01:41.07 | *** join/#arm-netbook ZaEarl__ (~malmrose@216.70.182.153) |
02:14.20 | *** join/#arm-netbook ZaEarl (~malmrose@216.70.182.153) |
03:10.31 | *** join/#arm-netbook CaCtus491 (~Kent@123-243-197-152.static.tpgi.com.au) |
03:12.06 | *** join/#arm-netbook QingPei (~qingpei@180.99.119.165) |
03:30.39 | *** join/#arm-netbook robertjw (~penguinis@tn-76-7-163-57.sta.embarqhsd.net) |
03:32.52 | robertjw | I just put Tom Cubie's Ubuntu image on an a1000 and am finding that this is useable, so now I'm starting to tweek and improve. One issue is with screen size. |
03:33.28 | robertjw | My tv slightly overscans and as a result the top/bottom and sides are slightly clipped. |
03:34.06 | robertjw | I thought xrandr could help, but it seems ineffective with resizing on this device. |
03:34.33 | robertjw | Anyone have any suggestions on techniques to look at for resizing? |
03:36.17 | DonkeyHotei | overscan is a general ubuntu problem. we discussed it at UDS |
03:36.37 | DonkeyHotei | maybe someone in #ubuntu-tv might have an idea |
03:39.58 | lundman | got to do something with my mele too |
03:41.29 | mikey_w | My TV allowed me to kill oer scan. |
03:41.53 | mikey_w | over scan |
03:42.06 | mikey_w | It had an adjustment. |
03:42.55 | mikey_w | Cleared up that problem. |
03:43.20 | mikey_w | But the ubuntu image is still unstable. |
03:44.16 | mikey_w | I added a new user and sound now works fine. |
03:45.58 | robertjw | Unfortunately my tv has no size adjustment, so I'm stuck. The mele android software comes with a nice size adjustment feature that worked perfectly for me. But a full gnu/linux distribution is the holy grail. |
03:46.21 | mikey_w | Sometimes I have to do both "depmod -a" and "ifup wlan0" to get wifi working. |
03:46.21 | lundman | i want to steal that video setting panel to use in ICS |
03:46.44 | lundman | but there was a code ioctl example on how to change the video |
03:48.17 | robertjw | The other thing I would like to get going is support for 1080i or 1080p. Is someone currently working on that? |
03:49.48 | lundman | changing the ioctl didnt work? |
03:50.40 | robertjw | I can try it. Do you have the command line in your head? |
03:50.47 | lundman | http://www.cnx-software.com/2012/04/28/how-to-create-your-own-debian-ubuntu-image-for-mele-a1000-allwinner-a10-based-stb/ |
03:53.39 | robertjw | hmmm, that c snippet is changing mode from hdmi to vga, but maybe it could be adapted also for this purpose. Now to find the doc on the ioctl codes for this device... |
03:54.05 | DonkeyHotei | do those exist? |
03:54.49 | robertjw | that would be in the A10/gpu doc I suppose. |
03:55.39 | DonkeyHotei | does that exist? |
03:59.36 | robertjw | I'm sure it exists, but whether it is available or not is the question. I'm not finding it. |
04:01.03 | robertjw | The other source of documentation would be the sources for the driver that supports the /dev/disp device. I don't have a source tree set up yet so I don't know if that is gpl or not. |
04:01.04 | DonkeyHotei | most useful people in this channel will start waking up in a couple of hours |
04:01.18 | DonkeyHotei | best to wait |
04:01.46 | robertjw | unfortunately in my part of the world its past time for sleep |
04:03.15 | mikey_w | Here too. |
04:03.25 | mikey_w | Virginia. |
04:03.29 | DonkeyHotei | for me, it's only 9pm |
04:03.42 | robertjw | thanks to everyone for their suggestions. I'll scan the channel in eight hours or so to see if anyone else leaves some ideas in the meantime. (TN) |
04:03.46 | mikey_w | La la lander. |
04:04.01 | DonkeyHotei | la la land is 400 miles away |
04:04.21 | mikey_w | Oh San Franciscian. |
04:05.05 | mikey_w | I lived in Milpitas for 30 years. |
04:05.47 | mikey_w | Silicone Valley. |
04:07.50 | robertjw | sheesh, I just had to read on to find the answer. |
04:08.04 | robertjw | "The value in this code are based on drv_display_sun4i.h in the kernel code." |
04:11.06 | CapnKernel | mikey_w: "silicone valley" (sic) is San Fernando Valley :-) |
04:12.05 | mikey_w | I was just south of SF Bay. |
04:12.18 | mikey_w | Nite. |
04:12.48 | mikey_w | Santa Clara county. |
04:14.18 | robertjw | I just grep'ed the whole android source tree and that file didn't show up. |
04:15.02 | robertjw | there must be a custom mele kernel involved here. |
04:16.11 | DonkeyHotei | milpitas is full of semiconductor factories |
04:21.22 | *** part/#arm-netbook robertjw (~penguinis@tn-76-7-163-57.sta.embarqhsd.net) |
04:22.33 | *** join/#arm-netbook robertjw (~ubuntu@tn-76-7-163-57.sta.embarqhsd.net) |
04:24.01 | robertjw | this is cool, I'm logged in now from my mele a2000 running tom's ubuntu image |
04:45.44 | *** join/#arm-netbook RITRedbeard (~RITReadbe@c-68-37-165-37.hsd1.nj.comcast.net) |
04:46.09 | RITRedbeard | I want an ARM netbook. |
04:46.54 | RITRedbeard | Namely, it should be very low power and have LVDS output. |
05:23.07 | RITRedbeard | I demand an ARM netbook. |
05:23.18 | RITRedbeard | You have twelve hours to meet my request. |
05:41.36 | destinal | lol |
05:42.42 | DonkeyHotei | i have an arm netbook. it has 128MB ram and runs WinCE 5.0 |
05:44.46 | *** join/#arm-netbook QingPei (~qingpei@180.99.38.75) |
05:47.11 | rm | "and runs WinCE 5.0" <- so, be careful for what you wish for :P |
05:47.17 | rm | -for |
05:47.25 | RITRedbeard | :( |
05:47.37 | rm | RITRedbeard, there are tons of ARM netbooks on Aliexpress |
05:47.47 | rm | 7" for $60, 10" for $120 |
05:48.13 | rm | but they run android, have slow CPUs and aren't really that hackable to run something else |
05:48.57 | rm | oh, you could also get a Toshiba AC100 |
05:49.02 | rm | but that's a whole another can of sh1t |
05:49.21 | rm | or Efika MX! |
05:49.28 | rm | but that one is a bit expensive for my tastes |
05:49.52 | rm | also not sure about LVDS output on all of these |
05:50.04 | rm | what it's for? |
05:52.25 | RITRedbeard | The AC100 is crap. |
05:52.28 | RITRedbeard | Overpriced crap. |
05:52.49 | RITRedbeard | Refitting Thinkpad 240 to thin client status. |
05:52.50 | RITRedbeard | :) |
05:55.19 | *** join/#arm-netbook cromo (~cromo@139-255-251-94.net.stream.pl) |
05:59.49 | RITRedbeard | Since it's the perfect size. |
06:01.02 | RITRedbeard | Best Keyboard In Class |
06:04.05 | *** join/#arm-netbook robertjw (~ubuntu@tn-76-7-163-57.sta.embarqhsd.net) |
06:11.11 | *** join/#arm-netbook hipboi (~hipboi@221.4.213.95) |
06:35.20 | steev_ | rm: actually... we're *really* low on stock for the MX smartbook (and smarttop but everyone keeps talking about netbook so yeah) |
06:35.38 | RITRedbeard | Who is *we*? |
06:35.39 | RITRedbeard | >:| |
06:36.27 | rm | Genesi, I suppose |
06:36.51 | rm | RITRedbeard, http://www.genesi-usa.com/products/smartbook |
06:37.19 | RITRedbeard | 1024 x 600 is a deal breaker immediately |
06:37.52 | rm | not going to get anything more in a foreseeable future |
06:37.57 | RITRedbeard | but the more that smartbooks, tops, netbooks are on the market the better |
06:38.01 | rm | and don't call it a netbook then |
06:38.04 | RITRedbeard | I'd rather do 800x600 |
06:38.09 | rm | if what you want has a hueg screen |
06:38.37 | rm | 800x600 is better than 1024x600, how? |
06:39.06 | RITRedbeard | you ever use XP/windows 7 in that messed up resolution? |
06:40.06 | RITRedbeard | it never works right |
06:40.23 | ssvb | RITRedbeard: wow, are you using XP/windows 7 on ARM netbooks? |
06:40.28 | rm | you're not going to run "XP/windows 7 " on an ARM netbook |
06:40.43 | rm | and GNU/Linux systems work fine with the resolution available |
