**** BEGIN LOGGING AT Sat May 22 02:59:57 2010 May 22 04:20:10 [florian]: r21522's md5sum is incorrect May 22 04:42:29 [florian], pm May 22 07:53:47 <[florian]> Strykar: I saw it and answered you May 22 07:55:42 florian * r21534 /trunk/package/iptables/Makefile: [package] update iptables md5sum May 22 07:55:47 <[florian]> swalker: fixed May 22 08:11:59 [florian], I never recd. it May 22 09:16:39 seems I found evil issue with 2.6.34 kernel on rt3052, wrong kernel command line May 22 10:00:59 heh, now I have hang on Calibrating delay loop... May 22 10:24:37 Linux OpenWrt 2.6.34 #5 Sat May 22 14:22:16 MSD 2010 mips GNU/Linux. Success :) May 22 10:24:39 I've got it running May 22 10:26:52 <[florian]> well done :) May 22 10:27:39 let's see, will rt20xx wifi drivers from 2.6.34 work with rt3052 May 22 10:27:56 <[florian]> rtz2: \o/ May 22 10:28:11 [florian]: ? May 22 10:28:21 <[florian]> rtz2: I am glad you just showed up May 22 10:28:35 <[florian]> rtz2: I was playing a bit with the watchdog driver this morning, and that thing is really strange May 22 10:29:08 <[florian]> rtz2: should not we only set the reset bit of the watchdog in the timer callback? May 22 10:29:19 [florian]: you wrote me a mail, give me a few minutes May 22 10:29:27 [florian], where to put patches ? on forum in my thread ? May 22 10:29:29 <[florian]> rtz2: yeah sure May 22 10:29:41 <[florian]> russiane39: better send them to the mailing-list or attach them to a trac ticket May 22 10:30:05 is it okay, that I haven't solved one issue and it's in dirtyhack state for now ? May 22 10:30:31 <[florian]> russiane39: send them, juhosg can still sort this out later May 22 10:34:12 hmm, why wifi driver rt2x00 compiles from compat-wireless, but not from kernel ? May 22 10:36:11 <[florian]> because compat-wireless is much more up to date May 22 10:36:30 this is not true for rt2x00 driver May 22 10:36:31 <[florian]> so this allows us to be close to the latest wireless developments May 22 10:36:50 <[florian]> this is also true for the rt2x00, compat-wireless is based on wireless-testing git tree May 22 10:37:46 strange, 2.6.34 reports rt2872 (used in rt305x) as supported, while for now driver from compat-wireless don't want to attach to device May 22 10:38:07 <[florian]> weird indeed May 22 10:39:00 [florian]: ok, I admit I can't follow you May 22 10:39:25 [florian]: rdc321x_wdt_reset only resets the timeout in the driver and doesn't touch the register at all May 22 10:39:52 [florian]: and reconfiguring the wdt in rdc321x_wdt_timer, I'm not sure how it's possible to avoid it May 22 10:39:58 [florian]: it's only one register May 22 10:40:00 <[florian]> rtz2: if we use read-modify-write that's true May 22 10:40:11 funny linux, there is no rt2x00 driver in kernel menuconfig at all! but I see it in source tree May 22 10:40:18 <[florian]> rtz2: right now we are blindly writing RDC_WDT_SETUP in the watchdog control register May 22 10:40:40 Device Drivers>Network Device Support>Wireless LAN , right ? May 22 10:41:15 [florian]: yes, but I don't see the problem with that, it's not like the timeout is adjustable May 22 10:41:18 <[florian]> rtz2: from my understanding of the datasheet, one would setup the watchdog correctly, and the timer callback would just set the reset bit to prevent a reset May 22 10:41:46 <[florian]> rtz2: well it kind of is adjustable by rounding up to the closest timeout requested May 22 10:41:46 [florian]: and we now the contents of the reg, reading it doesn't make much sense imho May 22 10:42:31 <[florian]> rtz2: ok, well it works like this but it does not seem consistent to me May 22 10:42:50 [florian]: well, maximum timeout is 5.37 s which is way to short to be useful May 22 10:43:14 <[florian]> rtz2: it's actually not so bad May 22 10:43:28 <[florian]> rtz2: firing a timer every 5 secs is not delirious, a couple of other watchdog do the same May 22 10:43:42 <[florian]> rtz2: one more thing, why did you go for "generate a nmi and reset the cpu" May 22 10:43:44 [florian]: what do you want to do, do a read-modify-write or write a value with only the reset bit set? May 22 10:44:15 <[florian]> rtz2: I want to read the register value, set the reset bit and write back this value May 22 10:44:31 <[florian]> rtz2: which seems like the "usual" way to do things if you do not want to loose your watchdog settings May 22 10:44:32 [florian]: well, I couldn't get it to reset the cpu at all, only create a nmi May 22 10:45:01 <[florian]> rtz2: that's also what I have noticed this morning May 22 10:45:28 <[florian]> rtz2: I was expecting that "generate a watchdog irq and reset the CPU" would somehow work May 22 10:45:35 <[florian]> which is setting bit 22, clearing bit 21 May 22 10:45:37 <[florian]> but it does not May 22 10:45:46 [florian]: nope May 22 10:46:19 [florian]: I have no clue, what the problem is, but no matter what you write in 22 and 21, it doesn't reset the cpu May 22 10:46:24 hmmm May 22 10:46:32 <[florian]> rtz2: yup, hardware is probably bogus May 22 10:46:35 <[florian]> rtz2: go to go now May 22 10:46:54 [florian]: will you be back this evening? May 22 10:50:01 [florian], maybe you know how to completely disable compat-wireless ? May 22 11:27:20 [florian] this still isnt added http://openwrt.pastebin.com/QPM1vhec just a reminder here maybe someone else will commit May 22 11:37:21 nbd: ping May 22 18:22:52 swalker * r21535 /packages/net/subversion/Makefile: [packages] subversion: update to 1.6.11 May 22 18:41:55 nbd: ping **** ENDING LOGGING AT Sun May 23 02:59:57 2010