<tuxbrain>
rafa: is there is no patent package , I will promote jlime in nano, in fact if there is a nand version to flash are is clean for before september the Nanowar version will be based on jlime
<viric>
larsc: the jz-2.6.35 kernel does not have a qi_lb60_defconfig
<larsc>
i know
<viric>
ok, I'll go for jz-2.6.34 then, right?
<larsc>
well, you can use the qi_lb60_defconfig from jz-2.6.34 and run a make oldconfig
<viric>
ok
<viric>
larsc: what is the difference between the uboot that resides in the nand, and that to be loaded with usbboot ? The 2nd is an ELF file directly?
<larsc>
what do you mean by "the uboot [...] loaded with usbboot"?
<viric>
larsc: hm I thought that usbboot can load an uboot image
<viric>
larsc: I can use usbboot to load a kernel into the nanonote and run it, right? Can't I do the same with an uboot image?
<viric>
(without flashing the nand I mean)
<larsc>
you should be able to do that, yes
<viric>
what image format usbboot wants? ELF?
<xiangfu>
viric, how you use usbboot load a kernel into nanonote??
<xiangfu>
viric, the command "xbboot" can do that.
<viric>
xiangfu: ah, well, I mix xboot and usbboot
<larsc>
viric: raw bytecode
<larsc>
i think
<xiangfu>
yes.
<viric>
And how can I build 'raw bytecode' ? The uboot nand image, for example?
<viric>
And for the kernel?
<xiangfu>
viric, when you build U-Boot, there will be "u-boot-nand.bin" "u-boot.bin"
<viric>
right
<xiangfu>
viric, "u-boot-nand.bin" is for nand image. "u-boot.bin" can load and run at 0x80100000 in NanoNote.
<xiangfu>
viric, for kernel you need zImage.bin. address at 0x80600000
<viric>
ah great
<wpwrak>
hmm, when I commit huge binaries (bz2) to projects, will they show up in all their glory on the commitlog list, overflowing everyone's mailbox ?
<xiangfu>
viric, for example "sudo xbboot -u 0x80100000 u-boot.bin"
<tuxbrain>
I have not done any significant work after that.
<tuxbrain>
My dream is to have same tools I have in debian avrdude, avr-gcc, in both in jlime and openwrt , I don't like very much to work with debian (please don't blame me is only matter of taste)
<tuxbrain>
I love to work with debian on big computers but I feel more confortable with jlime and openwrt on Ben
<zear>
bartbes, i read you're trying to fix path for gmenu2x?
<larsc>
bartbes: since gmenu2x is also hosted on projects.qi-hardware we should rather fix the bugs there instead of adding patches to the openwrt package
<zear>
because i thought this was fixed long time ago on qi-hardware projects
<bartbes>
I mentioned that on the developers mailing list
<bartbes>
zear: it wasn't
<zear>
bartbes, weird
<bartbes>
I just pushed a patch
<zear>
i haven't checked your patch, but just simply changing /card to / won't fix it ;)
<bartbes>
it almost did
<bartbes>
that and fixing the filter
<zear>
you also have to change the number of chars
<zear>
from 5 to 1
<bartbes>
I never saw number of characters listed as 5
<bartbes>
checks again
<zear>
hmm.. maybe then 4
<zear>
maybe it doesn't count / or starts with 0
<bartbes>
because my tests all succeeded anyway
<bartbes>
I believe it said 11
<zear>
bartbes, try to go up a dir
<zear>
it will crash and return to gmenu2x
<zear>
because the number of chars is incorrect
<bartbes>
right
<bartbes>
but it said 11
<bartbes>
so I figured that was unrelated
<zear>
ah, right, this was my mistake
<zear>
because i ported it stright from dingoo
<zear>
where the root location is : /boot/local/
<zear>
so it's probably 11
<bartbes>
well, I guess you could apply the patch and fix that
<zear>
yep
<bartbes>
push it to upstream
<bartbes>
and then update the package to point to the new commit (if it's 'stable')
<zear>
but i thought it was fixed long time ago because i was talking about it every second day for at least a month after i discovered it :D
<zear>
oh, and btw if the root is set to /, the auto searching option will lock up
<bartbes>
maybe you did fix it upstream
<bartbes>
but the 'downstream' package is still pointing to an old commit
<Textmode>
heh
<zear>
well, i tested it with / only on my x86 pc, maybe it just locked up on chmod problems (gmenu2x was written for gp2x which ran on vfat)
<zear>
but who knows, maybe it's locking up on symlink loops
<bartbes>
in any case, *you* are the one who can commit upstream, meaning *you* should fix it all :P
<zear>
no, i can't commit because i neither have an account nor i know how to do it ;)
<bartbes>
?
