<kyak>
xiangfu: hi! were you any luck getting rid of that "no job control" message?
<xiangfu>
kyak: no.
<kyak>
me to -\
<kyak>
i think gmenu2x is detaching itself in a way, that such tricks don't help
<xiangfu>
kyak: when we run 'bash' don't have such message. so is that maybe busybox problem ?
<kyak>
xiangfu: maybe.. or maybe just bash silently ignores it
<kyak>
did you try if the job control is actually woprking in bash?
<xiangfu>
kyak: oh.. job control not working in bash.
<kyak>
bash is a good software, makes you feel better by being silent :)
<kyak>
xiangfu:Â Â about that "alsa .." messagfes... i think they are due to the fact that load_modules in /etc/init.d/boot is called later then gmenu2x is started
<kyak>
and gmenu2x is trying to do some things to alsa.. maybe adjust the volume level
<kyak>
but the kernel modules are not loaded yet
<kyak>
xiangfu: yesterday i tried redirecting all this crap from gmenu2x in /dev/null, but unfortunately, the io of child processes (ie.. shells) is redirected there, too :)
<xiangfu>
kyak: ok
<xiangfu>
the job control on 'ash(default)' and 'bash' not working. but in fbterm works fine
<kyak>
fbterm is a terminal emulator itself, ash and bash is shell only
<xiangfu>
ok. so shell must attach to one device (tty1, ttyS0) for make the jobs control working. right?
<kyak>
indeed
<viric>
Hello back
<viric>
question...
<viric>
is anyone using an ubifs *Without compression* in the nanonote?
<rafa>
viric: for that experiment lzo as you use looks sexy ;)
<viric>
I mean, lzo was the default
<viric>
I did not change anything :)
<rafa>
there is not a lot of more space with zlib, and the speed looks a lot better with lzo
<viric>
no wonder it's their default
<viric>
what about size, in jlime... a usual installation on the nanonote, does it take much?
<viric>
rafa: those tests do not show if the speed is CPU-bound
<viric>
rafa: I'd expect that the read speed would be higher, if the bound is the nand-reading speed
<wpwrak>
would be nice if someone made a comparison
<rafa>
viric: jlime currently uses the whole nand as just one partition for rootfs.. I mean, bootloader, kernel and rootfs
<rafa>
no datafs
<rafa>
viric: those tests could be different in nn, but well, those shows something to think :)
<wpwrak>
rafa: i think we have very slow nand, so the question whether the increased amount of data or the processing time is more important may get interesting :)
<viric>
wpwrak: I'll try to test a bit
<viric>
it can be changed at runtime for new files... so it should be easy
<viric>
btw
<viric>
I already decided to make public how I play with the nanonote
<wpwrak>
kristianpaul:Â Â i think so, yes. and ram is only 16 bits.
<kristianpaul>
Where is the 32bits system!!
<kristianpaul>
:)
<kristianpaul>
bus always are the bottleneck
<viric>
wpwrak: a netcat to /dev/null gives 735kB/s. there is no nand or ubi in that
<kristianpaul>
:/
<viric>
(maybe I built something wrong. Who knows :)
<wpwrak>
viric: what happens if you just move data locally, without involving usb/ssh ?
<kristianpaul>
sdram can work at 32 bit data bus
<viric>
you mean to check the ubi cpu usage?
<wpwrak>
netcat seems indeed quite slow
<wpwrak>
yes
<viric>
sory
<viric>
netcat was in the PC
<kristianpaul>
ah?
<kristianpaul>
ahh
<viric>
in the nanonote I typed: cat /dev/zero > /dev/tcp/192.168.0.100/3000
<kristianpaul>
that explaina bit
<viric>
I used bash for the connection. But most cpu is taken by cat.
<viric>
(~15%.    75% idle says top, in the nanonote)
<wpwrak>
top may misrepresent idle time. a better test is to run something at a low priority in the background. see how many cycles that background task has left.
<viric>
so you think that cat+bash maybe at 100% of cpu
<viric>
?
<viric>
# time cp doom.wad doom.wad2
<viric>
real    0m11.947s
<viric>
user    0m0.060s
<viric>
sys    0m5.730s
<viric>
'sys' may be the ubi cpu time, not waiting for nand
<viric>
(12MB file)
<wpwrak>
viric: if cat, bash, and kernel don't eat all the cpu, that would mean that usb itself is the bottleneck. hard to imagine with high-speed usb. well, maybe check if the ben really registered as a high-speed device.
<viric>
I check...
<wpwrak>
also, is doom.wad cached when you start or not ? yuo can flush the cacke with blockdev --flushbufs
<viric>
mmm
<viric>
I forgot to mknod the mtd devices :)
<viric>
I can't flushubfs
<viric>
I don't know what we are testing, already :)
<wpwrak>
;-))
<wpwrak>
you could  dd if=/dev/zero of=file bs=1M count=100
<wpwrak>
do this to two different files. that should clear out the buffer cache too.
