hno changed the topic of #linux-sunxi to: Allwinner/sunxi development discussion - Don't ask to ask. Just ask! - See http://linux-sunxi.org | https://github.com/linux-sunxi/ | Logs at http://irclog.whitequark.org/linux-sunxi
bsdfox has joined #linux-sunxi
bsdfox has joined #linux-sunxi
bsdfox_ has quit [Ping timeout: 240 seconds]
tinti has quit [Excess Flood]
tinti has joined #linux-sunxi
Soru has quit [Ping timeout: 246 seconds]
<derethor> hi guys!! i can boot my a13
<derethor> i have a console!!!
<derethor> using a simple busybox at the moment
<derethor> kernel 3.4
<derethor> # uname -a
<derethor> Linux A13Micro 3.4.43 #1 PREEMPT Thu Jun 20 04:35:58 CEST 2013 armv7l GNU/Linux
<derethor> muahaha!! now it is time to play
<derethor> btw, i am writing a blog with my updates
<derethor> feel free to comment, etc
<derethor> the chip is very hot
[7] has quit [Disconnected by services]
TheSeven has joined #linux-sunxi
rz2k has joined #linux-sunxi
tinti_ has quit [Quit: Leaving]
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
bsdfox_ has joined #linux-sunxi
bsdfox has quit [Ping timeout: 248 seconds]
mturquette has quit [Ping timeout: 245 seconds]
rzk has joined #linux-sunxi
rz2k has quit [Ping timeout: 264 seconds]
mturquette has joined #linux-sunxi
<hno> derethor, the chip is normally barely warm.
<hno> 1.4v is normal core power.
BJfreeman has quit [Quit: had a good time]
rzk has quit [Read error: Connection reset by peer]
n01|away is now known as n01
rellla has joined #linux-sunxi
<oliv3r> mornin'
<rellla> morning
hipboi has joined #linux-sunxi
<rellla> hipboi \o/
<rellla> morning, yesterday we talked about new mali blobs and if there is a possibility to get new ones (e.g. r3p2-01rel2) as there is a new mali sdk v2.2 out, too.
<rellla> waht do you think?
Kadosch has joined #linux-sunxi
eebrah|away is now known as eebrah
<mripard_> Turl: what is it?
<oliv3r> hno how do I use i2c on our boards, on the A20 i get 'i2c probe' then nothingness
<oliv3r> mripard_: just to make sure, we don't even have a31-meminfo yet do we :)
<oliv3r> i don't think we do
tzafrir has quit [Read error: Operation timed out]
<hno> oliv3r, i2c probe is not implemented in my driver.
notmart has joined #linux-sunxi
<hno> and neither my driver or mvtwsi driver supports more than one I²C bus.
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
_BJFreeman has joined #linux-sunxi
BJfreeman is now known as Guest42920
_BJFreeman is now known as BJfreeman
Guest42920 has quit [Ping timeout: 252 seconds]
<mripard_> oliv3r: that can be worked out :)
<oliv3r> mripard_: of course, i'm writing it as we speak :)
<oliv3r> hno: do you know address or something so i can dump the eeprom on the oli quickly?
<oliv3r> mripard_: but first hurdle is allready to find the new register settings, quite tricky
_BJFreeman has joined #linux-sunxi
tzafrir has joined #linux-sunxi
BJfreeman is now known as Guest46875
_BJFreeman is now known as BJfreeman
Guest46875 has quit [Ping timeout: 252 seconds]
<oliv3r> register layout is extremly different
vicenteH has joined #linux-sunxi
<hramrach_> hello
<hramrach_> oliv3r: I checked out the cubieboard jellybean repo and replaced the kernel with the linux-sunxi kernel
<oliv3r> oh nice one
<hramrach_> and when I did it would not build with the prebuilt andriod tools
<hramrach_> you can use andriod tools to build andriod and emdebian tools to build kernel
<hramrach_> not sure it boots. The build spits out a livesuit image (presumably) but I have no livesuit
<hramrach_> also I winder where it takes the fex when building
<hramrach_> there are like 3 of them somewhere in the repo
BJfreeman has quit [Quit: had a good time]
<oliv3r> mripard_: got a moment?
<oliv3r> does anybody have a a31 fex file?
<oliv3r> i'm curious how the dram_para field looks like now, if its still the same or not
<Turl> mripard_: what is it what?
* Turl lost context overnight
<oliv3r> Turl: get mripard_ here, need 5 minutes with him
* Turl turns on his mripard_ magnet
Tartarus_ has joined #linux-sunxi
<Turl> oliv3r: is it working? it doesn't look like so :P
rz2k has joined #linux-sunxi
<oliv3r> Turl: work harder!
