<focus_it>
1000 is very small order for the kind of factories I know
<focus_it>
The PCB factories alone can produce 10,000 PCBs in 3 days
<libv>
focus_it: again, look at these devices
<libv>
focus_it: they look like bogstandard .cn hw
<focus_it>
They all are - but normally each order is still a full production run. They can't like buy for 2 runs because they don't normally have the working capital
<libv>
i haven't heard of this before, and i doubt that this person is actually involved with linux-sunxi at all
<libv>
i could be wrong though
<focus_it>
A lot of the time, you send a factory an LC, and they borrow money on the LC and make the products
<focus_it>
They have to do that because $100 is wages for 1 month for some factory workers!
<libv>
but to me it just seems like a scam to ride on the open source wave and sell cheap chinese imports early
<libv>
dearly
<focus_it>
In western equivalent it would be about $1000.
<focus_it>
libv: I kind of agree with you
<libv>
i wonder who this person is, and to what extent he has been active in our community, and what he means with all that integration work that he mentions
<focus_it>
If I had the knowledge, I would have made a $30 tablet add 30% to cover overheads and profit
<focus_it>
I think dranchensun? I vaguely remember something - don't take my word for it tho!
<focus_it>
libv: I was interested in seeing Linux on the flash. So I order the 7" with flash programmed with Linux. Then I hope to learn from that - shipping dates may be Feb for the Linux on flash version tho!
<libv>
focus_it: you haven't managed to do so from the linux-sunxi.org wiki yet?
<libv>
seems like this indiegogoer has managed
slash_random has joined #arm-netbook
<focus_it>
libv: My goal is to make an SO-DIMM module (SoM) that has A10 and connectors to wire from it to 5" LCD and then make the whole thing in KiCAD and release it for everyone to develop their own modules
<libv>
...
<libv>
anyway, my feeling is that you've been scammed
<focus_it>
libv: thank you for that tip - I want to do that next. I got Gemei G2 running with Lubuntu from uSD, need it to run from flash.
<focus_it>
libv: I almost don't care - I spent $1000 on witstech PRO A10 and then when it came to Linux support, they flatly refused saying they will only support Android.
<focus_it>
Thats after they say they support Linux
<focus_it>
Then when I say Android runs on Linux, they change mind, but I still not get anywhere. What a con / waste!!
<focus_it>
I only find out about this forum by accident
<libv>
what, google?
<focus_it>
Wish I had been here earlier.
<libv>
google being the accident
<focus_it>
It was some aliexpress store that sold Mele1000 that had a link to IRC channe arm-netbook. Came here and I was delighted!! :-)
<focus_it>
I got more support here for Linux than PRO_A10 rubbish!
<hno>
Curtman, odd.
<focus_it>
I agree pengpod sounds dubious - but I wish them success, and most of all, if I get my flashed tablet with Linux, I hope to learn from it, if not thanks to your tip, I may learn more from indiegogoer work.
<hno>
slapin, hope you slept well. Fixed some bugs, uncovered many more.
<libv>
urgh, this stuff actually hit slashdot
<hno>
slapin, can't gret our code running on my OLinuXino A13 where I have logic analyser connected.
<hno>
looks like we need to get NAND_CMD_PARAM (0xec) working first, but I can't get that one to wait for RB to clear before trying to chew the data.
<focus_it>
libv: I just see it. Good luck to them. :-)
focus_it has quit [Quit: Leaving]
merbzt has quit [Ping timeout: 268 seconds]
dedis18 has joined #arm-netbook
dedis18 has quit [Remote host closed the connection]
hg_5 has quit [Ping timeout: 240 seconds]
stefanro1 has quit [Ping timeout: 255 seconds]
stefanro has joined #arm-netbook
freakazoid0223 has joined #arm-netbook
<L84Supper>
PengPod is back on Slashdot
<drachensun>
I've got a complete, current work Plasma image running to.
<drachensun>
based on openSUSE, hopefully I'll have the video up tonight
<libv>
ah, the threading issue bit me on my wrapper lib too.
popolon has quit [Quit: Quitte]
<Turl>
libv: what's this threading issue you mention? :)
<libv>
Turl: i believe i talked about it on thursday evening
<libv>
will now write up a trivial test for it and open an issue
<libv>
i think that it is brokenness in our kernel
<Turl>
sounds good
<libv>
basically, threads are not being woken up when trying to lock a mutex which happens to be locked already, at the time when the already locking thread unlocks the mutex
<libv>
and this in the simplest mutex use case
<Turl>
never waken up?