<viric>
for sake of science
<wpwrak>
ah, and include a sync in the timing :)
<viric>
same numbers for both, and the sync took less than one second
<viric>
real    0m13.690s
<viric>
user    0m0.020s
<viric>
sys    0m5.240s
<wpwrak>
that's not so bad 7.3 MB/s
<wpwrak>
now, time dd if=file of=/dev/null
<wpwrak>
of course, nulls should compress really well ;-)
<kristianpaul>
xD
<kristianpaul>
how you get verbose mode in dd at openwrt..
<kristianpaul>
?¿
<wpwrak>
for the next test, maybe use 100 MB of data from /dev/urandom. will take a while to generate, though.
<wpwrak>
dd and verbose ? :)
<kristianpaul>
i meant resume
<kristianpaul>
it fiinshed no speed/stats..
<kristianpaul>
as in my computer..
<kristianpaul>
uramdom is heavy
<wpwrak>
just use  time dd  ?
<kristianpaul>
ah
<kristianpaul>
he
<kristianpaul>
1M count 100 takes real 15.856s
<kristianpaul>
sys 5.390s
<viric>
close to here
<viric>
then.. quite normal
<kristianpaul>
ah
<kristianpaul>
can i do FS on ram with the current kenrel?
<wpwrak>
kristianpaul: (random) use /dev/urandom, not /dev/random ;) /dev/urandom will make up fake random numbers if there's no real random data. that's good enough for what we need.
<wpwrak>
12 MB/s with random data ? not bad
<viric>
hm
<viric>
12MB/s ?
<viric>
why dd tells me 213KB/s ?
<viric>
you mixed seconds and minutes? :)
<viric>
rafa: uh. I lost again the link to the keyboard map
<viric>
rafa: but isn't jlime using the openwrt kernel?
<viric>
kyak: I am running the openwrt 2.6.35 kernel. Maybe it's a difference?
<rafa>
viric: openwrt kernel is something general to say..
<rafa>
viric: qi distribution uses 2.6.32 qi openwrt kernel
<rafa>
viric: lars works on 2.6.36 kerne I woud guess
<rafa>
viric: jlime uses 2.6.34 kernel from qi openwrt.. and blizzard from jlime is doing some work with latest lars qi openwrt kernel as well
<viric>
And all them have different keycodes from the keyboard matrix?
<viric>
How could all this become such a mess? :)
<rafa>
viric: and those kernel could be a bit different, maybe no with keyboard.. but i can not be sure
<rafa>
viric: maybe you need the keymap file from kyak :)
<viric>
Ok, so... I adapted that 'grave', and the shift-volume keys for scrollbuffer... and I'm happy.
<viric>
all the rest worsk
<viric>
Well, there is a circle over 'A'
<rafa>
cool
<viric>
which I can't imagine what it is.
<rafa>
an angel symbol? :)
<viric>
° ?
<viric>
5°C
<viric>
maybe this.
<wpwrak>
viric: it's a nordic character. i think something like a long "A"
<viric>
ahh
<viric>
in swedish it's an 'a' that sounds like 'o'
<viric>
(iirc)
<kyak>
viric: ok, maybe it's too late, but: target/linux/xburst/patches-2.6.32/500-modifier-keys.patch is a patch for linux-2.6.32.25/drivers/char/defkeymap.map, and then defkeymap.c should be generated, but it is in fact replaced..
<viric>
I'm not very interested in 'building openwrt'
<viric>
I think I cannot run that in my system
<viric>
I'm not using a LSB GNU system
<kyak>
then why did you clone both openwrt-xburst and wanted to clone openwrt-packages? :)
<viric>
I cloned openwrt trunk
<viric>
I use the xburst patches there for 2.6.35 to build the kernel
<kyak>
ok then
<viric>
I don't use the "qi openwrt", I use "openwrt trunk"
<viric>
how bad is it? :)
<kyak>
i got it
<kyak>
who knows?
<viric>
This is what worries me. hehe
<kyak>
it's rather weird
<kyak>
using something in the middle
<kyak>
not the stable backfire (2.6.32.25) or latest qi-kernel (2.6.36)
<viric>
I'm taking notes of all this... let's try to summarize with some links.
<viric>
kyak: I used 'the latest' when it was 'the latest', but it's no more 'the latest', and I did not care to follow 2.6.36 still, because I'm not sure of the benefits I'd get.
<viric>
And at that time I even did not know people where not using 2.6.35. :)
<viric>
I'm quite new in all this.
<viric>
kyak: so, how do you know that the stable backfire is 2.6.32.25 ?
<viric>
kyak: and, is 'stable backfire' the name of anything official? any official download image?
<kyak>
cause we are following the backfire, that's why
<viric>
- if there is anything 'official', I don't know
<viric>
backfire is an openwrt major version?
