Log from #neuros IRC channel

crweb: this is mgao, we are still having trouble with SAMBA on XP laptop, I understand you've spent sometime with jwu on this, it seems that we are not able to browser the SAMBA from XP that is natively installed, but we are able to access those on Linux, what more can we do to investigate? this is important for our CES demo ;-)

[02:21]

<crweb>

tell you what, give me.. 5-10 minutes to get into windows and we can step something out

[02:21]

<jwu>

we've tried Steven's ideas, checking the stuff he mentioned on ML, but didn't work.

[02:22]

<crweb>

jwu: and while I have you, I'd really like you guys to take some time and do a write up on the memory stuff jwu, terry, and i are talking about on the mailing list.

[02:22]

<crweb>

i'll brb

[02:23]

<jwu>

will do once we figure out things for CES. ;-)

[02:23]

<jwu>

short answer,

[02:23]

<jwu>

the memory reserved out of kernel was inherit from codec vendor's design.

[02:24]

<jwu>

haven't got a chance to look into whether we can completely remove that logic or not.

[02:24]

<crweb>

well, something happened between 0.35, and 0.49. and now I can't use memory that I once could

[02:25]

<crweb>

and its the network card, sd, ide, usb, etc that have stopped working if I go over 16M

[02:25]

<crweb>

they didn't do that before.

[02:26]

* jwu understood the frustration, but CES is the only thing that hovers over mgao's head now.

[02:27]

jwu: that's understandable, but please take some time after CES to give out some answers. .46 firmware and later left a lot of scary open questions

[02:28]

<jwu>

again, short answer for that: starting from .49, more memory might be reserved for MP4 encoding support, last comment on memory thing. ;-)

[02:28]

<jwu>

nerochiaro: ok.

[02:28]

* crweb is flashing, just a few more min

[02:28]

jwu: thanks

[02:29]

<jwu>

here we've got an XP laptop prof edition on DELL inspiron.

[02:30]

<crweb>

nerochiaro: my gui is as slow as the long list boxes, only mine doesn't screen update ;)

[02:30]

<crweb>

i get a lot of NWFPE errors

[02:31]

what's that ?

[02:31]

<crweb>

floating point errors

[02:31]

<jwu>

file share on this laptop can be seen by other XP desktop, but not by Linux machine.

[02:31]

<crweb>

jwu: excellent

[02:31]

<crweb>

jwu: when I get back in i'll be ready

[02:36]

crweb: probably there's something you have to tell to qt to make it know it's running on an integer-only machine

[02:36]

<crweb>

huh, would you look at that

[02:36]

<crweb>

my xp booted

[02:36]

<crweb>

nerochiaro: its designed for arm.. shouldn't have to tell it much

[02:36]

<crweb>

I probably do need to tweek my build though

[02:37]

<jwu>

more info: OSD is able to access this XP laptop if we know the IP beforehand, however OSD 'nmblookup' will not be able to _automatically_ find it, which is the problem.

[02:37]

<crweb>

ok

[02:37]

crweb: i doubt it's a problem in fp emulation in general. other stuff work fine

[02:38]

<crweb>

jwu: ok good news is, I have the same problem

[02:39]

<jwu>

good news

[02:40]

<crweb>

odd statement but, it means I can solve the problem

[02:41]

hi

[02:41]

i want to connect 5.1 channel speakers to my osd, any possible way ?

[02:41]

<crweb>

windows isn't broadcasting itself

[02:42]

<jwu>

it just reply the broadcasting

[02:42]

<crweb>

how do i make the osd refresh?

[02:43]

<crweb>

its not rescanning if I ext the menu and go back in

[02:43]

<jwu>

I think it not need to refresh

[02:43]

<jwu>

what vertion do you use?