<oliv3r> who has a31 with android and wants to test a bin for me :)
<hramrach_> for me awflash does not build and awimage crashes :/
<hramrach_> is awimage supposed to be 64bit clean?
<andoma> I remember when hacking A10 for a while ago (6 months?) that there were problems with displaying a full screen alpha blended opengl surface ontop of video output in 1080p60. It caused tearing or what looked like buffer underruns during picture scanout
<andoma> did that somehow get resolved?
<andoma> or does anyone know why that happened
n01 has quit [Ping timeout: 268 seconds]
tinti has quit [Ping timeout: 246 seconds]
<oliv3r> http://dl.linux-sunxi.org/users/oliver/a31-meminfo-static <- that would be the one that needs a test
<oliv3r> andoma: ssvb might know
n01 has joined #linux-sunxi
<andoma> ssvb: ping re above ^
<oliv3r> drachensun: actually if you have a31 with android running could test it too :)
Tartarus_ has quit [Ping timeout: 256 seconds]
n01 has quit [Read error: Connection reset by peer]
<hramrach_> andoma: low memory bandwidth most likely - so a hardware limitation
n01 has joined #linux-sunxi
<hramrach_> you can try 1080p50 or lower res and see if that helps
tinti has joined #linux-sunxi
<andoma> it gets better on p50 and disappears almost completely on p24
<andoma> so yeah, it sounds like a bandwidth issue
<hramrach_> hmmm, awimage overflows stack
<andoma> i was mostly thinking if this was due to some FIFO being setup incorrectly. ie. something that could be fixed in SW
<andoma> or if it's just: live with it
<hramrach_> looks like A10 can decode 1080p - so long as you don't hit a bug in cedarx lib - but not display
<hramrach_> maybe there *are* things to tune but no docs. Still one DDR2 memory for everything is going to be bottleneck
n01 has quit [Read error: Connection reset by peer]
n01 has joined #linux-sunxi
Tartarus_ has joined #linux-sunxi
Tartarus_ has quit [Client Quit]
Kadosch has quit [Remote host closed the connection]
n01 has quit [Ping timeout: 240 seconds]
piyushverma has quit [Read error: Operation timed out]
notmart has quit [Quit: notmart terminated!]
fldc has quit [Ping timeout: 245 seconds]
\\Mr_C\\ has quit []
piyushverma has joined #linux-sunxi
fldc has joined #linux-sunxi
<hramrach_> uh, what endianness is Intel and ARM?
<hramrach_> ok, the code *should* work on Intel
\\Mr_C\\ has joined #linux-sunxi
<hramrach_> how can it overflow the stack
<hramrach_> I don't see any recursion in awimage
piyushverma has quit [Read error: No route to host]
piyushverma has joined #linux-sunxi
n01 has joined #linux-sunxi
Soru has joined #linux-sunxi
<hno> hramrach_, there is many ways to overflow the stack..
<hramrach_> well, that's probably wrong diagnostic from vlagring
<hramrach_> it overflows the image
piyushverma has quit [Ping timeout: 264 seconds]
piyushverma has joined #linux-sunxi
<hno> should be easy to find with gdb.
<hramrach_> it somehow fails to load the symbol info
<hramrach_> I guess it tires to decrypt some files outside of the image
<hno> hramrach_, rebuild it with -O2 -g
<hno> err, -O0 -g
<hno> default Makefile is not good for debugging.
<hramrach_> tried already but did not work the first time :s
<hramrach_> now it finds the file
<hramrach_> num_files = 409721453
<hno> which image are you trying to decode?
<hramrach_> one built from cubieboard jellybean repo
<hno> one you built?
<hramrach_> yes
n01 has quit [Ping timeout: 268 seconds]
<hramrach_> I could unpack a prebuilt image I guess
<hramrach_> but I wanted something reproducible
n01 has joined #linux-sunxi
<hramrach_> yes, it works on prebuilt image
piyushverma has quit [Read error: Operation timed out]
<hramrach_> so the build produces bnogus results
<hno> maybe something changed in laer pack tools confusing awimage.
<hno> laer pack tools..
<hno> where is my t?
<hno> later pack tools
<hramrach_> actually the build produces unpacked image so it cannot be further unpacked
<hno> ?
<hramrach_> I have stufff like boot.img and system.img
<hramrach_> tried unpacking system.img and it dod not work, obviously
n01 has quit [Read error: Connection reset by peer]
<hno> ah, yes. awimage only works on the livesuit image.
n01 has joined #linux-sunxi
vicenteH has quit [Read error: Operation timed out]
<hno> should probably reject obviously non-livesuit images..
