<nebajoth>
has anybody compiled a working kernel for the nanonote that uses a more recent kernel?
<nebajoth>
I remember seeing the mailing list discussions regarding official inclusion
<nebajoth>
in the mainline, that is
<nebajoth>
ah yes, as of 2.6.36
<wolfspraul>
yes definitely there are more recent kernels, 34/35. but I don't know exactly where the patches are or how to build them. I'd say you need to look in openwrt upstream.
<wolfspraul>
everything based on the openwrt backfire release will stay with 32
<nebajoth>
that's fine
<nebajoth>
I'm just going to work out how to build it then
<wolfspraul>
36 should have a lot of Ben NanoNote stuff mainline, but not everything
<wolfspraul>
maybe one or two drivers are still missing, mmc if I recall
<nebajoth>
ls
<nebajoth>
haha
<nebajoth>
WRONG CMD LINE
<nebajoth>
just looking at the patch
<nebajoth>
the one linus merged
<nebajoth>
looks like mmc is there
<nebajoth>
I think
<nebajoth>
but I don't see anything for audio
<qi-bot>
[commit] Werner Almesberger: A new try at config.h management: moved it to ../common, along with io.h http://qi-hw.com/p/ben-wpan/57f3427
<qi-bot>
[commit] Werner Almesberger: Implemented AT86RF230 reset and access modes. (Completely untested.) http://qi-hw.com/p/ben-wpan/5447951
<qwebirc47408>
i have purchased new nanonote
<qwebirc47408>
want oto put it into usb-bbot mode
<qwebirc47408>
but when I shorten the two pins then nanonote doesnt bootup
<qwebirc47408>
plz help
<kyak>
just hold the "U" button during power on
<kyak>
no need to shorten if you have a working bootloader
<qwebirc47408>
if I am holding U button the it is not booting up
<wolfspraul>
qwebirc47408: you don't need to constantly hold it, just before and while you are pressing the POWER-ON button
<wolfspraul>
press the power-on (together with 'u') for a good 2-3 seconds
<wolfspraul>
then run 'lsusb' on your Linux host machine
<wolfspraul>
you should see an ID 0x601a:471a:ns
<wolfspraul>
sorry, - you should see an ID 0x601a:4740
<tuxbrain2>
qwebirc47408: in USB boot mode you will notice no reaction from nanonote , screen black , you hve to check if is in usb boot by lsusb in host computer and see a new device 0x601a:4740
<wolfspraul>
do you see that ID?
<qwebirc47408>
let me check
<qwebirc47408>
yes i m getting
<qwebirc47408>
how can I develop any C application for it and run over nanonote
<wolfspraul>
qwebirc47408: that's a great start! :-)
<wolfspraul>
if you see that ID, it means your NanoNote is alive, the CPU is running, and waiting for instructions
<wolfspraul>
it is not supposed to boot in that mode
<qwebirc47408>
ok
<qwebirc47408>
what to do next
<wolfspraul>
now you can run reflash_ben.sh, although I hate that script :-)
<qwebirc47408>
I want to reflash the  nanonote
<qwebirc47408>
how shall I run usbboot application
<wolfspraul>
you have usbboot installed already?
<wolfspraul>
then just run reflash_ben.sh
<qwebirc47408>
how to check it
<qwebirc47408>
i have installed openwrt-xburst
<wolfspraul>
you need to install the xburst-tools package
<wolfspraul>
that's the sources, that comes later
<qwebirc47408>
yes I have
<wolfspraul>
do you have reflash_ben.sh ? download and run
<wolfspraul>
I believe we do have some instructions in the wiki for that
<wolfspraul>
if you see the 601a:4740 ID all is fine. Now run reflash_ben.sh
<qwebirc47408>
I have downloaded reflash_ben and running it
<qwebirc47408>
what all will happen with this
<wolfspraul>
it will download the 0613 image and reflash the ben
<kyak>
weren't all these questions from beginner's guide?...
<bartbes>
btw, wasn't there a photo viewer?
<wolfspraul>
bartbes: imgv
<bartbes>
is it in the default image?
<bartbes>
yes, it is
<bartbes>
thanks!
<wolfspraul>
he, just trying
<bartbes>
ah well, my jpeg is too big
<bartbes>
(for the memory)
<wolfspraul>
we will cleanup the openwrt default app list and testing plan 'really soon', then a lot more should be included and stay included
<bartbes>
interestingly enough I have yet to run out of memory while running a game engine, while I run out wanting to view a simple photo
<wolfspraul>
I have imgv but if I run it without parameters it just hangs
<wolfspraul>
:-)
<wolfspraul>
scaling pictures can be very memory consuming, unless the scaling algorithm is specifically written to reduce memory footprint
<wolfspraul>
imgv --help or --version should at least say something, oh well...