<Turl>
never ever? :P
<libv>
2 threads, both lock a mutex, one comes first and does stuff and unlocks the mutex, the other thread should now wake up but doesn't
<Turl>
doesn't ever or doesn't quick enough?
<libv>
doesn't ever
<libv>
otherwise there would not be a problem
<libv>
suse toolchain guys say: kernel
<libv>
as so happens, i went boozing with one of them minutes after i fixed it on the main limare lib job handling code
<libv>
now i ran into the same thing again in my command stream capturing wrapper library
<Turl>
what kernel version btw?
<libv>
our 3.0
<libv>
anyway. copy pasting up the testcase right now
<Turl>
the awesome atheros driver failed >.<
<Turl>
[1477389.180000] ath: phy0: Failed to stop TX DMA, queues=0x004!
<libv>
this is all such a big heap of shit :)
<Turl>
this what?
<libv>
all the sunxi crap those allwinner people threw together
<Turl>
yeah
<Turl>
the atheros driver is not related to this tho
freakazoid0223 has quit [Quit: Leaving]
<libv>
grmbl, it works in the cut down case :(
<libv>
the only difference is that one involves mali, and the other doesn't :(
<Turl>
maybe your 'work' part needs more time?
<libv>
the thread is needed to wait for the kernel "job is done!" ioctl return
slash_random has quit [Ping timeout: 246 seconds]
<libv>
makes sense that it is only happening while using the mali dev, otherwise the system would not boot
itamarjp has quit [Ping timeout: 252 seconds]
jquip has joined #arm-netbook
Curtman has quit [Quit: Leaving]
Quarx has joined #arm-netbook
pwhalen has quit [Ping timeout: 260 seconds]
pwhalen has joined #arm-netbook
hipboi has joined #arm-netbook
ZaEarl has quit [Quit: Ex-Chat]
<slapin>
hno: good morning!
ZaEarl has joined #arm-netbook
<slapin>
gm, all!
sspiff has joined #arm-netbook
jquip has quit [Ping timeout: 244 seconds]
jquip has joined #arm-netbook
jquip has quit [Quit: jquip]
Ershov has joined #arm-netbook
<hno>
slapin, morning
<projectgus>
does anyone know if framebuffer console works on A10 yet? setting up my mk802 and the internet has a lot of older references to "once framebuffer console is supported"
<projectgus>
I'm guessing it's not as simple as just enabling support in kernel config, is there some magic modesetting that needs to happen or something?
<hno>
it should work,provided you get HDMI to work.
arnd_ has quit [Remote host closed the connection]
<hno>
slapin, having a bit of figth between u-boot mtd and the nand chip on my board with logic analyzer. Have not been able to test much. But squashed some bugs and think I cleaned some other up a bit.
<hno>
but likely need a different design like what is used in the tegra nand driver. mtd is a bit too focused on individual NAND operations instead of whole sequences to match well.
<hno>
on the good side the controller is easy to send commands to, so studying how the controller acts with logic analyzer is easy.
<oliv3r>
good morning all!
gzamboni has quit [Read error: Connection reset by peer]
Mazon has quit [Ping timeout: 265 seconds]
Mazon has joined #arm-netbook
<oliv3r>
hipboi: have you thought of my request to pay you via paypal?
<Mehhh>
is there a one image wheezy up yet that can auto add script.bin for OTG usb?
<jlj>
does framebuffer support mean you can see the entire boot sequence without a serial cable?
sspiff has quit [Read error: Connection reset by peer]
sspiff has joined #arm-netbook
MindBeat has quit [Quit: Leaving.]
Quarx has quit []
jquip has joined #arm-netbook
mripard has joined #arm-netbook
<mripard>
hi
<stefanro>
mripard: hi Maxime!
<stefanro>
mripard: i'm currently reworking the cubieboard support patch as requested...
<mripard>
stefanro: great, that would be nice to have it in 3.8 as well
<stefanro>
mripard: will send new version in a bit
<stefanro>
mripard: yes, that would be really nice
<mripard>
we still have some time before the opening of the merge window, so I think it will be ok.
<RaYmAn>
stefanro: did you submit it to lkml already? :)
<stefanro>
mripard: yep, Arnd seems eager to get it included as well :)
<stefanro>
RaYmAn: no, mripard did beat me by a few days - i didn't know he was working on it too
<stefanro>
mripard: did you already start in some other drivers? pinmux etc?