06:40.50 | RITRedbeard | That doesn't change the fact that it's not a good resolution. |
06:41.01 | RITRedbeard | 4:3 aspect ratio is better for 10" platform |
06:41.11 | RITRedbeard | You're going to have to take my word for it |
06:45.12 | RITRedbeard | The whole industry has gone a bit retarded in terms of LCD panels |
06:47.12 | *** join/#arm-netbook rellla (~chatzilla@p4FE5628D.dip0.t-ipconnect.de) |
06:48.39 | steev_ | RITRedbeard: i don't disagree that 1024x600 sucks, that's why i replaced mine with a 1280x720 lcd |
06:52.24 | DonkeyHotei | hipboi: is there any reason that serial over mmc could simply not work in one direction? |
06:52.40 | hipboi | yes |
06:52.56 | hipboi | serial can output to two pin |
06:53.06 | hipboi | but can not input from two pins |
06:53.18 | hipboi | neither will not work |
06:53.43 | rm | there's the Loongson 3 netbook coming out with 1376x768; but that's not ARM, not really low-power, and looks like it's going to be crazy expensive |
06:54.06 | DonkeyHotei | so serial over mmc is output-only regardless? zenitraM said it worked bidirectionally for him |
06:54.50 | RITRedbeard | I should have been an electrical engineer. |
06:55.10 | hipboi | serial over mmc, you must disable the original uart pin |
06:55.58 | DonkeyHotei | i changed the uart from PFx to the mmc pins; do i need to disable PFx somewhere else? |
06:57.38 | hipboi | you need to set pb(the default uart tx rx) as normal function |
06:58.12 | DonkeyHotei | i am assuming you're referring to the sys_config1.fex |
06:58.20 | hipboi | oh. no... |
06:58.22 | DonkeyHotei | however, uboot ignores that |
06:58.27 | hipboi | yes |
06:58.45 | hipboi | uboot is hard coded |
06:58.49 | DonkeyHotei | uboot has the serial pins hard-coded, and it behaves identically |
07:01.13 | DonkeyHotei | what could explain that? |
07:02.48 | hipboi | usually PB22 and PB23 is uesd for uart tx and rx |
07:03.19 | hipboi | PF2 and PF4 is for mmc signal |
07:03.28 | DonkeyHotei | a recent git commit to uboot hardcodes it to the mmc pins for serial |
07:03.38 | hipboi | oh, yes |
07:03.53 | hipboi | which repo are you talking about |
07:04.27 | hipboi | but PF2 and PF4 can also be used as uart tx and rx |
07:05.09 | hipboi | so if you want to serial from mmc |
07:05.25 | hipboi | just set PF2 and PF4 to uart rx and tx |
07:05.54 | hipboi | but at this time if PB22 and PB23 still as uart rx and tx |
07:06.09 | DonkeyHotei | git.hands.com/u-boot |
07:06.20 | hipboi | yes |
07:06.49 | hipboi | it's for tablet to get console on mmc easily |
07:08.30 | *** part/#arm-netbook hipboi (~hipboi@221.4.213.95) |
07:08.39 | *** join/#arm-netbook hipboi (~hipboi@221.4.213.95) |
07:13.27 | DonkeyHotei | i used that uboot and changed script.bin in nanda to use the mmc for serial, but both uboot and ICS ignore keystrokes over serial |
07:18.03 | zenitraM | robertjw: there's a tool here that will let you change to any display/resolution i think |
07:18.04 | zenitraM | https://github.com/allwinner-ics/lichee_buildroot/tree/master/target/skel/test/display |
07:18.39 | zenitraM | haven't tested it, but the constants for display change are the same than on the disp driver so you can change the ioctl code from the other program if it doesn't work |
07:19.14 | zenitraM | i found those on the *android* source, not on the kernel iirc, that's why you didn't find them |
07:21.19 | hipboi | the display resolution can be changed in script.bin |
07:21.38 | hipboi | there are comments in sys_config1.fex |
07:22.05 | DonkeyHotei | hipboi: any o |
07:22.22 | DonkeyHotei | hipboi: any idea why i get no response on serial? |
07:22.28 | hipboi | yes, |
07:22.32 | hipboi | it's a bug |
07:22.37 | DonkeyHotei | in what? |
07:22.43 | hipboi | you need add two more lines in uboot |
07:22.54 | hipboi | uart can not get input from two line |
07:23.04 | hipboi | but it can output to two line |
07:23.16 | DonkeyHotei | can you submit a patch? |
07:23.27 | zenitraM | hipboi: interesting, haven't thought of that |
07:23.41 | hipboi | yes, i was a problem i noticed before |
07:23.54 | hipboi | but i can only commit on my laptop |
07:23.56 | DonkeyHotei | also, why would uboot affect ICS's ability to respond on serial? |
07:24.05 | hipboi | then it was forgot... |
07:24.07 | zenitraM | hipboi: i modified manually uboot to go to sd before you put that patch on the repo and it worked |
07:24.27 | hipboi | yes. |
07:24.32 | hipboi | can you commit it? |
07:25.08 | zenitraM | i have no commit access, but the change in code was the same i think |
07:25.18 | zenitraM | (i no longer have it here...) |
07:25.29 | hipboi | ok |
07:25.48 | zenitraM | although as he says, if android doesn't receive input too.. then it wouldn't be a uboot problem, wouldn't it? |
07:25.59 | hipboi | no.. |
07:26.10 | hipboi | it's the problem of bootloader |
07:26.27 | hipboi | because the kernel and android don't set hardware pins |
07:26.59 | hipboi | DonkeyHotei:http://git.hands.com/?p=u-boot.git;a=blobdiff;f=arch/arm/cpu/armv7/sunxi/board.c;h=6db17db9cdcedec187576efc23b76e45cf8ee97c;hp=725aa7557825c67c9c0e8595a34b101ea4df5293;hb=f501a2cfd78fc65eed26c408159acef7e8f1a9b9;hpb=bacac0506e7b3a31e546050acc2a25c0cb0e8fbe |
07:27.05 | DonkeyHotei | hipboi: the "boot1" ? |
07:27.44 | zenitraM | but it worked both ways for me, at least on my tablet, on ICS just changing script.bin |
07:29.06 | DonkeyHotei | i used the uboot with that commit |
07:29.12 | hipboi | DonkeyHotei: yes |
07:29.12 | DonkeyHotei | no input |
07:29.24 | DonkeyHotei | and no input on ICS |
07:29.29 | hipboi | add two lines like that |
07:29.44 | DonkeyHotei | those lines were already there |
07:30.01 | hipboi | need two more lines |
07:30.05 | DonkeyHotei | ok |
07:30.16 | DonkeyHotei | but ICS has no input either |
07:30.22 | hipboi | to disable the PB22 and PB23 |
07:30.34 | hipboi | yes ICS still use this |
07:30.58 | DonkeyHotei | i thought ICS used the script.bin |
07:31.43 | hipboi | yes, ics uses script.bin |
07:32.14 | hipboi | but i won't disable the PB22 and PB23 |
07:32.28 | DonkeyHotei | changing the script.bin gives me output but no input in ICS |
07:34.27 | hipboi | i find i can not explain it |
07:34.44 | hipboi | can you paste your de-compiles script.bin? |
07:35.27 | DonkeyHotei | i am not at that machine now and you would likely be asleep when i can be there |
07:36.27 | zenitraM | everything worked for me without disabling anything just changing the ports... |
07:36.54 | DonkeyHotei | but there is nothing obvious in script.bin that i noticed to cause this, which is why i'm asking about boot1 |
07:37.27 | DonkeyHotei | zenitraM: in uboot, too? |
07:41.13 | zenitraM | DonkeyHotei: yes |
07:42.03 | DonkeyHotei | zenitraM: well, hipboi seems to say uboot shouldn't be working that way regardless |
07:42.27 | zenitraM | well, it worked for me, which leads me to think that the problem is somewhere else |
07:42.55 | DonkeyHotei | right, which was why i wanted to ask him |
07:43.37 | DonkeyHotei | what device was this on, btw? |
07:44.11 | zenitraM | ainol novo 7 advanced |
07:44.50 | DonkeyHotei | those are on aliexpress |
07:45.11 | DonkeyHotei | for USD$80 |
07:45.35 | zenitraM | yes, it's a bit old model |
07:45.43 | DonkeyHotei | plus ungodly shipping |
07:46.09 | zenitraM | although i only see Advanced II |
07:46.14 | zenitraM | mine is advanced I |
07:49.03 | DonkeyHotei | did you build your own uboot? |
07:49.56 | zenitraM | yes |
07:50.20 | lundman | what is needed for a working mele uboot? |
07:50.55 | DonkeyHotei | maybe the novo is just wired weirdly somehow |
