hno changed the topic of #linux-sunxi to: /Allwinner/sunxi development discussion - Don't ask to ask. Just ask and wait! - See http://linux-sunxi.org | https://github.com/linux-sunxi/ | Logs at http://irclog.whitequark.org/linux-sunxi
popolon has quit [Quit: Quitte]
<deasy> arokux2, it's from rhombus?
<deasy> so it's probably one
* deasy go in bed watch his movie and sleep
<deasy> good night people
<arokux2> night
<Turl> arokux2: according to hno, not yet
<arokux2> Turl, not yet? until release?
<Turl> arokux2: having a look at your patches, do you need style nitpicking or are you going to go through the details on a later date?
<Turl> arokux2: I dunno when, but it's not OSHW yet
<arokux2> Turl, not sure, I have no experience on what is better. comment on what is crucial to change in the first place maybe, then i'll do v2
<Turl> arokux2: I think your code editor has some weird notion of tab length
<arokux2> Turl, hm.. does it matter, tabs are not expanded
<arokux2> ?
<Turl> arokux2: well, it results on misindented code
<Turl> arokux2: 'git show' your first commit and you'll see it on the line wrapping of of_property_read_string_index and on the line wrapping on the dt clock names
<steev> mkdir -p lib/modules
<steev> rm -rf lib/modules/
<steev> wat?
<arokux2> steev, :)
<steev> i think that rm -rf is supposed to be cd
<steev> http://linux-sunxi.org/FirstSteps under setting up the rootfs
<arokux2> steev, fix our wiki :)
<arokux2> steev, yes, I'm aware of it
<steev> arokux2: i don't know what is being done there
<Turl> arokux2: + * FIXME: Should the credits be given to the original pile of code? <- yes, if you started from it
<arokux2> Turl, ah, there are extra spaces.. my bad.
<steev> yep, always
<Turl> arokux2: +/* FIXME: Remove this crap once everything works! */ <- needs fixing too
<Turl> arokux2: I think your editor is configured with tabs=4 spaces
<arokux2> Turl, maybe best you comment on ML?
<arokux2> Turl, no those spaces were added by hand or smth
<steev> more likely your editor expands tabs to spaces
<Turl> arokux2: I'm having a really quick look at it, I'll probably have a second look later and reply on ML
<arokux2> steev, no.
<steev> arokux2: there's a script to fix it either way
<Turl> steev: no, it's like it got a lot of extra tabs
<arokux2> Turl, "Remove this crap": will vanish once it passes all the reviews and gets testing..
<steev> oh
<Turl> arokux2: ok
<arokux2> Turl, ok, then feel free to drop you comments here
<arokux2> Turl, thanks a lot.
<steev> arokux2: the comment should be dropped if it's being worked on
<steev> cleanfile? i think is the script
<arokux2> Turl, does you mele run mainline?
<Turl> arokux2: dev_err(&pdev->dev, "failed to get I/O memory\n"); <- don't print needless messages, the kernel already does
<Turl> arokux2: no, it runs 3.4 as I need USB and a place to keep the rootfs
<steev> is this on the linux arm mailing list?
<Turl> steev: linux-sunxi on googlegroups
<steev> oh
FDCX_ has quit [Ping timeout: 256 seconds]
<arokux2> Turl, "don't print needless messages, the kernel already does": in this case it does not.
<Turl> arokux2: hm, well I take that back, it's not a kalloc :P
<Turl> yeah sorry
<Turl> noticed a bit too late
<arokux2> Turl, ok :)
<arokux2> Turl, no prob
Olaff has joined #linux-sunxi
<arokux2> Turl, so you'll have USB for Mele. where do you keep your rootfs?
<Turl> arokux2: mmc
<Turl> "fail1" could probably be just "fail" as there's just 1 of it
<Turl> +/* this is enabled by default, of not it should be
<Turl> + * enabled in U-Boot ---> hci_port_configure
<Turl> what is "this"? :)
<arokux2> Turl, hci_port_configure. yes this is weird comment. I wanted to change it actually but forgot :)
<Turl> arokux2: all those bits you use on sunxi_usb_passby could probably be #define's with proper names
Olaff has quit [Read error: Operation timed out]
<arokux2> Turl, ok
<arokux2> Turl, 8Gb USB Stick cost the same as 8Gb MMC, but we cannot boot from USB just now.
<Turl> yeah
<Turl> usb_phy_write is pretty unintelligible
Olaff has joined #linux-sunxi
<deasy> why these port are called mmc?
<deasy> yes yes i don't sleep :p
<Turl> deasy: multimediacard
<arokux2> Turl, do you mean readability? or why it is done the way it is done?
<deasy> there are little different with sd?
<deasy> difference*
<Turl> arokux2: I mean readability, yes; a lot of magic constants and stuff
<deasy> ow ow ow
<deasy> maybe it means mmc like compatible with several kind of memory card
<arokux2> Turl, I'm falling asleep, so leaving now. just send you comments here I'll see them tomorrow. thanks a lot.
<deasy> gn arokux2
<arokux2> deasy, gn
<Turl> night arokux2
<Turl> deasy: MMC and SD are two standards
<Turl> deasy: read the MultiMediaCard and Secure Digital wikipedia pages
<deasy> i know it
<deasy> i have in the past a nokia 6600 who use a mmc
<deasy> but how we can put sdcard on mmc on a10?
<deasy> i read it tomorow :)
<Turl> yes, you can use SD cards
<Turl> in fact most boards have microSD slots on them
<deasy> yes as my cubieboard
* deasy go to bed
FDCX_ has joined #linux-sunxi
ssvb has quit [Ping timeout: 256 seconds]
arokux2 has quit [Ping timeout: 260 seconds]
ssvb has joined #linux-sunxi
TheViking has joined #linux-sunxi
tinti has quit [Ping timeout: 276 seconds]
tinti has joined #linux-sunxi
FDCX_ has quit [Ping timeout: 245 seconds]
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
<steev> hmm
<steev> cb2 no ethernet with 3.4.61?
<Turl> steev: looks like you don't have the driver enabled in the kernel
<steev> Turl: possibly
<steev> i used sun7i_defconfig
<Turl> steev: look for EMAC on menuconfig and enable it
TheViking has quit [Ping timeout: 250 seconds]
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<steev> aha
<steev> ssvb: and you say that mtune=cortex-a7 doesn't actually perform as well as -a8 or -a9?
FDCX_ has joined #linux-sunxi
FDCX_ has quit [Ping timeout: 260 seconds]
<steev> the serial port for the cb2 is ttyS0 right?
setkeh has quit [Ping timeout: 240 seconds]
setkeh has joined #linux-sunxi
akaizen has quit [Remote host closed the connection]
setkeh has quit [Changing host]
setkeh has joined #linux-sunxi
jlj has quit [Ping timeout: 256 seconds]
setkeh has quit [Read error: Connection timed out]
setkeh has joined #linux-sunxi
setkeh has quit [Ping timeout: 256 seconds]
<Turl> Seppoz: it should be, yes
<steev> i figured it out
<steev> silly me forgot to change inittab from 9600 to 115200 :)
setkeh has joined #linux-sunxi
setkeh has quit [Changing host]
setkeh has joined #linux-sunxi
forcev has joined #linux-sunxi
tm512` has joined #linux-sunxi
Tartarus_ has joined #linux-sunxi
\\Mr_C\\ has quit [*.net *.split]
focus has quit [*.net *.split]
FunkyPenguin has quit [*.net *.split]
Tartarus has quit [*.net *.split]
tm512 has quit [*.net *.split]
focus_it has joined #linux-sunxi
tm512` is now known as tm512
jlj_ has joined #linux-sunxi
[7] has quit [Disconnected by services]
TheSeven has joined #linux-sunxi
akaizen has joined #linux-sunxi
mturquette has quit [Remote host closed the connection]
akaizen has quit [Ping timeout: 245 seconds]
JohnDoe_71Rus has joined #linux-sunxi
akaizen has joined #linux-sunxi
ynezz has joined #linux-sunxi
TheViking has joined #linux-sunxi
Amnience has joined #linux-sunxi
fldc has quit [Ping timeout: 245 seconds]
hipboi has quit [Quit: Leaving]
ibrah is now known as eebrah
eebrah is now known as Guest25610
Black_Horseman has quit [Remote host closed the connection]
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
FDCX_ has joined #linux-sunxi
n01 has joined #linux-sunxi
<TheViking> Would it be appropriate to add information about f2fs on the linux-sunxi wiki?