<bartbes>
true
<bartbes>
anyway, it was just a test to see if my camera supported SDHC
<bartbes>
well, 'my' imgv doesn't hang when I run it (even without parameters)
<bartbes>
and about why --help and --version don't help, they just don't exist ;)
<bartbes>
I doubt it does any kind of arguments
<kyak>
cp: cannot stat `./files/qmake.conf': No such file or directory
<kyak>
this is from recent qt
<kyak>
actually, i'm tired of things being broken every now and then upstream
<kyak>
i should better neve make package/symlinks
<kyak>
do they even test before they commit?
<bartbes>
kyak: testing is overrated :P
<wolfspraul>
mirko: how is imgv supposed to work? in the 08-15 testing image, I try 'imgv some_image.jpg' and it just hangs.
<wolfspraul>
ctrl-c will exit
<qwebirc26781>
reflashing is taking long time . is it ok ?
<wolfspraul>
how long?
<wolfspraul>
20-30 minutes should be ok.
<qwebirc26781>
it is already 1 hr gone but stuked at fetching rootfs
<wolfspraul>
that's not right
<wolfspraul>
oh well
<wolfspraul>
wait, maybe it is downloading? what is your internet speed?
<qwebirc26781>
512Kbps
<wolfspraul>
could take an hour maybe. not sure. also I don't know how to check the download progress with the script.
<qwebirc26781>
how can i develop simple C application for it
<qwebirc26781>
and test it
<rafa>
wpwrak: hey, after a while, yesterday, the bechnmark gives better values, no a lot more though.. : 65MB/s
<bartbes>
yeah, most tests you need to avarage as well
<wolfspraul>
zedstar: sorry I haven't gotten to replying to your mail yet. definitely will!
<bartbes>
sometime
<bartbes>
in about 3 years
<bartbes>
or later
<wolfspraul>
he
<wolfspraul>
I spend 50% of my time adding items to my todo list nowadays. Something is wrong...
<lars_>
you should add 'Stop adding items to my todo list' to your todo list ;)
<bartbes>
or if you want to cross something off
<bartbes>
'Add something to my todo list'
<wiwitu>
hey guys
<wiwitu>
quick question: whats the name for a powersupply with 16 pins, I cannot find one on the internet
<lekernel>
wtf? there are billions of power supplies with 16 pins
<lekernel>
if you're talking about PC power supplies this is the wrong channel
<viric>
he left, maybe realising that
<bartbes>
he ran out of power ;)
<wpwrak>
hmm, if anyone is running a recent kernel under openwrt, it would be interesting to try the memset benchmark there, too. to exclude kernel differences as possible causes.
<wpwrak>
lekernel: nice analysis, thanks ! so my estimates are basically sane.
<wpwrak>
(running the benchmark) just compile, maybe with -O9 to make sure the compiler doesn't do anything funny, run top to make sure that there's nothing that burns cpu, then run the benchmark a few times with time ./a.out
<wpwrak>
it just takes a few seconds
<lekernel>
try enabling RT priority too (ie context switches disabled)
<lekernel>
and write your own memset, in assembler if needed
<wpwrak>
lekernel: (rt) naw, i'd rather not do that. might upset the timekeeping.
<lekernel>
then use hwclock calls?
<wpwrak>
(own memset) i hope to get others interested in doing that, should it be necessary :)
<wpwrak>
(hwclock) uh, too complex for this little benchmark
<wpwrak>
it's not as if it would suffer high variability anyway
<lekernel>
apparently it does :)
<wpwrak>
right now, i'm not so much after maximizing memset, but after confirming that openwrt-based memset is slower than jlime-based
<wpwrak>
on the same system ? just a few percent
<wpwrak>
@#%. why does that silly device not respond to my control messages. almost the same code worked a gazillion times before.
<wpwrak>
hates chasing trivial bugs
<wpwrak>
oh, lovely. looks like yet another sdcc compiler bug :-(
<viric>
what do you need to benchmark?
<viric>
I'm not running openwrt, but I run a 2.6.35.1 iirc
<wpwrak>
viric: is your system uClibc-based ?
<viric>
wpwrak: no
<viric>
glibc.
<viric>
I can build that program for uclibc maybe.
<viric>
(not with glibc)
<wpwrak>
well, could be useful to have another data point anyway. what system is it ? debian ?
<wpwrak>
ah no. he wrote 65 MB/s. so it's the same.
<wpwrak>
what options did you use on gcc ?
<viric>
building gcc?