<stefanro>
mripard: please sync with other developers for them, especially mnemoc
<projectgus>
rm: thanks (delayed!)
<mripard>
stefanro: nope, didn't start anything yet, to not overlap like we already did.
<mripard>
stefanro: I guess it would be a good thing if you could send your cleaned up restart code
<libv>
hrm, could it be that the pengpod 700 is simply the already well-supported ly-f1?
<stefanro>
mripard: yes, let me include it in patchset as well...
<libv>
and the pengpod 1000 looks like the protab2 xxl
<RaYmAn>
it's pretty clear they are all just rebranded devics
<RaYmAn>
devices*
<libv>
yeah, but which :)
slash_random has joined #arm-netbook
<RaYmAn>
I must say, i still find it rather hypocritical that they sell it based on "real Linux" and "true Linux tablet"
<RaYmAn>
essentially they are just expensive, rebrancded android devices, with OS stuff mashed together on.
<RaYmAn>
but hey, it's certainly an easy way to make money :)
Mehhh has quit [Quit: Page closed]
tinti has joined #arm-netbook
<libv>
puts a bit of a bad light on the olinuxinos and cubieboards and such, which have full schematics available and rank as much more honestly open devices
<libv>
and knowing our codebase, i would not use it as the reason to sell bogstandard chinese hw at a markup, that just leads to disappointed buyers and a bad name for everyone
<stefanro>
mripard: if nobody has any further comments/objections to my sun4i patches, perhaps its easiest for Arnd that you include them in your git repo for pulling
<stefanro>
mripard: what do you think?
* stefanro
leaves for lunch...
<mripard>
stefanro: I'm fine with the changes, I'll take them and send a pull request to arnd
Quarx has joined #arm-netbook
popolon has joined #arm-netbook
alcides has joined #arm-netbook
alcides has joined #arm-netbook
alcides has quit [Changing host]
ZaEarl has quit [Ping timeout: 252 seconds]
jquip has quit [Ping timeout: 260 seconds]
cat_x301 has joined #arm-netbook
jquip has joined #arm-netbook
hipboi has quit [Quit: Leaving]
<stefanro>
mripard: great, thx
hipboi has joined #arm-netbook
<focus_well>
libv: I don't think of it as disappointing if they sell a tablet with Linux on it, there are none I know of yet at that price. The vivaldi is not selling as yet.
<libv>
focus_well: it will come with an sd-card with packaged up linux-sunxi.org things and other bits
<RaYmAn>
focus_well: they charge 10$ extra to ship it with linux preinstalled
<focus_well>
libv: $5 extra for linux on flash - that be what I want on a working tablet - no one else has it - have to wait till Feb for the flash version
<slapin>
hno: I think MTD fits very well with actual NAND driver, since it is command-based, too. It becomes very small and nice with it.
<libv>
no board added to u-boot sadly
<focus_well>
They mention somewhere more improved devices if they get their project - I for one want a 4 core for tablet with Lubuntu or Ubuntu on it in flash.
<rellla2>
hi all. is sunxi-3.4 ok to use? is it using mali r3p0 like sunxi-3.0, too? may it function quite as good as sunxi-3.0? so i must not compile needed dvb-drivers myself...
<libv>
focus_well: thank you for being relevant as ever
<libv>
focus_well: have you tried looking at the linux-sunxi page?
<libv>
wow, expensive one.
rellla2 is now known as rellla
<libv>
i see one at 76usd, with free shipping to getmany
<focus_well>
libv: I do look, a little overwhelmed with all the info - but I hope to have it cut down to make my SoM board asap
<focus_well>
libv: specs - cheaper ones are 512Mb and 4Gb flash
<libv>
not worth the 50EUR markup
<libv>
66%
<focus_well>
true - sad - it kills their sales
<focus_well>
if they do it at reasonable price, put ubuntu into flash and add tons of edu software into 8GB, they got a winner straight away
<focus_well>
and if they don't sell it, they can reflash it to Android and meet endless android demand
<slapin>
by the way, is there any SoM with A10, A13 or A10s under $50?
<libv>
focus_well: what part of "this is bog standard chinese hw" don't you get?
<focus_well>
slapin: not I know of - but if I succeed you will have plenty! :) - and the KiCAD files to make your own derivatives
<focus_well>
libv: hehehe!!
jquip has joined #arm-netbook
vgrade_ has joined #arm-netbook
cat_x301 has quit [Ping timeout: 248 seconds]
jquip has quit [Ping timeout: 256 seconds]
hipboi has quit [Ping timeout: 246 seconds]
jquip has joined #arm-netbook
hipboi has joined #arm-netbook
<slapin>
focus_well: I think quick hack on Olimex board should work perfectly well...