<oliv3r> TheViking: why?
<oliv3r> TheViking: you can put a link to wikipedia or kernel.org or wherver the information lives, but having random info on random topics would just be duplication would it not?
<TheViking> oliv3r: The cubieboards (and others) use SD cards for storage. F2FS would ideally improve SD card life and performance. The linux-sunxi kernel is only 3.4 and the earliest version of the mainstream linux kernel with F2FS is 3.8...
<oliv3r> TheViking: that's what i'm saying you can put a little bit like what you said up there (which kinda is moot, since its not even in our kernel, and our mainline work doesn't have mmc driver yet) with a link to the f2fs sites
<oliv3r> but much more would be overdone
<TheViking> oliv3r: My keyboard is acting up. Hold on please...
n01 has quit [Quit: Lost terminal]
<oliv3r> TheViking: hehe, sure
n01 has joined #linux-sunxi
<TheViking> oliv3r: There is a f2fs backport project here:https://github.com/nowcomputing/f2fs-backports
hipboi has joined #linux-sunxi
<TheViking> oliv3r: As you can see, it is void of linux 3.4 patchsets...
<TheViking> oliv3r: According to this issue ticket: https://github.com/nowcomputing/f2fs-backports/issues/1 A user by the name 01micko found a solution to adding f2fs to 3.4 via the patches on the repo...
<TheViking> oliv3r: I have tested that on the current linux-sunxi 3.4 and currently have 3 days uptime. No visible issues so far. ( I use it as my rootfs on an sd card )
<oliv3r> hehe, well if you port the patch back and make it work in our 3.4; good!
<oliv3r> until then i'd keep the noise on the wiki to a minimum
<oliv3r> but hey, it's an open wiki, you can write whatever you want :)
<oliv3r> i just shared my opinion as you asked for it :)
mcbrick has quit [Remote host closed the connection]
<TheViking> oliv3r: If you are suggesting I try and port f2fs for linux-sunxi's repo, I have never done anything significant with git or anything developer related for that matter.
Guest25610 is now known as eebrah_
<oliv3r> live and learn? :)
docice has joined #linux-sunxi
<TheViking> oliv3r: And what if I break something in the process?
<oliv3r> it will get reviewed :p
gnufan has joined #linux-sunxi
<oliv3r> but I don't expect huge issues
<oliv3r> git am >patch; done
<oliv3r> probably
<TheViking> I guess I could fork it and set the patches in order. Dirty but quick.
<TheViking> Like I need another account online :-P (GitHub)
<diego71> TheViking: But you don't really need a github account. You can clone on your pc, and make patch from there
<TheViking> How would I submit it to be merged into the linux-sunxi kernel?
<diego71> You can send your patch on the mailing list
<oliv3r> TheViking: git send-email
<oliv3r> or git format-patch
<oliv3r> ideally you'd metnion the commit hash(es) that needed to be cherry picked from upstream; and then a (small) patch to make it all work
panda84kde has joined #linux-sunxi
fredy has quit [Excess Flood]
fredy has joined #linux-sunxi
atiti has joined #linux-sunxi
massi_ has joined #linux-sunxi
eebrah_ is now known as eebrah
FR^2 has joined #linux-sunxi
<TheViking> So format-patch, zip it, put it in a mail list?
<TheViking> Is there a mail list?
<TheViking> I know how send-email works but um, there's a lot of patches.
rellla has joined #linux-sunxi
<JohnDoe_71Rus> try to boot cb2 from SD card. get this log http://pastebin.com/3Ag3uLmS How to build right partition table for SD card ?
wingrime has joined #linux-sunxi
book` has quit [*.net *.split]
hno has quit [*.net *.split]
book` has joined #linux-sunxi
hno has joined #linux-sunxi
<buZz> JohnDoe_71Rus: add something behind root=
<buZz> [ 0.000000] Kernel command line: console=ttyS0,115200 root= loglevel=8 panic9
<buZz> point it to the right partition
<rellla> morning
<wingrime> rellla: morn
<wingrime> mnemoc: [PATCH v4 3.4 0/4] sunxi-nand: incorporate sun7i AW nand driver code
<wingrime> mnemoc: I ACKed this V4
<wingrime> mnemoc: from patrick
<JohnDoe_71Rus> buZz: i use this script and settings https://github.com/shineworld/cubietools/blob/master/support/uEnv.txt
<rellla> wingrime: i need some help trying libvdpau. there's something i miss regarding X
<wingrime> rellla: ?
tzafrir_laptop has quit [Ping timeout: 246 seconds]
<rellla> no window manager here, i have to start everything from console. how do i tell mplayer to use X? mplayer is telling me, that file is played but i don't see anything. too far away i used X the last time ...
<mnemoc> wingrime: please paste me the subjects in /q so I can take a look and merge after leaving work
<rellla> wingrime: no cubie here atm, but sth. like "-vo output missing" is popping up
<mnemoc> rellla: export DISPLAY=:0.0 should do the trick
<oliv3r> rellla: export DISPLAY
<oliv3r> crap
<oliv3r> i'm to slow
<wingrime> mnemoc: first time V4 in linux sunxi?
<wingrime> rellla: they right
<mnemoc> wingrime: the cedarx unification was also v4 iirc
<steev> how to get rid of the stupid android loglevel garbage when running linux? it makes it extremely hard to read boot up
<mnemoc> build a not-android kernel...
<steev> i removed all of android from the config
<rellla> mnemoc: uahh. i'll try that, since everything with libvdpau-sunxi and xf86-video-sunxifb seemed to be right. mali test prog also don't show up ;) do i need ssvb's module for doing the mplayer test?
<steev> i also removed the "loglevel=${logleve}" from the setargs command, and loglevel isn't set in my boot.scr
<steev> loglevel*
<wingrime> mnemoc: with cedar I still have no clear decide
<rellla> oliv3r: de->de is way too fast :p
<steev> [ 0.000000] Kernel command line: console=ttyS0,115200 root=/dev/mmcblk0p2 rootwait panic=10
<mnemoc> wingrime: hey! you acked it, and I merged it...
<steev> setargs=if test -z \\"$root\\"; then if test \\"$bootpath\\" = "/boot/"; then root="/dev/mmcblk0p1 rootwait"; else root="/dev/mmcblk0p2 rootwait"; fi; fi; setenv bootargs console=${console} root=${root} ${panicarg} ${extraargs}
<wingrime> wingrime: ok
<wingrime> mnemoc: I talking about
<wingrime> mnemoc: blob
<mnemoc> ok
<oliv3r> rellla: hmm?
<steev> wtf, even setting loglevel=0 it still does it
<wingrime> mnemoc: we workarouded this in kernel but that should fixed otherway
<rellla> oliv3r: offtopic garbage forget it ;)
Super-noob has quit [Ping timeout: 260 seconds]
<wingrime> mnemoc: cache issue, can interfere with libvdpau
<mnemoc> ouch
<wingrime> mnemoc: dont care so much
<wingrime> mnemoc: blob should work also
<panda84kde> hi everybody. There seems to be an A23 SOC now from Allwinner. http://www.allwinnertech.com/en/awt/news/2013-09-10c.html
libv has quit [Ping timeout: 263 seconds]
<mnemoc> panda84kde: there is an interview from charbax to eva about it
<wingrime> mnemoc: so we HAVE this workaroud as we relay on users with blob
<mnemoc> sounds fair
<mnemoc> but maybe adding an option to disable it might be a good idea
<oliv3r> mnemoc: link
<mnemoc> youtube -> charbax
<mnemoc> Tsvetan forwarded it on G+ too
<steev> cnxsoft ?
<steev> easier to find his videos
<mnemoc> charbax = armdevices
<wingrime> mnemoc: where you pushed cedar
<wingrime> mnemoc: I don't see it in stage
Super-noob has joined #linux-sunxi
<mnemoc> the unification? to stage/sunxi-3.4
<mnemoc> git fetch? :)
<wingrime> mnemoc: not see there
<mnemoc> ow, it's g2d
<mnemoc> got confused
<wingrime> mnemoc: me too
<wingrime> mnemoc: facepalm
<wingrime> mnemoc: ok, then wait ssvb for decision
<oliv3r> charbax is sometimes a bit of a douche
<oliv3r> lol
<oliv3r> but i do enjoy watching the vids
<oliv3r> sorta
<mnemoc> charbax interviews are better than no interviews :p
<ssvb> wingrime: what decision? (I don't follow your chat)
<Tsvetan> oliv3r at least he present news :)
<oliv3r> Tsvetan: exactly, I did say I enjoyed watching :)
<wingrime> ssvb: leave cache w/o in a20 kerenl
<oliv3r> as he brings cool stuff
<oliv3r> badly reported news is still news :)
<Tsvetan> when I saw this video I asked Allwinner for A23 datasheet and they said we do not have yet :))) but Eva says the chips will be in production in October
<ssvb> wingrime: with cedar it would be nice for someone to 1) actually provide some benchmarks showing how slower it gets without cache 2) try writecombine mapping
<zumbi> Tsvetan: what's A23?
