00:06.52 | Hydrar | rzk: ok, kinda ,just generally lookes like a daunting task, but I was thinking that we sure could use a competitor to the expensive custom vendor ARM server solutions that's about to appear |
00:11.15 | A1Sheds | Hydrar, the problem has been getting ARM soc's that are worth developing for |
00:11.47 | A1Sheds | vendors don't want everyone's money, just money from some |
00:15.06 | Hydrar | A1Sheds: I see, hmm |
00:15.41 | Hydrar | Honestly got the idea when seeing the alwinner A10/mele A1000 |
00:17.05 | Hydrar | I thought the chips would be available though, not broadcom, but marvell and the others seem more sane |
00:17.43 | Hydrar | Although most seems like they'll be entrenched in IP hell regardless which is gonna drive up prices |
00:18.55 | *** join/#arm-netbook hp_ (~kvirc@220-136-106-59.dynamic.hinet.net) |
00:25.54 | *** join/#arm-netbook sepp_ (~seb@port-92-206-41-27.dynamic.qsc.de) |
00:29.59 | Xlab | 1024MB !!!! http://cl.ly/HPfB/Image%202012.06.15%204:26:48.png |
00:32.56 | Turl | Xlab: so? |
00:33.01 | Turl | what's so interesting? :P |
00:33.50 | Xlab | It's linked with my issue there https://github.com/hno/uboot-allwinner/issues/11 |
00:35.06 | Xlab | just a working way to change amount of detected DRAM memory |
00:37.05 | Turl | oh |
00:38.31 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
00:49.12 | hipboi | Xlab: so finally you got your 1G memory working? |
00:50.30 | Xlab | Exactly |
00:52.15 | hipboi | which CHIP_DENSITY |
00:52.39 | Xlab | I've done some tests |
00:52.40 | Xlab | DCR_CHIP_DENSITY_4Gb |
00:52.42 | Xlab | is mine |
00:52.47 | hipboi | ok |
00:52.58 | Xlab | DCR_CHIP_DENSITY_8Gb produces "2048Mb" value |
00:53.15 | hipboi | well, that's fake |
00:53.41 | Xlab | …and blocks count is just hardcoded in an "if" |
00:53.45 | hipboi | the dram controller doesn't know the dram density |
00:55.03 | hipboi | which block count |
00:57.04 | *** join/#arm-netbook leojrfs (~leojrfs@bl15-91-205.dsl.telepac.pt) |
00:57.50 | Xlab | oops |
00:57.54 | Xlab | *banks count |
00:58.51 | hipboi | well |
00:59.21 | hipboi | the dram parameters are in the script.bin for the stock firmware |
00:59.36 | hipboi | but in u-boot spl, the script.bin is not used |
00:59.49 | hipboi | so the dram parameters are hard coded |
01:00.09 | hipboi | we haven't found a better way to solve this |
01:04.28 | Xlab | If dram size isn't used in u-boot, maybe it's better to set it up from script.bin values at kernel init? |
01:06.10 | Xlab | Original u-boot fails too, it prints 512Mb, but sends to kernel density_4gb i guess |
01:39.04 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
02:10.37 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
02:56.45 | Xlab | usb host seems to be broken for sun4i_crane in allwinner-v3.0-android-v2 :( |
03:42.14 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
04:06.08 | *** join/#arm-netbook gimli (~gimli@xbmc/staff/gimli) |
04:58.10 | *** join/#arm-netbook tom_say (~quassel@cpe-72-178-202-3.stx.res.rr.com) |
05:04.34 | *** join/#arm-netbook gsilvis (~almostsix@50.12.163.241) |
05:12.50 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
05:21.10 | *** join/#arm-netbook naobsd (6fbc5fda@gateway/web/freenode/ip.111.188.95.218) |
05:36.15 | rm | http://linux.slashdot.org/story/12/06/14/1343205/ubuntu-1204-ported-to-the-allwinner-a10-mk802-mini-pc |
05:36.20 | rm | didn't take long, did it |
05:36.24 | lundman | too long |
05:59.50 | hipboi | <PROTECTED> |
06:01.56 | Triffid_Hunter | how different is the mk802 from the mele A1000? |
06:02.05 | hipboi | the size |
06:02.10 | lundman | how long is a string! |
06:02.10 | hipboi | and the dram size |
06:02.19 | lundman | they are pretty much the same thing |
06:02.33 | hipboi | yes |
06:02.33 | Triffid_Hunter | ah sweet |
06:02.42 | Triffid_Hunter | so those images should work basically unmodified on the mele? |
06:02.48 | hipboi | yes |
06:02.57 | lundman | where can I download the image :) |
06:03.00 | Triffid_Hunter | great, I know what I'm getting next week :D |
06:03.09 | Triffid_Hunter | lundman: page linked from article has them |
06:03.22 | lundman | I have to go get it myself? ah maan |
06:04.27 | lundman | ugh they used dd |
06:05.33 | Triffid_Hunter | I'm far more interested in the mele due to all the extra I/O ports on it, esp dual usb host, ethernet and the sata connector on top is a nice touch |
06:06.09 | naobsd | Known issues: Memory is limited to 512MB, possibly because it is using the u-boot from the mele |
06:06.17 | hipboi | yes |
06:06.35 | hipboi | : D |
06:06.48 | hipboi | it's all my fault |
06:07.22 | naobsd | porting: sometimes it's writing codes, sometimes it's using files from someone |
06:08.30 | rm | standing on the shoulders of giants! |
06:08.39 | rm | who also stand on the same |
06:08.44 | rm | it's giant's all the way down |
06:08.47 | rm | giants* |
06:09.54 | lundman | i thought the code was "return 512;" |
06:10.31 | lundman | s/the/\n/quit/g |
06:10.40 | lundman | oh well |
06:10.46 | hipboi | the big problem for daily usage of ubuntu is the xserver-video-mali |
06:12.38 | naobsd | hipboi: we need to stand on your shoulder about it :) |
06:12.56 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
06:13.22 | hipboi | naobsd: still can not figure out why mali is not working : ( |
06:15.06 | naobsd | oh, well, https://www.miniand.com/forums/forums/2/topics/1 "The current version of the images support Wifi and OpenGL" OpenGL? |
06:15.17 | hipboi | yes |
06:15.27 | hipboi | there are OpenGL libs |
06:16.00 | hipboi | but no applications use them |
06:16.37 | Triffid_Hunter | so the mali driver has gl but no hardware video decode? awesome |
06:17.09 | lundman | I wish I had something to do on the a10 |
06:17.14 | hipboi | the gl is not enough for a x11 driver |
06:17.40 | hipboi | lundman: can you have a look at the x11 driver? |
06:18.57 | hipboi | i will write a page on the wiki on how to compile the x11 driver of mali |
06:19.16 | lundman | I did x11 dev for many years, it'd not be pleasant to go back :) |
06:19.29 | lundman | but yeah i can do |
06:20.32 | lundman | hey, coding music http://www.mazemod.org/ |
06:20.39 | Triffid_Hunter | when I can make a working boot image for these things using gentoo I'll be very pleased |
06:20.56 | Triffid_Hunter | hm so mk802 has more ram and cpu speed but mele has more ports? |
06:21.59 | lundman | it has more ram, really? |
06:22.29 | naobsd | I want 1GB ver. of MK802... |
06:22.32 | hipboi | lundman: there are two types of mk802 |
06:23.04 | hipboi | mele will have products like mk802 soon |
06:23.36 | lundman | i personally dont care about the size, my mediaplayers stay under the tv, not in my pocket :) |
06:24.42 | naobsd | sometimes I want SATA port |
06:24.56 | lundman | slow download.. I'll put it on my mirror when done |
06:25.06 | lundman | then the two guys from japan will have it in seconds :) |
06:25.13 | Triffid_Hunter | torrent it |
06:25.49 | naobsd | ubuntu image on miniand.com? |
06:26.00 | lundman | ye |
06:26.14 | naobsd | I already got it so I can upload it on my server |
06:26.19 | Triffid_Hunter | these are unbrickable due to SD-boot built into the chip aren't they? |
06:26.23 | lundman | i got the recommended one |
06:26.25 | Triffid_Hunter | that's very clever, and about damn time |
06:26.37 | naobsd | my server is in japan |
06:26.54 | lundman | then we shoudl coordinate |
06:27.17 | lundman | wokring for an ISP does have advantages |
06:29.46 | Triffid_Hunter | well that download is maxing my connection, suppose working at an ISP gives you a quite different definition of "fast" too :) |
06:30.24 | lundman | sure, fast anywhere in japan, leaving the country and you get regular speeds though. but no bandwidth limit is where the win is |
06:32.15 | Triffid_Hunter | I got theoretical 8mbit down, semi-rural area which means actual speeds closer to 5-6Mbit are best we get.. I hear most of japan is wired with 100Mbit or faster |
06:32.49 | lundman | at home I have 100M fiber sure |
06:33.33 | naobsd | fibre is very common but ISP may have limit |
06:33.37 | naobsd | in japan. |
06:33.49 | naobsd | http://mirrors.androtab.info/miniand/ uploaded |
06:34.15 | naobsd | I keep it for a while, but I don't think maintain it up-to-date |
06:35.03 | naobsd | please don't use concurrent downloader ;) |
06:35.06 | lundman | 4MB/s |
06:35.11 | lundman | i dont |
06:35.48 | Triffid_Hunter | naobsd: I thought concurrent downloaders only existed to get around windows' shitty TCP/IP stack |
06:36.11 | Triffid_Hunter | never had trouble maxing my connection in linux |
06:37.16 | hipboi | anyone is interested in a10 notebook? |
06:37.21 | rm | to get around ISP's throttling aka "completely fair and just policy of sending people trying to download something longer than an average HTML page completely into the cr*pper" |
06:37.28 | rm | hipboi, sure |
06:37.29 | lundman | hipboi: sure, what does it do |
06:37.47 | lundman | triffid: you always tweak windows stack, right after installing it |
06:37.52 | hipboi | i have one on my desk |
06:38.00 | Triffid_Hunter | lundman: yeah I do, although haven't installed or used windows in years |
06:38.02 | naobsd | hipboi: sounds nice :) |
06:38.32 | Triffid_Hunter | lundman: speedguide tool still work on latest windows? so much easier than trawling through the registry |
06:38.32 | lundman | naobsd: i can recommend open-circuit btw (not my isp) they have no qos etc |
06:38.37 | hipboi | it's 700 rmb on tabao |
06:38.56 | lundman | i don7t use windows, so I dont know :) |
06:39.16 | naobsd | lundaman: open-circuit? |
06:39.37 | lundman | from the ISP-shit-list japan |
06:39.44 | naobsd | ah |
06:39.50 | lundman | http://isp.oshietekun.net/ |
06:40.03 | Triffid_Hunter | lundman: heh makes two of us.. I seem to be involved in more and more projects where the channel is fairly active, then someone jumps in and says "how do I do X on windows?" ... *crickets* ... amuses me every time |
06:40.09 | naobsd | my server is sakura's VPS. it's fast and no bandwidth limit |
06:40.25 | naobsd | and cheap, 980yen/month |
06:40.39 | rm | 2012-06-15 06:40:25 (6.01 MB/s) - \u201cubuntu-desktop-12.04-1-miniand.com.img.7z\u201d saved [1559312771/1559312771] |
06:40.56 | rm | naobsd, downloaded it to mine in Japan :) |
06:41.02 | lundman | なおsd:あl祖ア欄k、ごおdちょいcえ |
06:41.04 | lundman | baah |
06:41.10 | lundman | Also A-rank, so good choice |
06:41.22 | Hydrar | Triffid_Hunter: Oh, you hang around in here too :) |
06:41.29 | naobsd | rm: fast :) |
06:41.42 | rm | my VPS there downloads from youtube at 45 MB/sec |
06:41.42 | lundman | Another mirror http://www.lundman.net/ftp/mele/ |
06:42.00 | Triffid_Hunter | Hydrar: absolutely, mele or mk802 as a reprap high-end controller would be perfect, might even be able to slice on them |
06:42.06 | rm | too bad it's crappy in other aspects |
06:42.06 | lundman | rm: VPS from? |
06:42.09 | rm | TechieVPS |
06:42.13 | Hydrar | Triffid_Hunter: Yeah, agree :) |
06:42.15 | lundman | ah not ours :) |
06:42.59 | lundman | anyone know why they recommand lubuntu over ubuntu? size? |
06:43.28 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