07:51.12 | hipboi | DonkeyHotei: i found that two line |
07:51.22 | hipboi | sunxi_gpio_set_cfgpin(SUNXI_GPB(22), SUNXI_GPIO_INPUT); |
07:51.22 | hipboi | <PROTECTED> |
07:51.22 | hipboi | sunxi_gpio_set_cfgpin(SUNXI_GPB(23), SUNXI_GPIO_INPUT); |
07:52.49 | DonkeyHotei | would adding those to uboot allow ics to use serial? |
07:53.06 | hipboi | i think so |
07:53.43 | hipboi | there are no others i can think will cause that |
07:53.54 | DonkeyHotei | could you commit them with "#if 0" ? |
07:54.52 | hipboi | oh, i have commit it with a UART_PINS_TO_SD |
07:55.00 | hipboi | which is define in the sun4i.h |
07:56.06 | DonkeyHotei | just now? |
07:57.48 | hipboi | no.. |
07:57.51 | hipboi | before |
07:58.10 | DonkeyHotei | in the lichee-dev branch? |
07:58.15 | hipboi | yes |
07:59.05 | DonkeyHotei | i don't see it |
08:00.01 | DonkeyHotei | i don't see sunxi_gpio_set_cfgpin(SUNXI_GPB(22), SUNXI_GPIO_INPUT); |
08:00.08 | hipboi | yes |
08:00.16 | hipboi | i will commit it tonight |
08:00.30 | DonkeyHotei | thank you |
08:01.30 | DonkeyHotei | so it was just some gpio weirdness in zenitraM's tablet that made it work for him? |
08:04.45 | hipboi | i don't think so |
08:05.02 | DonkeyHotei | then why did it work for him? |
08:05.10 | hipboi | i think maybe you just changed one place in the script.bin |
08:05.44 | DonkeyHotei | i changed three places in script.bin, as instructed on the elinux page |
08:06.07 | hipboi | then it should work |
08:06.41 | DonkeyHotei | could uboot affect it? |
08:06.48 | hipboi | maybe it's your gpio weirdness that doesn't work |
08:06.56 | hipboi | uboot can effect it |
08:07.20 | DonkeyHotei | i tried on three different tablet boards |
08:07.21 | *** join/#arm-netbook CapnKernel (~mjd@204.37.233.220.static.exetel.com.au) |
08:07.23 | hipboi | but in the kernel, the uart driver will set the pin as script.bin |
08:08.23 | hipboi | at least one thing is sure |
08:08.37 | DonkeyHotei | and when i used the stock ICS uboot that doesn't use serial at all, ICS still had output but no input |
08:08.50 | hipboi | after i commit tonight, you can input in u-boot console |
08:09.00 | DonkeyHotei | i hope so |
08:11.09 | DonkeyHotei | ZaEarl: if you read scrollback, can you try a git pull tomorrow? it's unlikely i can get there before next week |
08:14.06 | lundman | anyway, I added ext4 and zfs to u-boot if that is interesting |
08:14.57 | DonkeyHotei | zfs? why? |
08:19.48 | lundman | well, imagine you have had a flushing toilet for your whole life, then when you buy a new house it only has an out-house. that's what its like to use something other than zfs |
08:20.49 | lundman | :) |
08:22.57 | *** join/#arm-netbook arokux_h (~roman@HSI-KBW-046-005-103-006.hsi8.kabel-badenwuerttemberg.de) |
08:23.11 | *** join/#arm-netbook drgreenthumb (~fletch@108-196-222-251.lightspeed.sntcca.sbcglobal.net) |
08:23.24 | mnemoc | lundman: if you want to extend uboot use hno's tree |
08:24.02 | lundman | github? |
08:24.15 | mnemoc | lundman: yes |
08:24.23 | lundman | great, so is mine, should be easy to merge |
08:30.52 | *** join/#arm-netbook hipboi (~hipboi@221.4.213.95) |
08:33.12 | mnemoc | just realized allwinnertech.com doesn't have an english version of their site... |
08:36.52 | mnemoc | hipboi: hi, btw, is your boss aware the sourcedrop of 3.0.8 from Qware included arch/arm/mach-sun5i (and several sun5i drivers)? |
08:37.08 | hipboi | i have no idea |
08:37.21 | hipboi | nor do i care |
08:37.25 | mnemoc | :) |
08:37.26 | hno | hi. Can talk more about u-boot in an hour or so. In a meeting. |
08:37.42 | mnemoc | lundman: ---^ |
08:38.15 | hipboi | thanks to Qware |
08:38.24 | hipboi | helped a lot of people |
08:38.44 | mnemoc | yeah |
08:38.55 | hipboi | including the A13 project |
08:39.00 | hipboi | : ) |
08:39.37 | mnemoc | but to mainline the stuff we still need a real release, or an statement telling everything there is GPL including the files without proper header |
08:40.45 | mnemoc | after finding what the hell is wrong with the sata driver I wanted to make a plat-sunxi and try to unify the stuff.... without the need of secrecy there is no point in keep everything duplicated |
08:41.06 | mnemoc | also forwardport the missing bits of sun3i |
08:51.43 | hipboi | yes |
08:51.53 | hipboi | plat-sunxi is needed |
09:00.44 | *** join/#arm-netbook QingPei (~qingpei@49.94.48.235) |
09:09.42 | lundman | heh, me too, i just finished hockey :) |
09:09.57 | *** join/#arm-netbook octet8 (~octet8@61.183.235.142) |
09:44.23 | zenitraM | mmm if we can't get such statement from allwinner directly we could try with qware |
09:45.28 | Turl | qware doesn't probably own the rights to it |
09:45.50 | mnemoc | the copyright owner is allwinner |
09:52.34 | zenitraM | i know, but they might be able to get an statement from allwinner easier than us, being a big client |
09:54.05 | mnemoc | ah, yes. |
09:59.53 | zenitraM | (i suppose having a link to them that says "GPL sources" is not enough) |
10:00.51 | mnemoc | zenitraM: please file a ticket about the mainlining and add a reference to the link there. It's the best we have :) |
10:01.40 | mnemoc | and maybe it is enough :) |
10:05.06 | mnemoc | http://www.x-powers.com/ ... I thought the AXP was from allwinner too... |
10:05.45 | Turl | heh nice |
10:05.48 | mnemoc | hipboi: is x-powers part of the same group or separated |
10:05.58 | Turl | I got someone complaining it charged way too slowly :P |
10:06.24 | hipboi | x-power is part of allwiner |
10:08.17 | mnemoc | Krosspower or cha-power? :) |
10:10.04 | Turl | hipboi: btw, is the android integration stuff from Cedar supposed to be out in the open? |
10:16.07 | hipboi | no |
10:16.18 | hipboi | eks power |
10:16.40 | mnemoc | MODULE_AUTHOR("Donglu Zhang, Krosspower"); <---- fancy |
10:16.54 | lundman | heh |
10:17.02 | lundman | anything with a K has to be Kool |
10:17.37 | mnemoc | they are kooler than allwinner, they have open datasheets! |
10:18.10 | Turl | allwinner is pure win, krosspower is pure kool |
10:19.16 | hipboi | intertesting |
10:19.33 | hipboi | Donglu Zhang has left allwinner |
10:19.35 | lundman | kinteresting. |
10:20.09 | mnemoc | all axp code is MODULE_AUTHORed by him |
10:20.37 | hipboi | yes |
10:20.52 | hipboi | i just haven't seen him for a while |
10:21.07 | hipboi | then someone told me he has left |
10:22.25 | mnemoc | :( |
10:26.36 | zenitraM | issue created |
10:26.37 | zenitraM | http://qware.nl/android4upgrade/index.html |
10:26.41 | zenitraM | https://github.com/amery/linux-allwinner/issues/12 |
10:33.06 | hno | is back |
10:36.00 | hno | damn, hipboi left. Needed to talk a bit to him. Well, another day. |
10:38.30 | hno | mnemoc, AXP209 seems to only have chineese datasheet? |
10:39.18 | mnemoc | yes, critical parts have texts in english |
10:39.24 | mnemoc | but+ |
10:40.23 | hno | Heh.. the datasheeds do say "confidential". in their footer. |
10:40.47 | mnemoc | :) |
10:43.26 | mnemoc | hno: but there is an generic AXP20x user guide in english |
10:43.48 | mnemoc | uh, and says "krosspower" |
10:44.06 | mnemoc | and in chinese :< |
10:46.03 | zenitraM | gotta love china. |
10:46.05 | zenitraM | http://www.dealextreme.com/p/ipega-0-9-lcd-digital-alcohol-tester-for-iphone-ipad-ipod-black-132174 |
10:46.28 | mnemoc | wtf |
10:51.28 | hno | the GPIO seems to be far more than one normally expect of GPIO.. |
10:53.39 | mnemoc | it's far more than one normally expect of a batter/power manager IC |
10:53.52 | hno | Wonder how much it differs from 202. Exact same pinout. |