<bartbes>
then who does have commit access?
<zear>
mth, larsc
<zear>
that guys
<bartbes>
well you sure beeped them now
<bartbes>
:D
<zear>
:D
<zear>
maybe they didn't change anything yet, because this repo is for gmenu2x for both the nanonote and dingoo (mth is working for dingoo fixes)
<tuxbrain>
I propose to disable the autosearching options
<zear>
so maybe they were planning to change it somehow else
<zear>
tuxbrain, that's not a good idea, because it's vital for dingoo
<bartbes>
well, yeah, but aren't we allowed to get a short-term fix?
<bartbes>
I'd be happy to edit the patches some more
<bartbes>
to get it working *today&
<tuxbrain>
mmmm, ok
<zear>
bartbes, there's also another reason this hasnt' been changed to /
<bartbes>
because tbh without these fixes gmenu2x is pretty useless
<zear>
because gmenu2x thinks this is SD card mount location
<zear>
and the icon indicating the free space has SD card icon
<bartbes>
minor inconveniences
<zear>
i think the guys were thinking of adding both / and sd mount location at the same time
<larsc>
bartbes: i added your account to the gmenu2x project
<zear>
and i think i saw some code towards it
<bartbes>
especially when compared to the current non-functioning
<bartbes>
larsc: wow, thanks
<bartbes>
but now I have to go read the code :P
<zear>
bartbes, the current non-functioning is thanks to me :D
<zear>
you can hit me in the face now :D
<bartbes>
btw, zear, I only changed the root in the file dialog
<bartbes>
just saying
<zear>
bartbes, there's more lines to change
<zear>
just grep -ir "/card" ./
<bartbes>
not if I just want the file dialog to work?
<bartbes>
:P
<zear>
:P
<bartbes>
larsc: btw, is there a reason why the filedialog is still filtering ".gpu,.dge,.sh"?
<zear>
bartbes, it's because i did a straight port from the dingoo :D
<bartbes>
...
<bartbes>
right
<bartbes>
the new code
<bartbes>
is more confusing
<bartbes>
(of course it is..)
<bartbes>
oh I see
<bartbes>
it has moved to browsedialog
<zear>
gpu is gp2x's extension for programs (as opposed to games which are .gpe), dge is dingoo extension
<zear>
just saying if you wondered what they are
<bartbes>
figured something like it
<bartbes>
no wait
<bartbes>
I knew that was the cause
<bartbes>
you know what? I'll fix the patches now
<bartbes>
get to know gmenu2x and then work on a long-term fix
<zear>
bartbes, don't remove the .dge, it's vital for the dingoo ;)
<mth>
I introduced CARD_ROOT as a constant, so you only have to modify one line
<zear>
mth, i see
<bartbes>
true
<bartbes>
but
<bartbes>
then the other errors pop up
<bartbes>
I'll update my patch to do dir-up correctly
<bartbes>
and then I'll start reading it all
<bartbes>
zear: actually
<bartbes>
due to the path being /
<bartbes>
the whole size comparison stuff can be killed
<mth>
the gmenu2x code is a big mess, but we (larsc, Ayla and I) have cleaned it to be less of a mess than it was at the start
<mth>
bartbes: why commit it as a patch instead of a change directly on the gmenu2x code?