06:43.30 | Triffid_Hunter | lundman: ram usage I'd say.. lxde is way lighter than kde or gnome |
06:43.54 | lundman | ah.. its faster to write to SD so that's what I am trying |
06:44.04 | lundman | :) |
06:44.25 | Triffid_Hunter | if you only got 512M ram, you don't want 300 of it taken up by the DE! |
06:44.52 | Triffid_Hunter | I switched my laptop from kde to lxde because it only had 1G and I got sick of kde taking a full 1/3 of it |
06:53.03 | naobsd | hipboi: how large screen size and resolution? |
06:54.09 | hipboi | naobsd: 10 inch, 1024 x 600 |
06:54.22 | hipboi | naobsd: sadly only 512 ram |
06:54.39 | naobsd | oh |
06:54.45 | hipboi | ubuntu 12.04 running on it seems nice |
06:55.17 | naobsd | 10" 1280x720 1GB will be nice ;) |
06:59.10 | mnemoc | hipboi: and no sata, right? |
06:59.21 | hipboi | mnemoc: yes |
06:59.50 | lundman | quite a few A10 devices are coming out then? |
07:00.33 | mnemoc | it seems to be slowing down in favour of rockchip A9s |
07:00.49 | CIA-121 | rhombus-tech: sahil master * r7ba88896811e /allwinner_a10/orders/sam.mdwn: |
07:00.50 | mnemoc | and next will be the imx6 boom |
07:02.27 | mnemoc | hipboi: personally I don't see much use for such netbook currently |
07:02.49 | hipboi | mnemoc: me too |
07:03.04 | *** join/#arm-netbook Quarx (~Quarx@46.233.218.196) |
07:03.10 | lundman | netbook for the kids maybe, but no 1080p hmm:) |
07:03.16 | mnemoc | better screen, 1GB and sata .... and a good sata, and the netbook becomes an inexpensive "smartbook" |
07:03.31 | hipboi | it's easy for them to make a notebook |
07:03.44 | hipboi | with the original intel notebook case |
07:03.49 | mnemoc | ah, yes, hdmi output is important to attach an external display |
07:04.05 | mnemoc | or projector |
07:04.48 | mnemoc | s/sata/case/2 |
07:06.16 | *** join/#arm-netbook rell01 (~chatzilla@p5B078F30.dip0.t-ipconnect.de) |
07:07.42 | mnemoc | hipboi: if it's easy for them, ask them to make a good one, even if it ends up been $75 more it's very possible it sells more too |
07:08.18 | mnemoc | hipboi: there is enough junk around and a notebook which is not better than a tablet with keyboard doesn't make much sense |
07:09.07 | Triffid_Hunter | hm, I don't have anything with hdmi input |
07:09.24 | Triffid_Hunter | any hints on smallish screens that are good, or what to avoid? |
07:09.27 | mnemoc | hipboi: people is sort of tired of buying cheap stuff they end up not using at all |
07:10.24 | mnemoc | Triffid_Hunter: motorola's atrix 4g lapdock seems quite interesting |
07:13.12 | Triffid_Hunter | mnemoc: can plug arbitrary devices into it? |
07:13.30 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
07:13.41 | mnemoc | with the right cables.... it has a male micro hdmi and male microusb |
07:13.48 | mnemoc | afaik |
07:17.00 | Triffid_Hunter | hm intriguing |
07:17.19 | Triffid_Hunter | sounds like perfect opportunity to make an mk802-lapdock adapter |
07:18.46 | naobsd | portable HDMI display/keyboard with battery |
07:20.43 | Triffid_Hunter | that 36Wh battery should last ages with an mk802, wonder how much the screen and usb hub use |
07:22.59 | *** join/#arm-netbook rell01 (~chatzilla@p5B078F30.dip0.t-ipconnect.de) |
07:24.31 | mnemoc | if you can solder, as proper cables are hard to find and overpriced, an small f/f board would work great |
07:25.19 | Triffid_Hunter | can solder, no worries there |
07:26.02 | Triffid_Hunter | was hand-soldering 0603 and LQFP the other day no worries, discovered that using a magnifier hurts my eyes, I do better without it |
07:27.09 | lundman | ok, the lubuntu image is also on mirror |
07:27.19 | mnemoc | if you make such board, count me in to buy you one :) .... I can't even solder loose wires :< |
07:28.34 | lundman | you tape lose wires! :) |
07:29.39 | mnemoc | I was 3h! trying to resolder the wire of the xmas tree lights last year... and lasted less than 20m connected |
07:29.49 | mnemoc | to the controlling board |
07:30.11 | naobsd | I want more hands |
07:30.27 | mnemoc | since then I decided I won't touch a soldering iron ever again |
07:31.50 | mnemoc | I have a bag with quite a few devices where one leg of the usb connector went loose waiting for me to have useful hands |
07:32.16 | mnemoc | taping didn't work |
07:35.45 | mnemoc | weeee... my (second) mele arrives today! |
07:36.45 | mnemoc | this one will go to the tv |
07:39.35 | lundman | seriously? I've not found a way for mele to be useful yet |
07:39.51 | lundman | which is why I'm trying this miniand image |
07:41.08 | naobsd | btw I finished to make initial ver. of small program...which just changes display output to HDMI for now... and added it to my CWM image. |
07:41.36 | Triffid_Hunter | I want to use it to control my 3d printer, with web interface and webcam.. having a touchscreen on the printer itself might be a neat upgrade further down the track if ever I find a cheap enough hdmi touchscreen |
07:41.40 | mnemoc | lundman: my daughters connect an A10 based tablet to the tv every morning and every afternoon to play videos.... a mele can do that job better |
07:41.41 | naobsd | now I can use CWM with HDMI display if HDMI cable is connected while power on |
07:42.26 | Triffid_Hunter | and yeah my kids would love a set-top box, does that image have support for the remote that dx package with it? |
07:42.27 | lundman | ah you mean, using the official mele firmware? |
07:42.43 | lundman | it can't stream videos over network (android 2.3 problem) so it is no good for me |
07:42.53 | lundman | I'm used to PCH, so all media is on NAS and gets streamed |
07:43.00 | Triffid_Hunter | of course, our videos are served over NFS and have lots of strange video formats so not keen on trying to play most of them in vanilla android |
07:43.04 | mnemoc | lundman: I planned to use Turl's android |
07:43.21 | lundman | ICS4 streams great, but too limited still, no video resolution selection etc |
07:43.29 | Triffid_Hunter | lundman: I stream videos over network with my android 2.3 phone using cifsmount and mxplayer |
07:43.48 | Triffid_Hunter | may not technically be streaming when done like that I suppose |
07:43.58 | lundman | samba is a windows monster |
07:44.01 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
07:44.03 | lundman | I could live with NFS, but better would be upnp |
07:44.04 | *** join/#arm-netbook rell01 (~chatzilla@p5B078F30.dip0.t-ipconnect.de) |
07:44.17 | lundman | since with PCH, whole house is upnp with llink |
07:44.18 | Triffid_Hunter | nfs gives me so many issues, samba actually works better here |
07:44.26 | mnemoc | I expect to use the mele's sata instead of streaming |
07:44.40 | Triffid_Hunter | we get incessant problems with things randomly becoming unavailable due to stale NFS handles |
07:44.49 | lundman | especially since with upnp you can send media from one device, to another device to play and all that |
07:45.08 | lundman | bubbleupnp on ICS4 is perfect, but alas, ICS4 for mele is not ready to be ysed |
07:45.11 | lundman | *used |
07:48.31 | mnemoc | lundman: tried using the open android trees for sun4i? |
07:48.50 | mnemoc | Quarx's for example |
07:48.58 | naobsd | oh I didn't know about bubbleupnp... seems nice! |
07:49.02 | lundman | first I've heard of it? |
07:50.11 | mnemoc | afaik Turl and ZaEarl base their android work on Quarx's repos |
07:50.18 | lundman | they are images? |
07:51.17 | mnemoc | i doubt |
07:51.22 | *** join/#arm-netbook RITRedbeard (~RITReadbe@c-68-37-165-37.hsd1.nj.comcast.net) |
07:51.44 | lundman | ok booting the mk image |
07:53.16 | Triffid_Hunter | someone tell me if that image works on a mele |
07:54.59 | mnemoc | you would probably need to recompile the kernel to add ethernet and sata |
07:55.45 | mnemoc | help testing the android-less 3.3 tree is very welcomed |
07:56.37 | mnemoc | only the usb gadget part is known to be brokwn, but that's not relevant for the mele |
07:59.05 | naobsd | about ICS for mele, I guess bootloader files except script.bin from another ICS device can be used. |
07:59.35 | lundman | ok not bad. the OS is nice and clean, but yes without video decoder you can't play anything really |
07:59.44 | naobsd | kernel can be compiled from source. other userland parts are portable |
08:01.10 | lundman | gets me both wlan0 and eth0 on mele |
08:01.27 | lundman | let me plug in the sata |
08:02.37 | lundman | hmm yes, no sata |
08:04.40 | Quarx | mnemoc: hi, about my problem with display(aurora) http://epsy.ath.cx:888/paste/?528 i added some printk to disp. |
08:05.54 | lundman | bah also dont let you chose the video resolution |
08:06.41 | lundman | quarx: so you have something I should try on mele? |
08:06.42 | mnemoc | lundman: unfortunatelly you need to edit your script.bin to change the resolution |
08:06.57 | mnemoc | Quarx: can you add a ticket on github? |
08:07.00 | lundman | mnemoc: ah , so you can at least change it? |
08:07.32 | mnemoc | lundman: sure, sepp_ did that last night and updated the wiki |
08:07.54 | lundman | wait, there is a wiki |
08:08.02 | mnemoc | yes... |
08:08.15 | lundman | why am I last to find out about these things |
08:08.37 | RITRedbeard | link? |
08:08.54 | Quarx | mnemonic: my english is bad, for good describe the problem.. |
08:09.09 | mnemoc | http://rhombus-tech.net/allwinner_a10/hacking_the_mele_a1000/ |
08:09.35 | mnemoc | Quarx: bad english is welcomed too ;-) |
08:09.40 | lundman | oh that page ok. dont feel like a wiki :) |
08:10.05 | mnemoc | lundman: it is |
08:10.14 | mnemoc | lundman: iki, but wiki |
08:13.38 | lundman | ok binary file, lets hope it doesnt have checksums |
08:13.43 | naobsd | https://github.com/amery/linux-allwinner/blob/allwinner-v3.0-android-v2/include/linux/drv_display_sun4i.h#L132 type |
08:14.05 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
08:14.05 | naobsd | https://github.com/amery/linux-allwinner/blob/allwinner-v3.0-android-v2/include/linux/drv_display_sun4i.h#L149 mode for TV/HDMI |
08:14.22 | mnemoc | lundman: get sunxi-tools from github to (de)compile it |
08:14.27 | naobsd | https://github.com/amery/linux-allwinner/blob/allwinner-v3.0-android-v2/include/linux/drv_display_sun4i.h#L149 mode for VGA |
08:14.49 | lundman | yeah that was was self explanatory, doesnt say how to edit script.bin, or that its called sys_config1.mele_mod.bin |
08:15.59 | naobsd | did anyone try ioctl(DISP_CMD_FB_RELEASE)? |
08:16.01 | lundman | actually, the variable is in all 3 .bin files |
08:16.32 | naobsd | I'm not sure what happen if I do release/request to change fb resolution... |
08:17.12 | mnemoc | an oops probably, those drivers are quite fragile |
08:17.45 | mnemoc | lundman: the bootloader takes one (usually script.bin) an loads it into memory, then the kernel reads it from there |
08:18.44 | lundman | there is no script.bin, but other docs say to use evb.bin |
08:18.46 | naobsd | bootcmd=run load1 boot_mmc |
08:18.54 | naobsd | load1=fatload mmc 0 43000000 evb.bin |
08:19.13 | naobsd | put modified .bin as evb.bin in 1st partition |
08:19.17 | lundman | but it talks about bin2fex to "decrypt" it, which bugs me, as it is clearly not encrypted |
08:19.28 | naobsd | it's u-boot env in that ubuntu img |