<Tsvetan> kinda reminds me Olimex we released imx233-NANO and next day people start asking where is the user manual which we didnt start :)))
<zumbi> Tsvetan: is A23 using SGX?
<Tsvetan> A23 is A20 but in TQFP package as A13
<Tsvetan> lower cost A20
<zumbi> nicee
<wingrime> Tsvetan: as expected
<zumbi> Tsvetan: btw, do you know if rockchip chips are available too?
<buZz> Tsvetan: A23 probably lacks sata?
<Tsvetan> zumbi - sure they are, Rk3188 is all around in TV boxes etc
<wingrime> mnemoc: so, than cedarx merge patch should stay for a while
<Tsvetan> buZz probably same as A13 - LCD and USB hosts
<buZz> hmhm
<Tsvetan> you have no much choice in 176 pin package
ykchavan has joined #linux-sunxi
<wingrime> Tsvetan: there only one cortex a in that package
<wingrime> Tsvetan: easy to solder
<Tsvetan> yes, and with good heat dissipation pad on back I hope
<oliv3r> oh god just watched his e-ink watch video and now i really don't like him lol
<oliv3r> Tsvetan: yeah they make chips and drivers without user manuals
<arokux> hi guys
<ykchavan> buZz, I am curious to know why you mentioned ' keep in mind berryboot is not a recommended boot image for A10 devices' in wiki.
<buZz> because its made for raspberry pi
<ykchavan> I see
<arokux> steev, still fighting with log messages? I know the cure.
<steev> arokux: i'm all ears
<steev> i figured it is because of LOGLEVEL=7 in the kernel
<rm> ykchavan, because it uses some sort of its own weird system for loading the kernel
<rm> and people fail to update their kernel following the usual sunxi howtos, then waste everyone's time trying to get help why
<steev> still waiting on the compile, since we like to enable everything and the kitchen sink
<rm> only a long time later it turns out they used berryboot
<rm> nuke that crap from orbit
<oliv3r> a23 is dualcore A13, probably sun8i
* mnemoc hopes for another sun7i
<mnemoc> they are running out of numbers
<oliv3r> how so?
<ykchavan> thanks for information rm.
<oliv3r> numbers kinda go into invinity
<wens> mnemoc: they can always go for 2 digits
<oliv3r> sun10i
<oliv3r> sun9i probably will be a31 followup
<oliv3r> so sun10i shoudl be sun47i brother
<arokux> steev, in kenel hacking turn of low level debugging
<oliv3r> what's the a23 allwinner video's title?
<oliv3r> as user/Charbax has many videos
<oliv3r> ohhh found it
<oliv3r> duh
<steev> arokux: er, interesting
<oliv3r> 2 days ago
<steev> arokux: are there a bunch of printk's hidden behind DEBUG_LL or something? you shouldn't have to turn off DEBUG_LL to get rid of duplicate messages
<arokux> steev, patches are welcome. I said I know the cure.. :)
<steev> arokux: i am talking the messages like <6> blah
<steev> not ALL the messages
<arokux> steev, those will be not all messages. the problem is when DEBUG_LL is on the messages get duplicated
<arokux> steev, to avoid duplication you need to disable "Kernel low-level debugging functions (read help!)"
<steev> arokux: yeah that's DEBUG_LL
<steev> i'll see if i can find some time for it on friday
<steev> later today i have to do amd64 stuff, and then tomorrow i promised someone i'd look into an issue with Kali on the ODROID-XU
<arokux> thanks steev no worries.
<JohnDoe_71Rus> Does anyone have working android SD cards image for cubieboard2 A20?
<JohnDoe_71Rus> only Linux
<rm> oh sorry
<JohnDoe_71Rus> for SD
shineworld has joined #linux-sunxi
<JohnDoe_71Rus> shineworld: hi
<shineworld> hi JohnDoe_71Rus :)
<JohnDoe_71Rus> did you get my email?
<shineworld> ... checking ...
<shineworld> PS: I was far home some day recently :)
<TheViking> linux-sunxi wiki search lacks an "I'm feeling lucky" button.
<shineworld> Ok, for actual AllWinner 4.2.2 android partitions I watch either original nand boot and what in xxxx.rc and all other android files which have some relation with /dev/...
<shineworld> the partitions table is changed than 4.0.4 and what declared in linux-sunxi wiki pages
<shineworld> a new env/databk are entered
<shineworld> I guess my new table is right, at least comparing any thing with xxx.rc files
<shineworld> I've left only /private which I don't meet in real declarations
PiyushVerma has quit [Read error: Operation timed out]
<JohnDoe_71Rus> shineworld: think it is secured
<shineworld> Tomorrow, I'm hopefully at home, I will restart by zero my analysis of android init and tools files to get a more sharp list
<shineworld> In my system all boot work and the only problem is in MM memory managemnt (kernel panic) throw by some zygote internal operation (I guess class pre-loading)
<shineworld> Unfortunately I've got very few time in these weeks to fix question (I was far to a custom plant for job)
notmart has joined #linux-sunxi
<shineworld> I need a CB2 working Android on SD fastly ...
<shineworld> the 11 partiotions question is because the /dev/sdc4 is an extended partion (id=5) You know primary are only 4 so first 3 are real, 4 is extended which implement the rest
<shineworld> this is the right way to have required 10 real partitions on SD
<shineworld> 3 real, 1 extended which inlcude remaining 7
<shineworld> About private: private compare in boot list but NEVER in android files ... so I've removed it
<JohnDoe_71Rus> shineworld: but at the boot, they 11 real with private
_enrico_ has joined #linux-sunxi
<shineworld> have you tried to add that ? and boot ?
<JohnDoe_71Rus> yes. my logs in mail with 12 partitions
<shineworld> try a "grep -lir "private" . in android root files to see if someone use it
<shineworld> I can be wrong in my analysis :)
libv has joined #linux-sunxi
soldoKyn has joined #linux-sunxi
<shineworld> I'm downloading sdk at now ... I don't have it at office... sigh
<soldoKyn> arokux: hi! I'm having this error trying binary driver's tests: http://pastebin.com/2DAQU3e8 do you know what it could be?
<arokux> soldoKyn, sorry ssvb and libv are the people working on this.
<arokux> soldoKyn, I can only point you to: http://linux-sunxi.org/Binary_drivers
<TheViking> Shineworld: "...MM memory managemnt (kernel panic) throw by some zygote internal operation..." I built an android 4.2 image for a no-name rockchip tablet a while back and had something very similar like this. Compiling with thumbcode disabled seemed to fix/get around the problem (--disable-thumb2 flag IIRC)
<shineworld> TheViking, how to change this setting ?
<oliv3r> arokux: i'm not sure if mainline folk will be happy to take a patch without a real name
<arokux> oliv3r, wingrime read a P.S. please :p
<soldoKyn> arokux: ok, thank you!
<TheViking> Shineworld: My build environment was on a VM to keep things tidy, but I no longer have that VM. I don't remember, I'm sorry.
<shineworld> I'm using a VM too (my host is a MAC)
<soldoKyn> ssvb: libv: can I turn the question to you?
tzafrir_laptop has joined #linux-sunxi
<shineworld> ... download of sdk require 23H... so slow server
<TheViking> Shineworld: Maybe adding "-disable-thumb2" to the CFlags?
<shineworld> I will tray this evening at home ... thank you for precious suggestion
<TheViking> shineworld: Thumb2 is a pretty useful optimization for android's JVM. I couldn't figure out why it broke stuff.
<arokux> is Charbax a noob or he's just playing one
<shineworld> some weeks ago someone suggested me a maillist with a patch but was too complicate for me to act it
<TheViking> act it?
<shineworld> do apply correctly
<shineworld> *to
<TheViking> git am?
<shineworld> a sec for recovery from old chat lines....