<bartbes>
2 reasons
<bartbes>
first, current gmenu2x code I haven't 'mastered' yet
<bartbes>
second, the package is lagging several commits anyway
<bartbes>
oh, and third, I will fix it upstream
<bartbes>
just wanted a short-term solution as well
<mth>
ok, as long as it's fixed in the gmenu2x repository it's fine with me
<bartbes>
I see I can easily change that repo to at least allow going up to root
<bartbes>
or CARD_ROOT
<bartbes>
whatever you want
<bartbes>
(though / is easier)
<mth>
I don't really understand the idea behind the multiple roots thing anyway
<mth>
well, to show free space it's useful
<bartbes>
yeah, that is the long-term solution
<bartbes>
turn everything into / and fix the space indicator
<bartbes>
I guess
<mth>
well, on the Dingoo there are three "roots": the actual system root, the SD card and the NAND
<mth>
the system root is a loopback mounted readonly file system, so there will be no apps on that
<mth>
but it would be useful to show the free space of both the SD and the NAND
<bartbes>
ugh, I'm stuck with the sound issue anyway
<bartbes>
I'll go look at that
<bartbes>
so what do you want me to do?
<bartbes>
make it all /, or /card?
<mth>
I think / is more useful: it would allow access to too many directories, but that's better than too few
<bartbes>
nvm, I know the answer
<bartbes>
oh
<bartbes>
I thought you'd say "keep it CARD_ROOT which then is defined at compile-time"
<bartbes>
anticlimatic here
<mth>
CARD_ROOT is better than "/card" hardcoded, that's true
<bartbes>
but browsing should go to /
<bartbes>
right?
<bartbes>
or what?
<bartbes>
I'm confused :S
<mth>
is the browsing for apps or also for data files?
<tuxbrain>
well actually gmenu doesn't rely apps with data files to open them so I guess is only for apps
<bartbes>
well
<bartbes>
the goal is to find apps
<bartbes>
but the filter is supposed to take care of other files
<mth>
in the case of the Dingoo apps could be on the SD or on the NAND
<mth>
so it would be good to be able to browse to both locations
<bartbes>
meaning /?
<mth>
one way of doing that would be to have multiple roots, but that's a big change
<mth>
so / would be a good compromise for now, I think
<bartbes>
okay
<bartbes>
I'll do that in a few mins
<tuxbrain>
rafa:ping
<tuxbrain>
rafa: ping
<rafa>
tuxbrain: hey morning
<tuxbrain>
good afternoon :)
<rafa>
well, here is almost noon in one hour.. just that I am late to start the day :)
<wpwrak>
rafa: you're early - the sun is still up :)
<rafa>
tuxbrain: ah.. I read .. cool.
<tuxbrain>
rafa wpwrak: good bless you with a child you will miss this wake up hours a LOT
<rafa>
tuxbrain: no news with theora yet, and our nand tries failed .. but let me chat with other jlimers about the idea ;)
<tuxbrain>
doesn't OE includes ubi as image format?
<wpwrak>
hmm, does anyone know if there's a place on projects where i can put an HTML file such that http://projects.qi-hardware.com/wherever/ will show it as HTML, not as "pretty" source ? or some magic option for the url ?
<rafa>
tuxbrain: kernel problem, no ubi
<rafa>
tuxbrain: but working on that ;)
<wpwrak>
tuxbrain: yeah, i figured out a while ago that reproduction and my lifestyle wouldn't go together too well :)
<bartbes>
mth: so have you replaced every occurance of /card with CARD_ROOT? (or only the ones you saw, so grep vs eye)
<wpwrak>
so, no idea for HTML on projects ? it's for the overview page for the scans
<mth>
bartbes: I used grep
<bartbes>
okay
<bartbes>
what platform uses .pxml
<tuxbrain>
rafa: due I also want to do a Makefile to have a ipkg package to openwrt to Nanowarize also openwrt users , I supouse that the ipkg than only will copyfiles to specified ubications containing no binaries, can be the same, just changing the gmenu menu files for whatever jlime is using now (desktop files?)