08:19.49 | mnemoc | lundman: who says "decrypt"> |
08:19.51 | mnemoc | ? |
08:20.01 | lundman | http://www.cnx-software.com/2012/05/06/editing-allwinner-a10-board-configuration-files-script-bin/ |
08:20.38 | mnemoc | the format is pretty dumb, it's more of a "serialize/deserialize" |
08:20.43 | lundman | yep |
08:20.51 | lundman | I'll just use emacs hex mode to change it |
08:21.27 | naobsd | binary is a cryptic ;) |
08:23.37 | naobsd | on miniand forum, someone mentioned bootloader part is same as ubuntu for mele |
08:23.54 | lundman | ah no, it is checksummed |
08:24.09 | mnemoc | lundman: it's not |
08:24.14 | lundman | 00003150: 6f75 7470 7574 5f74 7970 6500 0000 0000 output_type..... |
08:24.14 | lundman | 00003160: 0000 0000 0000 0000 af23 0000 0100 0100 .........#...... |
08:24.21 | lundman | what would the af23 be ? |
08:25.00 | naobsd | sunxi-tools must be easy... |
08:25.28 | mnemoc | lundman: offset to the data |
08:25.40 | mnemoc | lundman: use the decompiler, it's simpler |
08:25.49 | lundman | yeah, took all of 7s to setup |
08:26.30 | mnemoc | the file has the section names first, then the properties names, and then the properties values |
08:27.33 | lundman | ok booting |
08:28.15 | naobsd | https://www.miniand.com/forums/forums/2/topics/1#post-16 oh this is first time I saw "true" CPU spec... 1.5GHz = CPU 1GHz + GPU... |
08:28.24 | naobsd | err |
08:28.55 | lundman | hahah, ok, plugging in monitor as well |
08:29.05 | naobsd | first time factory or reseller guy mentioned truth |
08:29.15 | mnemoc | the cpu CAN be overclocked up 1.5GHz but above 1.2 becomes too unstable |
08:29.57 | naobsd | it's common formula for chinese tablets |
08:30.12 | lundman | sucksess! |
08:30.34 | naobsd | dual/manycore = CPU + GPU + any other units in SoC |
08:32.02 | lundman | but then, maybe I can change ICS4's resolution as well.. |
08:32.13 | lundman | although, it uses that phoenix thing to write to nand |
08:32.29 | naobsd | mount /dev/block/nanda |
08:32.29 | mnemoc | script.bin lives in /dev/nanda |
08:32.47 | *** join/#arm-netbook eebrah (~Ibrahim_K@41-139-199-130.safaricombusiness.co.ke) |
08:32.58 | lundman | you guys are full of good ideas |
08:34.43 | naobsd | you may try any ICS rom for another A10 device by replacing script.bin with yours |
08:34.52 | lundman | ah no dev nanda on the lubuntu boot |
08:35.16 | lundman | ah ram0 |
08:35.18 | mnemoc | lundman: you might be missing the sun4i-module |
08:35.20 | mnemoc | err |
08:35.24 | mnemoc | sun4i-nand module |
08:35.37 | lundman | miniand missed sun4i-nand module? |
08:35.57 | mnemoc | if you don't see the nand partitions... |
08:36.05 | lundman | looks like it is not in there |
08:36.26 | mnemoc | be brave, build your own kernel ;-) |
08:36.32 | lundman | tedious |
08:36.44 | mnemoc | then you can release better images |
08:36.47 | lundman | done nothing but build kernels my whole life |
08:36.49 | mnemoc | and make the world a better place |
08:36.55 | lundman | I'll just copy over the .ko |
08:37.52 | mnemoc | :( |
08:38.07 | mnemoc | no one seems to want to help me testing the 3.3 branch :< |
08:38.20 | lundman | 3.3 of ? |
08:38.27 | naobsd | Linux version 3.0.8+ (tangliang@MServer) (gcc version 4.5.1 (Sourcery G++ Lite 2010.09-50) ) #2 PREEMPT Fri Mar 2 14:28:08 CST 2012 |
08:38.36 | naobsd | this is kernel version string used in mk802 ubuntu |
08:38.47 | lundman | wish they'd stop with preempt too |
08:39.07 | naobsd | I guess it's not built by miniand guy |
08:39.10 | mnemoc | that one comes from allwinner's closed tree |
08:39.53 | naobsd | kernel modules have same timestamp (Mar 2) |
08:39.54 | mnemoc | with android's paranoic network etc |
08:40.01 | naobsd | lol |
08:40.06 | naobsd | it's not for linux ;) |
08:40.37 | mnemoc | use https://github.com/amery/linux-allwinner/ |
08:40.40 | lundman | dont think the eth0 works after all |
08:41.09 | mnemoc | the lichee-3.0.8-sun4i branch is compatible with the closes source one |
08:41.25 | mnemoc | allwinner-v3.0-android-v2 is our "stable" one |
08:41.47 | mnemoc | and wip/linux-3.3-sunxi/forward-3.0-v2 is "next" |
08:42.50 | mnemoc | most of those 3.0.8+ ubuntu images are based on hipboi's initial image, which comes with an script.bin (called evb.bin) with broken ethernet |
08:42.57 | naobsd | someone who wants ubuntu(or any other linux) should try mnemoc's 3.3 and help him :) |
08:43.23 | mnemoc | naobsd: thanks for your support :) |
08:43.33 | naobsd | porting: moving things to somewhere else |
08:43.41 | lundman | the kernel version seems to be irrelevant, since we dont have video decoder anyway? |
08:43.58 | mnemoc | lundman: we do... as a .a |
08:44.05 | lundman | does 3.3 kernel have nand? if so, I'm happy to boot it :) |
08:44.17 | lundman | stripped .a, or ? |
08:44.18 | mnemoc | lundman: allwinner-v3.0-android-v2 has functional nand too |
08:44.37 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
08:44.40 | lundman | mnemoc: ah hmm use adb then |
08:44.41 | mnemoc | lundman: not stripped. documentation and .h files are available too |
08:45.01 | mnemoc | lundman: 3.3 has broken usb gadget |
08:45.16 | lundman | all usb broken, or something called "gadget" ? |
08:45.21 | naobsd | probably I need to get serial console for playing linux ;) |
08:45.27 | mnemoc | lundman: but 3.0-v2 is quite decent |
08:45.58 | naobsd | usb gadget: kernel function for acting as USB device like as mass storage |
08:45.59 | mnemoc | lundman: usb gadget is the part were you device acts as a usb slave, for adb for example |
08:46.12 | lundman | ah ok |
08:46.15 | mnemoc | lundman: usb hosts works well here (tm) |
08:46.24 | lundman | that is precisely what I dont need, so thats something |
08:46.38 | lundman | where is this kernel for me to download :) |
08:47.15 | mnemoc | https://github.com/amery/linux-allwinner/ |
08:47.40 | naobsd | git clone git://github.com/amery/linux-allwinner.git |
08:48.03 | naobsd | -b wip/linux-3.3-sunxi/forward-3.0-v2 |
08:48.11 | naobsd | :) |
08:51.40 | mnemoc | back in 20m |
08:58.30 | naobsd | I'm trying to run ICS on CuBox... |
08:59.29 | lundman | never got any android to go on cubox myself |
08:59.34 | naobsd | only froyo binary is available for it |
08:59.36 | lundman | but I didnt try more than 2 mins |
08:59.45 | naobsd | I need to use some binary blobs from froyo |
09:00.08 | naobsd | currently I can see ICS home screen but it's unstable at all. no "usable" state yet :( |
09:15.11 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
09:45.42 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
10:16.10 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
10:34.36 | *** join/#arm-netbook Manvindar (irc2gowebc@101.210.34.75) |
10:35.45 | Manvindar | is there any other compiler than keil?? |
10:36.52 | Triffid_Hunter | lol, how about gcc |
10:36.54 | mnemoc | .o(what is keil?)o |
10:37.19 | Manvindar | keil IDE |
10:37.36 | Manvindar | i dont know how to use gcc |
10:37.42 | mnemoc | learn |
10:38.10 | Manvindar | any other soft. with good GUI |
10:38.51 | Triffid_Hunter | Manvindar: none that don't have a learning curve |
10:39.28 | mnemoc | and keep in mind that IDE != compiler |
10:44.18 | Manvindar | yup |
10:44.41 | Manvindar | u have any xperience with LPC2129 programming? |
10:46.10 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
10:46.14 | mnemoc | Manvindar: this is going way-too-offtopic |
10:47.06 | Manvindar | whats d topic of this irc?? |
10:47.21 | Manvindar | arm? |
10:48.40 | mnemoc | initially it was open source friendly arm netbooks, yes. |
10:49.12 | Manvindar | oh... |
10:49.52 | Manvindar | is there any leak of ubuntu for arm??? |
10:50.05 | hno | leak? It's all in the open. |
10:50.08 | Triffid_Hunter | Manvindar: leak? doesn't need to be leaked, there's plenty of it around |
10:50.36 | Triffid_Hunter | Manvindar: can only leak things that aren't supposed to be publically available.. open source doesn't really work like that |
10:50.37 | hno | Hard to leak something which is public. |
10:50.49 | mnemoc | hno: there is a ticket in your u-boot repo which might be interesting for you |
10:51.36 | hno | mnemoc, the 1GB memory one? Or something else? |
10:51.42 | mnemoc | yes, that one |
10:52.13 | hno | both me and Tom already responded and it's making good progress. |
10:52.16 | mnemoc | Triffid_Hunter: but it won't run on your LPC2129 (armv4) |
10:52.23 | mnemoc | hno: uh |
10:52.30 | Manvindar | hmmm |
10:52.32 | mnemoc | scratches his head |
10:52.45 | mnemoc | Triffid_Hunter: sorry |
10:53.20 | mnemoc | Manvindar: but it won't run on your LPC2129 (armv4), the armel repos need are armv5 |
10:53.37 | mnemoc | damn... /me goes for another coffee |
10:53.40 | Manvindar | in mobile world congress ubuntu had demo of android and ubuntu running at same time...donno when they will release it |
10:54.12 | Manvindar | @mnemoc ye i understand sir.. |
10:54.36 | mnemoc | Manvindar: http://www.ubuntu.com/devices/android |
10:54.44 | Triffid_Hunter | I've seen a vid of someone running linux on a lpc1788 |
10:54.45 | Manvindar | yep |
10:55.07 | Manvindar | 1788 is armv5?? |
10:55.16 | mnemoc | v7-m |
10:55.25 | mnemoc | m3 |
10:57.09 | hno | Right.. linaro did manage to get the kernel to build in Thumbs mode so should be possible to run Linux on that now. But there is no distribution built for thumbs mode. |
10:57.16 | hno | if there is sufficient memory. |
10:58.05 | Triffid_Hunter | 1788 supports external memory controller iirc which is what would make it possible |
11:01.11 | Manvindar | its not possible to interface xternal memory controller with 2129? |
11:02.25 | hno | But why would one want to build Linux on an Coretex-M class CPU? If one want's a low-end Linux ARM then Cortex A5 is meant to fill that space. |
11:03.22 | Manvindar | yea thats obvious.. |
11:05.34 | *** join/#arm-netbook leojrfs (~leojrfs@bl17-14-174.dsl.telepac.pt) |
11:11.12 | Quarx | mnemoc: i found problem |
11:11.36 | mnemoc | Quarx: source or script.bin? |
11:12.09 | mnemoc | (smells very script.bin-ish to me) |
11:13.21 | Manvindar | anyone has ported ubuntu on raspberry pi? |
11:13.33 | Quarx | mnemoc: drivers/video/sun4i/disp/de_bsp/de/disp_de.c 1701 line |
11:14.43 | mnemoc | https://github.com/amery/linux-allwinner/blob/allwinner-v3.0-android-v2/drivers/video/sun4i/disp/de_bsp/de/disp_de.c has only 248 lines... |
11:14.50 | Quarx | oh |
11:14.55 | Quarx | de_bsp/de/disp_lcd.c |
11:15.22 | lundman | ok |
11:18.06 | Quarx | if i comment this line, device wake up successfully |
11:18.33 | mnemoc | uhm |
11:19.36 | Quarx | i think it is not good for power consumption |
11:20.15 | Triffid_Hunter | Manvindar: isn't their main image an ubuntu or debian one? |
11:25.58 | mnemoc | Quarx: that sel seems to come from script.bin |
11:26.51 | mnemoc | .oO |
11:27.14 | Quarx | http://epsy.ath.cx:888/paste/?536 my script bin |
11:28.04 | mnemoc | Quarx: can you compare the lcd sections of yours with the one of another similar tablet that works fine? |
11:29.12 | mnemoc | Quarx: are you sure your script.bin matches perfectly that commented .fex you pasted? |
11:29.33 | Quarx | mnemoc: from stock aurora rom |
