--- Day changed Tue May 28 2013 | ||
@juiceme | morning | 07:16 |
---|---|---|
@juiceme | Jonni, I could maybe provide the whole log too, it's just split to dailies before formatting so that's no trouble. | 07:17 |
@juiceme | thedead1440, that's also an option. The corrent script even can provide a searchbox to the upper corner of the page but I turned it off since it requires additional PHP to actually make the search... Do you have any example I could look into on how to prvide that? | 07:19 |
@juiceme | Okay, now i added the full unsplitted log to the #ubiboot irclogs page. (for the intrested, just before splitting the log I just export the original log file to full html, copy that off to the server as well as the splitted batch, and fix the index.html with a sed oneliner) | 10:08 |
-!- rikanee [~rika@unaffiliated/rikanee] has quit [Ping timeout: 276 seconds] | 10:20 | |
-!- rikanee [~rika@unaffiliated/rikanee] has joined #ubiboot | 11:17 | |
-!- rikanee [~rika@unaffiliated/rikanee] has quit [Ping timeout: 256 seconds] | 11:24 | |
-!- rikanee [~rika@unaffiliated/rikanee] has joined #ubiboot | 11:25 | |
@juiceme | rikanee, hi | 11:35 |
rikanee | ohai juiceme. | 11:35 |
@juiceme | tried compiling omap_wdt and twl3040_wd as modules but it didin't help :) | 11:36 |
rikanee | juiceme: Nokia might have actually set the watchdog to not reset during a software reset | 11:37 |
@juiceme | I thought of deinstalling the watchdogs at kexec time, to do proper cleanup | 11:37 |
rikanee | juiceme: worth a try, but the watchdogs may still be watching during kexec, and if the reboot is too slow, WDT is fired | 11:38 |
@juiceme | would that change the behaviour? | 11:38 |
rikanee | IDK, I have no idea how Nokia switched around watchdog behavior after the N900 | 11:39 |
@juiceme | it's fairly difficult to find info about the watchdog implementation/behaviour, IMHO. it's regarded black magic :) | 11:40 |
rikanee | yup. | 11:41 |
rikanee | juiceme: how about throwing a really early watchdog pinger in kernel boot? | 11:41 |
rikanee | it may just be fast enough to satisfy the watchdog | 11:41 |
@juiceme | that coud work. | 11:42 |
@juiceme | but I think it already does that in preinit | 11:42 |
@juiceme | i'll have to check | 11:42 |
rikanee | juiceme: even earlier than preinit, in case the ~2 seconds it takes for kernel to load all modules is too long | 11:43 |
@juiceme | but the actual problems only are seen after preinit phase... | 11:44 |
@juiceme | how long it takes until watchdog resets? | 11:44 |
rikanee | no clue how long it takes on the N9. | 11:44 |
rikanee | Try making a ubiboot with the watchdogs off, and writes an incrementing second using text2screen to see how long we've got before timeout. | 11:45 |
@juiceme | yes, that might be a good test to find it out. I need to test that with several devices, it might be that the watchdog bite delay is one of the things that differ based on HWREV | 11:49 |
Generated by irclog2html.py 2.12.1 by Marius Gedminas - find it at mg.pov.lt!