10:54.00 | hno | and 202 have english datasheet. |
10:55.16 | mnemoc | kernel-wise they are the same |
10:57.51 | mnemoc | hno: the description in http://www.x-powers.com/product/axp202.html vs http://www.x-powers.com/product/axp209.html is identical but 209's doesn't have a features section :< |
11:00.57 | hno | notes to make sure all interesting pins of the power controller gets routed. |
11:43.41 | *** join/#arm-netbook QingPei (~qingpei@180.99.53.237) |
12:27.40 | hno | mnemoc, nice, updated A10 datasheets. |
12:28.27 | mnemoc | yup :) ... very fresh |
12:30.46 | mnemoc | https://docs.google.com/open?id=0B-xbPiTMurpSTTlRXzlaQ2lZTVE if anyone else wants it |
12:31.27 | hno | with confidental all over, but no NDA required. |
12:32.55 | mnemoc | I'm just passing it to the colleagues :) |
12:33.08 | mnemoc | confidentiality remains :p |
12:34.00 | mnemoc | and if AW requests it can have no troubles deleting _that_ copy |
12:34.15 | mnemoc | s/can/I/ |
12:35.16 | mnemoc | hno: and as you noted yourself the axp datasheets also say confidential all over and they can be freely downloaded. I'm just following the same policy |
13:01.54 | *** join/#arm-netbook hno (~hno@squid/developer/hno) |
13:02.39 | hno | Wtf.. "Excess flood" while being idle? |
13:08.15 | hno | maybe freenode was just 4 hours late in reacting. |
13:16.29 | mnemoc | :) |
13:41.28 | mnemoc | <6>sw_ahci sw_ahci.0: AHCI 0001.0100 32 slots 1 ports 3 Gbps 0x1 impl platform mode |
13:41.31 | mnemoc | <6>ata1: SATA max UDMA/133 mmio [mem 0x01c18000-0x01c18fff] port 0x100 irq 56 |
13:41.34 | mnemoc | <6>sw_ahci sw_ahci.0: flags: ncq sntf pm led clo only pio slum part ccc |
13:41.41 | mnemoc | f* uninitialized variables :< |
13:51.15 | mnemoc | anyone with a v2 tree wanted to test sata with a real disk? |
13:51.23 | mnemoc | wanting* |
13:59.41 | zenitraM | mnemoc: sure, i can try |
14:00.03 | mnemoc | zenitraM: http://dpaste.com/749327/ |
14:17.27 | DonkeyHotei | This A10 datasheet is the original work and copyrighted property of Allwinner Technology (“Allwinner”). |
14:17.27 | DonkeyHotei | Reproduction in whole or in part must obtain the written approval of Allwinner and give clear acknowledgement to |
14:17.27 | DonkeyHotei | the copyright owner. |
14:19.26 | mnemoc | DonkeyHotei: OT, you mentioned UDS before, are you an ubuntu developer? |
14:20.20 | DonkeyHotei | i was at UDS and people pointed at me with regard to the A10 tablet |
14:35.45 | L84Supper | I have one of those A10 netbooks |
14:36.33 | mnemoc | the sunwhatever one? |
14:36.37 | L84Supper | it's sitting with a friend in China, it works but it has keyboard and camera issues with Android |
14:37.08 | L84Supper | http://www.cnx-software.com/2012/04/16/sunlike-umpc-1021-85-usd-allwinner-a10-based-10-2-android-4-0-netbook/ |
14:37.17 | L84Supper | got it a week or two ago |
14:37.33 | L84Supper | they asked us to help fix it but we are busy |
14:37.51 | L84Supper | it will probably be fine with some tweaks under Linux |
14:38.41 | L84Supper | it's not in a very nice case |
14:39.52 | mnemoc | zenitraM: worked? |
14:41.05 | hno | mnemoc, so it was just pure chance that it worked before? |
14:41.28 | mnemoc | hno: they rewrote the script.bin parser |
14:42.44 | mnemoc | hno: it's much cleaner now.... but they seem to have forgotten to test some edge cases :) |
14:43.22 | zenitraM | mnemoc: trying to make the serial adapter work again... |
14:43.50 | mnemoc | zenitraM: uh |
14:43.52 | hno | ah |
14:44.03 | *** join/#arm-netbook QingPei (~qingpei@180.99.103.121) |
14:44.15 | hno | thought it was script.bin parser related from the trace. |
14:45.01 | zenitraM | also, don't touch the mele's case while barefoot. |
14:45.15 | mnemoc | i first assumed the way of using the parser had changed... |
14:45.29 | mnemoc | zenitraM: burns? |
14:45.43 | zenitraM | just a bit |
14:46.17 | traeak | saw the humorous mailing list posting....how these companies try to design and dump products without doing market research, but when you ask them to design something people might want they want to charge you to do it |
14:46.44 | mnemoc | hno: also the shrank the memory mapping from 4M to 3M ... that was the second thing I blamed |
14:50.24 | steev_ | traeak: what thread |
14:50.38 | hno | Hmm.. seems they have redone the A10 development kit significantly. http://www.wits-tech.com/pages/board-en.jsp |
14:50.50 | hno | much less flexible than earlier version. |
14:51.02 | hno | but 110% nicer packaged. |
14:51.25 | mnemoc | but ugly PCB color :p |
14:51.34 | traeak | http://article.gmane.org/gmane.comp.hardware.netbook.arm/2472 |
14:51.41 | traeak | specific posting, sry |
14:53.13 | mnemoc | I hope the first eoma68 laptop has a decent (not blending) case and keyboard |
14:53.23 | steev_ | "if you build it, they will come" is another way of putting it |
14:54.16 | mnemoc | bending* |
14:54.44 | traeak | mnemoc: i'm not sure the a10 has enough horsepower to make a good netbook though |
14:54.56 | steev_ | eh |
14:55.01 | steev_ | depends on what you run |
14:55.18 | mnemoc | the point of eoma68 is to be able to replace the core board and reuse the rest of the device |
14:55.18 | steev_ | if you expect a *netbook* to run like a *laptop*, then no, it won't have the horsepower to be a good laptop |
14:55.47 | mnemoc | but if the case is junk and the display is junk, what there is to reuse... |
14:56.03 | steev_ | in fact, a lot of the "slowness" you see in most of them running ubuntu is due to the murrine gtk engine |
14:56.32 | mnemoc | is very happy with with 11.6" AMD E350 thinkpad |
14:56.39 | traeak | comparing the a10 to the n270 systems i have |
14:57.00 | mnemoc | steev_: I thought the murrine was a *light* engine |
14:57.09 | steev_ | mnemoc: not even close |
14:57.18 | traeak | on my dev boxes i run windowmaker, on netbooks i run xfce (for my wife's sake) |
14:57.20 | steev_ | clearlooks is a light engine |
14:57.44 | traeak | i won't touch debian based distros anyways. i don't like debian atall |
14:57.50 | traeak | never have, it's always turned me off |
14:58.58 | steev_ | i started out with debian back in 2002-ish, but i had to rebuild the kernel because it didn't support my network card, and then i had to rebuild a few other applications, and i was ending up building more and more to have something recent, so i moved to the most logical distro for my uses, which was Gentoo, and here I am |
14:59.00 | hno | mnemoc, memory mapping of what? |
14:59.20 | zenitraM | mnemoc: |
14:59.20 | mnemoc | { SW_VA_IO_BASE, __phys_to_pfn(SW_PA_IO_BASE), (SZ_1M + SZ_2M), MT_DEVICE }, |
14:59.21 | zenitraM | [ 92.670000] scsi 0:0:0:0: Direct-Access ATA FUJITSU MHZ2160B 0081 PQ: 0 ANSI: 5 |
14:59.21 | zenitraM | [ 92.670000] sd 0:0:0:0: [sda] 312581808 512-byte logical blocks: (160 GB/149 GiB) |
14:59.21 | zenitraM | [ 92.680000] sd 0:0:0:0: [sda] Write Protect is off |
14:59.21 | zenitraM | [ 92.680000] sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00 |
14:59.21 | zenitraM | [ 92.680000] sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA |
14:59.22 | zenitraM | [ 92.700000] sda: sda1 sda2 |
14:59.24 | zenitraM | [ 92.700000] sd 0:0:0:0: [sda] Attached SCSI disk |
14:59.31 | mnemoc | zenitraM: \o/ |
14:59.41 | zenitraM | i can has sata root now :D |
15:00.09 | mnemoc | hno: { SW_VA_IO_BASE, __phys_to_pfn(SW_PA_IO_BASE), (SZ_1M + SZ_2M), MT_DEVICE }, |
15:00.17 | traeak | steev_: i got pissed at gentoo a few years back and moved over to arch ... |
15:00.24 | mnemoc | hno: and we used SZ_4M in the original 3.0 tree |