11:30.22 | mnemoc | Quarx: it's very common that the included .fex doesn't match the .bin. decompile yours and diff |
11:32.00 | Quarx | mnemoc: it is my decompiled script.bin |
11:32.23 | mnemoc | Quarx: it can't be. it has comments |
11:32.56 | mnemoc | Quarx: there is no way on earth the decompiler could reconstruct the comments |
11:33.07 | Quarx | mnemoc: i use img_depacker |
11:34.00 | mnemoc | pass your script.bin to bin2fex from sunxi-tools |
11:35.41 | mnemoc | Quarx: that tool only extracts the partitions packed inside the image |
11:36.03 | Quarx | mnemoc: ok, but it decompile script.bin too |
11:37.39 | mnemoc | Quarx: just to be sure, please `diff -uw` that .fex with the output of bin2fex |
11:38.43 | Quarx | mnemoc: http://epsy.ath.cx:888/paste/?537 |
11:38.57 | Quarx | the same, but withou comments |
11:39.20 | Xlab | Nope |
11:39.22 | Xlab | dram_clk= 360 |
11:39.51 | Xlab | I was wondering why the version with comments has dram_clk = 450 |
11:40.05 | Quarx | dram clock is my change :) |
11:40.09 | Xlab | %) |
11:40.31 | Quarx | it is increased gpu perfomance |
11:40.55 | mnemoc | Quarx: can you diff that to the script.bin of a similar tablet that works fine? |
11:41.13 | Quarx | i have only turl (zatab) config |
11:42.29 | CIA-121 | rhombus-tech: lkcl master * ra17bbe9dee13 / (evaluated_cpus.mdwn iMX6.mdwn): add iMX6 pricing and link to evaluated CPUs |
11:42.59 | mnemoc | Quarx: anyhow, I can't look at your problem until the weekend, so I would really really appreciate if you can collect the info in a ticket on github |
11:43.18 | Quarx | ok, i will |
11:43.33 | mnemoc | bad english is not a problem, mine is worse anyway |
11:52.01 | mnemoc | Quarx: btw, I wanted to try your ICS on an onda vi10 elite and a mele a2000, do you have documentation on how to start with it? (beside the generic android documentation) |
11:52.33 | mnemoc | want* |
11:53.51 | Quarx | mnemoc: this ICS will work on any a10 device.. need change only kernel modules |
11:54.13 | mnemoc | then I can try the script.bin of that tablet and compare with yours and try to debug the problem ;-) |
11:54.30 | mnemoc | Quarx: "this"? |
11:54.48 | Quarx | mnemoc: "my" |
11:54.52 | Quarx | :) |
11:55.41 | mnemoc | https://github.com/Quarx2k/android <--- so I just give "this" to repo and follow the standard procedure? |
11:56.01 | mnemoc | (ics branch) |
11:56.28 | Quarx | allwinner branch |
11:56.35 | Quarx | ics for motorola defy |
11:56.49 | mnemoc | ok |
11:58.24 | *** join/#arm-netbook eFfeM_work (~frans@D4B268BA.static.ziggozakelijk.nl) |
11:59.21 | mnemoc | `repo sync` running.... back to $work$ :p |
12:03.08 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
12:07.17 | eFfeM_work | yay, got my a2000 today! |
12:07.42 | eFfeM_work | anyone knows if there is already an english version of ICS for it (google didn't really help me with this one) |
12:09.22 | hno | eFfeM_work, better to learn how to build your own I think. |
12:17.28 | *** join/#arm-netbook leojrfs (~leojrfs@bl17-14-174.dsl.telepac.pt) |
12:17.52 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
12:18.54 | Quarx | mnemoc: another fix.. better than previous: cooment it LCD_CLOSE_FUNC(sel, LCD_power_off, 1000); |
12:21.06 | eFfeM_work | hno good plan but actually would like to start seeing if I can just get new sw on it |
12:25.45 | mnemoc | Quarx: ticket *please* |
12:32.53 | lundman | effem: just use the ics4 they released and change language to english |
12:35.30 | mnemoc | Quarx: do you have a tarball with the extracted files available for download? |
12:35.53 | Quarx | mnemoc: no, only github |
12:36.29 | mnemoc | Quarx: out/target/product/a10/obj/lib/libMali.so on github? |
12:36.37 | Quarx | yes |
12:36.45 | mnemoc | scratches |
12:36.55 | Quarx | in prebuilt |
12:38.11 | CIA-121 | rhombus-tech: Haakon master * r38f304e452ec /allwinner_a10/hacking_the_mele_a1000/script_for_installing_debian_on_sdcard.mdwn: V1.6 15.06.2012 HSt Bugfixes |
12:38.54 | mnemoc | Quarx: any particular command to copy them where `make` expects them? |
12:39.38 | hno | eFfeM_work, A10 are unbrickable. You can always get new software on them. Even the ones with locked bootloaders can be reflashed completely. |
12:39.51 | Quarx | mnemoc: it automatically copied..blobs mk |
12:40.14 | eFfeM_work | hno yeah read that |
12:40.25 | eFfeM_work | lundman do you by any chance have a url ? |
12:40.50 | eFfeM_work | in the end I'll proobably put ubuntu on it with some server sw |
12:41.03 | eFfeM_work | planning to port amahi |
12:47.54 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
12:52.39 | lundman | to the image? sure |
12:52.46 | lundman | http://lundman.net/ftp/mele/ |
12:59.12 | mnemoc | Quarx: ok, I'll try to build it again |
13:15.04 | eFfeM_work | lundman: thanks (was afk) |
13:16.42 | *** join/#arm-netbook mikey_w (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
13:16.57 | *** join/#arm-netbook mike_ (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
13:17.55 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
13:21.45 | mnemoc | Quarx|Away: suggestions? http://pastie.org/4092290 |
13:22.41 | Quarx|Away | mnemoc: copy android_device_a10 / include / system /window.h to /system/core/include/system |
13:23.03 | Quarx|Away | oh no.. |
13:23.11 | Quarx|Away | add defines to it |
13:23.16 | Quarx|Away | from this file |
13:25.13 | *** join/#arm-netbook eebrah (~Ibrahim_K@41-139-199-130.safaricombusiness.co.ke) |
13:33.06 | mnemoc | Quarx|Away: thanks |
13:47.28 | mnemoc | Quarx|Away: now it's ranting about CAMERA_CMD_SET_SCREEN_ID which is not defined anywhere... do I have to change the branch of something? |
13:59.14 | mnemoc | Turl: does your tree have these things fixed? |
14:06.24 | *** join/#arm-netbook MMlosh (~MMlosh@2001:718:1e03:5168:a1b3:dbf1:3b5b:3c77) |
14:09.50 | Quarx | mnemoc: http://epsy.ath.cx:888/paste/?539 |
14:10.03 | Quarx | mnemoc: ystem/core/include/system camera.h |
14:10.19 | Quarx | only 2 headers missing on github |
14:11.48 | mnemoc | Quarx: was I don't something wrong or those are "known issues"? |
14:12.13 | Quarx | mnemoc: it is known issues |
14:12.19 | Quarx | for my repo |
14:13.33 | mnemoc | meh, now native_window_set_buffers_geometryex not declared :< |
14:14.07 | Quarx | mnemoc: yeah.. u should not replace window.h |
14:14.38 | mnemoc | I didn't... only added the missing enum entries |
14:14.53 | Quarx | http://epsy.ath.cx:888/paste/?540 |
14:14.56 | Quarx | my window.h |
14:16.57 | mnemoc | ok... i'll overwrite that one too... |
14:17.48 | A1Sheds | I wonder what the lead times will be for imx6 devices ones they get the next spin of silicone done? |
14:18.13 | mnemoc | Quarx: http://dpaste.com/759994/ <--- resulting diff |
14:18.47 | A1Sheds | has anyone seen the tree yet for anything imx6 related? |
14:19.12 | Quarx | mnemoc: i will fix it on github |
14:19.27 | mnemoc | A1Sheds: yes, but steev is our local expert for imx6 |
14:19.31 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
14:22.17 | mnemoc | A1Sheds: there are kernel and u-boot repos for imx6 on github, and linaro supports their devkit |
14:23.01 | A1Sheds | https://android-build.linaro.org/builds/~linaro-android/imx6-ics-gcc47-freescalelt-stable-open-12.05-release/#build=3 android build |
14:23.54 | *** join/#arm-netbook leojrfs (~leojrfs@233.163.108.93.rev.vodafone.pt) |
14:25.34 | mnemoc | Quarx: I think i'll better pospone my ICS experiments until I actually have time to fix problems myself.... now it was: |
14:25.37 | mnemoc | frameworks/base/ethernet/java/android/net/ethernet/IEthernetManager.aidl:19: couldn't find import for class android.net.ethernet.EthernetDevInfo |
14:25.41 | mnemoc | make: *** [out/target/common/obj/JAVA_LIBRARIES/framework_intermediates/src/ethernet/java/android/net/ethernet/IEthernetManager.java] Error 1 |
14:25.59 | Quarx | oh |
14:26.51 | A1Sheds | it looks like the imx6 might be the first ARM soc that is freely available and worth developing on |
14:27.04 | Quarx | mnemoc: forgot about it... /build/core/pathmap.mk add ethernet \ to FRAMEWORKS_BASE_SUBDIRS := \ |
14:27.15 | mnemoc | A1Sheds: s/is/will be* |
14:27.18 | Quarx | should be last problem |
14:27.24 | A1Sheds | i.MX 6Quad Consumer Ext. Temp 1.2GHz w/ VPU&GPU ~$29 |
14:27.36 | traeak | "freely available" you mean purchasable in small quantities ? |
14:28.23 | A1Sheds | mnemoc, I'm wondering what allocation is going to be like. Maybe it's time to start lining up suppliers in China |
14:28.52 | A1Sheds | traeak, yes, any size orders through major distributors |
14:29.18 | mnemoc | A1Sheds: I'm lining up for the efika mx6 laptop :p |
14:29.55 | traeak | they're using a vivante gpu thought right? |
14:30.02 | traeak | so on the software side it's a bit dicier |
14:30.03 | A1Sheds | http://pastebin.ca/2161667 imx6 pricing |
14:32.09 | A1Sheds | I'm wondering what the associated weirdness will be for imx6. 6 month lead times, more bugs, etc |
14:37.38 | A1Sheds | https://github.com/dirkbehme/linux-imx6 |
14:38.25 | mnemoc | Quarx: thanks for the ticket |
14:39.19 | Quarx | mnemoc: :) but i can't set labels |
14:39.38 | mnemoc | Quarx: next error are a gazilion of messages like: packages/apps/Settings/res/xml/testing_settings.xml:42: error: Error: No resource found that matches the given name (at 'title' with value '@string/testing_wifi_info'). |
14:40.20 | *** join/#arm-netbook rell01 (~chatzilla@p4FE562F1.dip0.t-ipconnect.de) |
14:40.59 | Quarx | mnemoc: strange error.. maybe it happened after merge with upstream.. i will check it now |
14:50.33 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
14:50.42 | Quarx | mnemoc: https://github.com/Quarx2k/android_packages_apps_Settings/commit/aa723c9313f3406c9bde482a817217a3e703ef3f yea fail merge :) |
14:51.40 | mnemoc | Quarx: should I backup my changes to include/system before `repo sync` or those are fixed now too? |
14:52.47 | Quarx | mnemoc: repo sync will not remove your changes in build/ and system/ but anyway i will fix it now too |
14:53.21 | Quarx | mnemoc: oh new error in settings app.. |
14:53.31 | mnemoc | last time I played with android `repo sync` loved to destroy all my changes |
15:03.38 | libv | general Q: mele A2000 is a mele A1000 with less nand and in a smaller casing, right? |
15:04.19 | Quarx | mnemoc: all errors should be fixed now |
15:04.29 | libv | i am thinking about taking a dremel to the A1000 case and putting in a serial connector |
15:04.30 | mnemoc | libv: both have 4GB of nand |
15:04.39 | libv | A2000 has no room for that apparently |
15:04.41 | libv | ok |
15:05.22 | libv | so it is all about the case then |
15:05.39 | mnemoc | libv: and the a2000 doesn't come with the sleeve for the sata disk :| |
15:05.50 | mnemoc | libv: so you dock it naked |
15:06.13 | mnemoc | but there is one amazing difference in the board itself! |
15:06.27 | mnemoc | the uart pins are horizontal instead of vertical :p |
15:06.53 | mnemoc | I suppose that is what makes the a2000 more expensive |
15:07.04 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
15:07.11 | *** join/#arm-netbook Tsvetan (d55bf1f8@gateway/web/freenode/ip.213.91.241.248) |
15:07.14 | Tsvetan | hi |