<bfree> cubie2, "stuck" for now with just https://github.com/hno/u-boot/tree/wip/a20 and https://github.com/cubieboard2/linux-sunxi/tree/sunxi-3.3-cb2 ? could I cherry-pick the u-boot stuff onto sunxi-current? guess no hope yet for Linux 3.4 or 3.10? obviously I'm thinking about deb packages ;)
n01 has quit [Read error: Connection reset by peer]
piyushverma has joined #linux-sunxi
eebrah is now known as eebrah|away
n01 has joined #linux-sunxi
n01 has quit [Read error: Connection reset by peer]
<hno> bfree, the A20 stuff will go into sunxi-current after a little more verification.
<hno> the wip/a20 branch is kept up to date with sunxi-current meanwhile.
n01 has joined #linux-sunxi
<oliv3r> hno: i think I did the best i could, for 'phase 1' anyway https://github.com/oliv3r/u-boot-sunxi/commit/867ae94fde7c0dfa6c1e22775919a1134779d61c it boots so should be okay :p
<hno> can't comment on kernel support just yet, but 3.4 will come, and also quite hopeful that mainline support for a20 is also coming.
<hno> oliv3r, have you read the changes an extra time?
n01 has quit [Read error: Connection reset by peer]
<bfree> hno: yep, thanks, I noticed it was also syncing up to denx master, hence wondering if I could pick it across to add to my sunxi-current based package pending it getting the support itself (new cubie2 here, just trying to decide what to do with it first and obviously I'd "just" like to update my packages)
<oliv3r> hno: i may have not pushed yet
<oliv3r> hno: going over github now to double check
<hno> oliv3r, that commit is there. Just wondering if you have manually read the changes one extra time?
<oliv3r> reading it now
<oliv3r> but if i made more stupid mistakes; feel free to rub them in :)
n01 has joined #linux-sunxi
n01 has quit [Client Quit]
Soru has quit [Ping timeout: 246 seconds]
* hramrach_ building andriod SD card
<hno> oliv3r, I think you have mixed up DRAM_DLLCR_RESET and DRAM_DLLCR_RESET.. mctl_enable_dllx makes no sense, leaving the state as !ENABLED, RESET
<oliv3r> let me check
<oliv3r> L139; but i didn't change it
<oliv3r> - clrsetbits_le32(&dram->dllcr[i], 0x1 << 31, 0x1 << 30);
<oliv3r> + clrsetbits_le32(&dram->dllcr[i], DRAM_DLLCR_ENABLE, DRAM_DLLCR_RESET);
<hno> Values are correct. But names do not make sense..
<oliv3r> ohh
<oliv3r> erm, i got the name from:;
<oliv3r> 31 0x80000000 ENABLE 30 0x40000000 nRESET
<oliv3r> of course if nRESET is
<oliv3r> '!RESET' :)
<oliv3r> then yes, i got the name wrong
<oliv3r> so set 1 to NOT reset, 0 to reset?
<hno> I don't know. In such case enable is also nENABLE.
<oliv3r> well
<oliv3r> check that out
<oliv3r> i think that's how it should end up in the end
<hno> what are the defines set to there?
<hno> DLL_DISABLE, DLL_RESET, DLL_ENABLE
<oliv3r> sec
<oliv3r> arch/arm/include/asm/arch-sunxi/dram_sun4i.h:128:#define DLL_DISABLE (1)
<oliv3r> arch/arm/include/asm/arch-sunxi/dram_sun4i.h:129:#define DLL_RESET (1)
<oliv3r> arch/arm/include/asm/arch-sunxi/dram_sun4i.h:129:#define DLL_RESET (1)
<oliv3r> arch/arm/include/asm/arch-sunxi/dram_sun4i.h:127:#define DLL_ENABLE (0)
wigyori has quit [Read error: Operation timed out]
wigyori has joined #linux-sunxi
<hno> Makes sense if the bits are DISABLE and nRESET
soul has joined #linux-sunxi
<oliv3r> hno: the debian paste is from the code I mentioned in my email
<hno> That code is what tom wrote for the SPL. #ifdef:ed out in aw u-boot.
<hramrach_> heh
<oliv3r> could be it exists there aswell
<ssvb> andoma: nobody knows for sure what is the root cause of this shaking screen effect, but it indeed looks like the scanout is getting starved and does not have enough memory bandwidth
<oliv3r> as i said in my email; it may not work at all
<hramrach_> is there an andriod application similar to cu?
<oliv3r> hno: so rename it to NRESET?
<hno> and DISABLE.
<hramrach_> I would like to see serial output without running a pc
<oliv3r> rgr
<hramrach_> and initial image is probably android
<hno> hramrach_, then you need something else with a UART that can read the output..
<oliv3r> now i'll probably have to break my own git with a froced push, since i branched a tree off of my wip/a20 :)
<ssvb> andoma: reducing refresh rate, reducing color depth, reducing resolution, enabling "scaler mode" and overclocking dram helps
<hramrach_> it won't have the serial uart driver anyway
<hno> hramrach_, what?
