qi-bot | The build was successful: http://fidelio.qi-hardware.com/~xiangfu/build-nanonote/openwrt-xburst.minimal-20120827-2243 | 01:09 |
---|---|---|
larsc | mth: to be honest I have no idea, I guess it does this because everybody else does it as well | 11:22 |
mth | that was exactly my reason for doing it for the JZ4770 :) | 11:43 |
wpwrak | cargo cult programming at its finest ;-) | 11:44 |
wpwrak | (and no, i don't know either :) | 11:45 |
whitequark | wpwrak: just like the cache tricks magic in the initialization code | 12:20 |
whitequark | I've stripped them and nothing has changed | 12:20 |
wpwrak | those mistreated caches will wait for you in the afterlife ;-) | 12:21 |
whitequark | haha | 12:21 |
Action: whitequark imagines several poor cache lines in the Charon's boat | 12:22 | |
mth | gdb aborts the process I'm debugging because it allegedly receives signal 128 pretty close to startup | 12:46 |
mth | but if I run the process outside gdb it starts up fine, the thing I'm debugging happens much later | 12:46 |
mth | and there doesn't seem to be a singal 128 | 12:46 |
mth | s/singal/signal | 12:46 |
mth | maybe gdb is seeing ghosts: the signal does not come from __send_signal in the kernel | 13:12 |
mth | could it be a tortured cache line? | 13:13 |
DocScrutinizer51 | is that process maybe skype? ;-P | 13:14 |
mth | no, gmenu2x | 13:14 |
mth | it is certainly something inside gdb: when trying to attach to a non-existing pid, the same "signal" occurs | 13:30 |
lindi- | mth: on ARM? | 13:33 |
mth | on MIPS: JZ4770 | 13:35 |
Fallenou | wolfspraul: I've just read an email from you dating from 2009 about marvell combo chip 88w8688 , do you have more intel about what bugs are encountered or what's wrong with the chip etc ? | 13:35 |
Fallenou | or some guy I could shoot an email about this chip ? | 13:35 |
mth | but it happens on JZ4740 too | 13:35 |
lindi- | mth: ok, no such hardware here | 13:35 |
lindi- | mth: it doesn't happen on arm or x86? | 13:35 |
Fallenou | on my daily job we happen to use this chip (connected to an OMAP 3630), and indeed we have a lot of troubles with bluetooth :/ | 13:36 |
Fallenou | I am referring to this email http://lists.en.qi-hardware.com/pipermail/discussion/2009-November/001108.html | 13:36 |
mth | I don't have an ARM device, on my x86_64 desktop everything is fine but that's an entirely different system (different gdb version, glibc vs uClibc) | 13:36 |
mth | it even happens on /bin/ls (busybox), so it's not related to the application that is being debugged | 13:37 |
wolfspraul | Fallenou: oh my, no I don't :-) | 14:51 |
wolfspraul | that's a looong time ago and I learned many things, primarily to stay away from such chips :-) | 14:52 |
wolfspraul | if you are still fixing bugs on that chip today, that's itself a bug already. but other than such non-helpful 'suggestions' I don't have anything, sorry about that | 14:53 |
Fallenou | yes marvell shipped a new firmware a few days ago to us | 14:54 |
Fallenou | and it has a lot of regressions | 14:54 |
Fallenou | it's a nightmare | 14:54 |
wpwrak | ;-) | 14:54 |
wolfspraul | all I can say about these things will not help you | 14:55 |
wpwrak | there's indeed a bit of a message in "oh, we'll fix [many flaws] in the next firmware update", when that comes for a chip that's been around for so many years :) | 14:56 |
wolfspraul | but: GOOD LUCK! | 14:56 |
wolfspraul | wpwrak: the chip was never designed and made for companies like fallenous, imho | 14:56 |
wolfspraul | so by now both marvell and fallenou's company are in a loose-loose situation on this | 14:57 |
wolfspraul | fallenou can just "try his best" and be happy about his pay check, which is hopefully comforting... | 14:57 |
wolfspraul | Fallenou: don't start drinking :-) | 14:58 |
Fallenou | ahah I won't | 14:58 |
Fallenou | it's just annoying | 14:58 |
wolfspraul | you are paid for the annoyance? | 14:58 |
wolfspraul | the enjoy the money... | 14:58 |
Fallenou | I don't care that much deep inside | 14:58 |
wolfspraul | it's all you will get :-) | 14:58 |
Fallenou | Marvell seems to only have 2 automotive chips | 14:59 |
Fallenou | and the 88w8688 is one of them | 14:59 |
Fallenou | too bad it's just crap | 14:59 |
Fallenou | I wonder what will happen when we will try to use te 3.0 features :) | 14:59 |
Fallenou | the* | 14:59 |
wolfspraul | it probably works for the 1 or 2 customers that are being directly supported | 14:59 |
Fallenou | well we have kind of support | 14:59 |
wolfspraul | you are not bmw, afaik | 14:59 |
Fallenou | we have conf calls and firmware fixes/kernel driver shipping | 14:59 |
Fallenou | wolfspraul: no we are not but we may ship to bmw/volvo/mclaren/whatever :) | 15:00 |
wpwrak | reminds me of the times at openmoko ... an epic struggle with that wireless chip, yet it never quite worked. in the end, it came down to fatal firmware bugs nobody could be bothered to fix. | 15:00 |
wolfspraul | the economics are tough for everyone | 15:00 |
Fallenou | add Android on top of all this, and coexistence between wifi and bluetooth | 15:00 |
Fallenou | and there you go | 15:01 |
wolfspraul | anyway, all blurb that won't help fallenou, I'm afraid | 15:01 |
Fallenou | head banging against the wall :) | 15:01 |
wolfspraul | in a few years it's all forgotten | 15:01 |
Fallenou | anyway thank you for the human conforting support ! | 15:01 |
wolfspraul | read the leaked apple genius thing today | 15:01 |
wolfspraul | "I understand why you feel that way" | 15:02 |
wpwrak | you should tell your bosses to find some small company that makes some halfway decent wifi chip, buy that company, then open the design - at least at the register level, deeper if you can - and let the hacker world work on it for a few years | 15:03 |
Fallenou | that sounds like a good idea | 15:03 |
wpwrak | then you could hope for having, say, decent 802.11g in a couple of years | 15:03 |
Fallenou | I'm not high enough in the company hierarchy yet to give this kind of advice | 15:03 |
wpwrak | apple genius ? | 15:04 |
--- Thu Aug 30 2012 | 00:00 |
Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!