15:07.40 | ssvb | mnemoc: yes, I guess it is just a cheaper hardware with higher price tag |
15:07.44 | libv | ok, but no difference otherwise from where you sit? |
15:08.05 | libv | (as in, with respect to kernel development) |
15:08.30 | mnemoc | libv: they are the same |
15:08.42 | mnemoc | even the same revision written on the pcb |
15:08.58 | libv | ok, A1000 it is then, as there is space for a serial port below the vga connector |
15:09.03 | libv | mnemoc: thanks :) |
15:09.06 | mnemoc | :) |
15:09.39 | traeak | libv: just rip out the vga connector? |
15:09.44 | mnemoc | but I think you can get the uart wires out without drilling |
15:10.01 | mnemoc | using the ventilation holes on the bottom |
15:10.41 | traeak | mnemoc: the holes are just touch too small for that |
15:10.52 | libv | traeak: i have been doing display/modesetting work for the best part of the last decade, i do want the vga connector to be available as well :) |
15:10.54 | mnemoc | :( |
15:10.55 | A1Sheds | https://www.miniand.com/forums/forums/2/topics/1 MK802 Ubuntu Images, the site is back up |
15:12.48 | A1Sheds | http://liliputing.com/2012/06/how-to-re-install-mk802-firmware.html |
15:13.04 | traeak | since the bottom of the mele is raised using the stamped steel of the case bottom drilling out one of the ventilation holes and runing the serial wires out that will will work |
15:13.04 | A1Sheds | winderz only tool |
15:13.20 | traeak | i don't keep a drill at work either |
15:15.18 | A1Sheds | https://www.miniand.com/forums/forums/2/topics/3 some links to uboot scripts |
15:15.48 | traeak | https://www.miniand.com/forums/forums/2/topics/1 |
15:15.50 | traeak | oops |
15:16.52 | A1Sheds | http://www.cnx-software.com/2012/06/13/hardware-packs-for-allwinner-a10-devices-and-easier-method-to-create-a-bootable-ubuntu-12-04-sd-card/ |
15:17.20 | *** join/#arm-netbook Xlab1 (~Adium@home-pool-188-244-36-2.com2com.ru) |
15:20.04 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
15:23.52 | *** join/#arm-netbook rell01_ (~chatzilla@p4FE562F1.dip0.t-ipconnect.de) |
15:27.26 | hp_ | A1Sheds: the pre-build image for the mk802 looks also nice.. just downloaded it.. |
15:27.30 | hp_ | i might give it a try |
15:28.43 | *** join/#arm-netbook tom_say (~quassel@cpe-72-178-202-3.stx.res.rr.com) |
15:30.30 | hp_ | A1Sheds: i wonder what the lubuntu will bring |
15:33.49 | A1Sheds | hp_, let me know how well it works, Ubuntu is almost like Linux |
15:35.17 | hp_ | Lubuntu |
15:35.24 | hp_ | its the lightweight thingy |
15:36.32 | mnemoc | lxde based |
15:36.55 | hp_ | Win32 Disk Imager under windows 7 ugh!! |
15:39.54 | Quarx | mnemoc: build finished? |
15:40.24 | mnemoc | Quarx: uh, didn't know you had finished fixing it :) .... repo sync now |
15:42.54 | mnemoc | repo is more polite now, it didn't trash my changes, it aborted the sync :p |
15:43.39 | Quarx | mnemoc: i added this changes to github... |
15:44.18 | mnemoc | cd core |
15:44.20 | mnemoc | err |
15:48.54 | mnemoc | Quarx: I was asked to `make update-api` after some errors |
15:49.13 | mnemoc | trying again now |
15:52.48 | Quarx | ahh latest merge, makes many errors in build |
15:54.00 | mnemoc | it's still compiling... good sign :) |
15:54.43 | Quarx | new error in NMEAParser.java |
15:56.29 | mnemoc | ./external/cedarx/CedarXAndroid/IceCreamSandwich/CedarXMetadataRetriever.h:26:29: error: GetAudio_format.h: No such file or directory |
15:57.25 | hp_ | mnemoc: just wondering what are you guys trying to do.. get android 4.0 running? |
15:58.00 | mnemoc | hp_: I'm trying to compile Quarx's android tree, yes |
15:58.56 | Quarx | i fixed all locally, not pushed.. one moment |
15:59.15 | mnemoc | `repo status` is your friend ;-) |
16:19.51 | Quarx | mnemoc: ok, all errors fixed (i hope):)) |
16:20.09 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
16:20.22 | *** join/#arm-netbook eFfeM (~frans@g235028.upc-g.chello.nl) |
16:21.23 | mnemoc | Quarx: let's see ;-) |
16:22.19 | Quarx | mnemoc: for me building is done Package complete: /home/quarx/Android/alwinner/out/target/product/a10/CM9-NIGHTLY-120615-A10.zip |
16:30.58 | mnemoc | make: *** No rule to make target `out/target/product/a10/obj/lib/libMali.so', needed by `out/target/product/a10/obj/SHARED_LIBRARIES/gralloc.sun4i_intermediates/LINKED/gralloc.sun4i.so'. Stop. |
16:31.39 | *** part/#arm-netbook Xlab1 (~Adium@home-pool-188-244-36-2.com2com.ru) |
16:31.44 | mnemoc | Quarx: --^ |
16:32.24 | Quarx | mnemoc: prebuilt dir exist? |
16:32.30 | Quarx | in device/a10 ? |
16:32.53 | mnemoc | Quarx: seems to be a parallelism issue, running plain `make` instead of `make -j4` gets it moving |
16:33.30 | mnemoc | the rule that creates out/target/product/a10/obj/lib/libMali.so not been well defined |
16:33.40 | Quarx | libMali.so should copied automatically to /product/a10/obj/lib/ |
16:33.56 | Quarx | i run build with -j12 :) |
16:34.47 | mnemoc | I started this... experiment with -j16.... but have been shrinking it after each failure |
16:35.42 | mnemoc | make: *** No rule to make target `out/target/product/a10/obj/lib/libMali.so', needed by `out/target/product/a10/obj/SHARED_LIBRARIES/gralloc.sun4i_intermediates/LINKED/gralloc.sun4i.so'. Stop. again.... wtf |
16:36.02 | mnemoc | device/allwinner/a10/prebuilt/ exists |
16:37.00 | Quarx | rule to copy lib in a10-blobs.mk |
16:38.10 | mnemoc | will try again tomorrow having more brain cells to dedicate to it |
16:38.45 | Quarx | lunch cm_a10-userdebug ? |
16:38.56 | mnemoc | full eng |
16:39.04 | mnemoc | full a10 eng |
16:39.23 | Quarx | hm, i'm always do cm_a10-userdebug |
16:40.06 | mnemoc | will try that one next, but I need to get some $work$ done |
16:41.12 | mnemoc | Quarx: make running, let's see. thank you :) |
16:41.53 | mnemoc | make: *** No rule to make target `vendor/cm/proprietary/Term.apk', needed by `out/target/product/a10/system/app/Term.apk'. Stop. |
16:41.56 | mnemoc | haha |
16:42.21 | *** join/#arm-netbook naobsd (7c294571@gateway/web/freenode/ip.124.41.69.113) |
16:42.52 | eFfeM | just unboxed my a2000, interesting experience, especially if you cannot read chinese, got the language changed but still lots of chinese stuff :-) |
16:43.31 | Quarx | mnemoc: it is normal |
16:43.48 | Quarx | mnemoc: run //vendor/cm/get-prebuilts |
16:43.53 | Quarx | for first build |
16:44.33 | eFfeM | upgrading an img from linux is that just a matter of dd-ing the image to an SD card and put it in the A2000 ? |
16:44.46 | eFfeM | and reboot |
16:44.51 | eFfeM | ? |
16:44.55 | eFfeM | df |
16:45.45 | mnemoc | Quarx: running again |
16:49.53 | Quarx | mnemoc: how i can connect uart to a10 without open device? |
16:50.30 | mnemoc | Quarx: playing with the script.bin you can get an uart out of the card0 (uSD) pins |
16:51.18 | Quarx | mnemoc: i read about it, on mimi-usb no way? |
16:51.28 | mnemoc | nope |
16:51.42 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
17:01.00 | Quarx | mnemoc: whats mean "bus is busy, status = f9"? driver of mma7660 spamming in dmesg and not working |
17:01.06 | Quarx | on new kernel |
17:04.51 | mnemoc | eFfeM_work: if the image fits in tour card I suppose `dd` will work fine |
17:05.56 | *** join/#arm-netbook eFfeM (~frans@g235028.upc-g.chello.nl) |
17:07.31 | sepp_ | eFfeM: what kind of image? |
17:08.02 | eFfeM | hi sepp, wanted to go to ics, figured I'd just reboot into win7 and use phoenix |
17:08.14 | eFfeM | hoped it was possible to do this fully from linux |
17:09.28 | eFfeM | guess I could; tossed the sd card in the pc and it does recognize partitions, however dd of the image file to the raw sd card (dev/sdc in my case) did not work |
17:09.30 | eFfeM | oh well |
17:10.22 | sepp | for the images from mele.cn you need that stupid win32 phoenix thing, sadly |
17:11.30 | eFfeM | I think it is possible to reroll them into an image suitable for dd |
17:12.21 | sepp | that shuold work |
17:14.53 | mnemoc | "official" images are not dd-compatibl |
17:14.55 | mnemoc | e |
17:16.38 | sepp | but one could use the win32 tool and make an dd-able image |
17:16.38 | sepp | after you have a card ... |
17:16.41 | mnemoc | ah, yes, sure |
17:17.06 | sepp | i guess that was the idea |
17:17.18 | mnemoc | and also RE it so we can make those SDs from linux |
17:19.28 | sepp | have to go now ... |
17:19.56 | mnemoc | Quarx: I have no idea what can cause that error, but comes from an infinite loop in i2c_sunxi_do_xfer() in drivers/i2c/busses/i2c-sunxi.c |
17:20.01 | mnemoc | sepp: cu :) |
17:20.15 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
17:20.48 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
17:22.10 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
17:24.04 | eFfeM | sepp, yes that was the idea |
17:24.09 | eFfeM | and thanks for yourhelp |
17:26.01 | mnemoc | eFfeM: please don't forget to share your findings :) |
17:26.39 | eFfeM | mnemoc: happy to do so, what is the preferred location ? (wiki ?) |
17:30.55 | mnemoc | eFfeM: yes |
17:31.23 | eFfeM | which wiki? I've seen several |
17:32.03 | mnemoc | http://rhombus-tech.net/allwinner_a10/ |
17:35.58 | eFfeM | ok |
17:36.55 | *** join/#arm-netbook technologov (~technolog@bzq-79-181-165-99.red.bezeqint.net) |
17:37.03 | technologov | hi dudes ! |
17:37.35 | technologov | In the past I though of ARM processors as CPUs for toasters, which had limited interest from my side, but now it all changed |
17:37.46 | technologov | with some real Linux OS demo on ARM boards |
17:38.41 | technologov | Now I'm mailing several manufacturers to manufacture good ARM PC boards, to replace the traditional x86 Intel PCs |
17:39.06 | technologov | IMO the performance of modern ARM is sufficient for this goal (to fight low-end Intel PCs) |
17:39.32 | technologov | After reviewing half-a-dozen products I really liked the concept of the new VIA APC |
17:40.10 | technologov | as it features a lot of I/O ports. Otherwise the price advantage of ARM is quickly eaten by lots of converters required |
17:40.40 | technologov | such as USB Audio card, USB powered hub, USB speakers, HDMI-to-DVI cables, etc... |
17:41.57 | technologov | Basically I believe that a board similar to VIA APC in idea, but better spec would sell like crazy |
17:44.27 | hno | technologov, we mostöy look at Allwinner A10 here. Coretex A8 with more I/O capabilities than you ever need from a SoC of the class. |
17:44.49 | eFfeM | have to leave now, back later |
17:44.56 | technologov | I also look in the same direction, the A10 :) |
17:45.34 | technologov | Just I ask for a VIA-APC-like board (read: Neo-ITX sized), but with A10 CPU and Linux support |
17:46.11 | technologov | The VIA board is great in idea, as it has all the I/O I will need: VGA, analog audio, etc... |
17:46.30 | technologov | Just under-spec'ed (720p is full stop) |
17:47.16 | technologov | so building something like the VIA APC, but with more RAM and A10 CPU and SATA port will be the product many people will buy |
17:47.36 | hno | Have you seen the Mele A1000 board we use as dev platform? Not exactly ITX form factor but.. |
17:47.53 | technologov | heck... Neo-ITX form factor is even big enough for a RAM slot, which is great |
17:48.01 | technologov | It is my prediction, that some form upgrade-ability in the form of RAM |