<TheViking> Or did it require a different tool to patch?
fredy has quit [Excess Flood]
<shineworld> I'm a firmware/kernel designer for custom products but work with linux and related things is a very new activity for me
<shineworld> so you can guess how a simple operation could be hard to do ...
fredy has joined #linux-sunxi
<oliv3r> arokux: you only mention EHCI bindings, what about OHCI?
<oliv3r> e.g. usb 1.1
<arokux> oliv3r, will come later. I want to gather experience from reviewers first.
<oliv3r> i'm just curious at this point
<oliv3r> i compiled a new 3.11 kernel for my server the other day, and idsabled pci-platform for usb
<oliv3r> and all of a sudden, my USB 1.1 devices stopt working
<oliv3r> only USB 2.0 devices worked
<oliv3r> due to the missing usb 1.1 controller
<oliv3r> so you say you tested keyboards, mouse and stick; which means usb 1.1 and 2.0 devicese (unless the usb stick is 1.1 too :p)
<oliv3r> well it IS backwards compatible
<arokux> oliv3r, yes, the mouse and the keyboard I have worked :)
<oliv3r> in other words, on a PC, you need USB 1.1 (ohci) drivers for 1.1 devicse, EHCI driver ONLY does 2.0
<oliv3r> so your patch makes it _sound_ that it's *only* EHCI
<arokux> oliv3r, usb stick is 2.0 i think..
<oliv3r> but i don't know how it works with this aw device of course
<arokux> oliv3r, yes, commit messages say explicitly *only* EHCI support.
<arokux> oliv3r, you are welcome to test it, just say what your device is i'll add dt bindings and you can easily test.
<oliv3r> cb2 and olimexino :)
<arokux> oliv3r, olimexino - which one?
<oliv3r> so (i'll ready your patch in a mminute) do both drivers? or does the hardware do both simultaniously?
<oliv3r> a20
<oliv3r> cb1 i have too :)
<oliv3r> but don't worry
<oliv3r> i'll add the devicetree nodes :)
<arokux> oliv3r, that would be cool, but can take some time, however mele dt is a good basis.
<arokux> oliv3r, "so (i'll ready your patch in a mminute) do both drivers? or does the hardware do both simultaniously?" I do not get the question.
<arokux> oliv3r, so you have? A20-OLinuXino-MICRO
<soldoKyn> ssvb: libv: for completeness: that's my Xorg.log http://pastebin.com/QWh3jYVV and I'm using a Debian (from olimex) with a kernel compiled from sunxi 3.4 branch and with graphics driver compiled from http://linux-sunxi.org/Binary_drivers (libdri2 compiled from git repo listed there and mesa-egl dir moved like listed in "common pitfalls") thanks!
<shineworld> TheViking, what I get is "kernel BUG at include/linux/mm.h:791" during zygote execution
shineworld has quit [Quit: Sto andando via]
<oliv3r> arokux: from what I understand, and my experience on x86; if you want USB support you need OHCI and EHCI drivers
<oliv3r> I know usb 2.0 is backwards compatible with 1.1 devices
<oliv3r> but you still need an OCHI driver to have this backwards compatibility
<oliv3r> if you compile ECHI without OCHI, you only have 2.0 devices working (storage etc)
<arokux> oliv3r, I see. maybe it is needed for ARM SoCs too. but having EHCI only works so I thought I'm fine
<oliv3r> but you can't use a mouse, as it requires 1.1
<oliv3r> now you say you used a keyboard + mouse; which worked, so thats strange
<oliv3r> what does lsusb say? USB 2.0 hub and USB 1.1 hub is what you need to see
<oliv3r> if you only see USB 2.0 hub; you don't have USB 1.1 support afaik
<arokux> oliv3r, hm.. well cat /dev/input mouse has output some noise, I've considered it working :)
<arokux> oliv3r, so I will do more thorough testing
<oliv3r> lol
<oliv3r> yeah lsusb should give you a lot interesting information as well, also /sys/*/usb/* something gives info
<oliv3r> lsusb -v basically
atiti has quit [Ping timeout: 276 seconds]
<arokux> ok thanks oliv3r I'll take a close look at them.
PiyushVerma has joined #linux-sunxi
<oliv3r> arokux: you can look at my ahci driver for the regulator useage bit for the US gpio; it's really easy :)
<oliv3r> arokux: basically, I don't know how OCHI and EHCI work together on arm, or how it works with the AW code
<arokux> oliv3r, have I done smth wrong?
<oliv3r> btw, i only see your 0/4 mail, where's 1/4 - 4/4?
<oliv3r> arokux: i'm just reading alexey's review, where he says 'FET gate to usb'
<arokux> oliv3r, yes. Turl have seen them
<oliv3r> maybe it's ower on the list :)
<arokux> oliv3r, I haven't understood the FET bit, have no idea what it is, will look into it.
<libv> soldoKyn: what application are you running?
<arokux> oliv3r, then click on subject
<arokux> oliv3r, wrong link
<soldoKyn> libv: the test from sunxi-mali git...
<libv> soldoKyn: did you run gdb on it
<soldoKyn> libv: I'm doing it right now...
ykchavan has quit [Quit: Leaving]
<ssvb> soldoKyn: do you have the right permissions for /dev/mali and /dev/ump (correct ownership, and you user in the 'video' group)?
<libv> would that really cause a segfault?
<ssvb> don't know, but the first line of the log implies that there is something wrong with DRI2
rz2k has joined #linux-sunxi
<ssvb> also he got EGL Vendor: "Mesa Project", that's not the mali binary driver
atiti has joined #linux-sunxi
<libv> ah, right, explains the dri stuff, seems that binary drivers wasn't followed
<ssvb> soldoKyn: you can try to run ldd for libEGL.so and libGLESv2.so to see what is really used in your system
<libv> this is probably still there: /usr/lib/arm-linux-gnueabihf/mesa-egl
<oliv3r> arokux: the FET bit is basically 1 gpio to enable/disable power; e.g. a regulator, if you look at the ahci patch (or the hackberry emac) both use a 'fet switch' to enable power too
<arokux> oliv3r, have you seen my code? I use regulator_enable stuff
gnufan has quit [Ping timeout: 245 seconds]
<soldoKyn> ssvb: libv: uhm...I think I've broken something trying to build glmark for a performance test (I had to install some packages with apt and maybe some of them were just from mesa). I tried ldd command with this result: http://pastebin.com/r1cBiUNv
<ssvb> soldoKyn: do you also have /usr/lib/libEGL.so ?
<libv> a20?
<libv> so you're running r3p2-01-rel2
<ssvb> soldoKyn: it is possible that your version of debian has reshuffled the shared libraries placement and the "common pitfalls" section needs to address this
<ssvb> libv: not yet, sunxi-3.4 kernel is still using r3p0 mali module
<libv> ah, ok
<soldoKyn> ssvb: yes, I've got that "chain": libEGL.so -> libEGL.so.1, libEGL.so.1 -> libEGL.so.1.4, libGL.so.1.4 -> libMali.so
<ssvb> soldoKyn: looks right, now you need to move mesa out of the way
<libv> lib_GL_.so.1.4?
<ssvb> soldoKyn: ldd for your gles application can reveal which libraries are going to be used
<wingrime> arokux: about FET
<soldoKyn> libv: excuse me is libEGL.so.1.4 -> libMali.so (not GL!!!)
<arokux> wingrime, it would be best if you show me some example
<wingrime> arokux: some vendors use FET transistor like key that enable or disable usb
<libv> soldoKyn: ok then :)
<wingrime> arokux: port
<wingrime> arokux: you have trun some gpios to 1 for enable usb host
<arokux> wingrime, ok, so how that applies for sunxi? I need to put a GPIO to out and let it output "1" so that the VBUS is powered. it is not for enabling of the usb host.
<arokux> hipboi, you have even more spam on http://cubieboard.org
<wingrime> arokux: check some board schenematics
<wingrime> arokux: mars bord have this keys as I remeber
<arokux> wingrime, I do not understand what are you trying to tell me. USB EHCI works as is.
rz2k has quit []
<wingrime> arokux: I mean some boards need more love for enable host
<ssvb> techn_: btw, what was the maximum mali clock speed that you could get on a13? is it higher than 380mhz?
<arokux> wingrime, ah, ok. well, I do not have the hardware :p so patches are welcome. tonight I'll add bindings for the cb2 and olinuxino a20 so ppl can test.
rz2k has joined #linux-sunxi
<oliv3r> arokux: no i haven't, i only see 0/4 (the cover letter)
<arokux> oliv3r, so wingrime is talking about something else as you see.