<hno>
slapin, its not about size. The driver needs a redesign to fit MTD interface. NAND controller really want to know size to transfer with each transfer command, MTD issues command first and then size of transfer.
<hno>
but it's all manageable. we are not alone in having this interface issue.
<slapin>
hno: well, all commmandw, which require size have one (look at read command). I think we can try to stay at current API, and look how ugly this will be. Then we can propose API change.
<slapin>
hno: look at API properly - it really contains all we need, we can replace read/write functions, etc.
<focus_well>
slapin: true but the $ company partly funding this needs ethernet and possibly sata for its products which is A10 territory. Cubieboard release their full schematics http://dl.cubieboard.org/hardware/cubieboard_schematic_2012-08-08.pdf so its a short while before I get that into KiCAD and then cut it down for SoM board.
arokux has quit [Ping timeout: 246 seconds]
antonl has joined #arm-netbook
slash_random has joined #arm-netbook
hipboi has joined #arm-netbook
hg_5 has joined #arm-netbook
* slapin
needs to start with A13, as making 8-layer boards is too expensive for him
* slapin
hopes focus_well will make $20-$30 modules...
grimes99 has joined #arm-netbook
sspiff has quit [Remote host closed the connection]
<mnemoc>
that's the point of SoMs
<mnemoc>
people can do cheap base boards for them
hg_5 has quit [Ping timeout: 260 seconds]
<focus_well>
SoMs require 4 to 8 layer boards while base boards and mboards can be cheap 2 layer
<mnemoc>
as what slapin needs
<focus_well>
Counting down the days :-)
hg_5 has joined #arm-netbook
<L84Supper>
focus_well: what do you use for layout?
<focus_well>
Ki-CAD does schematics and PCB layout - I normally manually route.
<focus_well>
For complex boards, auto layout is hideous.
<L84Supper>
just wondering, I use Allegro
cheng has joined #arm-netbook
* slapin
uses Eagle and KiCAD
<focus_well>
My belief is Ki-CAD is a game changer - you use Allegro, I used to use Edwin, then Protel, then Altium but we all not able to cooperate
<focus_well>
Ki-CAD is free - so we can all use it and do open hardware development if we all learn it
<focus_well>
youtube videos plenty
<slapin>
focus_well: KiCAD is quite hard to use sometimes
<L84Supper>
it's not about learning how to use the tools, it's about properly designing boards
hipboi has quit [Ping timeout: 276 seconds]
<focus_well>
Agree - but I get used to it forcefully as in FORCEFULLY!!, now its just lightening quick
<slapin>
focus_well: a year ago I hired hw engeneer to make me schematics+pcb in KiCAD, his background was old PICAD version, or something like that.
<slapin>
focus_well: and I wanted to see how things are by this experiment
<slapin>
focus_well: thet guy did the job, so it was more or less success, but I never heard so many insults...
<cheng>
among these open source routing, which one is comparable to altium/protel?can suggest a few for me?
<slapin>
focus_well: actually, he mostly missed a small helpers to make task easier, not as whole, and ergonomics.
<focus_well>
L84Supper: properly designing board requires great knowledge about how signals travel on PCB - I have that knowledge
<focus_well>
wait...
<slapin>
cheng: that depends. All of them do their job.
<L84Supper>
so why do you use Kicad?
<slapin>
so if KiCAD gets its interface improved (I mean things like "shooting" to pad for clean connection, semi-automatic routing (which is better than autorouter because you draw single line and select variants of auto-placement), etc. to improve routing speed, people will really use KiCAD for their $$ work, like they use Eagle.
arokux has joined #arm-netbook
<slapin>
actually, not much needs to be done software-wise
<rm>
I hate git with a passion
<rm>
$ git checkout sunxi-v3.0.42-r3
<rm>
error: pathspec 'sunxi-v3.0.42-r3' did not match any file(s) known to git.
<focus_well>
L84Supper: so I hope to put into practice good design - luckily everything I build works - but it gets harder at high frequency.
<mripard>
rm: did you update your local tree ?
<mripard>
git fetch allwinner-remote
<jquip>
mnemoc: erm.. if he hasn't done that already ..