17:48.01 | technologov | slot and SATA port will narrow the gap vs. the traditional Intel PC, |
17:48.02 | technologov | and will drove PC users to ARM en-masse. |
17:48.26 | hno | The Mele do have SATA. |
17:48.35 | technologov | I seen A1000, but unfortunately it has too little RAM (512 MB is not enough to run Linux) |
17:48.49 | technologov | same problem for VIA |
17:49.10 | technologov | clean install of Debian 6 + KDE already takes 700 MB RAM |
17:49.49 | technologov | ... and this is without Open/LibreOffice and without Chromium... |
17:49.49 | hno | Agreed. 1GB would be much better. In theory A10 can address 2GB even but it's unknown if it works reliably (no one have built products with such memory configuration). |
17:50.25 | *** join/#arm-netbook toxicpsion (~toxicpsio@h216-45-98-91.dynamic.platinum.ca) |
17:50.33 | technologov | There is Rikomagic MK802+ device, which does have 1GB RAM, but no analog audio. Full stop. |
17:50.34 | hno | For more than 2GB you need to step up several steps in CPU class. |
17:51.11 | hno | The MK802 only have USB & HDMI. |
17:51.17 | technologov | (USB Audio card cost >$100 here, and no audio drivers for ARM Linux) |
17:51.18 | hno | and micro-SD. |
17:51.18 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
17:51.22 | technologov | yea |
17:51.38 | traeak | technologov: 512 isn't enough to run linux? maybe not enough to run a full browser... |
17:51.50 | hno | Any standard USB audio device should just work. No speical drivers needed. |
17:52.26 | technologov | basically my idea is to ask manufacturers to build a VIA-APC-sized board, but on better SoC, so this could fly... |
17:52.35 | technologov | What is the status of drivers for the A10 ? |
17:52.41 | hno | 512 (384 or what it is on the A10 when framebuffer, GPU, VPU etc have eaten their share) is not sufficient for an acceptable desktop performance. |
17:52.43 | *** join/#arm-netbook Guest6198 (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
17:52.46 | technologov | esp. Xorg drivers and hardware video decoder ? |
17:52.56 | *** join/#arm-netbook mikey_w (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
17:53.05 | hno | technologov, as sad as in pretty much all the socs. |
17:53.18 | traeak | the a10 cpu itself isn't fast enough to provide a full desktop experience |
17:53.59 | technologov | What is the name of their VPU ? (in the desktop world: NVIDIA GPU also includes the display controller _and_ the VPU inside the GPU, but not so in the ARM world, where pieces are taken from different designers...) |
17:54.15 | mnemoc | allwinner's vpu is called cedarx |
17:54.16 | hno | technologov, Allwinner. |
17:54.32 | hno | HD Video is their prime technology. |
17:54.58 | technologov | traeak, According to the benchmarks I seen, the A10 CPU is close to Intel Atom in performance, which is enough for a full KDE desktop. |
17:55.30 | technologov | Surely it will not beat Core i7 in benchmarks, but it could work... |
17:55.31 | traeak | technologov: from the testing i've been doing it's noticeably slower than an atom |
17:56.17 | technologov | traeak, Can you please upload your test results for us to study ? |
17:56.24 | traeak | and you may be mistaking pandaboard ES bechmarks for a10 ones |
17:57.02 | technologov | the Cortex A8 and A9 have similar performance, right ? |
17:57.28 | technologov | (I could mistake, yes, but I seen some, and they seem "good enough") |
17:59.18 | technologov | traeak, If I mistake, please show me benchmarks of Cortex A8 vs single-core Atom |
18:00.05 | traeak | technologov: i've only done some simple ones about loads i care about...pure cpu stuff like compilation, etc |
18:00.15 | mnemoc | the Atom plays alone. a cortex-a8 has the help of a gpu and a vpu |
18:00.47 | technologov | mnemoc, you can do computations on ARM, compiled in C/C++, which will show only CPU performance |
18:01.13 | mnemoc | technologov: sure, but the results will be lower than the perceived performance |
18:01.51 | technologov | Well, first we compare the CPUs, then the systems overall. |
18:01.58 | mnemoc | a "desktop experience" is usually combined by the usage of the three. |
18:02.13 | technologov | yes |
18:02.40 | technologov | ... it is very difficult to get benchmarks |
18:02.43 | traeak | kde will probably kill you...running something more reasonable is fine |
18:02.45 | mnemoc | that imo gives an important advantage to arm over atoms |
18:02.54 | technologov | but I can get from geekbench |
18:02.59 | traeak | one problem is: how to get good repeatable benchmarks |
18:03.17 | technologov | GEEKBENCH is repeatble |
18:03.42 | technologov | For some reason I can't get SPEC CPU benchmarks for ARM... anyone knows why ? |
18:04.48 | technologov | benchmarks of ARM A8 @ 600 MHz vs Atom @ 1.6 GHz: |
18:04.49 | technologov | http://browser.primatelabs.com/geekbench2/694040 |
18:04.53 | technologov | http://browser.primatelabs.com/geekbench2/691945 |
18:05.28 | technologov | The hypothetical desktop ARM A8 @ 1.2 GHz would be about 2x faster, so pretty close to Atom. |
18:06.23 | mnemoc | iirc steev got benchmarks of a quad imx6 at full perfomance similar to an i5 |
18:08.13 | traeak | be nicer to see more systems performance: compilation times (exercises more of the system) |
18:08.13 | traeak | etc |
18:08.33 | traeak | but then there are some things the SOC may be slightly better at due to better integration |
18:08.58 | technologov | mnemoc, come on. It can't come close to an i5. Really. |
18:09.15 | technologov | on geekbench an ARM A9 scores 400 points. |
18:09.21 | technologov | Guess how much i7 does ? |
18:09.26 | technologov | 12,000 ! yay. |
18:09.34 | traeak | i'm mostly miffed about the desktop claims for even 384MB ram, mostly because it's a software bloat problem from the amd/intel wars and not a hardware problem |
18:10.00 | technologov | traeak, I agree, but optimizing KDE will take _years_ |
18:10.10 | technologov | and eveything else is just below usable |
18:10.17 | traeak | kde is a POS |
18:10.42 | traeak | and a poster child of that frivolity is all |
18:10.48 | traeak | chase windows to bloat |
18:11.45 | traeak | anyways, it'd be better to try to solve that than to complain |
18:11.51 | traeak | and then i notice i have to get back to coding, sry |
18:15.08 | technologov | If here is any manufacturer listening: *please* *please* *please*, build a quality Neo-ITX board, similar in concept to VIA APC, but better spec, for the ultimate Linux ARM PC. |
18:19.43 | mnemoc | imx6 (having pcie, ddr3 and sata) is perfect for such board |
18:20.15 | technologov | PCIe is not necessary for low-end boards, but ddr3 and SATA are very important |
18:20.31 | technologov | what is the price of it ? |
18:20.45 | technologov | Where can I get prices of VIA VM8950 SoC ? |
18:21.48 | mnemoc | imx6 goes from $11 to $30 in qty 10k according to a mail posted to the ML today |
18:22.10 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
18:22.46 | technologov | $11-to-$30 ... yay ! 3x times diff, for same quantity ? |
18:23.00 | mnemoc | from 1 to 4 cores |
18:23.09 | technologov | ahh ! makes sense ! |
18:24.04 | technologov | imx6 has no SATA, but has GigE. |
18:24.37 | mnemoc | that's what I thought, but it seems the datasheet says it has sata... |
18:25.18 | technologov | What about "cedarx" VPU ? Who develops it ? |
18:25.55 | mnemoc | allwinner. it's their most important technology |
18:26.36 | technologov | he he :) I'm intrigued. First I thought that SoC manufacturers only combine logic from other vendors on their chip |
18:26.36 | hno | technologov, you asked the same thing 30 mins ago. |
18:26.46 | technologov | ..so they _actually_ develop some logic ? |
18:27.03 | technologov | (ARM logic is not theirs... but licensed from UK, both CPU+GPU) |
18:27.35 | technologov | hno, nope... 30 min ago I asked the name of the VPU. Now I ask a bit more about the VPU. |
18:28.10 | technologov | Does ARM licenses also include some VPU ? |
18:29.09 | hno | technologov, you license the pieces you need individually from ARM. But I do not think they have a VPU. But there is several providers of VPU blocks in the ARM ecosystem. |
18:31.48 | technologov | ARM eco-system just moves the lego of various components decisions from the end-users (as in x86 PC) to the manufacturers... they build the lego now. |
18:32.05 | technologov | But it's good. |
18:33.25 | technologov | (good for low-end sector anyway) |
18:33.57 | hno | It's a different ecosystem than in X86 indeed,. highly centered around integrated SoC designs. |
18:35.25 | technologov | ARM arch is limited to 4GB of RAM ? (excluding the ARM v8) |
18:35.36 | technologov | ARM v7 |
18:36.08 | hno | but it's growing "up". We now see PCIe beeing very common on high-end ARM SoCs which opens for more PC like solutions. |
18:37.03 | technologov | but no actually PCIe x16 slots for GPUs I guess... just for on-board internal I/O, right ? |
18:37.35 | hno | Latest ARMv7 can address huge amounts of memory with PAE type extensions. But you don't see that in SoCs targeting consumer segment. Those are generally limited to 2 or maybe 4gb. |
18:38.06 | hno | Even X86 is moving away from external GPU. |
18:38.24 | hno | there is no reason for ARM to move in that direction. |
18:38.44 | technologov | the x86 is divided... so everyone does what he likes ^^ |
18:39.15 | technologov | Biggest problem of Intel: closed-source components. Impossible to re-use Atom cores in own SoCs |
18:39.24 | hno | Not really, in X86 there is only two meaningful CPU manufacturers which then everyone builds their systems around. |
18:39.56 | hno | In ARM there is only one ARM but very many different kinds of SoCs for various purposes with various performance balances. |
18:40.12 | technologov | yes, but no 3rd party can use Intel or AMD cores... which prevents high level of integration |
18:40.41 | technologov | if some company develops new ideas, it will be separate chip, which drives up prices |
18:40.50 | hno | exactly, which limits creativity. |
18:40.53 | technologov | in ARM, it is all integrated |
18:41.00 | technologov | which is good ^^ |
18:42.55 | hno | but it also means there is many more VPU/GPU/xxx providers which all consider their techology the most important and must be kept uber-secret.. so driver side is very sad story. |
18:43.42 | technologov | GPU is less of a problem (theoretically), because official ARM sells Mali-400 which is kind of de-facto standard. |
18:43.43 | hno | but it honestly isn't very good on the X86 side either. |
18:44.18 | hno | ARM MALI is just one GPU provider. They compete with the others on a relatively level play field. |
18:44.20 | technologov | on x86 only 3 players: Intel / AMD/ NVIDIA. The rest dies out (Cirrus Logic, 3dfx, VIA, SIS...) |
18:44.58 | hno | and SoC designers are not very loyal... often jumping from one to another. |
18:45.03 | technologov | Also what is a GPU for ARM ? It seems only the 3D part. |
18:45.38 | technologov | Mali is only 3D part, lacking 2D acceleration, video decoding, and such... |
18:45.49 | Quarx | mnemoc: kernel works fine:) all hw parts works |