<oliv3r> i should see it :)
<oliv3r> but i haen't found it yet
<oliv3r> still need to read some mails
<ssvb> soldoKyn: btw, after you get mali binary drivers working, the performance is not going to be great in your setup, but it can be improved significantly with some tweaks
<arokux> oliv3r, I've given a link to gmane... please give further comments after actually looking at the code :p
<oliv3r> arokux: lol yeah if i have some more time
<arokux> oliv3r, no worries. just comment on the code :)
<soldoKyn> ssvb: libv: I confirm that was mesa the problem. removing some mesa packages used for building glmark leads to a working mali binary driver! thank you so much!!
<soldoKyn> ssvb: where I can get these tweaks?
<oliv3r> wingrime: while still backreading, the regulator framework should take care of the FET
<panda84kde> soldoKyn: ssvb: are you referring to line 284 of Xorg.0.log or even something else? http://pastebin.com/QWh3jYVV
TheViking has quit [Ping timeout: 250 seconds]
<oliv3r> arokux: more descriptive commit messagse would be nce to have :)
<wingrime> oliv3r: no
<arokux> oliv3r, nothing to describe there...
<arokux> oliv3r, i'm not fixing stuff, just adding support.
<wingrime> oliv3r: thats does not related with axp
<oliv3r> arokux: even so, a more details commit message is nice to have :) you will get a comment on that :)
<oliv3r> wingrime: AXP? i'm talking about USB EHCI
<oliv3r> wingrime: the FET is triggered via a GPIO, which is controlledv ia the regulator framework
<oliv3r> arokux: i'm not sure if i'm too happy with the usb1 -> usb0 alias
<oliv3r> but i'll leave maxime to comment on that as he knows it all much better
rz2k has quit []
<oliv3r> arokux: Allwinner sunxi on-chip/onboard EHCI support :)
<oliv3r> to us sunxi is 'normal' to others it may be confusing
<oliv3r> +#include "ehci-sunxi.c"
<oliv3r> that's a big fat nono
<oliv3r> your list of includes looks awefully shor
<arokux> oliv3r, this is how it should be done, however.
<oliv3r> including a C file?
<arokux> oliv3r, yes.
<oliv3r> O.o
<arokux> oliv3r, check out ehci-hcd.c
<oliv3r> i am :)
<arokux> oliv3r, "your list of includes looks awefully shor" mm.. could be, forgot to clean it
<wingrime> arokux: sort includes list in alphabet order
<arokux> wingrime, is there a requirement like this??
<oliv3r> but as far as I know, you should never, ever include a C file :)
<libv> oliv3r: depends
<libv> oliv3r: i sometimes do that for big tables
<arokux> oliv3r, from every rule there is an exception...
<oliv3r> libv: exceptions prove the rule :)
<oliv3r> arokux: I see, oh that's fugly :(
<wingrime> libv: if you have time you can review my new patches, still no comments
<oliv3r> wingrime: which ones?
<wingrime> oliv3r: for disp
<arokux> oliv3r, patches against mainline usb subsystem are welcome, I think :p
<wingrime> oliv3r: delay replace patches, cos/sin table divide . tvenc regs
<oliv3r> wingrime: ah ok i replied to some of those; and libv did reply to some of them too?
<oliv3r> wingrime: the cos/sin table devision sounds sensible i suppose
<oliv3r> tvenc regs you never replied wether we really need _REG suffic
<oliv3r> suffix*
<oliv3r> arokux: yeah I guess it would require some cleanup; cause right now it's very ugly :)
<wingrime> oliv3r: _REG suff for somethins make it mainline
<oliv3r> arokux: i think OHCI did get cleaned up recently
<oliv3r> wingrime: mainline suggest adding _REG to register defines?
<oliv3r> lemme se some of maxime's patches
<wingrime> oliv3r: no, but I think SUNXI_TVENC_ names space may include not only regs
<oliv3r> he actaully renamed _REG registers and dropped it i think
<oliv3r> wingrime: personally, I think the rename is good; but _REG is a little too much
fldc has joined #linux-sunxi
forcev is now known as FunkyPenguin
FunkyPenguin has quit [Changing host]
FunkyPenguin has joined #linux-sunxi
<oliv3r> arokux: rename your defines to SUNXI_CCM_AHBMOD_OFFSET, not SW_VA_
<wingrime> oliv3r: than send ACK, I will contine tvenc hacking
<arokux> oliv3r, have you read the comment to that function?
<oliv3r> arokux: i'd rename all USB to USB0 and USB1, as that's logical; I know the datasheet uses USB1 and USB2 for host, as it names USB0 for gadget, but in the host driver itself, it should be just what it is imo, USB0 and USB1
<arokux> oliv3r, mainline ohci uses the same "include C file" scheme.
<oliv3r> arokux: yeah that's why i said 'omg that's fugly' :)
<arokux> oliv3r, but then you said: <oliv3r> arokux: i think OHCI did get cleaned up recently
<oliv3r> arokux: and your comment wasn't very clear on what the crap was :)
<oliv3r> arokux: well i saw the seperation between platform and PCI seperation for OCHI, so SOMETHING was done ther e;)
<arokux> oliv3r, I would like you to post your comments in e-mail reply to my patches, this is much more effective. and we can keep track of the issues.
<oliv3r> arokux: make sure you tripple check that writeb actually works as expected. I know for sid, the registers are read-aligned on 32 bit boundries so you _have_ read 32 bits at a time
<oliv3r> arokux: i don't have them (yet?)
<arokux> oliv3r, re: writeb: ok.
<Amnience> what's the speed of the USB0 port when using it as a host?
<arokux> oliv3r, ok, I think you'll get them.. or use gmane to reply. irc isn't appropriate place for comments like this.
<Amnience> I'm getting 24MB/s download and only 12MB/s upload
<oliv3r> arokux: ew magic values! :p 'magnitute' :p
<oliv3r> Amnience: USB ethernet?
<Amnience> no, removable storage device
<arokux> Amnience, tell me how to measure it and I will
<Amnience> i.e. USB HDD
<oliv3r> Amnience: that doesn't sound too bad
<oliv3r> Amnience: did you try the regular host port?
<arokux> Amnience, your USB HDD can be a bottleneck too...
<Amnience> erm dd if=/dev/zero of=/mnt/disk/test.img bs=16k count=8192
<Amnience> regular host ports work fine
<Amnience> 24/24 both ways with dd
<oliv3r> Amnience: it's quite possible that the OTG port is slower, are you using 3.4 kernel?
<Amnience> I was copying a 100MB file ftom tmpfs to disk and back
<Amnience> 3.4.61
<oliv3r> Amnience: did you flush your cashes inbetween?
<oliv3r> Amnience: well the mUSB is probably better; but that's only in mainline
<Amnience> I copied a different file
<Amnience> bottom line is this USB0 only uploads 12MB/s from device to HDD.
<arokux> Amnience, we were discussing USB host by the way.....
popolon has joined #linux-sunxi
atiti has quit [Remote host closed the connection]
<Amnience> Question. Could it be the USB cable?
atiti has joined #linux-sunxi
<Amnience> I got this USB mini B to host A from ebay
<Amnience> I use it to interface with a sata HDD dock via another USB A to B cable
<Amnience> When I plug in the cable from ebay into cubieboard
<Amnience> I notice some dmesg output
<Amnience> like a host driver being loaded or something along those lines
<Amnience> doesn't happen when I plug a normal USB A to mini-B cable
<oliv3r> could be a ton of things, but I would personally be not supprised at all, if OTG is slower
<ssvb> panda84kde: yes line 284 of Xorg.0.log at http://pastebin.com/QWh3jYVV is definitely a performance problem
<ssvb> soldoKyn: setting fb0_framebuffer_num >= 3 in the fex file is one tweak
<Amnience> I'm going to try to rule out the cable by making my own cable with no hidden circuitry
<Amnience> then again I was wondering... I saw there are some options for OTG support in the kernel config
<ssvb> soldoKyn: but also the r3p0 mali drivers can use only one PP core instead of two in Mali400 MP2, and the mali clock frequency is set a bit too low right now
<Amnience> and I saw some options to configure USB0 in the FEX file
<Amnience> do you think changing the fex options to host only along with the kernel will help?