<hramrach_> USB serial
rz2k has quit [Read error: Connection reset by peer]
<hramrach_> the andriod image
<hramrach_> they have no drives, ever
<hramrach_> so no point getting applications for them :/
<hno> there have been some serial drivers in the Allwinner images I have looked at. And easy to build your own image with more drivers if needed But I don't use Android.
<hramrach_> easy to build ..
<oliv3r> nRESET or RESET? (lowercase in a define looks iffy)
<oliv3r> nRESET or NRESET*
<hramrach_> when it comes to building imges it's easier and more useful to build Linux
<hramrach_> now I have something that is Android and *should* boot on cubieboard unless broken
edeloget has joined #linux-sunxi
<hramrach_> oliv3r: NO_RESET ;-)
hipboi has quit [Quit: Leaving]
<hno> oliv3r, pick one. But signal name in hardware designs is nRESET
<hno> or #RESET depending on notation. But here it does not matter.
<oliv3r> i'll use NRESET to stay with the capitals for defines rule
<oliv3r> _!RESET :D
<hramrach_> does that work as define?
<hramrach_> NRESET is somewhat unintelligible
<hramrach_> does it mean number of resets?
<hramrach_> not that the original nRESET is better
<oliv3r> negative_reset :p
<oliv3r> doubt you can use _!RESET
<oliv3r> stupid!
<hramrach_> it's quite likely that nRESET will be challenged by some nitpicker
<hramrach_> but until it is at least it matches the docs
<oliv3r> not really an issue with u-boot i suppose; but yeah it should be nice
<oliv3r> we have no docs :)
<hramrach_> or schematics or whatever
<hno> hramrach_, these are internal signals within the SoC, in a part for which we do not have any docs what so ever other than some random pieces of code..
<oliv3r> hno: anyway, is it worthy for the ML to submit for review? or do i need to hack more on it
<hramrach_> hmm, I have some drm error
<hramrach_> [ 64.880000] drm: disagrees about version of symbol kobject_uevent_env
<hramrach_> [ 64.890000] drm: Unknown symbol kobject_uevent_env (err -22)
<hramrach_> maybe need a reboot
rellla2 has joined #linux-sunxi
<hramrach_> actually need some more patches to make the modules buildable :s
<oliv3r> hno: one last bother; i don't understand this bit of code (i link sun7i; but sun4i lower is identical (why do we need 2?)) https://github.com/hno/u-boot/blob/wip/a20/arch/arm/cpu/armv7/sunxi/dram.c#L583
<oliv3r> para_zq is always 1 byte, '0x7b' or something is the facto default
<oliv3r> so how can you shift that to the right, and have anything left over?
<oliv3r> what am I overlooking?
<oliv3r> unless it's a really silly way to do reg_val = 0;
notmart has joined #linux-sunxi
paulk-desktop has joined #linux-sunxi
<andoma> ssvb: thanks. at least i know there have been no real breakthroughts related to that issue
<hno> oliv3r, I have seen other values.
vicenteH has joined #linux-sunxi
<ssvb> hmm, NEON seems to be significantly downgraded in A7
<hramrach_> or rather a8 is good at neon
<ssvb> yes, neon in a8 is very good, a9 can be up to 20% worse per cycle on hand optimized code
<ssvb> but a7 seems to be twice narrower and has horribly large latencies even for basic arithmetic instructions
rellla2 has quit [Remote host closed the connection]
* ssvb wants to cry
rellla has quit [Quit: Nettalk6 - www.ntalk.de]
<ssvb> integer instructions also don't seem to dual issue as much as they used to be
piyushverma has quit [Ping timeout: 240 seconds]
<hramrach_> well, you have two cores
<hramrach_> but together they have about the same perf :/
<hno> oliv3r, I would probably replace https://github.com/oliv3r/u-boot-sunxi/commit/867ae94fde7c0dfa6c1e22775919a1134779d61c#L0R120 by bus_width = (bus_width >> 6) & 0x7 +1; n = bus_width / 8 + 1 but that is a code structure change and can be done separately.
<hramrach_> anyone has a tip for fast USB SD reader?
<hramrach_> the a10 mmc tops at ~ 10MB/s but the USB card readers I get with some cards top at ~ 2MB/s
<hramrach_> a built-in card reader in a Dell desktop does ~12MB/s so it's possible to improve
<hramrach_> it's hdparm readonly numbers so not exactly awesome benchmark but still much difference
rellla has joined #linux-sunxi
Tsvetan2 has joined #linux-sunxi
Tsvetan has quit [Ping timeout: 245 seconds]
n01 has joined #linux-sunxi
edeloget has quit [Ping timeout: 250 seconds]
cnf has joined #linux-sunxi
cnf has left #linux-sunxi ["Bye"]
bsdfox_ has quit [Quit: Leaving]
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
<techn_> hey.. anyone know how mnemoc is doing?