15:00.43 | traeak | steev_: not sayign arch is the best, the maintainers are asses in some ways |
15:00.54 | hno | Ah, the simplified I/O mapping? |
15:00.59 | steev_ | arch is good from what i hear, i've never had any issues with gentoo myself. i say use what you're most comfortable with |
15:01.06 | mnemoc | hno: yes, only 3 lines now |
15:01.17 | steev_ | of course, it's harder to move to a different distro when i'm a gentoo dev |
15:01.51 | traeak | steev_: the whole openssl upgrade that went bad was irritating, and i got tired of chasing use flags (they have way too many of them) |
15:02.07 | mnemoc | zenitraM: can you test once more reverting the change in core.c ? |
15:03.02 | hno | Wonder why they map the BROM in Linux. It's nice as it allows dumping it, but why? |
15:03.31 | hno | Have you seen any drivers makign calls into BROM? |
15:04.06 | zenitraM | building... that shouldn't need to change the modules right? |
15:04.20 | mnemoc | zenitraM: right, only uImage |
15:04.28 | mnemoc | hno: grepping |
15:04.49 | mnemoc | hno: nothing in our tree |
15:05.15 | zenitraM | mnemoc: this happened when running the tool to enable vga |
15:05.16 | zenitraM | http://pastie.org/3925829 |
15:05.22 | hno | Hmm.. they don't seem to map the sram any more? |
15:05.47 | mnemoc | { SW_VA_SRAM_BASE, __phys_to_pfn(SW_PA_SRAM_BASE), (SZ_128K + SZ_64K), MT_MEMORY_ITCM }, |
15:06.02 | mnemoc | SRAM, IO, BROM |
15:06.05 | hno | Right, saw it now. Misread it as the DRAM mapping first.. |
15:06.12 | mnemoc | :) |
15:06.39 | hno | memory fades too quick. Been many hours in this code earluer. |
15:06.54 | mnemoc | zenitraM: reply the ticket about sata and file a new one about lcd :) |
15:07.28 | mnemoc | zenitraM: or it worked before fixing sata? |
15:07.48 | zenitraM | didn't try |
15:08.16 | zenitraM | booting with the core.c reverted.. |
15:08.25 | mnemoc | that fix really shouldn't have side effects |
15:08.27 | zenitraM | okay, forget about vga |
15:08.34 | zenitraM | i forgot to modprobe disp |
15:08.39 | mnemoc | doh |
15:08.53 | mnemoc | modprobe lcd should bring disp too |
15:09.27 | zenitraM | s/disp/lcd/ |
15:09.39 | mnemoc | :) |
15:09.42 | zenitraM | also sata seems to work |
15:09.56 | zenitraM | with the original core.c |
15:09.57 | mnemoc | zenitraM: great, reply the ticket please |
15:10.16 | traeak | which kernel tree is this for? |
15:10.33 | mnemoc | 3.0.31 |
15:10.44 | mnemoc | based on the source drop by qware |
15:11.08 | traeak | they dropped 3.0.8 though right? so there was soe porting? |
15:11.13 | traeak | where is this tree? |
15:11.30 | mnemoc | https://github.com/amery/linux-allwinner |
15:13.43 | mnemoc | zenitraM: so, does the display work for you? |
15:15.26 | zenitraM | yup, seems so |
15:17.06 | zenitraM | also the fb_test.dat tool works |
15:21.31 | mnemoc | amazing, I'll tell the list |
15:24.20 | zenitraM | also mac_addr works |
15:35.29 | *** join/#arm-netbook cromo (~cromo@139-255-251-94.net.stream.pl) |
15:46.06 | *** join/#arm-netbook t0dbld|work (~t0dbld@unaffiliated/t0dbld) |
16:07.05 | traeak | grr this damn kernel source is too big for my sd card |
16:08.17 | mnemoc | mine is 23M include uImage and .ko files... |
16:08.23 | mnemoc | including* |
16:08.35 | traeak | 1GB for trying to pull the git repo |
16:08.49 | traeak | any better way of pulling that? |
16:08.50 | mnemoc | 1.5 :) |
16:08.58 | traeak | well i ran out of room before that :-p |
16:10.20 | mnemoc | traeak: if you have another recent clone linux you can copy that .git in and add the new remote |
16:10.34 | mnemoc | then only the new commits will be fetched |
16:11.03 | *** join/#arm-netbook eebrah (~Ibrahim_K@212.49.88.77) |
16:11.30 | mnemoc | git clone local is also very helpful when you have more than one checkout of the same remote |
16:11.41 | traeak | or i can cross compile it |
16:12.06 | mnemoc | cross compiling the kernel is the normal way.... |
16:12.56 | mnemoc | my test checkout is actually 2.4GB |
16:14.53 | traeak | my sd card has everything needed for native compilation |
16:14.59 | traeak | should scare up a usb drive then |
16:17.54 | t0dbld|work | even if your sd card is set up for ram , what ever is doing the processing is going to have to :-) |
16:18.15 | t0dbld|work | an amd64 isnt natively going to compile arm :-) |
16:18.31 | t0dbld|work | unless your going to compile on your smart phone :-P |
16:18.45 | traeak | of course...setting up the cross compiler is fun :-p |
16:19.00 | t0dbld|work | jsut making sure |
16:19.17 | t0dbld|work | as mnemoc said it is the normal way to compile and kernels :-) |
16:28.08 | traeak | t0dbld|work: well i would try to compile it on the mele itself |
16:30.09 | t0dbld|work | lol whats the specs on that machine |
16:30.11 | traeak | hehe, clone onto a usb drive with ntfs on it |
16:30.25 | traeak | allwinner a10 cortex a8 |
16:30.32 | traeak | should be molasses i'm sure |
16:30.44 | traeak | have to deal with all the missing ram of course |
16:30.49 | zenitraM | that's when having a sata drive plugged to the mele is helpful |
16:30.53 | zenitraM | both for swap and space |
16:31.03 | traeak | don't have one of those available (yet) |
16:31.15 | traeak | can the mele power a 3.5" drive? |
16:31.28 | t0dbld|work | I prefer to have my i7 do my building :-) |
16:31.48 | zenitraM | no it can't |
16:31.50 | traeak | not enough voltage |
16:31.53 | zenitraM | you need external power for that |
16:32.25 | traeak | t0dbld|work: well i do have a dual i7 xeon here, but i would use the dual c2q xeon for this stuff |
16:33.02 | t0dbld|work | that why would you build on mele ? |
16:33.20 | traeak | i haven't set up the cross compiler yet :-p |
16:33.45 | traeak | all my work is highly threaded HPC stuff :-p |
16:33.54 | zenitraM | apt-get install gcc-4.6-arm-linux-gnueabi |
16:34.11 | traeak | a couple more steps with arch |
16:36.38 | t0dbld|work | not many |
16:36.46 | t0dbld|work | like replacing with pacman |
16:39.36 | t0dbld|work | there ya go http://archlinuxarm.org/developers/distcc-cross-compiling |
16:39.45 | t0dbld|work | should take about 5 min :-) |
16:44.58 | traeak | sort of :-p |
16:45.51 | mikey_w | I understand what uimage is but what are evb.bin, mele.bin and sys_config1.mele_mod.bin and how do you build them? |
16:47.49 | t0dbld|work | mikey_w: best way to learn is to DO .... use these and investigate https://github.com/amery/sunxi-tools |
16:52.25 | traeak | i think the evb.bin is the one that's really used |
16:56.47 | mnemoc | traeak: check your uboot env |
16:57.04 | traeak | mnemoc: thx |
16:57.55 | mnemoc | printenv and see what is been loaded at 0x43000000 |
16:59.18 | mnemoc | should read the backlog before answering |
17:00.24 | traeak | wiki nazis would be helpful :-p |
17:00.47 | mnemoc | ^^ |
17:18.26 | mikey_w | t0dbld, I compiled the tools and ran bin2fex on the .bin files and seem to have gotten a bunch of device parameters? |
17:18.51 | t0dbld|work | yep |
17:19.21 | mikey_w | Am I correct in thinking that uboot uses these to configure the hardware? |
17:20.43 | mnemoc | mikey_w: the kernel does, not uboot |
17:21.28 | mikey_w | ok they are device trees passed to the kernel. |
17:21.45 | mnemoc | yes |
17:22.18 | mikey_w | Which make the kernel useful for many types of A10 implementations. |
17:22.27 | mikey_w | s/make/makes |
17:24.05 | mikey_w | Maybe even different SOCs. |
17:24.07 | mnemoc | yes. if .config-ured correctly the same sun4i kernel can run any a10-based device |
17:24.17 | mikey_w | Very cool |