<oliv3r> Amnience: i don't think we use the musb driver in 3.4 yet; so that driver is qualitivly poor to begin with; i wouldn't invest too much time there
<oliv3r> Amnience: fex options probably only influence poweron GPIO etc
<rm> <Amnience> bottom line is this USB0 only uploads 12MB/s from device to HDD.
<rm> that's correct
<rm> cable is unrelated
<rm> the OTG port is speed-limited and some devices (like webcams) may not work in it at all
<Amnience> is there a way to use it host-only without the limits?
<Amnience> yeah, couldn't get an external usb sound card to work on it.
<rm> says: "This is a hardware limitation which we
<rm> will likely not be able to work around."
<rm> so maybe we can't
<Amnience> hmm, well that's good to know so that I won't waste time on pointless things
<arokux> rm, is this hardware limitation?
<rm> hansg say so
<Amnience> though why is the speed limited only in one direction?
<oliv3r> hardware error
<oliv3r> :p
<oliv3r> design mistake :)
<Amnience> :-/
<tgaz> is there a list of known issues with usb0 somewhere?
<Turl> maybe it has no DMA implemented
<Turl> the driver is pretty crappy after all
<oliv3r> with the mbus driver, it might be worth investigating
<Turl> oliv3r: does your ahci driver work btw? :p
<oliv3r> Turl: well i still need to add the dt
<oliv3r> need your clocks
<oliv3r> and lots of time to actually test it
<oliv3r> but it shoudl work, in theory
<Turl> oliv3r: clocks are there on sunxi-clk branch waiting for you :p
<oliv3r> Turl: :p
<oliv3r> then there is lack of time
<oliv3r> very busy $work this month
<oliv3r> next month should be good again
<oliv3r> hopefully next week
FDCX_ has quit [Ping timeout: 264 seconds]
\\Mr_C\\ has joined #linux-sunxi
<ssvb> panda84kde: mali can run up to 381mhz according to the documentation, and up to 408mhz based on my tests with overclocking (but we should always keep some safety margin, so ~380mhz or even a bit less seems reasonable)
<ssvb> panda84kde: right now mali is clocked only at 312mhz in cubieboard2 (pll4 with divisor 1)
wingrime has quit [Quit: Lost terminal]
<panda84kde> ssvb: I see. Should I look at ccmu_regs.h? http://www.sysfwlab.com/?p=345
<ssvb> panda84kde: I'm going to clean up the mali kernel code and resolve all these issues :)
atiti has quit [Read error: Operation timed out]
<panda84kde> ssvb: wonderful, my hero :) Is it something high priority in your list?
<ssvb> panda84kde: yes, I'm working on it now, the problems are: sane clock settings for sun5i hardware (techn_ has been doing some tests with it), and upgrade to r3p2-01rel2 without breaking Android
<oliv3r> arokux: instead of using BIT(10) /* comment */ make a define for it :) #define SUNXI_EHCI_AHB_ICHR8_EN BIT(10)
<panda84kde> ssvb: great! I'll keep an eye on the mailing list then. Good luck and my sincere thanks for your great work!
Montjoie has joined #linux-sunxi
<arokux> oliv3r, ok
<oliv3r> arokux: your failing of cloks; i'd check if all 3 work; then just fail all 3 in one go
<oliv3r> you only need to enable clks to make it work?
<Montjoie> hello, from which version can I use vanilla kernel on cubieboard2(A20) thanks?
<oliv3r> Montjoie: 3.11/3.12
<oliv3r> Montjoie: but you won't have much support, only initramfs + enthernet
<arokux> oliv3r, sunxi_ehci_enable should tell you what is needed.
<oliv3r> arokux: yeah that's the function i talk about
<Montjoie> So for SATA I still need linux-sunxi sources ?
pitillo has joined #linux-sunxi
pitillo has quit [Changing host]
<oliv3r> arokux: you only have to enable clocks; normally they tend to have an enable bit etc or something
<oliv3r> Montjoie: if you want a usable kernel; yes, 3.4
<arokux> oliv3r wingrime: Ian Cambell on ML has asked about SATA
<oliv3r> Montjoie: though i RFCed very early sata a few days ago
<oliv3r> arokux: allready replied :)
<arokux> oliv3r, yes, this is one-bit set, you mean I do not need "prepare"?
<oliv3r> arokux: well you need to prepare your clocks etc, but in 'enable' i'd expect the enable bit to be set aswell i guess
<arokux> oliv3r, everything is set by clk_prepare_enable
<oliv3r> if its only about clocks, sunxi_ehci_enable_clk() would make more sense?
<arokux> oliv3r, sunxi_ehci_enable is not only about clocks.
<oliv3r> ah yeah it calls init
<oliv3r> hmm that's confusing :)
<arokux> oliv3r, it calls whole lot of stuff.
<oliv3r> yeha i see now, i overlooked that
<oliv3r> and sets up your regulator \o/
Black_Horseman has quit [Remote host closed the connection]
<oliv3r> arokux: you forgot module_author :)
<oliv3r> brb
<arokux> oliv3r, have you got the e-mails with the patches? :p your comment are extremely welcome, but better as reply to the patches.
atiti has joined #linux-sunxi
<oliv3r> i still don't have them
<arokux> oliv3r, :$
<oliv3r> use module_platform_driver() Instead of module_exit/init
<arokux> oliv3r, check it out, it won't fit my needs.
<oliv3r> maybe your needs are wrong :)
<oliv3r> but could be yeah
<oliv3r> USB looks like an ugly beast anyway
<arokux> oliv3r, no, they are not. everything is fine with this particular issue.
<arokux> oliv3r, smth isn't ugly only because you haven't see / used it the way it is used.
<arokux> seen*
<oliv3r> the C-includes is really guly
<oliv3r> i know it's valid C, but it's really ugly
gnufan has joined #linux-sunxi
<oliv3r> popolon: what's that
<popolon> a library with optimized functions for neon
<popolon> made by arm
<popolon> found here
<popolon> * Math - Vector/Matrix
<popolon> * DSP - FFT/IFFT/FIR/IIR
<popolon> * Imgproc - Image resize/rotate
<popolon> * ARM v8(64bit)
<popolon> * OpenCL
<popolon> * Physics engine
<ssvb> popolon: who is already using it? is it better than the existing traditional free software libraries?
<popolon> that's free
<ssvb> emphasis is on traditional
<popolon> BSD 3-clause
<ssvb> to me it seems like a solution looking for a problem
<popolon> :)
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
<arokux> oliv3r, gotos are ugly too, but there is niche for them.
<popolon> lot of asm optimized function on dsp, neon
<arokux> popolon, are we actually right ppl to talk about this library? not that we are not interested, but you may find more professional replies in linaro channels
<popolon> just found the information, and speak about
<arokux> oliv3r, you may want to backread up to this point: http://irclog.whitequark.org/linux-sunxi/2013-09-17#4985063;
<popolon> but you are right, could be more intersting on linaro
<ssvb> popolon: I would be more convinced if they could claim something like "we have replaced the shitty arm asm code in ffmpeg with our new shiny arm asm and got X% performance improvement" ;)
<oliv3r> arokux: ah _defconfi
<popolon> :)
<merbanan> ssvb: good luck writing better neon code then in FFmpeg
<soldoKyn> ssvb: thank you so much! I'll try the suggested tweaks!
<arokux> oliv3r, yes. it would be nice we set the policy and document it.
tinti_ has joined #linux-sunxi
tinti_ has joined #linux-sunxi
tinti_ has quit [Changing host]
<popolon> but ffmpeg is only for dsp functions
<popolon> not general purpose I suppose
<merbanan> correct
<oliv3r> arokux: aye, but i'm still not sure what was decided in the USB camp; i still say 'all usb devices' a random user should be able to plug their USB device in a snormal
<oliv3r> what i'm curious about, some drivers have been allwinnerified (fex)
<oliv3r> but what if I don' thave the onboard version, but an external USB stick :)
<atsampson> if you have an offboard rtl8192, then you plug it in, udev loads the module and it powers up the internal one too ;)
<ssvb> merbanan: as a matter of fact, not all the neon code in FFmpeg is perfect
<oliv3r> atsampson: doesn't the fex check for usb_para or some shit?
<merbanan> Lengths supported by the transform: Internally, the functions utilize a radix-4 decimation in frequency(DIF) algorithm and the size of the FFT supported are of the lengths [16, 64, 256, 1024].
<merbanan> not impressed
<arokux> oliv3r, the people including me tend to agree that the _defconfigs should be for all: tablets and boards. also it should include modules for the hardware that is supplied by EOMs - wifi modules etc.
