01:29.14 | CIA-91 | rhombus-tech: Dan master * r0438c07e2ee6 /allwinner_a10/orders/danh.mdwn: |
02:11.54 | *** join/#arm-netbook hp__ (~hp_@p5B38B3F1.dip.t-dialin.net) |
03:06.56 | *** join/#arm-netbook QingPei (~qingpei@124.64.123.28) |
08:18.18 | *** join/#arm-netbook ibot (~ibot@rikers.org) |
08:18.18 | *** topic/#arm-netbook is EOMA: Embedded Open Modular Architecture - Don't ask to ask. Just ask! - http://elinux.org/Embedded_Open_Modular_Architecture/EOMA-68 - ML arm-netbook@lists.phcomp.co.uk - Logs http://ibot.rikers.org/%23arm-netbook - http://rhombus-tech.net/ |
11:39.53 | *** join/#arm-netbook j1nx (~IceChat7@163-62-223.ftth.xms.internl.net) |
11:41.23 | j1nx | gents, does anybody of you can succesfully boot the 3.3 kernel in QEMU |
11:41.53 | j1nx | would like to find the right parameters to start debugging the kernel on the server console through QEMU |
11:42.31 | mnemoc | I've only tried on the mele |
11:43.31 | mnemoc | but the 3.3 crashes on the irq initialization. 3.0's works fine, so I assume there is something missing in the crappy porting I did |
11:44.41 | j1nx | I can't get the earlypintk to be redirected to the server console when starting QEMU |
11:45.04 | mnemoc | i'm not sure if the a10 is qemu-compatible at all |
11:45.24 | mnemoc | because of it dependency in script.bin |
11:45.38 | j1nx | ? please explain? |
11:46.31 | mnemoc | there is important work done by the bootloader before linux can start |
11:47.17 | mnemoc | the a10 is very flexible, the meaning of the pins is defined in a external file |
11:47.21 | j1nx | hmmm OK. But I think we can start u-boot in qemu as well |
11:47.28 | mnemoc | and that's very board-specific |
11:48.04 | *** join/#arm-netbook penguin42 (~dg@tu006.demon.co.uk) |
11:48.37 | mnemoc | a "proper" script.bin has to be preloaded into 0x43000000 before the linux kernel can do anything, even earlyprintk |
11:48.42 | j1nx | so, I have to start looking at what the allwinner u-boot is exactly doing. Simulate that in qemu first and start the krenel from within that u-boot |
11:49.17 | mnemoc | yes |
11:49.39 | j1nx | the u-boot is loading the script.bin into 0x43000000 and then start the kernel from the fat partition? |
11:50.06 | mnemoc | by default they do that in boot1 |
11:50.27 | mnemoc | but when booting from mmc there is no boot1, so uboot needs to do so |
11:50.48 | j1nx | I think there was a page on rhombus-tech explaining the whole u-boot process wasn't it? |
11:51.06 | mnemoc | yes |
11:51.21 | j1nx | <off searching>? |
11:51.56 | mnemoc | there is http://rhombus-tech.net//allwinner_a10/a10_boot_process/ and http://rhombus-tech.net//a10_mmc_boot/ |
11:52.08 | j1nx | got it! |
11:52.31 | mnemoc | but I seriously doubt you'll be able to boot a sunxi kernel there |
11:52.36 | mnemoc | it's just too fancy |
11:53.35 | j1nx | Oh in the end, ussually I can get it done ;) The question is, can I figure it out before tomorrow morning as I will then fly out to Abu Dhabi for a month. |
11:53.56 | j1nx | I have doubts about the second one :) |
11:54.25 | mnemoc | your energy would be far more useful fixing the video drivers to compile on 3.0 :p |
11:54.56 | mnemoc | and waiting for runtime tests until you actually have hardware after returning from abu dhabi |
11:55.04 | j1nx | But let's first see if I can inject the u-boot-mmc.bin into the QEMU environment up to the u-boot console |
11:58.47 | j1nx | hmm, indeed. It requires to much time to dig into the u-boot code. |
11:59.32 | j1nx | So Alejandro, please give me something else to play with during my boring hotel evenings? |
12:01.20 | mnemoc | https://github.com/amery/linux-allwinner/tree/allwinner-v3.0-android compiles fine and reaches userspace, with sound and video drivers disabled |
12:02.00 | j1nx | compile fails when audio and video are enabled? |
12:02.07 | mnemoc | yes, API changes |
12:02.25 | j1nx | who is working on Video a nd who on audio? |
12:02.35 | mnemoc | that sounds like a very nice boring-hotel-evenings task ;-) |
12:02.51 | mnemoc | no one |
12:03.42 | j1nx | indeed? I can perfectly compile on my server, reachable from the hotel :) |
12:06.43 | mnemoc | the drivers are made for 2.6.36 API |
12:07.17 | mnemoc | up to 3.0 there were changes in the v4l2 and soc_dai interfaces |
12:08.18 | mnemoc | I can gladly do runtime tests for you.... unfortunatelly too busy to do the necessary research myself |
12:08.32 | j1nx | hmm, crap nothing I have worked with before. That will require a lot of reading ;) |
12:09.02 | mnemoc | perfect for the hotel evenings :p |
12:09.19 | j1nx | my thoughts exactly! |
13:07.53 | MaDMaLKaV | shit, two weeks and I still haven't got a valid bootloader.bin for this bloody tegra tablet. I need to get that done before I can start playing with the mele :/ |
14:11.20 | traeak | hmm |
14:11.46 | traeak | nvidia suck? |
14:18.43 | MaDMaLKaV | actually, the fault is of the chinese factory that promised to release the files and never did it and somewhat locked the bootlaoder |
14:26.20 | mnemoc | chinese factories don't do software, neither know how to lock bootloaders |