<tuxbrain>
bartbes: openpandora
<bartbes>
ah
<bartbes>
btw, scanning would add loads and loads and loads of executables on openwrt
<bartbes>
what extensions should I set it to?
<rafa>
tuxbrain: ah yes, I think so. Current jlime uses Matchbox.. so it uses freedesktop.org spec. So, it is so easy to add an icon with two files /usr/share/applications/your_file.desktop and /usr/share/pixmaps/your_file.png (icon)
<rafa>
tuxbrain: then we just need to add your ipk file two files to have the launcher icon on the matchbox desktop
<rafa>
(or commands in your Makefile)
<larsc>
yes, we should add support for .desktop file to gmenu2x
<tuxbrain>
well it will be some icons I want to add various Nanowar content (albums, lirics, history, artwork... ) and each of it will have it icon
<tuxbrain>
In openwrt will be a new tab, in jlime a folder
<bartbes>
mth: so how do I build it?
<tuxbrain>
larsc:Â Â this will be awesome
<mth>
bartbes: this is what I use for Dingux: ./configure --host=mipsel-linux --with-sdl-prefix=/opt/mipsel-linux-uclibc/usr
<mth>
the SDL prefix would be different for the NN
<bartbes>
and how do I set the toolchain location?
<mth>
I have the toolchain in the $PATH
<mth>
the host= argument tells autotools to cross compile
<bartbes>
right added it to my path
<bartbes>
ah
<bartbes>
there we go
<bartbes>
I needed ot set host to mipsel-openwrt-linux
<bartbes>
hmm
<bartbes>
and sdl fails
<bartbes>
ah, ~ is invalid
<bartbes>
thanks
<viric>
larsc: sorry, I just came back. For the upstream linux, you can ask "make uImage", and it calls mkimage after building the image.
<bartbes>
mth: interestingly enough it fails to accept input
<larsc>
viric: i guess it depends on your target architecture
<wpwrak>
sigh. where is wolfgang when i need him ... ?
<mth>
bartbes: I always start gmenu2x by letting init respawn it
<mth>
maybe that makes a difference (vs starting from telnet/ssh)
<bartbes>
I never do
<bartbes>
:P
<bartbes>
oh, I just use the console on the nanonote
<mth>
ah, the Dingoo doesn't have a keyboard, so that's not an option there
<bartbes>
well, I run it from the dir the current version is installed
<bartbes>
so I guess it should pick up on that input.conf
<mth>
I don't know whether it uses the current dir or looks up the dir of the executable
<bartbes>
I guess current dir
<bartbes>
because when I ran it in executable dir it segfaulted
<bartbes>
;)
<bartbes>
though I could copy the confs over
<bartbes>
that helped
<bartbes>
well, I guess I got it running
<bartbes>
I can browse everything
<bartbes>
hmm I messed up somewhere..
<bartbes>
I hadn't added that yet :P
<tuxbrain>
xiangfu: there are plans for new release?
<xiangfu>
tuxbrain: hmm. we need ask Mirko Vogt :). we have done some preparation work. I even start to write changelog about next release.
<tuxbrain>
mirko: there are plans for new release? :P
<xiangfu>
tuxbrain: there a lot of new packages. and the 32bit font work really nice in NanoNote.
<tuxbrain>
that's why I propose it's time to do one :)
<xiangfu>
tuxbrain: yes.
<bartbes>
mth: found the problem, the one who added those std::string.compare statements reversed the check :P
<tuxbrain>
also due I want to release a special version  for september It would be better to work with the last one :)
<bartbes>
or.. not..
