there, > but that just changes a couple of printks.) If you can test vanilla Linus and post a bug report that would help for sure. I do not think we are aware of existing pending issues on 2.6.35, and in fact the regression list often posted by Rafael shows 0 regressions for 2.6.34 and 2.6.35-rc so far so you're issue is certainly not reported until now, but we need more details, like chipset, how to reproduce, etc. >> Reporting issues is the way to go which BTW i have yet to get to your >> other e-mail, sorry just got back from vacation and still catching up. > > No problem. (I didn't think you worked on iwlwifi anyway...) No, but if its regulatory yes. >>> I'm getting tempted to just revert the whole driver back to 2.6.34 and >>> use the rest of 2.6.35 when it comes out. I can also try to bisect, >>> either the normal way or just the iwlwifi driver, but that'll have to >>> wait until the Monday after next at least because I'm leaving town. >>> If any of you have any ideas to test *today*, I can do it. > > Do you know how hard it is to bisect compat-wireless? Sure by date, it also has the specific dates it pulls stuff from linux-next.git so you can use that as reference. > Bisecting just > the iwlwifi directory is a bit of a pain because I have to manually > fix up everything that doesn't compile, and it's also a pain to bisect > the whole tree because I need some local i915 patches to get a working > system. (Also, IIRC, 2.6.35 was very broken around -rc2 or -rc3.) I would just say to use wireless-testing directly and bisect that, I did that yesterday night. You bisect wireless-testing by the ^master-$(date -I) tags. A little of a pain but its not so bad. As John noted also you can just use his wireless-next.git too and you should be able to bisect that regularly. I understand your requirements for i915 changes.. but maybe you can test with vesa in the meantime? Luis -- To unsubscribe from this list: send the line "unsubscribe linux-wireless" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