[02:43]

<crweb>

it does if it didn't catch a broadcast right when it was updating

[02:43]

<crweb>

0.49

[02:44]

<jwu>

I think it is new enough

[02:46]

<crweb>

well, its not updating

[02:46]

<crweb>

the osd still thinks my old linux box is up

[02:50]

<crweb>

well, its networking isn't working now

[02:52]

<crweb>

yeah, networking had died on me for some reason

[02:52]

<crweb>

couldn't ping, or telnet

[02:53]

<jwu>

OSD network stopped working?

[02:53]

<crweb>

yeah, i had to reboot it

[02:54]

<jwu>

it will rescan if you XIM->refresh.

[02:54]

<crweb>

ah ok

[02:54]

<crweb>

what kind of hub are you using?

[02:56]

<jwu>

a really cheap one bought locally, and a router bought locally as well.

[02:56]

<crweb>

ok, well, it for sure won't go through the router

[02:56]

<crweb>

so make sure both machines are on the internal side

[02:56]

<jwu>

with the router, we tried the DHCP idea.

[02:57]

<jwu>

ok, OSD-->HUB<--XP, using static ip

[02:57]

<jwu>

is that what you meant?

[02:57]

<crweb>

yes, thank you

[02:58]

<crweb>

it works with win 2000 though?

[02:58]

<jwu>

then?

[02:59]

<jwu>

not tested with 2K yet.

[03:01]

<jwu>

2K is OK here

[03:02]

<crweb>

its because windows 2k is setup to be the master

[03:02]

<crweb>

I have 1 more idea, i'll brb

[03:04]

<crweb>

as you can see, he was involved in my idea

[03:13]

<jwu>

so, what is the verdict?

[03:13]

<jwu>

no XP support at this point, and we'd need to bring a 2K laptop to vegas? ;-)

[03:13]

<crweb>

my network card froze again.

[03:14]

<crweb>

how long do we have?

[03:14]

<crweb>

till CES?

[03:15]

<jwu>

a few hours...

[03:15]

jwu: are many of you Xiamen folks going to Vegas ?

[03:15]

<jwu>

none.

[03:16]

ah, ok. just you Michael ?

[03:16]

<jwu>

neither me.

[03:16]

so it will be just Joe and some cute stand girls ?

[03:16]

<jwu>

that is why we need to make sure this thing 100% work without hacks.

[03:17]

<jwu>

Joe will be there for sure, and our marketing/sales guys, not sure how cute the stand girls are though.

[03:20]

<crweb>

damn microsoft

[03:24]

<crweb>

it can't be done

[03:24]

<crweb>

you have to nmap for open port 139's

[03:24]

<crweb>

then nmblookup -A all those ip's

[03:25]

<crweb>

windows xp isn't using netbios on tcp/ip

[03:25]

crweb: not even professional version ?

[03:25]

<crweb>

nope

[03:25]

<crweb>

just 2000, and 2003 server

[03:25]

bleh

[03:27]

<jwu>

do we have nmap on OSD?

[03:27]

<crweb>

we can cheat

[03:27]

<jwu>

if that is the case, it seems doable in a few hours?

[03:27]

<crweb>

if you want to do that...

[03:27]

<jwu>

how to cheat?

[03:27]

<crweb>

:)

[03:27]

<crweb>

edit /etc/hosts

[03:27]

<crweb>

put <ip address of xp machine> <netbios name of xp machine>

[03:27]

<crweb>

save it.

[03:28]

<jwu>

damn

[03:29]

<jwu>

we dont' have nmap on OSD as part of busybox?

[03:29]

<crweb>

no

[03:29]

can't even add it ?

[03:29]

<crweb>

not easily

[03:29]

<crweb>

it would be like trying to recompile samba

[03:30]

<crweb>

whats the exact command you are using for nmblookup?

[03:31]

<jwu>

nmblookup -B 192.168.1.255 "*" -S

[03:34]

<jwu>

crweb: it seems like the rootfix will be getting nmap support on OSD?

[03:35]

<crweb>

or, getting a wins server

[03:35]

* jwu will add a hack just for CES to search a given IP.