17:24.19 | mnemoc | script.bin is allwinner-specific |
17:24.36 | mikey_w | ok |
17:24.37 | mnemoc | but using all their recent socs |
17:24.42 | mnemoc | used* |
17:25.42 | mikey_w | script.bin? |
17:26.26 | mnemoc | this binary data structure similar to device tree |
17:27.03 | mikey_w | I only have evb.bin, mele.bin and sys_config1.mele_mod.bin. |
17:27.06 | mnemoc | script.bin is the common name, but doesn't need to be called like that as long as your bootloader knows what file to load |
17:27.16 | mikey_w | ok |
17:27.30 | mnemoc | check your uboot env |
17:27.34 | mnemoc | <PROTECTED> |
17:27.56 | traeak | is there a canned "config" to go with that linux-allwinner kernel ? |
17:28.18 | mikey_w | I will have to actually open the mele and attach the serial dongle to do that. |
17:28.37 | mikey_w | I have been running in the dark, so to speak. |
17:28.40 | mnemoc | traeak: I use sun4i_defconfig + tweaking |
17:29.14 | mnemoc | mikey_w: you can also ask in the ML which .bin is the one used in that image |
17:29.34 | mikey_w | ok |
17:29.40 | traeak | anything special about the "tweaking" ? |
17:29.54 | mnemoc | traeak: my current "tweaking" http://dpaste.com/749428/ |
17:30.17 | mnemoc | traeak: but that's actually for debugging broken drivers, not for user use |
17:31.02 | traeak | understood |
17:31.39 | mikey_w | So that is the kernel make configuration? |
17:32.07 | mnemoc | traeak: newer http://dpaste.com/749429/ |
17:32.38 | mikey_w | or the diff for it. |
17:32.47 | mnemoc | a diff against the default |
17:34.31 | mikey_w | yes, understand. |
17:34.59 | mikey_w | Overwhelmed at the moment. |
17:37.23 | hno | regarding cross compile. Fedora ARM is all native compiled on ARM boxes. |
17:37.37 | traeak | i believe arch may do that as well |
17:37.47 | mnemoc | some use qemu |
17:38.37 | mnemoc | many packages are hard or impossible to cross compile :( |
17:38.59 | traeak | x86 dominated world sadly |
17:39.13 | traeak | although i'll say arm isnt' much better :-p |
17:40.18 | hno | x86 or arm do not matter. crosscompiling do not work at that scale. |
17:43.57 | *** join/#arm-netbook QingPei1 (~qingpei@180.99.27.177) |
17:44.28 | zenitraM | tries to add Mali DRM to the kernel |
17:46.37 | traeak | hno: not effectively...if the system were built cross compiler friendly from the beginning... |
17:47.05 | traeak | as long as the stupid configuration utilities don't directly bang on the hardware |
17:47.18 | traeak | (blas/lapack grrrrrrrrrr) |
17:48.32 | mnemoc | zenitraM: problem is to find a version compatible with the android and linux closed source libraries we have... |
17:49.00 | mnemoc | zenitraM: a ticket would be nice :) |
17:49.13 | t0dbld|work | hno: mnemoc i wish we could get that ddk |
17:52.21 | zenitraM | mnemoc: i'll do |
17:52.47 | zenitraM | but at least for the drm kernel module we have all the different versions to try |
17:52.50 | zenitraM | and also for x11 |
17:52.58 | mnemoc | nice |
17:58.48 | hno | t0dbld|work, the mali one? |
17:59.00 | mnemoc | zenitraM: it might be good to see the extracted sources we have in modules/mali with the original releases, to see what allwinner has done to them |
17:59.20 | hno | we have all needed for mali softfloat linux x11. |
17:59.32 | t0dbld|work | hno: yeah ddk for mali |
17:59.46 | t0dbld|work | not just the OS half stack on site |
17:59.52 | hno | the ddk is not distributed outside ARM licensors. |
18:00.05 | hno | only in binary form. |
18:00.08 | hno | which we have. |
18:00.53 | traeak | i guess android uses softfloat for everything right? |
18:01.21 | hno | yes. but that's irrelevant. |
18:01.28 | hno | mostly anyway. |
18:01.38 | t0dbld|work | well i know im having hell of a time with it still :-P |
18:02.02 | traeak | makes it hard to verbatim steal drivers is all |
18:02.32 | hno | we can't steal much userspace from android. Different OS stack only using a similar kernel. |
18:02.52 | t0dbld|work | android im not as concerend with |
18:03.06 | t0dbld|work | the linux side and x11 stuff is bigger issue for me |
18:03.32 | t0dbld|work | the android side will only become more of a problem some jelly bean maybe |
18:03.48 | t0dbld|work | some == come |
18:04.57 | t0dbld|work | hmm never noticed Hexxeh in here before ... you get an arm-netbook ? |
18:05.05 | t0dbld|work | how goes the chrome hacking ? |
18:05.11 | Hexxeh | nah, just idling anyway though |
18:05.15 | Hexxeh | i was considering an a10 though |
18:05.20 | Hexxeh | eh, not too good |
18:05.29 | t0dbld|work | :-( |
18:05.36 | Hexxeh | the board overlay works good, but chrome itself is proving a bitch |
18:05.47 | Hexxeh | see my thread here: https://groups.google.com/a/chromium.org/d/topic/chromium-dev/7mIFBxMKcCY/discussion |
18:05.58 | t0dbld|work | i was im pressed with the chrome os on the mac air :-) |
18:06.05 | t0dbld|work | would have tried had i not sold my mackbook air |
18:06.26 | t0dbld|work | yeah a10 been fun so far , we grabbed a bunch of tablets |
18:06.47 | t0dbld|work | im having a hell of a time getting display on gentoo though :-( |
18:06.51 | t0dbld|work | other distros np really |
18:07.43 | hno | t0dbld|work, what issues do you have with the Linux X11 mali driver? Apart from the mali GPU part being a softfloar binary blob (4 .so files and headers) |
18:09.16 | t0dbld|work | well i wish i could be more specific , but i dont havea serial console so right now all i can tell you is it is not loading properly |
18:09.18 | t0dbld|work | :-P |
18:09.32 | *** join/#arm-netbook menomc (~amery@shell.opensde.net) |
18:09.51 | t0dbld|work | im in middle of trying an alternative way that may provide a few more answeres in a little bit ... compiling isnt the fastest on the a10 :-P |
18:10.42 | *** join/#arm-netbook lundman (~Accident@194.109.206.190) |
18:10.44 | *** join/#arm-netbook specing (~specing@unaffiliated/specing) |
18:11.13 | traeak | agreed on speed of compilation...however it is nice that i can touch the SOC package and not get burned |
18:12.51 | t0dbld|work | well im doing gentoo ATM no other option |
18:13.46 | t0dbld|work | i didnt do kernel or anything on it |
18:14.02 | t0dbld|work | but subversion and xorg were not quick by any means :-) |
18:14.21 | menomc | try C++ stuff |
18:14.47 | t0dbld|work | haha well of course i am getting some of that as well :-) |
18:15.09 | t0dbld|work | right now doing through chroot though so its even slower |
18:15.18 | t0dbld|work | but i got sick of doing compiles blind |
18:15.21 | traeak | menomc: c++ is okay unless you pull in boost/qt :-p |
18:15.23 | t0dbld|work | and jsut checking logs later |
18:16.11 | traeak | all of fltk2 compiles in ~9s here |
18:16.41 | t0dbld|work | I have had worse ... not sure if Hexxeh ever did but i compiled a few linux kernels on CR-48 .... now that was slow |
18:16.54 | t0dbld|work | and if it failed with error that really ruined your day |
18:17.29 | t0dbld|work | mostly cause it took half of it already |
18:31.43 | traeak | note to self: remember to use screen next time |
18:35.29 | menomc | traeak: screen is dead, long life to tmux |
18:35.57 | Turl | screen <3 |
18:36.47 | menomc | suddenly the dungeon collapses - you die |
18:36.59 | t0dbld|work | lol |
18:37.29 | Turl | I use screen because it does what I need, and I know how to use it :) |
18:39.36 | menomc | me too :p |
18:40.28 | rm | does anyone have an A10 device (Mele Ax000 perhaps), with Debian installed, right now? to run a quick benchmark |
18:44.15 | zenitraM | rm: will Ubuntu armhf work? |