<oliv3r> if I hack on it, and i build my own kernel; I know how to do make menuconfig and idsable what i don't want
<oliv3r> joe user, who just wants to compile his kernel because its cool; wants a working _defconfig
<oliv3r> distro builder, could run a custom kernel config, but most of the time would rely on the joe user config anyway
<arokux> oliv3r, sunxi_defconfig should not include support for all sound cards out there, do you agree?
<oliv3r> arokux: only USB sound cards + soc_sunxi
<oliv3r> and possibly all i2s ones :p
<arokux> oliv3r, you are serious?
<ssvb> merbanan, popolon: still I just mean that this Ne10 has to prove its usefulness in real software, with real benchmarks demonstrating the improvements
<popolon> yes, just need to know its existence to do benchmarks :)
<merbanan> popolon: there is no neon assembly in this lib
<popolon> I seen asm at least in dsp part
<popolon> well this is neon asm I suppose
Amnience has quit [Quit: Leaving]
<merbanan> there we go
<popolon> :)
<popolon> there are in imgproc too
<popolon> this is what claims the doc
<merbanan> might be useful afterall
<popolon> at least for efficient math function, between auto-vectorization and perhaps more optimized one ???
<ssvb> well, another similar library (liboil) did not seem to have much real success
<atsampson> if you look at what hansg's done for the fedora kernel, he's split the config up into multiple chunks...
<atsampson> you could have a "sun4i generic" config chunk and an "all USB devices" separate one?
<arokux> sun4i generic - is what we want to have in sun4i_defconfig
<arokux> at least me, hno wingrime and steev :)
<oliv3r> arokux: of course; my A10 device doesn't have a 3'5mm jack, so i can't connect it to my stereo, but I do have an USB sound card
<oliv3r> arokux: just some random use case
<oliv3r> arokux: but i can give you a few more :)
<oliv3r> there's no reason to limit the amount of modules users have available
<arokux> oliv3r, well for me that is not a problem as modules won't blow up the kernel image. however me and I think most ppl here have no idea what USB sound cards should modules be built for. some of them are not used anymore. also, you cannot see *all* the use cases ppl will have. the other way i.e. including the most probable hardware support seems to be more reasonable. it is not a distro config after all.
ganbold_ has joined #linux-sunxi
<oliv3r> exactly, you don't know what users will want or need
<arokux> oliv3r, so? should we build a distro-like config?
<arokux> how come EOMA68-A20 isn't open source hardware???
<buZz> why should it be?
Mithrian has joined #linux-sunxi
<arokux> buZz, mm.. it should't I just thought it is, because the community is backing it so much
<buZz> the community is backing using proper power supplies, still, not many power supplies are open source hardware ;)
<arokux> buZz, I meant irc, mailing lists and stuff
<buZz> yes
<buZz> i know what community means
<popolon> didn't know liboil, that's used by MLT
<popolon> and so gstreamer
<arokux> oliv3r, it will then be called "distro-config", we however want to come up with sunXi_defconfig
<oliv3r> arokux: yes, but USB opens a load of options for drivers
<oliv3r> since every user can pickup any USB device and just plug it in
<oliv3r> why would you NOT wanna put it in? you even said so yourself, it only takes some extra initial compile time, and it's only a little bit of filesystem space for the modules
<oliv3r> it doesn't really get in the way
<arokux> oliv3r, you are 100% right and I agree.
<arokux> oliv3r, so maybe we just build everything that depends on USB? or how should we proceed?
<oliv3r> yeah i'd think so, any usb driver should be M
<oliv3r> for the rest, there aren't as high as a requirement
<arokux> oliv3r, following your logic we should build support for all major filesystems too.
<oliv3r> i suppose so yes :)
<oliv3r> but only ext4 as Y for now
<arokux> oliv3r, why should be any fs be built as Y?
<oliv3r> arokux: so you can boot your kernel?
<tgaz> with ramdisk, that isn't really needed, right?
<arokux> oliv3r, why not as module in initramfs?
<oliv3r> but i don't want an initramfs! :p
<oliv3r> but FS is probably a lesser issue
<oliv3r> most users don't randomly plug in a special fs
<oliv3r> it'll be either fat, or ext most of the time, if not nearly always
<oliv3r> but yeah, in theory, all modules
<oliv3r> but then the debate goes hardware vs other features, a fs isn't a hardware driver I guess :)
<oliv3r> but athat's up for debate yeah
<oliv3r> arokux: but yeah there is a line that needs to be drawn somewhere of course
<arokux> oliv3r, well either it is a distro-like config or a more minimal config.. :) you want it to be distro like. it is ok for me.the only problem I see is to predict what devices will be connected to the boards.
<oliv3r> well we can't include every possible driver, tehre's GPIO drivers that we can't anticipate, but USB, we can anticipate that anything is game, since it's just plug and play
<oliv3r> also, GPIO drivers are board specific really, you don't really plug that normally into a STB or tablet
deasy has joined #linux-sunxi
ZetaNeta has joined #linux-sunxi
ZetaNeta has joined #linux-sunxi
<ZetaNeta> afaik A13 is pincomaptible with A10?
<oliv3r> ZetaNeta: no
<oliv3r> A13 -> A23; A10 - > A20
<ZetaNeta> Well, i remember there were 2 allwinners which are
<ZetaNeta> oh.....
<ZetaNeta> 4 allwinners :D
<libv> 5 actually, but the 5th is the uncle we do not like to talk about :p
<ZetaNeta> Can you atleast call the name?
<oliv3r> A31 and A31s
<libv> well, due to lifestyle choices, that uncle contracted a rather nasty disease
<oliv3r> the uncles retarded problem
<specing> PowerVR?
<oliv3r> brother!
<libv> specing: sssh!
<ZetaNeta> I am doing a list of "unique retarded features" i am going to implement in my ARM book
<arokux> does this PowerVR have at least good performance?
<libv> arokux: perhaps, but as with things powervr, it might be even harder to get at
<libv> this performance
<oliv3r> arokux: who cares :)
<oliv3r> don't count on linux binaries; android only if your really lucky
<arokux> oliv3r, you seem to be very biased. users who buy tablets care.
<oliv3r> ZetaNeta: EOMA68-A20 netbook :)
<Montjoie> Does someone have a working .config for linux-sunxi kernel sources for Cubieboard2 (A20)
<oliv3r> arokux: PowerVR is closed source propriatary crap
<specing> arokux: doubt it
<oliv3r> arokux: if you are 'okay' with that, then your choices are weird :)
<specing> arokux: if they did, they would buy Intel tablts
<oliv3r> specing: or samsung tablets :p
<ZetaNeta> oliv3r, EOMA68 is the one of the greatest inventions ever
<ZetaNeta> in arm industry
<specing> oliv3r: samsung has good chips underneath?
<oliv3r> specing: if your lucky and get an exynos
<specing> My friend is always bitching about the thing being a clusterfuck
<oliv3r> exynos isn't really bad is it?
<specing> its still ARM
<oliv3r> specing: hehe, yeah i know team-hacksung and replicant aren't tooo happy with samsung/exynos
<ZetaNeta> oliv3r, Is allwinner any open? Can i write that it is, or not for the glory of satan (strikethrough) for marketing purposes?
<arokux> specing, well price vs perf
<oliv3r> ZetaNeta: a10, A13, a20 and a23 could be considerd 100% open
<oliv3r> ZetaNeta: after we're done RE-ing stuff :p
<specing> oliv3r: lmao @team-hacksung
<ZetaNeta> :DDDDD
<oliv3r> ZetaNeta: like lima and cedarX
<specing> arokux: think about all the time you waste hacking on ARMs
<oliv3r> a31 has powerVR, which is a huge problem :)
<oliv3r> also, always remember, libv said, powerVR is the biggest utter crap of GPU's :)
<specing> arokux: if I looked at the difference between an ARM tablet that I bought and an Intel one, its only 200$, and if I look at how much my time is worth hacking those (and I still haven't made them work), it would be 0.1$ / hour at best
<arokux> specing, and? i'm not a "tablet user" I love to hack. but i'm not biased. for hacking I'd pick Mali, for my brother that plays 3D game i'd buy PowerVR
<libv> and, like robclark, libv is still in therapy for having been exposed to pvr
<oliv3r> libv: lol
<specing> hhahaha
<oliv3r> arokux: some people actually do care about open source and it being out in the world
<oliv3r> arokux: and powervr is probably the worst choice in that regard
<arokux> oliv3r, well you must know I do, but there are lines that should be drawn.