18:45.57 | technologov | lacking DAC, lacking 2D output |
18:46.22 | mnemoc | Quarx: you mean 3.0-v2? what was that i2c busy bus thing? |
18:46.52 | mnemoc | technologov: in the arm world each of those is chosen separatedly |
18:46.57 | lkcl | technologov: there's also the RDC IAV100xx series of x86 SoCs (see norhtec's gecko "edubook"). it's not much, but it works |
18:47.27 | Quarx | mnemoc: yea 3.0-v2, incorrect order of loading modules |
18:47.50 | technologov | Google can't find "RDC IAV100xx" |
18:47.59 | lkcl | 1sec... |
18:48.18 | lkcl | it's an x86 CPU, one of which has a PCIe bus |
18:48.51 | lkcl | http://www.rdc.com.tw/en/product/iad.php |
18:49.05 | lkcl | no GPU / VPU though (oops...) |
18:50.53 | hno | MALI 2xx/4xx is only 3D. MALI 6xx moves into 2D, VPU and OpenCL computing areas as well. |
18:51.22 | hno | it's an entirely new GPU core, marketed under the same brand name. |
18:52.11 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
18:52.45 | lkcl | that would be superb to have VPU and cluster-computing [reverse-engineered or a proper software libre driver] in future ARM SoCs. |
18:53.12 | technologov | yea :) |
18:53.47 | technologov | The x86 can't stay alive for some reason... take VIA for example |
18:53.53 | *** join/#arm-netbook leojrfs (~leojrfs@233.163.108.93.rev.vodafone.pt) |
18:54.15 | technologov | They build $300 VIA EPIA x86 platform vs.$49 VIA APC ARM-based platform |
18:55.37 | technologov | Why can't VIA build $50 x86-based PC board, but can ARM is beyond me... |
18:56.34 | mnemoc | blame the business schools... |
18:56.37 | toxicpsion | hno: what's the progess on that FEL-usb loader you rere working on? |
18:56.45 | toxicpsion | s/rere/were |
18:56.57 | hno | Mali T658 is 8-core GPU, No hard perfomance numbers other than 10x MALI-400MP which is four times the MAIL-400MP single core you find in A10. |
18:57.09 | toxicpsion | argh. can't type. |
18:57.10 | technologov | Also the SoC is not a true system-on-chip: Flash and RAM chips are off the chip. |
18:57.11 | hno | toxicpsion, very good progress. |
18:57.37 | hno | can boot a stripped down SPL. But need to write another SPL to actually do something useful. |
18:58.31 | technologov | Can flash and RAM be integrated into SoC ? Why no one does it ? |
18:58.39 | toxicpsion | that's great news! |
18:58.55 | technologov | And another question: how much transistors the Allwinner A10 has ? (approx.) |
19:00.13 | toxicpsion | technologov: probably because of production yield issues.... maybe package-on-package or flip-chip, but not usually on same die. |
19:01.46 | toxicpsion | hno: you have loader code posted somewhere? |
19:01.46 | technologov | Also strange thing is that the SoC vendors like old manufacturing tech. 55 nm for A10. 22nm for Ivy Bridge. Yay ! 2.5 levels of difference ! |
19:01.58 | toxicpsion | ^^price :p |
19:02.11 | technologov | companies say that new process is cheaper... |
19:02.25 | toxicpsion | the process yes, the equipment - no |
19:02.25 | technologov | maybe only for complex chips, dunno. |
19:05.42 | toxicpsion | new process = higher yield for same silicon area, but the one-time equipment cost must be higher. |
19:05.56 | technologov | Can we approximately calculate amount of transistors of the A10 ? |
19:06.04 | technologov | at least order-of-magnitude |
19:06.13 | technologov | is it closer to a million or a billion ? |
19:06.31 | technologov | (estimate, not calculate) |
19:07.04 | technologov | (unlikely to be a billion on 55nm for $7... but just curious) |
19:08.31 | mikey_w | <PROTECTED> |
19:09.04 | mikey_w | Getting frustrated. |
19:10.17 | mikey_w | I need some hand holding here. |
19:15.49 | mikey_w | The 1st partition has evb.bin LOST.DIR mele.bin sys_config1.mele_mod.bin uImage |
19:16.20 | mikey_w | I dd(ed) u-boot into the MBR. |
19:16.47 | mikey_w | I hate my mele at the moment. |
19:20.42 | toxicpsion | technologov: see http://www.engineersgarage.com/articles/arm-advanced-risc-machines-processors?page=2 |
19:20.50 | toxicpsion | puts cortex A9 at ~26M trans |
19:21.30 | toxicpsion | closest relatice that i could find numbers on :( |
19:21.38 | toxicpsion | s/relatice/relative/ |
19:29.58 | technologov | toxicpsion, thx ! |
19:32.12 | mikey_w | Help I've fallen down and can't get up. |
19:34.23 | traeak | get rid of the extra instruction sets and cut the transistor coint |
19:35.01 | technologov | come on.. majority of transistors on a real CPU (read: Core i7) are cache, not CPU logic. |
19:39.35 | hno | toxicpsion, yes, on github as announced on the mailinglist. |
19:46.59 | Hexxeh | anyone got the link to that allwinner box? |
19:47.10 | Hexxeh | chinese wholesaler site or something where you could buy them |
19:47.24 | mnemoc | that allwinner box? |
19:47.39 | Hexxeh | the a10 or something? |
19:48.46 | hno | Hexxeh, A10 is a SoC. Are you thinking of the Mele A1000 or MK802 perhaps? |
19:48.49 | technologov | Allwinner A10 CPU is used in Mali A1000 and Rikomagic MK802 devices, which you can order. |
19:48.56 | Hexxeh | a1000, that was it, thanks |
19:49.19 | technologov | sorry. Mele A1000 ^^ |
19:49.21 | hno | A1000 is the box we use as development platform. MK802 is an HDMI stick. |
19:49.22 | mnemoc | there is a games console named a1000 |
19:49.26 | technologov | Mali is the GPU ^^ |
19:49.34 | toxicpsion | hno: was looking at github and wasn't seeing it... mus've been a cached page copy here or something.... thx. |
19:49.56 | hno | toxicpsion, you found it? |
19:50.09 | toxicpsion | yea |
19:50.26 | hno | should import it into sunxi-tools as promised. It's in quite good shape already. |
19:50.48 | mnemoc | yes, pleeeeeease :) |
19:51.05 | toxicpsion | :) |
19:51.45 | toxicpsion | ...any links to the stripped SPL you're using either? |
19:52.17 | hno | toxicpsion, it's the same SPL. Only disabled MMC to cut down size. |
19:52.23 | toxicpsion | ah... k |
19:52.52 | hno | which means it can initialize the DRAM but then there is no boot method available. |
19:52.57 | technologov | Does the A10 has WiFi controller ? (seems not n spec, but shown on the diagram...) |
19:53.03 | hno | technologov, no |
19:53.20 | technologov | sad :( |
19:53.27 | mnemoc | technologov: connected over usb ;-) |
19:53.43 | technologov | Also the network part 10/100 -- it lacks PHY. Why ? |
19:53.52 | hno | technologov, which diagram? |
19:53.56 | technologov | Isn't it good idea to integrate PHY on board ? |
19:54.15 | technologov | A10 Datasheet - v1.21 (2012-04-06).pdf diagram |
19:54.46 | mnemoc | the mele uses an rtl8201cp |
19:55.17 | mikey_w | traeak, you here? |
19:55.37 | *** join/#arm-netbook eFfeM (~frans@g235028.upc-g.chello.nl) |
19:56.21 | toxicpsion | hno: so no NAND either then. ? :( |
19:56.44 | hno | technologov, A10 is the center square box only. The rest are external. It also do not include IR remote, display, or even battery |
19:57.49 | technologov | well... this is a bit strange... I mean ... > 80% of A10 users will likely need WiFi chip, so why no integrate it on-chip ? |
19:58.04 | hno | No NAND, no DRAM, no cellular modem, no G-sensor, no amplifier, no power management. |
19:58.30 | technologov | and no network PHY (for 10/100) |
20:00.57 | hno | you normally have the PHY external with MII interface. Why do you expect to see it integrated? |
20:01.55 | technologov | Why integrated ? Because it seems to me, that the higher the integration, the cheaper the product (until you reach 2 billion or so transistors, ... more than that and yields drop...) |
20:02.51 | technologov | (products more complex than NVIDIA Geforce GTX560 cost a ton of money... at exponential rate, and NVIDIA starts on complain on yields...) |
20:03.25 | technologov | (GTX 560 is 2 billion transistors, and is highest complexity chip, on which NVIDIA does not complain... they do on GTX580/GTX680 yields...) |
20:05.40 | technologov | But those are just my ideas of efficiency (based on observation and 3rd party speculation). Of course I might be exactly wrong... |
20:06.18 | hno | technologov, it's not so much about the transistors in the PHY case. It's simply hard to mix them on the same die. |
20:06.35 | technologov | perhaps same for Audio DAC, right ? |
20:06.49 | hno | Audio DAC is no issue. |
20:07.08 | hno | and is internal to A10. |
20:07.15 | technologov | aha, right. |
20:07.17 | hno | or pretty much any other SoC. |
20:07.47 | hno | audio amplifier is not included in A10. |
20:08.01 | hno | mostly because one size do not fit all. |
20:08.33 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
20:08.43 | hno | the mele box do not have any amplifier at all, only line out. |
20:09.24 | hno | and your intergrated home cinema system would need a quite different amplifier from a tablet. |
20:11.09 | technologov | they have "HP AMP", which seems enough for a head-phones audio out |
20:12.26 | hno | yes.right, forgot about that. And makes sense as it's a common requirement for a wide range of products. |
20:13.50 | technologov | WiFi logic prolly will get integrated into next core... it is very logical... |
20:13.56 | technologov | many ppl need WiFi |
20:14.11 | mnemoc | in the case of head-phones you do have a "a size fits all" |
20:14.47 | mnemoc | in the others there isn't |
20:16.18 | hno | mnemoc, would you mind me using GPLv2 for fel.c? |
20:16.25 | hno | Well, GPLv2+ |
20:16.47 | mnemoc | fine with me |
20:17.09 | mnemoc | I can change the rest to GPLv2+ too if you prefer |
20:18.37 | mnemoc | to avoid having multiple licenses in the same package |
20:18.45 | eFfeM | hi, installed ICS on A2000 and want to enable debugging but the option does not seem to be there under development options i see only usb error detection, not android debugging; any idea how to enable it? |
20:18.52 | hno | Have no opinion on your code. And not likely they need to mix any code. |
20:20.30 | mnemoc | it makes no difference to me |
20:21.27 | *** join/#arm-netbook gimli (~gimli@xbmc/staff/gimli) |
20:23.17 | mnemoc | hno: from my perspective both are equally good and equally useless anyway... |
20:24.10 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
20:24.12 | mnemoc | hno: so you define which to use and I relicense my code to the same |
20:24.25 | hno | then it's better to use GPLv2+, GPLv3+ does not mix well and is more a political standpoint. |
20:25.01 | mnemoc | ok, give me 5m and I'll push the change |
20:26.10 | toxicpsion | hno: what do you get for 'fel ver', and what device you using? |
20:26.46 | hno | toxicpsion, fel version should be the same in all A10 I think. |
20:27.10 | toxicpsion | 'should' and 'are' are very different, thus the question/ |
20:27.19 | toxicpsion | lol. |
20:27.37 | hno | mine is in the USB-protocol.txt |
20:27.41 | toxicpsion | k |
20:28.32 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
20:28.33 | hno | raw form, but the printed form is not much differnt (mainly byteorder) |
20:28.46 | eFfeM | is it not possible to enable adb on a2000 ics ? can't find it |
20:34.28 | mnemoc | hno: done |
20:45.31 | hno | pushed |
20:45.49 | mnemoc | \o/ |
20:49.26 | toxicpsion | :) |
20:49.54 | hno | and gone from Allwinner-Info. |