<mripard>
if it's a new tag and didn't fetch that, that might explain :)
<Triffid_Hunter>
focus_well: I hear that the biggest barrier to entry with kicad is that it's stupidly difficult to design new parts esp their footprints
<mnemoc>
it's not new at all
<Triffid_Hunter>
I'm still using eagle until that's significantly easier
<focus_well>
cheng: autorouter in KiCAD is useless - it uses an external one
<rm>
the latest commit in mine is from Oct 3
<cheng>
noted, thanks for the advice
<focus_well>
Triffid_Hunter: its about knowledge - you can in KiCAD design everything in text files!
<slapin>
a problem is when you need to produce 3d model for your board (e.g for case designers or heat spreading calculations), thay there is really a lot of work in KiCAD...
<focus_well>
slapin: Well done! I made some 100 pin arms and some very complex products - no problem - done it in record time
<rm>
mnemoc, btw I noticed the mainline is up to 3.0.52 already, no plans to merge that?
<Triffid_Hunter>
focus_well: argh robo-voice causes ears to bleed
<focus_well>
slapin: I agree - anything beyond PCB is not something KiCAd can do well - its free project :-)
<mnemoc>
rm: i wish to... but $work$ is taking me over 90h/w currently :(
<focus_well>
Triffid_Hunter: need to listen several times and practice over and over to get the swing of things :)
<slapin>
hno: what resume re NAND is by now?
* slapin
makes his SCSI controller project to learn KoCAD
<L84Supper>
how does kicad handle tweaks to trace and space? Let's say you have the DDR3 section routed with all the traces mitered and match for length and then you wish to optimize or shove the location of the RAM by 1-3mm.
<slapin>
L84Supper: badly, of course, as Eagle, too, and most other programs.
<slapin>
L84Supper: some clever autorouter is needed to fix these things
<L84Supper>
that's one of the main reasons I don't use it
<focus_well>
L84Supper: A large number of things it can't do. The external autorouter is more capable but I don't know its limits.
<mnemoc>
rm: hopefully other can do it too... there are many commiters. I don't want to hold the repo back due to my lack of availability
<focus_well>
But it is the only common one all open source developers can work with
<slapin>
L84Supper: IIRC Altium is most tolerant to that, but there is a lot of exceptions in there, too.
<rm>
well I dunno 3.0.42 works... also maybe it's better to abandon 3.0 and focus on 3.4
<focus_well>
L84Supper: None of the PCB packages can share data - if you Allegro, and I got Altium and someone else can Orcad, we all stuck and can't cooperate, but if everyone learn KiCAD, we all can work together
<mnemoc>
we can't abandon 3.0 until the uart and usb-gadget hackery gets ported to 3.4
<slapin>
focus_well: Eagle <-> KiCAD is mostly interoprable
<focus_well>
slapin: Eagle hate that! :-)
<slapin>
well, the main thing about gEDA and KiCAD, is that you really can do yur full PCB development cycle using OSS, from ideas to gerbers.
<slapin>
focus_well: I don't see how this can be bad, both products are nice, and iteroperability works well for them
jquip has quit [Ping timeout: 260 seconds]
<L84Supper>
IIRC the a10 dev or reference boards were done with PADS
<focus_well>
slapin: yes - and at least with KiCAD the files are all text based, so it is possible to script up a large number of things
<slapin>
focus_well: I can migrate libs here and there, play around various things, can make 8-layer design in KiCAD and 4-layer in Eagle as parts of the same project, which can be done by different people with various preferences not affecting the speed of project by interoperability issues
<focus_well>
slapin: eagle loose customers for their 600 euro pro packages
<L84Supper>
an old version from ~2005
<focus_well>
slapin: that is good! I really really learn all the intricate details of KiCAD because I don't want to go back to proprietary and semi-proprietary packages no more
<focus_well>
Its not that I hate them but the boards are all over the place locked up in all kinds of file formats
<focus_well>
Its bad for business
<focus_well>
I train engineer to KiCAD
<slapin>
focus_well: they provide fully-featured educational versions for $100+ (don't remember), also for most designers an issue is lack of pre-made libs, unlike what is mostly spoken as good taste of PCB/schematics design (you should make your own libs of everything to be consistent both on your board due to various factory requirements and on your schematics for good looks of it).
<focus_well>
slapin: I keep extending my libraries now - and also semi routed modules - so I can copy and paste and make systems more quickly
<L84Supper>
found them .... PADS-POWERPCB-V2007.0-BASIC-250L
<slapin>
focus_well: for average designer joe in most companies I seen it is better to take pre-made library and very hastely make PCB (omitting schematics step), then, later, for docs, to generate ugly junky schematics with one chip per page and interconnections via bus labels
<RaYmAn>
slapin: s/better/cheaper/
<focus_well>
slapin: :)
<slapin>
this is how it looks, anyway.