[03:35]

<crweb>

thats a good idea

[03:36]

<jwu>

this still sounds weird to me, there is no fix by just patching the current nmblookup tool?

[03:37]

<crweb>

nope, its a microsoft thing

[03:37]

<jwu>

damn microsoft.

[03:39]

<jwu>

so current Linux desktop uses the same nmap trick to access XP share?

[03:39]

<cr-tablet>

hold that thought, might be onto something. Don't get hopes up though

[03:40]

<jwu>

so current Linux desktop uses the same nmap trick to access XP share?

[03:40]

<cr-tablet>

yeah, thats what it looks like

[03:40]

<cr-tablet>

only they use rpc

[03:40]

<jwu>

anyway, I'll leave some time for real jwu to implement the cheat idea for CES...

[03:41]

<jwu>

and we can come back for a real fix.

[03:41]

<jwu>

thanks a lot guys, see you in tomorrow's meeting.

[03:41]

<cr-tablet>

wiat

[03:41]

<cr-tablet>

2 more min

[03:45]

<cr-tablet>

nope

[03:45]

<cr-tablet>

i hate microsoft

[03:45]

<cr-tablet>

excellent work though guys

[03:45]

<cr-tablet>

i'm impressed

[03:46]

<cr-tablet>

i'll keep reading

[03:47]

<cr-tablet>

i have to go pass out now

[03:47]

<cr-tablet>

night

[03:47]

jwu: and good luck for CES, folks !

[03:56]

<jwu>

nerochiaro: thanks

[06:54]

hey. are there any non-nightly recent tarballs somewhere? I tried the latest nightly, but my OSD just goes into endless reboots (might be faulty HW)

[07:35]

beaufour: there were some faulty units being sent out recently. when did you get it ?

[07:36]

nerochiaro: yeah, I know. around Christmas

[07:44]

beaufour: might be that then. does it reboot right after the kernel finishes initializing ?

[07:45]

beaufour: try sending an email to check with neuros, or get ahold of JoeBorn, but he might be super-busy for CES at this point

[07:56]

nerochiaro: hmm, it gets to some usb init, and then reboots

[07:57]

now I have 0.49 on it, and it does not reboot

[07:57]

any tips for how to open the device, except from brute force? I was told to check for the presence of some resistors

[07:58]

honestly i don't know. never tried it, but it seems mine is just held togheter without screws. but i have an older model, not the black ones

[07:59]

oh wait, maybe there's screws below the rubber feets

[08:01]

mkay, I have a black model

[08:01]

try lifting one of the rubber feets, see if there's a screw below

[08:01]

* beaufour tries to look under the rubber feet

"look ma, screws!"

:)

there you go ;)

how about the tarballs? is there newer ones, or is checking out specific tags just the way to go?

[08:05]

the idea is that you download a tarball, even a very old one. then you unpack it and then use svn switch to move to the tag you want

[08:05]

yeah, tried that, but something must have gone wrong in the process. I guess I have to try again. I had missing header files

[08:06]

that's a weird issue, many included me had that. only solution right now is to manually pick these from trunk from svn web and put them where you need them

[08:07]

i have absolutely no idea why they keep not getting included in tags

[08:07]

actually for me it was just one header

[08:07]

is it more for you ?

[08:11]

I did not check further than the error

[08:12]

hmm, yes I have resistors at the given spots... now the million dollar question is: Is that good or bad :)

[08:12]

they were added, as far as i understand from what i've heard here in chat. so it must be good. but not sure

[08:13]

as for the build error, you probably have only one header file missing, like the others.

[08:13]

'k, will try again

[08:15]

if you do, can you please report in ML the steps you followed and what the missing header was ? so the folks at neuros might be able to fix it this time

[08:19]

ml == mailing list?

[08:19]

yes

[08:21]

it's a bit weird that the root dir lacks: 1. readme fil, 2. a clean script, and no 3. a "switch to another tag" script

[08:22]

beaufour: which is the root dir ?

[08:22]

beaufour: i suppose you're following derobert's guide ?