<techn_> also.. cubieboardii owners.. what was your pre-installed image?
<techn_> and did usb host work oob?
<bfree> techn_: linaro on flash. usb keyboard+mouse (well ps2 via adapter) working fine ootb
<techn_> linaro here too.. but mine usb kb+mouse is not working :(
techn_ has quit [Ping timeout: 252 seconds]
techn_ has joined #linux-sunxi
<[hawk]> Yay! I did it!
<[hawk]> Turl: ^^^
<[hawk]> it works
tkoskine has quit [Ping timeout: 276 seconds]
<[hawk]> goodix 801_2plus1 works like the goodix_touch_3f on the 907
<techn_> bfree: seems that usb port is working at some level
<techn_> but may have bugs that block my kb working
<techn_> [ 49.951586] logitech-djreceiver 0003:046D:C52B.0006: claimed by neither input, hiddev nor hidraw
<techn_> [ 49.960440] logitech-djreceiver 0003:046D:C52B.0006: logi_dj_probe:hid_hw_start returned error
<techn_> but I'm wondering how Tom&co made linaro boot so fast :D
<techn_> boot in 5 seconds or so :)
<jelly-home> everything builtin, delayed setup of network and ntp?
<techn_> lsmod shows only 6 modules
tkoskine has joined #linux-sunxi
<techn_> so.. no one knows how's mnemoc.. anyone know how's mnemoc's a20 merge? :p
<oliv3r> hno: ok, let me rephrase; all values seem to be in the 0-256 range, but that shift doesn't make no sense whatsoever
<oliv3r> techn_: you finally got it?
<oliv3r> still nothing here :(
<oliv3r> hno: i'll send it off to the ML then; i'll be identifying more bits with a 2nd patch; then we can start looking at restructuring
<oliv3r> i have a few idea's too
<oliv3r> but i think a31 support would e nice to have first
eebrah has joined #linux-sunxi
eebrah is now known as Guest19108
Guest19108 has quit [Client Quit]
<ssvb> techn_: usb keyboard & mouse is working fine here on a cubieboard2 with the linaro image from nand
<hramrach_> ssvb: there is simething wrong with blending of some windows in gnome
<hramrach_> when I select profile preferences in Termainal and set color the color seems ok in color picker but blueish in hte preferences dialog
<hramrach_> eg. green in color picker becomes cyan in preferences
<hramrach_> ok, I don't have teh latest driver so I guess I could try upgradeing, too
<hramrach_> actually, the terminal itself looks odd too but that might be gnime default terminal palette
<techn_> oliv3r: yeah I got it.. I was afraid that it couldn't make to mid summers eve.. but it made :)
<ssvb> hramrach_: also please try the original mali driver
<oliv3r> lucky you!
<techn_> "mid summers eve" aka "juhannus" is one day holiday in here :)
<hramrach_> mid-fry eve
<techn_> celebration for nighles night or somehting
<hramrach_> it's not like anybody would be doing anythein anyway
sanka has joined #linux-sunxi
<hramrach_> in this weather
<techn_> yeah.. here's rainy
<hramrach_> here is 35 deg
<techn_> lucky you ;)
<hramrach_> lucky you :)
<techn_> it would be perfect weather if you have air condition :p
<hramrach_> it is like 3 days a year on average so nobody has air conditioning for that
<hramrach_> and when you have air cinditioning weather is irrelevant overall
<hramrach_> plus it still sucks to go out into that even when you do have air conditioning
<techn_> quite many ppl has air condition here in finland.. during hotest days (30C) it cools and between -20 - +20 it saves electicity
<techn_> well it's not full air conditioner.. bu "air-source heat pump"
<oliv3r> hno: what other values have you seen? All A10 boards in git have zq = 0x7b. A20 has 0x7f as the only one. So again reg_val = ((para->zq) >> 8) & 0xfffff;
<oliv3r> makes absolutly no sense to me
<oliv3r> so I must overlook something, I just don't know what
notmart has quit [Remote host closed the connection]
gzamboni has quit [Read error: Connection reset by peer]
gzamboni has joined #linux-sunxi
<soul> I have a question. I'll be able to install a GNU/Linux distro on my tablet If I don't know how to enter on the bootloader?
<mripard_> Turl: your magnet doesn't work really well I'm afraid :)
eebrah|away is now known as ibrah
tzafrir has quit [Ping timeout: 248 seconds]
<techn_> mripard_: how's mainline a20 progressing?