18:44.16 | traeak | menomc: i don't use either enough to justify getting fancy :-p |
18:44.56 | traeak | rm what type of benchmark did you do ? the old unixbench stuff ? |
18:45.18 | rm | mbw |
18:45.21 | rm | zenitraM, yes |
18:45.36 | zenitraM | i have it on ssh, tell me what to do |
18:45.39 | *** join/#arm-netbook ZaEarl (~malmrose@173-11-125-162-SFBA.hfc.comcastbusiness.net) |
18:45.57 | rm | install the mbw package (should be in the repository), tell me which version |
18:46.18 | rm | then run "mbw x", where x is a number that is about 1/3 of your RAM |
18:46.25 | menomc | "I'll use screen because it does what I need, and I know how to use it" ---> work work work -> "Suddenly the Dungeon collapses!! - You die..." ---> F! F! F! I really need to start using tmux --> loop |
18:46.30 | rm | e.g. "mbw 100" |
18:48.15 | Turl | menomc: my use case is running make or an IRC bot and disconnecting :) |
18:49.17 | t0dbld|work | ONE hour later i have subversion !! |
18:49.30 | menomc | what for? |
18:49.37 | menomc | masochism? |
18:49.42 | t0dbld|work | haha |
18:49.44 | t0dbld|work | no |
18:49.51 | t0dbld|work | i want tablet to run e17 gentoo |
18:49.57 | t0dbld|work | needed fopr layman |
18:50.20 | DonkeyHotei | in other words, yes, masochism |
18:51.01 | t0dbld|work | LOL |
18:51.45 | *** join/#arm-netbook lundman (~Accident@194.109.206.190) |
18:51.52 | traeak | rm: http://pastie.org/3926858 |
18:52.29 | traeak | rm version 1.1 sry |
18:52.42 | rm | traeak, thanks, and which device? |
18:53.20 | traeak | rm: mele a1000 |
18:53.41 | traeak | rm: kernel 3.0.8+ on arch linux, compiled gcc-4.7.0 default make |
18:54.10 | menomc | 3.0.8+? :( |
18:54.45 | traeak | menomc: copped from the basic debian install |
18:56.23 | menomc | 3.0.8+? :( |
18:56.27 | traeak | mbw was compiled gcc-4.7.0 |
18:56.41 | hno | Not sure if tmux can do serial terminal program. That's my main use of screen. |
18:56.53 | hno | screen /dev/ttyUSB0 115200 |
18:57.39 | rm | traeak, in case you're interested http://romanrm.ru/en/loongson/benchmarks-mbw-reference |
18:57.41 | traeak | wowsers ntfs-3g is prettry brutal |
18:58.39 | traeak | rm: didn't see mbw 1.2.2 as a download |
18:59.12 | rm | 1.2.2 is in debian testing |
18:59.22 | rm | I don't think it's that relevant |
19:00.44 | zenitraM | rm: http://pastie.org/3926891 on mbw 1.1.1-2 |
19:01.09 | zenitraM | on menomc's kernel tree |
19:01.16 | zenitraM | (menomc?) |
19:01.47 | traeak | zenitraM: trying to compile his tree now...got a failure drivers/media/video/sun4i_csi/csi0/sun4i_drv_csi.c |
19:02.04 | zenitraM | how are you compiling? mine compiles cleanly |
19:02.29 | traeak | could be gcc 4.7 |
19:02.30 | hno | zenitraM, is this softfloat or hardfloat (armel or armhf) |
19:02.39 | zenitraM | hno: armhf |
19:03.26 | traeak | pissed about 'inlining failed in call to always_inline bsp_csi_set_buffer_address’: function body not available' |
19:03.31 | traeak | trying to track that down |
19:05.12 | traeak | for mbw i wish i had that intel 3930k system here :-p |
19:08.04 | Brandon15811 | http://arstechnica.com/gadgets/2012/05/new-74-android-mini-computer-is-slightly-larger-than-a-thumb-drive/ |
19:08.09 | Brandon15811 | new a10 device |
19:08.22 | traeak | Brandon15811: yup, been discussed...it SHOULD be 50USD and not 74usd |
19:09.04 | specing | Is the A10 1GHz, 1.2 GHz or 1.5 GHz? |
19:09.24 | Brandon15811 | "The small computer has an AllWinner A10 single-core 1.5GHz ARM CPU, a Mali 400 GPU, and 512MB of RAM." |
19:11.05 | menomc | specing: 1GHz |
19:11.08 | hno | specing, it's marketed as 1.5GHz capable, but all designs I have seen is 1.2GHz or 1GHz. |
19:11.47 | menomc | they add the Hz of the GPU |
19:12.01 | L84Supper | specing, people have been running them stable at up to 1.2GHZ |
19:12.05 | traeak | rm: dual xeon e5520's http://pastie.org/3926931 |
19:12.39 | L84Supper | somebody said they had 1.3GHz going for a bit, but they needed to figure out how to bump the core voltage up higher |
19:12.50 | traeak | rm: mbw can't seem to handle a very high number on the ram, must be 32bit limited or something |
19:13.12 | rm | 1GB should be enough to test reliably |
19:14.02 | traeak | Mem: 99157804 80772152 18385652 0 64704 75173076 |
19:24.00 | t0dbld|work | hno: any time i tried going above 1.2 i freeze and die / crash |
19:25.14 | L84Supper | t0dbld|work, did you change to core voltage at all? |
19:26.01 | *** join/#arm-netbook rm (rm@fsf/member/rm) |
19:26.48 | traeak | turn off camera stuff i guess |
19:29.30 | *** join/#arm-netbook eebrah (~Ibrahim_K@212.49.88.71) |
19:31.43 | traeak | or better yet, don't mark stuff as inline that can't be inlined |
19:37.06 | traeak | or even better, make a 3rd copy of th exact same inlined function (!) |
19:39.12 | *** join/#arm-netbook nikk (~somebody@nikk.starlink.ru) |
19:39.21 | nikk | hi guys |
19:39.56 | nikk | I have a question about A10 sysconfig.fex and related stuff (fex2bin) |
19:41.44 | CIA-91 | rhombus-tech: Zackary master * rc6fd13809bee /allwinner_a10/orders/Zacknamara.mdwn: |
19:41.49 | hno | nikk, what about it? |
19:41.55 | nikk | the problem is I can't compile it back from ауч with sunxi-tools |
19:42.09 | hno | what happens? |
19:42.51 | nikk | error line in fex is- ir0_rx= port:PB04<2><default><default><default> :pmu_irq_id ---irq= 1 |
19:43.46 | nikk | and error is: sysconfig.fex:837: invalid character at 1. |
19:44.08 | hno | where is the fex from? |
19:44.15 | nikk | ainol A7 |
19:44.32 | nikk | generated with bin2fex |
19:44.38 | Guest43253 | Hey, have any of you guys seen this? http://bit.ly/J10omS $74 Allwinner A10 "thumb drive sized" PC |
19:45.01 | hno | ok, so you have a bin, converted it to fex and now want to convert it back to bin and the tool barfs at you? |
19:45.13 | nikk | hno: yup ;-) |
19:45.30 | hno | did you edit the fex in any way? |
19:46.05 | nikk | yes, i'm figting with goodix touch params and module loading. but without any mods the output is the same |
19:46.14 | hno | menomc ^ |
19:47.29 | traeak | menomc: although i can't swear to it, it seems like gcc-4.7 is more anal about wanting inline function bodies to be available if a function is declared as inline. inside: drivers/media/video/sun4i_csi/csi0/sun4i_drv_csi.c expects bsp_csi_int_clear_status and bsp_csi_set_buffer_address to be available and they aren't. hack is to copy the functions into the file...however this just propagates more copies. |
19:47.34 | nikk | hno: this is the strange line: it starts with :pmu_irq_id ---irq= 1 |
19:48.08 | hno | traeak, probably right. Later gcc versions is more strict about many things. |
19:48.56 | nikk | hno but as i compared sysconfig.fex with old one, generated with other stuff from 2.3 firmware, it was the same earlier. it's IR receiver params section |
19:49.09 | menomc | nikk: can you upload the script.bin of your device somewhere? |
19:49.52 | menomc | the original one |
19:49.55 | nikk | yup: http://dl.dropbox.com/u/2258618/NOVO7A/script.bin |
19:52.37 | nikk | and old one from 2.3 firmware: http://dl.dropbox.com/u/2258618/NOVO7A/script-2.3.bin |
19:53.31 | menomc | nikk: can you file a ticket on github? I can reproduce, but can't debug it right now |
19:54.04 | menomc | traeak: ticket please :) |
19:54.11 | traeak | menomc: that's fine. to truly hack on this i really really really need to get cross compilation to work |
19:54.20 | traeak | menomc: page for ticketing? |
19:54.24 | nikk | i made diff and found that goodix touch section not the same. I try to make it work with goddix module from 3.0 kernel. OK i'll make ticket 4 it |