<wpwrak>
no, when invoking it
<viric>
gcc -static -O9
<wpwrak>
ah, nothing special for the cpu then
<viric>
well
<viric>
the gcc was built with "--with-arch=mips32"
<viric>
so it assumes that arch
<viric>
at every invocation
<wpwrak>
hmm. mine seems to be mips1, too
<wpwrak>
the mips gcc should support several mips targets. you can switch them with -march=mips1|mips32|etc. no need to rebuild gcc just for that
<viric>
right.
<viric>
well, for my case it's not any additional effort
<viric>
and this way I don't have to set anything special building every package. I get mips32 everywhere
<wpwrak>
one man's excessive effort is another man's laziness ;-)
<viric>
:)
<wpwrak>
now, let's see how this goes on my ben ...
<wpwrak>
i should probably remove gmenu2x if it likes to play with the clocks ...
<viric>
I don't have it
<tuxbrain2>
I think the newest version should not play with clocks it was removed due flickering screen problems, but I'm not totally sure about this
<viric>
do you use any NES emulator in the ben?
<wpwrak>
tuxbrain2: yes, but this ben of mine still has the factory installation. i'll save the system upgrade for another day.
<wpwrak>
viric: me ? no. so far, i hardly "use" it at all. it suffers considerable abuse on the hardware side, though ;-)
<tuxbrain2>
so if you have the old gmenu2x yes remove it to do the performance tests
<viric>
wpwrak: ok :)
<wpwrak>
after getting rid of gmenu2x, i get ~60 MB/s. too. no difference between mips1 and mips32 here. mips32r2 is a bit faster, though. (~2 MB/s)
<viric>
wpwrak: lars_ said the cpu should be mips32, not mips32r2
<viric>
maybe it is something in between
<viric>
wpwrak: the difference between mips1 and mips32 is at the time of compiling the libc, not the program
<wpwrak>
(mips1) ah, okay
<wpwrak>
(mips32r2) maybe it turns into a mips32r2 only for benchmarks ;-)
<viric>
tuxbrain2: can you paste somewhere your asoundrc configuration?
<tuxbrain2>
viric actually using jlime , it can be valid?
<tuxbrain2>
also if you can tell me where this file is sopossed to be will be helpfull as well, I using opkg that drain resources as hell and not able do do a #find
<viric>
yes
<viric>
I only want the soft volume
<viric>
tuxbrain2: ~/.asoundrc?
<viric>
or /etc/alsa/asoundrc? I'm not sure
<tuxbrain2>
I have no found any asoundrc file in jlime---?
<viric>
oh.
<viric>
and you have softvol?
<viric>
Btw, I'm trying to run mpg321, and it does not work fast enough for the ben
<viric>
although it uses the same libmad 'madplay' should use.
<turtlee>
I got my nn on Monday and have been enjoying it.
<wolfspraul>
same here - what takes you here?
<wolfspraul>
oh nice! well thanks for buying one first of all!
<wolfspraul>
you just entered a massive construction site :-)
<turtlee>
Hehe. I can see that! But it's well worth it. It takes me back to the fun of discovery I enjoyed as a child in the late 1980s.
<turtlee>
I'm here tonight because I've been playing with the startup sequence and I'd like to understand it better.
<wolfspraul>
turtlee: "I am a craftsman and I live and work within the belief that fine, durable workmanship combined with good design helps create a better world. I am passionate about quality and have been fortunate to study its expression in wood, metal, leather, and stone."
<wolfspraul>
that's you?
<turtlee>
Indeed it is.
<wolfspraul>
great! we wish we get to those standards one day... it's a very good mission statement.
<wolfspraul>
Werner has been doing some 3D scanning work lately, have you seen that?
<turtlee>
Thanks very much. I also like this quote from Anthony Bourdain that a friend sent me the other day:
<turtlee>
People will continue to pay for quality. They will be less and less
<turtlee>
inclined, however, to pay for bullshit."
<turtlee>
woops.
<turtlee>
"If there's a new and lasting credo from the Big Shakeout, it's this: People will continue to pay for quality. They will be less and less inclined, however, to pay for bullshit."
<turtlee>
Well, close enough.
<wolfspraul>
well exchanging these nice words is good, but for the Ben we have to keep our expectations realistic. It's a long way.
<wolfspraul>
for example several connectors really suck, sorry about that
<wolfspraul>
as a mechanical guy it's painful for me to imagine that you take a closer look one day
<wolfspraul>
battery connector for example
<wolfspraul>
microSD connector, especially insertion pin
<turtlee>
Fair enough, but it's still a pleasure to hold, and that's a start.
<wolfspraul>
thanks, yes. it's not a hopeless start.
<wolfspraul>
did you see werner's work?