<mripard_> techn_: hmmmmmmm, not well :)
<mripard_> I haven't started it yet :)
<techn_> ok.. so that's one todo item :)
<techn_> mripard_: are you or someone else planning to start it?
<mripard_> yeah, somewhere around this week end
<techn_> ok.. great.. first impression is that it is simple? :)
<mripard_> I have put all the patches out of my queue and in 3.11, so now, I'll be able to focus on A20, A31 and SDIO
<hramrach_> soul: you can boot from SD card
<mripard_> techn_: it looks like an A10 with 2 cores and a GIC instead of allwinner's interrupt controller
<mripard_> we already have a gic driver in the kernel
<mripard_> so I think I can have some patches working very quickly
<soul> hramrach_: But It will load the distro instead of android?
<mripard_> however, it won't be smp at first
<hramrach_> it wil load whatever is on the sd card
<soul> Ohm... hramrach_, ok thanks
<oliv3r> mripard_: you got a few
<mripard_> oliv3r: a few what?
<oliv3r> single core a20 will be good
<oliv3r> pm
<mripard_> ah
<mripard_> yeah, dual core will come later
<mripard_> I'm pretty sure it will trigger some additionnal problems we didn't catch before
<mripard_> synchronisation, mostly.
notmart has joined #linux-sunxi
notmart has joined #linux-sunxi
BJfreeman has quit [Read error: Connection reset by peer]
<oliv3r> lkcl: you've did awesome work in getting allwinner-boot (boot0/1) to hno, but there's bits missing GPL headers, and it's only 1 branch. Any chance to legally clear the other branches (A31 mosty)?
<oliv3r> if you could run that
<oliv3r> also i'd love to see your fex file :)
_BJFreeman has joined #linux-sunxi
<oliv3r> (most likly since we don't have meminfo tool)
_BJFreeman is now known as BJfreeman
<hramrach_> ssvb: it happens with latest sunxifb and mali too
<hramrach_> laso the window blending is somehow crude
<hramrach_> also
<mripard_> oliv3r: I need to have a working system back first :)
<mripard_> drachensun: I just finished downloading images, thanks!
<hramrach_> when you look at the profiles dialog in terminal you will see an odd fram that is not present on 'normal' hardware
<oliv3r> mripard_: it might not even work :p
<hramrach_> maybe the Gnome GLES backend error or Mali error. hard to tell
<oliv3r> mripard_: but yeah, if you and drachensun could get some fex files posted; it would help a little
<hramrach_> there is single-core a20?
<oliv3r> if you disable the second core; sure :)
<mripard_> oliv3r: rather, if you don't enable it :)
<ssvb> hramrach_: I'm afraid that this might be hard to debug
sanka has quit [Quit: Leaving]
<bfree> I don't see a cubie2 fex up there under a20 yet either? only the olinuxino
<drachensun> mripard: no problem
<hramrach_> ssvb: in absence of other GLES hardware
<drachensun> oliv3r: Can do
<oliv3r> it might blow up things
<bfree> fex2bin of script.bin from cubie2 is http://paste.debian.net/11640/ ... any more needed?
<oliv3r> i don't have a cubie yet :(
<oliv3r> if you can push it to the repo, cool; otherwise i can copy/paste that for you
<ssvb> hramrach_: about the mali related glitches, there is an interesting bug reported against opengles-book-samples - https://code.google.com/p/opengles-book-samples/issues/detail?id=8
<oliv3r> well on the upside, the DRAM para's are still identical
<oliv3r> oh nvm
<bfree> oliv3r: if you were talking to me, go ahead and grab it from there ;)
<oliv3r> was thinking a31 :S
<oliv3r> rgr
<hramrach_> hmm, should get a browser working on this
<ssvb> hramrach_: the code from opengles-book-samples is wrong, but nobody cares to fix it, maybe because it "works" on other hardware
<drachensun> well I don't see an easy way to upload the test file to the wiki
<drachensun> so I justed pasted the fex into this
<oliv3r> paste is good
<oliv3r> if it's ok with you i'll put it in git
<oliv3r> oh wow, i think layout is messed up somehow
<oliv3r> http://linux-sunxi.org/Special:Upload can't you use that?