14:26.52 | mnemoc | (or care) |
14:28.28 | MaDMaLKaV | unfortunately, your wrong in this case |
14:29.23 | MaDMaLKaV | in fact, instead of giving us the nvflash files, they sent us a software made by them that ofuscates all the nvflash process for recovering, and the recovering only starts after the app have got authorization from their server |
14:30.09 | mnemoc | what device is this? |
14:30.57 | MaDMaLKaV | nothing you perobably know, it is a chinese device an spanish rebrander is selling here, but the chinese factory as I said now is not provinding the files they promised |
14:31.04 | MaDMaLKaV | smartlvt is the name of the chinese maker iirc |
14:31.57 | mnemoc | for the records, /me is also in spain |
14:32.37 | MaDMaLKaV | then perhaps you know the device as nvsbl gemini |
14:33.09 | mnemoc | never heard of it :-/ |
14:33.40 | MaDMaLKaV | well, it is just another chinese tegra rebranded tablet |
14:35.14 | zenitraM | that sounds like too much effort for a chinese manufacturer for a chipset that's not even from them |
14:35.41 | MaDMaLKaV | yeah, it is absolutely stupid by their side |
14:36.55 | mnemoc | unless they need to protect the spyware injected by the chinese gov :p |
14:37.57 | MaDMaLKaV | I stopped trying to understand these chinese factories long ago, they do weird stuff that only gets them to lose international customers |
14:38.28 | mnemoc | that's why I love the a10 :p |
14:39.11 | mnemoc | it doesn't matter what the manufacturer or vendor does, the device is still yours |
14:40.52 | MaDMaLKaV | yeah, they are releasing an a10 tablet soon, also following with deep interested the posibility of running xbmc in mele |
14:43.25 | MaDMaLKaV | in fact hy provided me a free mele to have fun with |
14:43.42 | mnemoc | o_O |
14:43.54 | MaDMaLKaV | but then this situation with the bootlaoder in the tegra tablet appeared and I want to help them to solve that before we start play witht he mele |
14:44.12 | mnemoc | fair enough |
14:44.27 | traeak | the fun...the chinese factories do hardware, hot software. Contract out the software and the contractor does what they please, leaving the factory with the licensing and legal mess when it's all said and done |
14:45.28 | traeak | i guess i have 2 days left before the mele expires |
14:45.49 | MaDMaLKaV | yeah, but this stuff is over my head. using the dumped bootloader partition won't work. using the nvdia sdk bootloader will enter in an infinite bootloop. he software they provided for recovery have a file *I KNOW* it have all the partitions, if I disasemble the software I even see the calls to nvflash, but I can't figure what fucking format that shit have |
14:47.56 | traeak | yuk, no one else to help here i gather? |
14:48.40 | MaDMaLKaV | traeak, to help with what? |
14:49.55 | traeak | MaDMaLKaV: fun with boot loaders |
14:50.33 | MaDMaLKaV | people on the toshiba ac100 irc channel have probed quite helpful, but I still couldn't figure out how to solve it with their tips |
14:50.56 | MaDMaLKaV | there is some guy that unlocked the bootlaoder of the acer a500 but no rereference anywhere about how he did it |
14:51.53 | *** join/#arm-netbook mikey_w (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
14:52.01 | *** join/#arm-netbook zenitraM (~zenitraM@2607:f358:1:fed5:11:0:1018:14) |
14:52.02 | *** join/#arm-netbook Guest25667 (~mike@c-71-63-115-202.hsd1.va.comcast.net) |
14:52.54 | MaDMaLKaV | I need some coffe and I have nothing at home, going to the bar, brb |
14:53.26 | traeak | heh |
14:53.29 | zenitraM | MaDMaLKaV: for when you come back, if the nvflash executable is standalone you could try replacing it with something else that dumps the file |
15:07.19 | MaDMaLKaV | no, it is embedded on the software. nvflash with nvdidia sdk botoloader will work ok, but that bootloader programed on the device won't work ok. also, rewriting the bootloader partition dump from the device will also fail, weird |
15:41.47 | *** join/#arm-netbook hp_ (~hp_@p5B38B3F1.dip.t-dialin.net) |
16:09.00 | MaDMaLKaV | ok, time to try some stuff from zero, I think I can be missing files from different rom versions and messing the stuff hard |
16:47.02 | *** join/#arm-netbook j1nx (~IceChat9@163-62-223.ftth.xms.internl.net) |
17:02.02 | *** join/#arm-netbook j1nx (~j1nx@cl-623.ams-04.nl.sixxs.net) |
17:41.57 | *** join/#arm-netbook eebrah (~Ibrahim_K@212.49.88.34) |
17:58.12 | *** part/#arm-netbook QingPei (~qingpei@124.64.123.28) |
18:05.58 | *** join/#arm-netbook arokux (~arokux@merkur178.inf.uni-konstanz.de) |
18:07.28 | MaDMaLKaV | almost got it, now it boots "ok", it fails to mount the partitions, but it boots both to recovery kernel and normal kernel. why it fails to mount the partitions, I still don't know |
18:08.05 | *** join/#arm-netbook arokux (~arokux@merkur178.inf.uni-konstanz.de) |
20:46.55 | *** join/#arm-netbook hp_ (~hp_@p5B38B3F1.dip.t-dialin.net) |
22:09.59 | specing | D'oh my phone still doesen't work |
22:16.44 | CIA-91 | rhombus-tech: Florin master * r6d59f09057c6 /allwinner_a10/orders/z11design.mdwn: |