<kyak>
yes
<viric>
(sorry, I really know little all this... I don't want to annoy)
<viric>
kyak: qi-hardware openwrt-packages branch is following backfire. It has 'features' one day to be merged into openwrt?
<kyak>
qi-hardware openwrt-xburst is following backfire, and contains additional features
<kyak>
then these features are picked and commited upstream
<kyak>
and then they return back :)
<viric>
ahh
<viric>
and what is openwrt-packages?
<viric>
in qi-hardware
<kyak>
there are packages, that are not ready yet to go upstream
<viric>
but in the usual openwrt, packages and 'xburst' are in a single tree, right?
<kyak>
openwrt base system is differenet from "feeds"
<kyak>
no
<kyak>
that's the point
<kyak>
"OpenWrt uses so called "feeds" which provide packages for software beyond the default minimal system."
<viric>
ah
<viric>
openwrt-packages is a feed for that?
<kyak>
yes, it's one of the feeds
<viric>
openwrt-xburst is also a feed?
<kyak>
no, it's following backfire, the base system
<viric>
Ahhh
<viric>
I think now I get it :)
<viric>
is openwrt-packages any related to 'backfire'?
<viric>
openwrt stable is not openwrt, trunk, right?
<kyak>
well, you can add this feed
<viric>
ok
<kyak>
and use it in backfire
<kyak>
openwrt stable is backfire release
<kyak>
trunk is trunk
<viric>
ok
<kyak>
:)
<viric>
It may look evident to that used to it... but a few minutes ago I had a completely distorted view compared to now
<viric>
And where is written, what kernel openwrt-xburst takes in?
<viric>
(I'll try to find by myself)
<kyak>
it is located in target/linux/
<dandon>
hey
<viric>
kyak: yes, got it. thank you
<viric>
And then... rafa, there is jlime. the jlime now distributed in the web uses that same 2.6.32 kernel, right?
<viric>
kyak: openwrt trunk has many patches for xburst, for 2.6.34, 2.6.35, 2.6.36, 2.6.37 ... who maintains that and with what purpose?
<dandon>
does anyone know a website that reviews apple notebooks? but websites like xbitlabs or bit-tech or similar that also have non apple laptops
<wpwrak>
ah, sega platform. published by absolut entrtainment
<wpwrak>
absolutE. important :)
<zear>
if sega, then it must be sega genesis :)
<zear>
ah, it's for sega CD
<zear>
still, there's a quite decent sega genesis/megadrive/CD emulator for the dingoo
<zear>
i don't see a reason why it couldn't be ported to the nanonote
<wpwrak>
zear: have you looked at the video ?
<zear>
of the game? I know about this game :D
<zear>
this is actually a sort of game i would enjoy
<zear>
if it wasn't THAT long and boring, of course
<zear>
but a simple bus/track simulator where you don't have to do much but watch the road would be interesting
<zear>
just would have to have a very varying scenery, so you can't get bored that easily
<viric>
wpwrak: (that review was quite bad, yes! So... I know some fans of that game. I'm one of them!)
<zear>
about that tokyo bus game, this is quite a funny coincidence, but i got a dreamcast few days ago and was about to try this game out :)
<viric>
wpwrak: at university, a friend and me were given a board with an fpga, and pins for a VGA connector and a PS2 input. We made the videogame in an Altera fpga. :)
<viric>
making the vga signals (digital, no DAC... but it gave enough colours :)
<viric>
anyway, guten nacht
<zear>
speaking of simulation games, you guys ever played Densha de go? It's a japanese train simulator, a really cool thing
<wpwrak>
zear: combine it with OSM, and maybe you have something usable
<zear>
OSM?
<wpwrak>
zear: ah ... you commented on it. hadn't seen it
<wpwrak>
zear: OpenStreetMap
<zear>
ah
<zear>
open street map is a little.. incomplete :D
<zear>
at least the city where i live :D
<zear>
so i basically use google maps on my freerunner
<zear>
i know, it's a shame
<zear>
but they have better maps
<zear>
and with the street view of their, you could pretty much hook it up to a bus driving game like this one
<wpwrak>
viric: (game in fpga) nice :)
<zear>
wpwrak, they even have some sort of 3d models they generate for every street view pic
<zear>
so they can know if the part of the pic is a building or not
<zear>
and the transition between pictures is very smooth, sometimes it feels like you're really moving from place to place and not changing pictures
<zear>
here's an example: http://mapy.google.pl/maps?f=q&source=s_q&hl=pl&geocode=&q=japan,+tokyo&sll=35.423819,139.936552&sspn=0.016262,0.023453&ie=UTF8&hq=&hnear=Tokio,+Japonia&ll=35.768835,139.766049&spn=0.008061,0.011727&t=h&z=17&layer=c&cbll=35.769187,139.766419&panoid=M1SNVuzoFvYEstH7nCnyUg&cbp=12,167,,0,5