<oliv3r> otherwise use paste.debian.org or something
<oliv3r> drachensun:
<oliv3r> try:
<oliv3r> cat fex | curl -F 'sprunge=<-' http://sprunge.us
<oliv3r> :
<oliv3r> btw, cubieboardii is officially called cubieboard2 btw i think
<drachensun> I can't upload, only allows image formats
<drachensun> I guess its probably not smart enough to check the content
<drachensun> neat trick
<oliv3r> :)
<drachensun> you can put that in the git, thats fine
<n01> mripard_ oliv3r: http://pastebin.com/TQBKQ4hw
<oliv3r> rgr
<n01> :D it needs _huge_ cleanup now
paulk-desktop has quit [Quit: Ex-Chat]
<oliv3r> nice one n01
<mripard_> n01: cool :)
soul has quit [Ping timeout: 264 seconds]
<oliv3r> how do I get the chinese characters from a text file to show up in my terminal, through screen
<oliv3r> now what :S
<oliv3r> fatal: remote error: You can't push to git://github.com/linux-sunxi/sunxi-boards.git
<oliv3r> use https instead
<oliv3r> bs
<oliv3r> i _know_ i've pushed before
<drachensun> ugh
<drachensun> so I unpacked the image file provided by wits with the EVB
<drachensun> and the script.bin doesn't match the one that comes with the build tools
<drachensun> which is almost surely why I can't get the USB port to work
<oliv3r> default fex files are horrible most of th etime
<drachensun> this was included with the EVB and supposed to be for it specifically
<hramrach_> oliv3r: you need ssh. git is readonly.
edeloget has joined #linux-sunxi
<drachensun> its just unreal how bad they are about this kind of thing
<oliv3r> they are just as lazy
<oliv3r> hramrach_: yeah but i didn't change the url
<oliv3r> i think
<oliv3r> lemme check
<oliv3r> (my git powers are wak)
<drachensun> seriously, its like they give anti-help, its worse that just trying to reverse engineer their work
<oliv3r> lol aye
<oliv3r> good thing fex will be a thing of the past
Soru has joined #linux-sunxi
<oliv3r> DT is 'harder' to get right i think
<oliv3r> fex is way to easy to fuck up
n01 has quit [Ping timeout: 255 seconds]
<drachensun> well it sounds to me like DT is a big XML type set of data passed in by the bootloader
<oliv3r> sorta; yeah
<drachensun> is that right?
<drachensun> if so I honestly can't see how its different from fex
<oliv3r> but its harder to actually write i suppose, so you have to know what youa re doing?
<drachensun> seems like an easy conversion
<oliv3r> not again the fex vs dt thing :p
<oliv3r> read the ML if your into that :)
<drachensun> I'm not arguing
<mripard_> drachensun: it's similar to some aspect
<mripard_> well, at least, the idea is the same
<drachensun> I was a little curious what all the fuss was about, again given how it appears to be almost the same, just a question of the data format it seems to me
<oliv3r> iterseting read at the laest
<mripard_> "let's have a generic kernel and a hardware description file"
<oliv3r> drachensun: does your a31 board boot?
<mripard_> the only thing that fex lacks to be similar to the dt, really is the driver to load
<oliv3r> can you download that and run it?
<oliv3r> curious as to what it prints
<mripard_> oliv3r: we even had a LWN article about us!
<drachensun> ok, maybe a little later, if this new gmac setting in fex fixes my ethernet I can try it
<drachensun> mripard_ So the DT info should cause the driver to load you mean?
<oliv3r> mripard_: seriously? link!
<mripard_> drachensun: yeah
<mripard_> fex only tells that if you have a mmc driver, it should use these parameters
<mripard_> but you have no informations about *which* mmc driver to load
<drachensun> mripard_ that sounds way better, every time I went to a new distro I had to figure out 'their' way to setting the load by default drivers
<drachensun> of course I think at this point I think I know all the typical ways of doing it
<mripard_> well, it doesn't mean that your driver will get loaded
<mripard_> it only describes the hardware
<oliv3r> i want x86 with dt :D
<drachensun> lol
<drachensun> mripard_ so does it still need to detect it or call it like it does with built in modules?
<mripard_> oliv3r: it's called BIOS :)
<oliv3r> yeah my bios is utterly shit
<mripard_> drachensun: actually, it doesn't really relates to what driver is in the kernel
<mripard_> it just describes the hardware
<oliv3r> once i have some time (when not doing sunxi) i want to do coreboot
<oliv3r> but, i need to get a better job first; ideally one that does linux work :)
<mripard_> so that if a driver is there for this hardware (compiled statically or a module just loaded), it will get probed with the right parameters
<oliv3r> it's there now
<oliv3r> the easy ones are -1; but who cares :)
<oliv3r> thanks :)
<oliv3r> btw, for my productivity it's better if nobody is here :p but it's 22:30; so it's good :)
<oliv3r> hmm, why did I miss that lwn; i check lwn form time to time
<mripard_> it's been published today.