<slapin>
so they mostly concentrate on PCB designer, and KiCAD's is quite ok. Not nearly perfect, but OK for these casual jobs to be done without too much hussle,
<L84Supper>
heh, PCB manually routed first, then make the schematics
<L84Supper>
and then when it doesn't work they ask me to fix it, this sounds familiar :)
<Triffid_Hunter>
board first then schem? sounds like a complete clusterf*ck from a QA perspective for anything even mildly interesting
<slapin>
L84Supper: aha, en masse the real design process is the same everywhere
<focus_well>
slapin: I'm different - I make modules and put them into boxes and keep as reference to go back to. I choose best modules to make a sample. Then I laser cut acrylic and bolt down the PCBs and chips and make the product. Usually one to two days if no software involved.
<L84Supper>
somebody once called us with a hand routed board without schematics, they wanted schematics and a working design LOL
<Triffid_Hunter>
L84Supper: that's what the special "reverse engineering fee" is for
<L84Supper>
there were some hand drawn schematics that used similar parts however :)
<focus_well>
You were lucky L84Supper! Some guy turned up with machine code (Hex listing) and wanted a gas analyzer repaired!
<slapin>
L84Supper: this is so common that is no longer funny, they simply don't know the proper way, and they really don't want to know.
<L84Supper>
the PCB didn't work either, BTW
<slapin>
L84Supper: even when designing "possibly Open Source products" (too bad I seen this a lot).
<L84Supper>
it's sad, most of this is getting lost
<slapin>
tha bed thing in this way of design is not only the reversed process, but also in this cases boards are not well thought and are hardly possible to show to people who don't really involved with design, but can read schematics and add suggestions, also, the whole picture is only in one designer's head, which might be windy.
<slapin>
s/bed/bad/
<ibot>
slapin meant: tha bad thing in this way of design is not only the reversed process, but also in this cases boards are not well thought and are hardly possible to show to people who don't really involved with design, but can read schematics and add suggestions, also, th...
<L84Supper>
few seem to understand thermal management as well, I'm not sure what EE's even cover in school
<L84Supper>
then again Apple throws a 1000 monkeys at iphones designs that are very difficult to assemble and only last 1-2 years
<slapin>
L84Supper: Apple is mostly marketing and looks, they don't care much for actual electronics inside. This should look good, period.
<slapin>
I can say this for most tablets on aliexpress, too, including A10 ones. A7HD is good esclusion - it does have average-good electronics, but bad software...
grimes99 has quit []
pwhalen has quit [Quit: Leaving]
<slapin>
s/esclusion/exclusion/
<ibot>
slapin meant: I can say this for most tablets on aliexpress, too, including A10 ones. A7HD is good exclusion - it does have average-good electronics, but bad software...
mikey_w has joined #arm-netbook
<L84Supper>
aren't the Orcad Capture files on the wiki or on the ML for the A10 reference design?
<focus_well>
Takes about 3 days to do up something like this
<focus_well>
All made from separate modules that are all individually working and perfected outside of specific requirements
<focus_well>
I hope to do the SoM modules in the same way
<focus_well>
The software can take months - but if software modules available, then its hours to days to knit everything together
<slapin>
focus_well: OMG, this one is large!
<focus_well>
I got something 4x large!
* slapin
uses the same technology for junk projects, but with carbon-based black plastic, can take 1-2 days from idea to assemble
* slapin
has i386SX-based project using this technology still somewhere collecting dust
<slapin>
also a good use for such plastic - you can use soldering fan to glue things together, no actual glue involved
<slapin>
and no drilling, etc.
<Triffid_Hunter>
focus_well: hint: hot glue on the high voltage section esp where cables enter the board.. apply hot glue using SMD rework hot-air gun for best results ;)
<slapin>
Triffid_Hunter: do you mean thermal glue they use to attach everything in cheap kid's toys?
<focus_well>
I got the holes in dxf format and the window cut outs in dxf all already made up to slap together to make acrylic board design in about 1 hour. Then 30 minutes to cut on laser, and the board building can begin!
<slapin>
focus_well: do you really need laser for that?
<Triffid_Hunter>
slapin: yep
* slapin
use it to make wire connections stronger
<focus_well>
To be quick laser as I move from project to project within days some times. I also box the frames - for that I make software to make custom boxes in acrylic - type in dimension and out pops all the box contours and attachment flanges with correct screw holes - 3 minutes to design box and 1 hour to cut box!!!