<wolfspraul>
(the url)
<turtlee>
I did. It's wildly cool. Thanks very much.
<turtlee>
So, were there not solid models before the molds were made?
<wolfspraul>
we bought the design, and the tooling process was done without any input or oversight from us
<turtlee>
Ah, that makes sense.
<wolfspraul>
our volumes are too low for our own professional mechanical design right now
<wolfspraul>
but now we have work like Werner's so hopefully we get this process all opened up slowly
<wolfspraul>
werner is hanging out here in irc, btw, his nick is wpwrak
<turtlee>
At least you found a way to produce at your low volumes.
<wolfspraul>
yes
<wolfspraul>
very hard work, but it's moving
<turtlee>
I preordered a Pandora last December. Ultimately gave up.
<turtlee>
Cancelled and ordered a BeagleBoard and an nn.
<wolfspraul>
yes our project is very different from those, in priorities
<wolfspraul>
for me manufacturability comes first
<wpwrak>
welcome, turtlee ! you came to the right place :) mechanical is the weakest area of the whole process (and the device has its flaws there, too)
<turtlee>
Thanks, Werner. :)
<turtlee>
I look at the frustration on the Pandora boards, and I think if they'd started from manufacturability at the outset, they could have arrived at this point with a lot more polish on the software and a lot more enthusiasm, even after the long wait.
<nebajoth>
hell yeah
<nebajoth>
wolfspraul is a pragmatist
<nebajoth>
and it makes all the difference in the world
<wolfspraul>
the pandora guys are fighting a massive uphill battle
<nebajoth>
just like torvalds is a pragmatist
<wolfspraul>
first they use more 'high-tech' than we have in the NanoNote, like the OMAP chip, and Wi-Fi
<turtlee>
I've been cruising around playing Flashback (via REmeniscence) and Monuments of Mars (via dosbox) on my nn.
<wolfspraul>
then they do some work steps in China (mechanical), but they seem to have zero experience in actual manufacturing, so they learn painful lessons, and slowly.
<wolfspraul>
now they are moving more work to the UK, and sooner or later they will realize that costs are going through the roof.
<wolfspraul>
so I don't know.
<wolfspraul>
I make my little NanoNote.
<wolfspraul>
maufacturing under control
<turtlee>
And I enjoy your little nanonote.
<wolfspraul>
can sell profitably at 99 USD, actually working towards reducing prices
<wolfspraul>
I take the beating and laughing over the hardware specs, no problem. Others can wait for their Pandora, position whatever in the queue :-)
<turtlee>
Heh, and they're just another layer. Look at the comments at Engadget about Pandora, and it's the same story.
<wolfspraul>
people don't like the hw specs anymore?
<wolfspraul>
engadget is typically just a hw spec pissing match
<turtlee>
Exactly.
<wolfspraul>
"hey, my LCM has 40,000 dpi, how about yours?"
<turtlee>
So, here's my experience:
<wolfspraul>
it's hilarious :-) also engadget has how much percent news that is just announcements? I stopped wondering.
<wolfspraul>
of course if you have to compete over numbers eventually you find out the numbers in the announcements are so much higher
<wolfspraul>
we should announce a new NanoNote, with specs that beat the iPhone
<wolfspraul>
I'm sure engadget would take the 'story' (ahem)
<turtlee>
Hah! That'll net you some press, at any rate.
<wolfspraul>
yeah, embarassing, isn't it.
<wolfspraul>
this will never happen here.
<turtlee>
Bravo.
<wolfspraul>
I rather build it slowly and through actual strength rather than announcements strength.
<turtlee>
You and I are of the same mind.
<wpwrak>
wolfspraul: you're too late. there's that video about the "xphone" ...
<turtlee>
I took my foray into app-phones with a Droid, and am nonplussed. I think it's fair to say that I am not the target audience for convergence devices.
<turtlee>
Anyway,
<turtlee>
I do have some specific tech questions.
<turtlee>
(and will be back here to talk shop again. you guys are fantastic.)
<nebajoth>
what
<nebajoth>
its a trap!
<wolfspraul>
wpwrak: hah yes, it's great! thanks for the link didn't see it before
<turtlee>
I've already deactivated the gp2x menu
<turtlee>
per the new user guide instructions on the wiki.
<nebajoth>
real men run debian
<turtlee>
nebajoth: Ah! So they do. Do we have a good keymap for debian yet?
<nebajoth>
well
<nebajoth>
yes
<nebajoth>
not integrated into the rootfs yet though :D
<nebajoth>
but nothing keeping anybody from downloading it and using loadkeys
<nebajoth>
and even making it automatic
<turtlee>
Brilliant.
<turtlee>
Because that would probably induce answers to my other questions right away.