<oliv3r> lol
<oliv3r> ok
<oliv3r> lets see if it's better then phoronix journalism
<mripard_> I sent a mail yesterday to the phoronix mail
<drachensun> alright, new fex and USB is up
<mripard_> *guy
<mripard_> about that article
<oliv3r> intersting
<oliv3r> drachensun: your memory section has all the usual -1 things omitted
<oliv3r> but tons of new parameters
<oliv3r> :S
<oliv3r> yikes
<oliv3r> was affraid ofthat
<drachensun> yeah the old tools choke on it now I think
<oliv3r> wow only 240 MHz memory bandwith?
<oliv3r> ouch
tzafrir has joined #linux-sunxi
<ssvb> how much L2 cache is A20 supposed to have?
<oliv3r> 240 MHz for dual cores and mali 400 mp2?
<oliv3r> the memory will be chocked to death
<oliv3r> allthough
<oliv3r> 64bit (dual channel 32bit)
<oliv3r> would that matter at all ssvb?
_BJFreeman has joined #linux-sunxi
<hramrach_> rellla: this one crashes player with native cedar and fails to decode with libhybris. Cannot test on android. And I cannot decode it so don't know what it is. use at your own risk ;-) http://dl.linux-sunxi.org/users/hramrach/non-decode.mkv
BJfreeman has quit [Ping timeout: 252 seconds]
_BJFreeman has quit [Ping timeout: 252 seconds]
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
<oliv3r> Allwinner seems to have been uninterested in the switch to Device Tree (DT)
<bfree> http://paste.debian.net/11658/ is printenv from cubieboard2. can anyone explain wtf it means? i.e. what is the "partitions" stuff? what is boota? where is the kernel actually? My actual first aim is just to put a rootfs on sata (with the linaro /lib/modules) and boot in to that
<oliv3r> sounds a bit too harsh
Soru has quit [Remote host closed the connection]
<mripard_> oliv3r: and your sid driver even reached LWN as well :)
<oliv3r> how so?
<oliv3r> i'll read the article first
<mripard_> in the usual patches-worth-mentioning section of LWN
<mripard_> not in the article
Soru has joined #linux-sunxi
<oliv3r> reading
<edeloget> oliv3r: http://lwn.net/Articles/554849/ ; not sure if it needs an LWN account or not
<edeloget> mripard_: is the A10S patch queued for 3.11 enough to run the A10S-based card by Olimex
<edeloget> (is it necessary?)
<mripard_> yes, and yes :)
tkoskine has quit [Ping timeout: 260 seconds]
<oliv3r> ohh i saw that
<oliv3r> i googled for my patch as i needed to know something :)
hramrach_ has quit [Ping timeout: 240 seconds]
<mripard_> oliv3r: you're beginning to frighten me if you don't know about your patch :)
<oliv3r> haha
<oliv3r> commments :p
<oliv3r> was looking for reviews while mailbox wasn't available
hramrach_ has joined #linux-sunxi
tkoskine has joined #linux-sunxi
<edeloget> (looking at the patch on infradead. their software transform serial@01c28000 to serial at 01c28000 ; that doesn't help much)
<mripard_> edeloget: 2s
<mripard_> https://github.com/mripard/linux, branch sunxi-next-a10s
<edeloget> mripard_: don't worry, that was just a commentary on their mail-to-web software which is a little too aggressive with what it thinks is a mail address.
<mripard_> anyway, it's there if you want it :)
<edeloget> mripard_: I want it, for sure :) I'll receive my board in a few days now, and I'm quite excited (not that it's a wonderfull board but it has some merrits)
<mripard_> you won't have much out of it though
tkoskine has quit [Read error: Connection reset by peer]
tzafrir has quit [Ping timeout: 260 seconds]
<mripard_> so don't expect too much :)
tkoskine has joined #linux-sunxi
ibrah is now known as eebrah|away
rellla has quit [Quit: Nettalk6 - www.ntalk.de]
notmart has quit [Quit: notmart terminated!]
Soru has quit [Ping timeout: 264 seconds]
<oliv3r> bed time
<oliv3r> someone try to run the a31-meminfo :
<oliv3r> ;)
<edeloget> oliv3r: do you know any low-cost A31 board ?
tinti has quit [Read error: Connection reset by peer]
edeloget has quit [Quit: Page closed]
G[3z] has joined #linux-sunxi
vicenteH has quit [Ping timeout: 264 seconds]
G[3z] has quit [Remote host closed the connection]
Soru has joined #linux-sunxi
<drachensun> edeloget: The new mele is the cheapest one I have seen that has a serial port and all
piyushverma has joined #linux-sunxi
<drachensun> oliv3r: I just tried a31-meminfo I don't think it is cross compiled right
<drachensun> -bash: ./a31-meminfo-static: cannot execute binary file
<drachensun> root@linaro-alip:~# file a*
<drachensun> a31-meminfo-static: ELF 64-bit LSB executable, x86-64, version 1 (GNU/Linux), statically linked, for GNU/Linux 2.6.9, not stripped