<focus_well>
It used to take 1 day to design box - I get fed up and spend 3 weeks getting software working
* slapin
abuses work 3d printer for that at nights...
<focus_well>
I keep telling work to buy 3D printer to abuse it at night - no luck yet
<focus_well>
:-)
<focus_well>
:-(
<slapin>
and when really lazy I make small single-use enclosures out of thermal glue and some cheap toner
<slapin>
not really repair-friendly, but really cheap
<hno>
slapin, I am dissecting the controller a bit, now that I have it under a logic analyser.
<hno>
can't test run our code at the moment due to MTD seemingly not liking the Hynix flash chip on that board.
<hno>
seems it's an odd which saying it's ONFI but witout any ONFI parameter page.
<focus_well>
Makes a hell of a difference with development work being able to connect together boxes at leisure!
<slapin>
hno: IIRC ONFI doesn't require that page anyway
<hno>
slapin, did fix a number of issues yesterday. See my git repo.
<hno>
slapin, possible. But MTD seem to require it.
<slapin>
hno: when you enable that define...
<slapin>
hno: actually nand config block is elsewhere
<slapin>
hno: the one with pageshift, etc. settings
<hno>
anyway, the troble is that ends up with "NAND bus width 8 instead 16 bit" and rejects the chip.
<slapin>
hno: ah, I had that one
<focus_well>
No glue holding the boxes together - just tight fit precision laser cut panels that snap together, no screws if avoidable. Everything can be taken apart if boards blow up.
<slapin>
hno: it was when I it read garbage from configuration block because readid malfunctioned
<hno>
readid seems fine. And the chip info reported just before is correct.
<slapin>
hno: it is checked in some function, let me look
<hno>
just before it sasy NAND device: Manufacturer ID: 0xad, Chip ID: 0xd7 (Hynix NAND 4GiB 3,3V 8-bit)
<slapin>
hno: that's ok, it was there also
<slapin>
it just misred some byte of readid due to some ready-busy issues or something
<slapin>
ah, yeah, I read 6 bytes instead of 8, and also r/b was wrong
<slapin>
so that really was readid issue, yes.
<slapin>
Manufacturer ID: was printed properly, too
<hno>
tried looking at the function but did not manage to unwind the cause much, and lots of other interesting things to decode.
<slapin>
hno: yeah, nand_base line 2783
<slapin>
s/nand_base/nand_base.c/
<ibot>
slapin meant: hno: yeah, nand_base.c line 2783
<hno>
readid is not using RB. But the ONFI params page read do and got that fixed.
<stefanro>
slapin: u-boot code is mostly copied from linux (some old version)
* Marex
knows nothing
rm has joined #arm-netbook
<slapin>
stefanro: especially isn't there some work re splitting NAND commands with reading data from them, becaus eon modern NAND controllers these things are tight together, and separation looks a bit vulgar?
<slapin>
stefanro: e.g NAND controller, which automagically fetches data to some memory location (and size for these data should be specified in command via writing some register)?
<stefanro>
slapin: i don't really understand what you mean - sorry
<slapin>
stefanro: imagine NAND controller like A10's - you specify commands, specify read data size, and execute everything in one batch. Voila - you get data somewhere in memory location
<stefanro>
slapin: whats the current status of your NAND work? you can read the NAND ID etc correctly? (i didn't follow the development really)
L84Supper has quit [Quit: <puff of smoke>]
<slapin>
stefanro: it has some limits though
L84Supper has joined #arm-netbook
<slapin>
stefanro: it can read NAND ID, and can do some things, except no reading yet (but close).
<stefanro>
slapin: and where do you specify this command, read size etc? this sounds like DMA based transfer
<slapin>
stefanro: hno fixed lots of stuff, will testa that today and will see.
<slapin>
stefanro: might not be DMA.
<slapin>
stefanro: in registers
<slapin>
stefanro: and data in dedicated SRAM
cheng has quit [Quit: Leaving]
<slapin>
stefanro: you specify commands, flags, settings, sizes, then execute -> data
<stefanro>
slapin: do you have the code in some git repo right now?