[08:22]

svn.neurostechnology.com

[08:22]

in lack of other guides, yes :)

[08:22]

there is no other one, indeed

[08:23]

and well, if you're following the guide, that's your readme. i agree with the lack of switch script, it should be added, it's a good idea. and what do you mean with "a clean script" ?

[08:25]

the equiv. to a "make clean"

[08:25]

there's a script in the guide too, I know

[08:25]

ah, that one. last time someone checked, it didn't work completely well

[08:26]

well, one more argument for a good one to be in SVN :)

[08:26]

indeed

[08:26]

these are all good points, it's all stuff that someone should do

[08:27]

as some folks would say "patches welcome" ;)

[08:27]

== the story of open source :)

[08:27]

heh, exactly

[08:27]

yep :)

[08:27]

and also "i would do that, but i'm already too busy right now with other stuff"

[08:28]

heh

[08:29]

maybe I'm not getting it to work because I've failed to follow the troubleshooting tips

[08:30]

* beaufour looks around for a towel ;-)

[08:31]

* nerochiaro hands a clean towel

[08:33]

is there not something about a corner of it being soaked in something?

[08:33]

* beaufour cannot remember his Hitchhikers...

[08:33]

probably it just needs to be wrapped

[08:33]

can't remember either

[08:54]

<anders_>

PAL broke in 0.49 it seems.

[08:54]

<anders_>

Colors disappeared.

[09:04]

* beaufour is not having much luck with getting the source ... svn just died switching to another tag .... grrrr

[09:04]

ouch

[09:05]

died how ?

[09:05]

not having done anything for around half an hour ...

[09:05]

network connection timed out in the background is a guess... dunno

[09:05]

already tried again ?

[09:06]

it's continuing with the rest of the dirs now. let's see

[09:06]

if usually does it in a minute at most, more often a few seconds

[09:09]

yeah, I just forgot all about it... so when I flicker over to the term, I saw it doing nothing at all...

[09:27]

grrrr "./00_Build.sh: line 99: pushd: linux-r3-main-app/build: No such file or directory"

[09:28]

but is it there ?

[09:29]

nope, but the line above says "BUG: Main app installs in the wrong location (in ~/neuros-bsp)"

[09:29]

nah, that's not it

[09:30]

if the directory doesn't exist, something is wrong with your sources. it should be there

[09:31]

* beaufour is so much not impressed by this

[09:31]

i can imagin

[09:31]

svn.neurostechnology.com

[09:31]

there is no build dir

[09:32]

then they changed it in .49 ... it was there at least up to .35

[09:32]

no wonder the script breaks

[09:32]

<anders_>

That looks familiar. Just remove /build from that path.

[09:33]

yeah, but they sholdn't do that kind of changes, without warning.

[09:33]

it happened when they split nms server and and osdmain

[09:34]

should add some conditional logic in the script to handle both cases

[09:34]

anders_: looks right, will try

[09:34]

yeah, a simple test -f could fix it

[09:43]

wee, at least it boots now, without eternal rebooting

[09:43]

problem is that the network seems to die suddenly... nfs craps out

[09:43]

another known issue. you are using dhcp, right ?

[09:44]

yes

[09:46]

well, what happens is this: at boot an address gets assigned to load the rootfs via nfs. then osdmain tries to change it, and in doing it brings net down, which in turn brings nfs down, which in turn kills the root fs

[09:46]

bam, dead

[09:46]

so have both use static addressing and it will work

[09:48]

* beaufour swears some more

[09:49]

yeah, it's ugly. anyway, reboot on cramfs, in osdmain go to configure network settings, set static addressing. save. it gets saved to flash. now reboot with nfs and should work

[09:51]

* beaufour had envisioned using 2 hours on this today, and have ended up using most of the day :(

[09:52]

the dev env initial setup is still the most ugly part of the OSD

[09:52]

i agree

[09:57]

nerochiaro: woohoo. Finally succeded. thx a bunch!