19:55.01 | menomc | traeak: the issues tab in github, where the repo lives |
19:56.09 | menomc | nikk: please attach both .bin files, the one from the gingerbread firmware and the one from ics |
19:56.47 | menomc | I'll take a look at both tomorrow morning |
19:58.06 | menomc | nikk: in the mean time just remove that line |
19:58.29 | menomc | nikk: :pmu_irq_id ---irq = 1 is clearly garbage |
19:59.00 | menomc | probably a bug in their compiler, but need to deal with those |
19:59.08 | nikk | menomc: in amery github account, yeah? thanks, i'll try this! |
19:59.26 | menomc | nikk: yes, that's me |
20:02.05 | zenitraM | nikk, mnemoc: my decompiler returns exactly the same line |
20:02.18 | zenitraM | so it sounds like garbage left by the compiler or something |
20:02.23 | nikk | mnemoc: nice 2 meet you here! ;-) https://github.com/amery/sunxi-tools/issues/1 |
20:02.58 | zenitraM | also, the closed-source script compiler compiles the file ok |
20:03.26 | mnemoc | zenitraM: the : in the original source was probably a ';' |
20:03.32 | mnemoc | (comment) |
20:03.46 | zenitraM | probably |
20:03.47 | mnemoc | but their compiler didn't notice |
20:04.34 | mnemoc | but I need to be consistent, if I validate keys when compiling I need to validate them when decompiling too, and generate a usable .fex |
20:05.13 | mnemoc | nikk: thanks |
20:06.47 | nikk | zenitraM: true, I was able compile this fex back before |
20:08.50 | traeak | hmm...'c' and inlining must use different rules than with c++ |
20:09.24 | hno | traeak, why? |
20:09.47 | hno | well ther is some differences in legacy but.. |
20:10.42 | mnemoc | hno: safest is to use `static inline` |
20:10.46 | mnemoc | err |
20:10.52 | mnemoc | traeak: safest is to use `static inline` |
20:12.18 | hno | yes static inline behaves the same in most versions. extern inline is a mess. And inline without static/extern somewhere inbetween but mostly works as expected. |
20:12.51 | traeak | mnemoc: i figured that out :-p |
20:13.24 | traeak | sorry i'm just a c++ coder ... i don't like object hierarchies though |
20:14.00 | hno | extern inline was quite nice in C for a long while, but now crippled. |
20:14.35 | hno | or was that only GCC, don't remrember. |
20:15.00 | hno | hard to say some times when one only uses GCC. |
20:16.06 | traeak | or just provide the one line inline implementation in the header and be done with it |
20:16.57 | traeak | we ended up making ".inl" files an include those in the header |
20:17.03 | traeak | just one way to do crap like this (argh) |
20:17.04 | nikk | yahoo, adding [ctp_para] section fixed goodix_touch.ko loading! |
20:17.44 | mnemoc | \o/ |
20:17.58 | hno | traeak, we use .cci and .ci. (c++ / c) |
20:18.17 | mnemoc | _inl.h :p |
20:18.52 | hno | in our case we have a --disable-inline option, which makes the .cci /.ci files compile separately. |
20:18.54 | mnemoc | nikk: please comment both changes in the ticket |
20:25.46 | traeak | hno: youch, extra stuff |
20:33.01 | *** join/#arm-netbook acfrazier (~acfrazier@opensn0w/developer/acfrazier) |
20:36.06 | nikk | mnemoc: ok. I fixed goodix_touch.ko issue with script.bin, but can't fix hv2605.ko (it's touch buttons module). Both from stock firmware and from 3.0 source loaded fine with the same output, both registered with /dev/input/event3, but stock one works, and compiled from source doesn't, I cant see nothing on "cat /dev/input/event3" from it... |
20:38.00 | mnemoc | nikk: same .bin works in closed source 3.0 but fails in open source 3.0? |
20:39.01 | mnemoc | nikk: in that case it's a ticket for linux-allwinner, please attach the .bin and the .config there :) |
20:41.29 | nikk | mnemoc: tab loaded with modded .bin (with 2 ctp sections), so touchscreen is not an issue anymore. both modules work. But for hv2605.ko, there is no error on loading module, fetch params (the kernel output the same for stock and self-compiled). But stock works (send events via /dev/input/...) and self-compiled doesn't ;-) I don't have any ideas except use stock one ;-) |
20:43.01 | mnemoc | I don't even know what those modules are for, but I can take a look tomorrow if you add enough info in the ticket |
20:43.41 | traeak | yeah, past all the video stuff |
20:43.42 | traeak | :-p |
20:43.47 | *** join/#arm-netbook QingPei (~qingpei@180.99.84.179) |
20:46.43 | nikk | mnemoc: it's Anol's Novo7A touckscreen buttons. As it "just works" with stock module, it's not a big issue, and I don't have any thoughts how to describe an error: "it's loaded, but doesn't work" not a good description ;-) Thanks for info about fex2bin, it solved main problem with touchscreen! |
20:47.42 | *** join/#arm-netbook ibrah (~Ibrahim_K@212.49.88.65) |
20:49.18 | mnemoc | nikk: it's very possible ainol did changes/fixes to the driver we don't have.... but we can trace the problem an fix it in the open tree too |
20:50.54 | *** join/#arm-netbook traeak (~bolsen@75-151-91-165-Colorado.hfc.comcastbusiness.net) |
20:56.47 | *** join/#arm-netbook drgreenthumb (~fletch@drupal.org/user/72475/view) |
21:04.36 | nikk | ok, i'll compare 3.0 source with 2.6.36 ainol's tree and report back. Other 3.0 kernel issue is "incomplete xfer" messages in dmesg, they foget to comment printk in i2c/busses/i2c-sunxi.c:809 |
21:05.24 | mnemoc | nikk: what 3.0 branch are you using? v2 or lichee-3.0.8-sun4i? |
21:06.45 | nikk | 3.0.8 (leaked one). Just downloaded v2 and will switch to it |
21:07.42 | mnemoc | :) |
21:09.03 | mnemoc | v2 is based in the leak, but updated and with fixes... also new fixes will go there |
21:28.38 | ZaEarl | What is SUN4I_GPIO_UGLY? When I build the leaked kernel it asked how to compile it (y/n/m). Should it be added to sun4i_crane_defconfig? |
21:31.26 | mnemoc | ZaEarl: it's just a backdoor, you don't _need_ it |
21:33.32 | DonkeyHotei | i like the name |
21:36.44 | mnemoc | :) |
21:37.56 | traeak | hmmm |
21:38.02 | traeak | lets see if this thing boots :-p |
21:45.28 | *** join/#arm-netbook eebrah (~Ibrahim_K@212.49.88.66) |
21:46.36 | mnemoc | traeak: did it? |
21:49.49 | traeak | mnemoc: i guess i don't know how to make a boot image...time to go research which one to use |
21:52.58 | traeak | mnemoc: so of course now i'm looking around trying to figure out how to boot whatever i created and frankly, which of the images to use :-p |
21:58.42 | traeak | i guess "make uImage" might help |
22:16.12 | mnemoc | for uboot? yes, that's the way |
22:16.35 | mnemoc | I thought you were talking about something android-ish |
22:21.44 | traeak | nah |
22:21.48 | traeak | and 3.0.31+ boots fine |
22:22.17 | traeak | now to fix that bouncing dhcp addy |
22:29.33 | *** join/#arm-netbook arokux_h (~roman@HSI-KBW-046-005-103-006.hsi8.kabel-badenwuerttemberg.de) |
22:29.42 | mnemoc | traeak: ethernet or wifi? |
22:30.23 | traeak | mnemoc: ethernet...the wifi device is loaded but i hvaen't tried to connect yet |
22:30.27 | mnemoc | traeak: if ethernet pass mac_addr=... from the bootloader |
22:31.09 | traeak | mnemoc: something to add to uboot i gather? |
22:31.19 | mnemoc | yes |
22:31.19 | traeak | i'm now interestedin getting 'X11' going |
22:32.39 | mnemoc | modprobe lcd for vga |
22:33.25 | traeak | funny, i wonder if i have vga devices around |
22:33.57 | mnemoc | :) |
22:35.53 | lundman | :) |
22:59.50 | traeak | so lcd creatsa framebuffer? |
23:44.00 | *** join/#arm-netbook QingPei (~qingpei@49.90.178.243) |
23:47.55 | CIA-91 | rhombus-tech: Brendan master * rb08111b8f1ba /allwinner_a10/orders/sstrifeau.mdwn: |