<slapin>
stefanro: so command then address then more commands then read dat approach is not that effective
<mripard>
stefanro: I might have an idea about how to not share the timer_base variable, yet not write anything into the timer driver.
tinti has quit [Read error: Connection reset by peer]
* stefanro
has only working with the old-style NAND controller that execute commands one-by-one "manually"
<stefanro>
slapin: i might find the time to look into this NAND stuff as well tomorrow (no promise though!)
slash_random has quit [Ping timeout: 246 seconds]
<slapin>
stefanro: it is better to extend board support, we with hno play with low level stuff using logic analysers at the moment :) but improving mtd interface, or ideas on how implement this cleanly with current mtd are very very welcome.
n6pfk has joined #arm-netbook
* slapin
gone
<Marex>
it kinda sucks there's no documentation
slash_random has joined #arm-netbook
<Marex>
stefanro: you know ... the thing I don't understand is how the NAND engine handles various feedback from the NAND chip during the execution of chain of commands
<Marex>
stefanro: it can be a branching DMA, but that's really sick
<stefanro>
Marex: yes, really strange
<stefanro>
i'll keep my fingers of this driver/chip for now :)
wim has quit [Ping timeout: 272 seconds]
wim has joined #arm-netbook
SouL_ has left #arm-netbook [#arm-netbook]
mikey_w has quit [Read error: Operation timed out]
<slapin>
hno: crash is triggered by data_fetch_flag = 1 in RNDOUT command, checking
<rz2k>
vinifm: expect anything to be missing from allwinner code
grimes99 has joined #arm-netbook
<slapin>
hno: and without this bit same thing happens, hmmm...
<vinifm>
rzk2: sorry, i dont understand?
<vinifm>
rzk2: sorry, i dont understand
SouL_ has left #arm-netbook [#arm-netbook]
<vinifm>
i mean... I did not understand
ibrah has joined #arm-netbook
<Mehhh>
Is there a single wheezy image now that isn't made from the 3 images as per the website?
<Mehhh>
possibly that allows for script.bin or has the OTG enabled already?
vinifm has quit [Ping timeout: 265 seconds]
ibrah has quit [Ping timeout: 240 seconds]
vinifm has joined #arm-netbook
<slapin>
READ0 works, that's fantastic, seen very well on oscilloscope. But there's something wrong with reading of buffer.
* slapin
gone sleeping
alcides has quit [Quit: man woman || $> Segmentation fault (core dumped)]
fredy has quit [Ping timeout: 260 seconds]
fredy has joined #arm-netbook
fredy is now known as Guest57439
SouL_ has joined #arm-netbook
tuliom has joined #arm-netbook
rellla has quit [Quit: rellla]
alcides has joined #arm-netbook
alcides has joined #arm-netbook
alcides has quit [Changing host]
gzamboni has joined #arm-netbook
<Curtman>
Well I'm much closer to getting this A3700 up and running.. It's still hanging somewhere during kernel init. It mounts the rootfs described in inittab, it brings swap online, mounts the home directory, then nothing. :( No /sbin/init. Anyone have a clue why?
<rz2k>
and also, how do you mount swap if you have no init?
<rz2k>
kernel automounter?
ibrah has joined #arm-netbook
<Curtman>
This one is gentoo arm-hardfloat-gnueabi stage 3. I tried the linaro one that libv suggested, it didn't do any better. I'm not really sure if init is running, but I'm not seeing it. Maybe the serial driver is still messed up, and console is getting confused.
<Curtman>
arm-hardfloat-linux-gnueabi I mean.
myfluxi has joined #arm-netbook
sv has quit [Read error: Connection reset by peer]
<rz2k>
interesting
<rz2k>
3700 must be same as cubieboard
<rz2k>
are you sure that both linaro and gentoo spawn a tty @ ttyS0?
<rz2k>
because for some reason ArchLinux-ARM doesnt have the serial output on my Mele A2000 too
* specing
wakes up at the Gentoo highlight
<rz2k>
when Linaro works okay
sv has joined #arm-netbook
<specing>
Curtman: I got mine to the login prompts but had no rx pin to actually login
<rz2k>
you still have crappy mk802? :p
<specing>
No, I have an A10 tablet
<specing>
I never got the mk802 or a cubieboard or a mele because I have no time to work on them
<rz2k>
oh, yeah, I forgot, I thought you've found the rx at some of the caps on the board.
<Curtman>
specing, I have rx pin. I can interact with u-boot fine. My original problem was hardware flow control enabled in minicom. ONce I turned that off, rx was working fine.
<Curtman>
My problem is no login prompt ever comes. It should be at least doing dhcp on the ethernet. I should at least see a ping response from it, but it never comes. :(
<specing>
I also tried getting console on a pl2303 converter but it didn't register