20:51.18 | hno | Still plenty of unknowns in FEL. For example what is 0x7e00-0x7EFF used for? It's filled with 0xCC by FEL and livesuite changes first word to 0. |
20:51.45 | hno | 7E00 is also in one of the fields in the version response. |
20:52.24 | hno | and livesute asks for version just before each operation to 0x7e00. |
20:52.37 | toxicpsion | mnemoc: sunxi-tools/README -> fel: most devices (tablets) enter fel by holding vol+, plugging usb, and then pressing power 5+ times. |
20:53.02 | hno | toxicpsion, that's a feature of boot1. |
20:53.15 | hno | which we are getting rid of. |
20:53.34 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
20:53.35 | toxicpsion | sucky. have fel switch unpopulated on my pcb... :( |
20:54.30 | hno | silly, that means you need to use sd-card boot as recovery when all else fails. |
20:55.19 | toxicpsion | ah.. well no worries then. carry on.. nothing to see here :P |
20:55.30 | mnemoc | if boot1 can trigger fel I suppose we should be able to do so too.... |
20:55.34 | hno | mele also do not have a FEL switch, but it also do not have an USB OTG port.. |
20:55.55 | hno | mnemoc, quite likely. |
20:56.27 | hno | unless u-boot changes to much so it's hard to return to the bare metal state of FEL. |
20:56.59 | hno | It do not work jumping to FEL from FEL. |
20:57.46 | hno | but maybe just needs some preparation first. |
21:01.19 | Hexxeh | where's the best place to buy an mk802? |
21:04.55 | technologov | Hexxeh, some say aliexpress |
21:05.08 | technologov | anyway - get the MK802+ model with 1GB RAM, if you want one... |
21:05.39 | technologov | standard MK802 has 512 MB RAM |
21:06.24 | A1Sheds | I asked 2 distributors in the USA for pricing on the ST-Ericsson a9500 and 9540 ARM SOC's..... they can't even figure out how to get pricing |
21:07.32 | A1Sheds | same for the imx6.... it seems that it's just too late for US to even get anything done anymore |
21:07.59 | A1Sheds | ask any disti in China for prices and they don't stop calling you with pricing and delivery info |
21:08.11 | mnemoc | soon you won't even be able to get fat |
21:08.29 | A1Sheds | that would be a plus |
21:08.54 | mnemoc | :) |
21:09.18 | A1Sheds | look for the revolution in streets when the porn, light beer and televised sports get interrupted |
21:10.37 | *** join/#arm-netbook technologov (~technolog@bzq-79-181-149-182.red.bezeqint.net) |
21:12.19 | mikey_w | BTW "lubuntu-desktop-12.04-1-miniand.com.img" boots up on the mele a1000. |
21:13.02 | mikey_w | It sees my wifi router but won't connect yet. |
21:13.32 | mikey_w | Ethernet is not seen. |
21:14.24 | mnemoc | funny how they keep "repurposing" the old image from hipboi with that broken evb.bin |
21:15.09 | hno | mikey_w, you need the right script.bin for your hardware. |
21:15.40 | mikey_w | I'll need some help with that. |
21:16.13 | mikey_w | It took me forever to make a usable sd card. |
21:16.44 | hno | it's in nanda on your hardware and need to be copied to the boot partition of your sdcard. |
21:17.58 | mikey_w | And how do I do that? |
21:18.08 | hno | mount, cp, umount |
21:18.38 | mikey_w | Mount the nand? |
21:18.58 | mnemoc | iirc that kernel doesn't have a functional nand driver either |
21:19.28 | mikey_w | Another catch 22 for the mele? |
21:20.41 | mnemoc | there is plenty documentation on how to build your own a10 kernel, even automatized scripts |
21:21.32 | mnemoc | mikey_w: get the stock script.bin from the mele from https://github.com/amery/sunxi-bin-archive/tree/master/mele-a1000/stock-nanda |
21:23.04 | mikey_w | ok |
21:23.10 | mnemoc | mikey_w: but instead of blindly using that lubuntu-desktop-12.04-1-miniand.com.img you could have used cnxsoft's scripts to build your own bootable ubuntu card from open source software |
21:23.37 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
21:24.53 | *** join/#arm-netbook technologov_ (~technolog@bzq-79-181-147-123.red.bezeqint.net) |
21:25.14 | mikey_w | It works for me. i'm happy. |
21:33.16 | technologov_ | What is "sun 4i" ? |
21:33.22 | hno | A10 |
21:33.39 | technologov_ | LOL, why ? Sun looks like SPARC workstation, no ? |
21:33.53 | ManoftheSea | technologov_: it's a branch name for code that supports the A10 processor |
21:34.01 | ManoftheSea | Which is an ARM A8, I believe. |
21:34.11 | technologov_ | Cortex A8 |
21:34.32 | technologov_ | bad name. Very confusing with SPARC. |
21:34.52 | technologov_ | but huh... nvm. We can't fix it. |
21:34.53 | mnemoc | soon no one will remember sparc at all |
21:35.57 | ManoftheSea | A10 is confusing with the version of arm, too |
21:36.04 | technologov_ | true |
21:36.09 | mnemoc | ManoftheSea: chinese marketing |
21:36.14 | technologov_ | he he ^^ |
21:36.18 | ManoftheSea | also, the confusion between allwinner, boxwinner, etc |
21:36.39 | mnemoc | and their cortex-a9 will most probably be called A15 |
21:36.50 | technologov_ | ha ha (@_@) |
21:37.07 | ManoftheSea | mnemoc: that's where the 2160p and 1.5GHz come from. |
21:37.38 | technologov_ | 1.5 GHz is pure bull****. All testers report 1 GHz. |
21:37.52 | specing | 1.5 GHz is CPU + GPU :) |
21:38.01 | toxicpsion | my gemei G2 tablet is clocked 1.2 stock. |
21:38.07 | mnemoc | it CAN be overclocked up to 1.5, but it's stable up to 1.2 |
21:38.13 | ManoftheSea | that's what the code says. |
21:38.29 | ManoftheSea | specing: which is bull**** |
21:38.38 | specing | indeed |
21:38.49 | ManoftheSea | and anyone who repeats it knowing it's bs is a bs'er |
21:39.47 | mnemoc | Turl has a kernel branch where he played with cpufreq to enable >1GHz |
21:40.35 | ManoftheSea | the code received from allwinner has numbers for > 1.0 |
21:40.46 | mnemoc | commented out |
21:40.51 | ManoftheSea | for 1.45 and 1.5V core |
21:40.57 | ManoftheSea | are they? |
21:41.00 | mnemoc | yes |
21:41.02 | ManoftheSea | hmm, I forgot that then |
21:41.15 | mnemoc | it works up to 1.2 |
21:41.22 | mnemoc | GHz |
21:41.30 | ManoftheSea | yeah, that's what I remember |
21:41.31 | technologov_ | MK802 = Z802 ? |
21:41.36 | technologov_ | Z900 has A10 CPU ? |
21:41.49 | technologov_ | http://www.zero-devices.com/products.php |
21:42.52 | technologov_ | mm ... Telechips TCC8925 |
21:43.45 | ManoftheSea | hmm? What's good about that? |
21:44.18 | technologov_ | The Z900 and Cotton Candy have unique feature: male HDMI... just plug-into your TV and go. No cables at all. |
21:44.26 | hno | technologov_, Z802 and MK802 is the same, except that there have been seen 1GByte versions of MK802 but have not heard anything about 1GByte versions of Z802. |
21:44.28 | technologov_ | Feels like USB flash disk :) |
21:44.53 | technologov_ | sadly I have no HDMI at home. |
21:45.10 | hno | neither have I |
21:45.12 | technologov_ | ..and not feeling like buying a new TV. |
21:45.58 | ManoftheSea | no tv, no monitor? |
21:46.14 | specing | Arent there like $10 hdmi to vga on ebay? |
21:46.21 | technologov_ | nope. only DVI monitor and old analog video on TV |
21:46.24 | ManoftheSea | I expect DVI |
21:46.29 | ManoftheSea | HDMI->DVI |
21:46.37 | technologov_ | doesn't work. no Audio. |
21:46.47 | ManoftheSea | ah, for that device, yeah. |
21:46.51 | technologov_ | I need analog audio, due to $500 speakers here. |
21:47.41 | hno | DVI monitor, but no HDCP -> Black screen from most HDMI devices. |
21:47.57 | ManoftheSea | damn DRM |
21:48.09 | ManoftheSea | so, when do we get the TVs with the EOMA slots? |
21:48.22 | technologov_ | black screen even on boot ? yay ! I thought black screen only on DRM content, no ? |
21:48.34 | hno | Yes.. very important to DRM protect a Linux desktop. |
21:48.51 | ManoftheSea | well, secure boot |
21:48.53 | technologov_ | DVI monitors, BTW, do _not_ support HDCP by definition. Only (some?) HDMI do. |
21:49.01 | ManoftheSea | means having to DRM protect linux |
21:49.20 | hno | technologov_, devices with HDCP tend to have it always on. |
21:49.45 | technologov_ | How many devices are HDCP devices ? Mele A1000 ? |
21:49.46 | hno | toggling it on/off based on content is "too dangerous" |
21:50.04 | hno | seems so. |
21:50.05 | specing | HDCP... just another way to make your TV setup fail |
21:50.30 | technologov_ | well, if TV supports HDCP, then it won't fail, right ? |
21:50.33 | hno | I thought it was another way to milk the TV industry on money. |
21:51.19 | hno | technologov_, not if your HDMI cable is in good shape and your TV have not been blacklisted. |
21:52.16 | hno | and there is plenty of DVI monitors with HDCP support. |
21:52.28 | hno | HDCP != HDMI. |
21:52.33 | technologov_ | how ? DVI does not support HDCP, even in theory. |
21:53.14 | technologov_ | i.e. HDCP can work only via HDMI, not over DVI, not over DisplayPort, not over D-sub. (According to what I read anyway) |
21:55.24 | hno | technologov_, then you read wrong. |
21:55.41 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
21:57.03 | hno | http://en.wikipedia.org/wiki/High-bandwidth_Digital_Content_Protection |
21:57.28 | hno | HDMI is only supported in HDCP version 1.1+. 1.0 was DVI only. |
22:05.08 | traeak | mnemoc: correction, the cortex-a9 will be called a16 not a15 :-p |
22:05.40 | ManoftheSea | it's 1 better |
22:08.28 | mnemoc | :D |
22:31.38 | *** join/#arm-netbook bkero (~Ben@osuosl/staff/bkero) |
22:35.06 | steev | what will their a15 chip be/ |
22:35.10 | steev | a25? |
22:41.23 | mnemoc | I doubt they will live that long |
22:47.02 | lundman | ok |
22:47.28 | mnemoc | I mean, they target at a very low cost market |
22:48.27 | mnemoc | it will take very long until that market thinking about running A15... if ever |
22:48.47 | mnemoc | +starts |
22:54.59 | mnemoc | hno: https://github.com/amery/linux-allwinner/blob/wip/linux-3.3-sunxi/plat/drivers/misc/sunxi-plat.c <--- too spagethish? the idea is to turn it into plat-sunxi/devices.c spawning platform-devices and doing the gpio initialization |
22:55.57 | mnemoc | spagettish* .... I should go to sleep :< |
22:56.43 | *** join/#arm-netbook nibb (~jyuyu@41.202.218.87.dynamic.jazztel.es) |
23:02.58 | hno | mnemoc, personally I am not too fond of script.bin in kernel. Hardware is supposed to be wired up by the time the kernel starts. |
23:03.53 | hno | I think we can find a solution where kernel do not need script.bin, instead having bootloader and/or userspace do the config part. |
23:04.33 | ManoftheSea | what does script.bin do? |
23:04.46 | ManoftheSea | tweak bits on and off? |
23:04.49 | mnemoc | hno: I'm not fond of it either. it's just a cleanup step to remove related stuff from the drivers, and prepare the ground for devicetree support |
23:05.13 | mnemoc | ManoftheSea: describes the board and configures the multiplexed pins of the chip |
23:05.51 | ManoftheSea | that's the chip on the EOMA card, with the storage/bootloader, right? |
23:05.57 | ManoftheSea | Does that side need devicetree? |
23:06.03 | hno | and a bunch of general parameters used by userspace even.. |
23:06.11 | ManoftheSea | I thought it applied to the other side of EOMA |
23:06.53 | hno | ManoftheSea, the A10 starts with basically every pin as GPIO. Not even serial port is enabled by default. |
23:06.57 | ManoftheSea | the EEPROM and such. |
23:07.01 | ManoftheSea | oh. |