<bartbes>
wait.. now it's getting weird
<tuxbrain>
how perverse is marketing.. special version=some image and pinping _P
<bartbes>
tuxbrain: marketing is manipulation
<bartbes>
mth: there we go
<bartbes>
time for a commit
<qi-bot>
[commit] bartbes: Changed the file browser to use / as root and added ability to see extensionless files (i.e. executables) http://qi-hw.com/p/gmenu2x/ba9af53
<mth>
bartbes: eh, setting CARD_ROOT to / means the free space indicator will be wrong on Dingoo
<bartbes>
right
<larsc>
bartbes: I think the "|| path.length() <= CARD_ROOT_LEN" check is superfluous, at least if it is done after the path has been compared with CARD_ROOT
<bartbes>
larsc: it is not though
<mth>
I thought you were going to let the browser go to / while CARD_ROOT would stay the same
<bartbes>
because
<bartbes>
it checks *before* changing dirs
<bartbes>
so if you're at /the/root
<bartbes>
and press ..
<bartbes>
it allows you to go to /the
<bartbes>
with this code you at least have to be at /the/root/<something?
<bartbes>
mth: was I?
<bartbes>
so
<bartbes>
what are the problems with CARD_ROOT /?
<mth>
apparently not, but that was what I assumed :)
<bartbes>
just the sd card indicator and stuff?
<bartbes>
*- and stuff :P
<mth>
I don't know what else CARD_ROOT is used for
<bartbes>
greps
<bartbes>
it's used in selector
<bartbes>
what is selector?
<bartbes>
I'll revert CARD_ROOT
<bartbes>
and then change those related to file browsing to a hardcoded /
<bartbes>
is that okay? (mth)
<viric>
larsc: then to build a kernel, I better look at the openwrt makefiles?
<mth>
bartbes: yes
<larsc>
viric: yes
<bartbes>
so what is selector.cpp?
<mth>
I have no idea
<rafa>
tuxbrain: booting from nand now ;)
<mth>
larsc: are the changes from the jz-2.6.34 branch merged into jz-2.6.35?
<tuxbrain>
rafa: GREAT!!! giveme giveme
<rafa>
tuxbrain: okey, okey! just for you, nobody else right? :D
<mth>
anyway, if it has to stay, is it possible to replace "5" by "strlen(CARD_ROOT)"?
<mth>
in the definition of CARD_ROOT_LEN, I mean
<bartbes>
guess that works
<bartbes>
but then again
<bartbes>
you could change all occurences of CARD_ROOT_LEN to that
<bartbes>
yeah, that might look better
<bartbes>
the compiler should optimize that out anyway
<larsc>
mth: it should be, yes
<mth>
does the compiler replace strlen calls?
<bartbes>
it should
<bartbes>
well, if its input is hardcoded
<mth>
but strictly speaking strlen() is not part of the language but of the standard lib
<larsc>
gcc has a built-in version of strlen
<bartbes>
which you can see if you ever forget to include <string.h> when you build code with strlen
<larsc>
nowadays it will give at least a warning
<bartbes>
I meant the error tells you that
<bartbes>
test.c:2: warning: incompatible implicit declaration of built-in function strlen
<mth>
actually CARD_ROOT_LEN is currently not equal to strlen(CARD_ROOT) because of the trailing slash
<mth>
is that intended or a bug?
<bartbes>
bug
<hallo99>
I replaced the rootfs. when booting I now only see openwrt image but no text. How can I tell the kernel to display text messages instead of the image, so I can debug?
<mth>
do I have to produce some kind of signature or is it just a line you can add?
<larsc>
just a "Signed-of-By: ..."
<larsc>
but i already send the mail
<larsc>
you are in CC
<mth>
ok
<mth>
yes, this is fine
<wpwrak>
rafa: of course, this should be over here. let future idbg users find the solution to their future problems in the archives :)
<wpwrak>
rafa: well, solution, except that i'm changing the pin in version 2 of the board :)
<rafa>
haha.. okey. I think that reset did the trick
<rafa>
But no sure.
<rafa>
I did like 1000 tries
<rafa>
then I ran like you suggested (to enable usb boot mode)
<rafa>
idbg-gpio P0_3=0
<rafa>
but no luck
<wpwrak>
the reset is essential. without it, nothing happens
<rafa>
then perhaps I did a reset
<rafa>
but I did not see anything in console
<rafa>
so I do not remember if I did or not
<wpwrak>
i think the usb_boot signal is only sampled at hardware reset
<wpwrak>
maybe if you do a software reboot too, but i'm not sure
<wpwrak>
(don't have any system up and running now)
<rafa>
wpwrak: all was black and dark :).. But well, I mean, I did all but without signals to go well. For example, I did reset, almost sure. But console just shows me : [ Connected ]
<rafa>
so I do not remember now exactly
<rafa>
and that [ Connected ] message is before to the reset
<viric>
larsc: I noticed the 'make uImage' works for some architectures (arm, avr32, powerpc, sh, ...), but not for mips
<viric>
arch/*/boot/Makefile
<wpwrak>
"console connected" only means that you're talking to idbg
<wpwrak>
rafa: idbg doesn't know what's happening at the other end of the serial line
<rafa>
yes, I just would like to see one nn led or something :)
<rafa>
when resseting
<rafa>
as feedback
<wpwrak>
;-)))
<wpwrak>
just trust idbg ;-) if you idbg-reset and you don't get an error, it did reset the cpu
<viric>
ahhh I remember this from the mailing list. ok.
<viric>
thank yo
<viric>
u
<wpwrak>
viric: yup. np :)
<viric>
larsc: openwrt-xburst builds the kernel 2.6.32, having patches that allow 'make uImage', and that was added by you I think :)
<viric>
also the git kernel 'master' branch has that
<kristoffer>
Has the flashing improved any lately? Is there any point in upgrading xburst tools?
<tuxbrain_away>
kristoffer: I will try the SD flash approach in short, I can report if you are interested
<kristoffer>
tuxbrain, please do
<tuxbrain>
in fact is to flash jlime :)
<tuxbrain>
did rafa talk to  you about my proposal?
<kristoffer>
tuxbrain, yeah we talked about it, sounds very interesting. But we need to get a good grasp of the details (atleast me), so rafa was to mail you and Wolfgang (and CC me)
<kristoffer>
tuxbrain, so expect the mail whenever rafa wakes up. Btw, that SD flash approach is a way to flash onto NAND by supplying files on the SD?
<tuxbrain>
kristoffer: yes
<tuxbrain>
kristoffer: then until tomorrow morning for me , lunch time for rafa :)
<kristoffer>
tuxbrain, timezones suck :) thx for link, Im having great problems with usbboot atm.
<tuxbrain>
mean while you have a u-boot able to to boot from sd, I think is the less anoying solution, I never have serious problems with usbboot but I find that sd aproach a little bit more confortable
<kristoffer>
tuxbrain, yeah. My issue currently is that it refuses to flash the kernel correctly (looks like 2 bad blocks inside kernel area). Trying to figure out some smart solution but so far no luck
<tuxbrain>
I will tell you now compiling and openwrt image, when it ends surelly tomorrow morning I will do some test
<tuxbrain>
that was one of my units, isn't it?
<tuxbrain>
I don't know if nandwrite will be less sensible to that and will allow you to smash that bad blocks
<kristoffer>
tuxbrain, no idea, might have been. Ive had issues like this before but it sorted themselves out after extensive nerase' s, but not this time.
<tuxbrain>
well only way is trying :)
<kristoffer>
Ive been at it for an hour or so, so at this moment it looks semi-bricked (atleast until I can figure out where the bad blocks are and adjust u-boot or kernel)
<viric>
talking about flashing...
<viric>
I use the latest xburst-tools (I think) and I have troubles flashing the first blocks
<viric>
(those of uboot)
<urandom__>
viric everyone has trouble flashing sometimes ;)
<rafa>
tuxbrain_zzzzzz: I am able to play theora.. but the video delays several seconds until it goes properly, so sound starts immediately and video no. After a seconds both are fluent but no sync
<rafa>
tuxbrain_zzzzzz: if you have jlime you can try. I converted to ogv with ffmpeg2theora, in jlime I used : mplayer movie-sample2.ogv -correct-pts