<oliv3r> arokux: so hell yeah i'm biased, as I care about OSS, from that warm and fuzzy inside place
<oliv3r> arokux: annhilate all propriatery users? :D
<deasy> i confirm than powervr is crap
<deasy> even their proprietary driver is a crap
<deasy> they are even not able to do performance in 2d
<deasy> (in my experience)
<arokux> oliv3r, i'm sacrificing a lot, but you sacrifice everything..
<mripard> arokux: where did you send it to?
<Turl> mripard: linux-sunxi
<oliv3r> arokux: if I have to; yeah
<merbanan> libv: is the actual hardware crap also, or only their software ?
<oliv3r> merbanan: libv worked on the actual drivers while working for ... I forgot, i think nokia
<deasy> merbanan, i think than both is crap
<deasy> as they are even themself not able to do a good driver for their hard
<deasy> (by good driver i mean decent performance at least in 2d)
<deasy> are*
<deasy> but if people want to reverse powervr , do it
<deasy> all thing reversed and open are good
gnufan has quit [Ping timeout: 264 seconds]
<deasy> and maybe someone in this world is able at least to do better 2d perf than the owner himself...
<oliv3r> Turl: where are the ahb_gates numbers defined?
<oliv3r> Turl: i bet it's not related to bit numbers :)
<arokux> deasy, why do you think powervr is bad?
<mripard> 
<deasy> arokux, because i have get one in the past...
<Turl> oliv3r: they are, read the fine binding documents :)
<deasy> and even old igp intel as 915(or something like it) are better in 2d
<oliv3r> Turl: i was looking at the wrong kernel dir so couldn't find them :p
<deasy> (far better)
<oliv3r> Turl: hmm, do we even know ahci muxes? is it even mux or something that's just always tehre'
<deasy> their driver give the sensation to be slower than vesa rendering...
<oliv3r> Turl: looks like they are dedicated pins!
<deasy> <=== poulsbo user
<deasy> all poulsbo user can say than powervr is crap
<arokux> deasy, and now you nave a Mali thingy and see it is better?
<deasy> i have not hdmi screen :)
<arokux> mripard, ?
<deasy> when i get one i can say you it
<Turl> oliv3r: let me check the wiki :p
<deasy> but the fact than powervr is against open driver as nvidia is not encouraging for buy their crap
<arokux> oliv3r, I just try to be more pragmatic you know. that is all. i see it the same way Torlvalds saw it while using BitKeeper
<oliv3r> yeah T20, T21, U20 and U21 aren't defined on the wiki; i checked he olimexino schematic for those
<oliv3r> arokux: yeah and torvalds got flamed for that plenty :) and he corrected his ways :)
<oliv3r> and its no secret that torvalds isn't really a HUGE oss enthousiast :)
<arokux> oliv3r, not because of flames ;)
<oliv3r> arokux: think more like RMS
<deasy> oliv3r, links! or it not happens
<arokux> oliv3r, how do you think programmers will earn there livings by doing everything open source?!
<deasy> oh no sorry
<deasy> misunderstand
<oliv3r> arokux: i'm sure all these kernel devs must be dirt poor by now
<deasy> it's ok
<Turl> oliv3r: maybe mripard recalls, he wrote all of pinctrl :)
<oliv3r> Turl: i think it's fixed pins
<oliv3r> Turl: like usb etc
<Turl> oliv3r: yeah, poor people with chromebook pixels :)
<oliv3r> arokux: you CAN make money using OSS; personally I think oss is the only proper future :)
<oliv3r> but i also think knowldge should be available to anyone willing to learn
<arokux> oliv3r, why you do not do it then? :p
<oliv3r> an education should not be something only the rich can afford
<oliv3r> arokux: look at mripard ; how do you think he earns his living?
<oliv3r> arokux: because i'm still a student :)
<arokux> oliv3r, not everything can be open source -- this is my point.
<oliv3r> arokux: yes it can :)
<deasy> of course it can
<arokux> oliv3r, how can something exactly like Photoshop, AutoCAD etc. be open source?
<deasy> and for an hacker...everything MUST be open...
<oliv3r> arokux: ask the gimp developers
<deasy> except if he like challenge :D
<oliv3r> deasy: also when you want security, it has to be
<arokux> oliv3r, gimp != photoshop
<oliv3r> propriatery software is by default insecure
<deasy> i know oliv3r
<oliv3r> arokux: that's all momentum
<oliv3r> arokux: hollywood uses gimp for their movies
<deasy> arokux, how? simple...be open source that's all
<oliv3r> arokux: well they forked it to filmgimp which is now ... i forgot tis name, but still opensource
<deasy> it's not as it hard thing to do
<oliv3r> arokux: REAL software, opensource OR closed, gets licensed b use
<oliv3r> arokux: e.g. a company using your product has to take a license for using it
<oliv3r> arokux: look at red-hat, you have to pay big money to use red-hat inside a company
<arokux> you simply cannot provide this kind of support and documentation quality. take a look at Matlab, for example
<deasy> in multimedia the better exemple of piece of software is Blender
<oliv3r> arokux: take a look at red-hat linux
<deasy> Blender is used in France for do LOT of publicity
<deasy> you can't imagine how much
<oliv3r> if you have a good product, you make companies buy a license to use your prodcut, it can sitll be fully open source
<oliv3r> worst case, dual license, GPL + commercial license
<deasy> or buy support ;)
<arokux> oliv3r, how is red hat open source and I still need to pay for its use? cannot i just build it from source and use it?
<oliv3r> arokux: you can, but you can't ask red-hat for support
<deasy> most case is buy for support
<oliv3r> arokux: but if you, as a use want that, you use cent-os :)
<oliv3r> arokux: but how do you think red-hat became a billion dollar company?
<deasy> arokux, you can, centos is rhel open
<deasy> the sources are available
<deasy> just you have no support
<oliv3r> but as a company, you woul dnever run cent-os
<deasy> when you paid redhat, you get support
<oliv3r> you want to be able to call red-hat and say 'look, i have this problem, fix it'
<arokux> oliv3r, by offering support - that one I know. but they've used software that was developed prior to their existence
<oliv3r> but I don't know the rhel/centos licensing
<deasy> oliv3r, an internet provider here use centos for hosting ;)
<oliv3r> they could very well state 'cooporate users have to pay licensing fee's
<arokux> oliv3r, do you think the could start from scratch with oss?
<deasy> if you don't need support, as you have your internal techies
<oliv3r> deasy: exactly
<oliv3r> arokux: just dual license your prodcut, say 'GPL for private use, for commercial uses inquire by email'
<oliv3r> arokux: IF a company decides to use it anyway, they are just as illegally using your software, as if they would download photoshop
<oliv3r> which is illegal aswell
<Turl> arokux: have a look at asterisk
<deasy> my sentence was wrong
<deasy> CentOS is a build of sources RHEL
<ZetaNeta> How aprox is the price for A31? And how different they are from A31s?
<deasy> oliv3r, most part of rhel is under GPL
<arokux> ZetaNeta, check olimex website, look for components on the left
<ZetaNeta> tnx
<mripard> arokux: seriously, since when linux-sunxi has become the proper mailing list to send mainline patches? Not even CCing the actual maintainers and mailing lists
<Turl> ZetaNeta: and remember than buying direct from china/in bulk is cheaper :p look at aliexpress
<mripard> you will *never* get a patch merged that way.
<Turl> mripard: I think he was just looking for a pre-review to clean up any big issues
<Turl> considering the TODOs and FIXMEs
<mripard> oliv3r: I don't really recall wether the USB pins are muxed or not
<mripard> I think there is at least some of them iirc, but I'm not definitive
<arokux> mripard, yes, Turl is right, if you say it is wrong, then I'll resend.
<mripard> Turl: arokux: well, either the patches are ready to be mainlined, or there not. If they are, then send them to the proper people. If there not, don't claim it is then.
<Turl> mripard: I don't recall anyone claiming so
<ZetaNeta> oliv3r, Would you pay for a "cheaper than EEE PC" netbook, that have 2 motherboards with "Quick Switch" stuff?
<ZetaNeta> I mean, 2 EOMA's in a netbook
<Turl> ZetaNeta: I recall someone doing that
<Turl> they added an arm board inside a notebook and had kind of a switching mechanism
<arokux> mripard, they are ready from my point of view. next is mailing list. I've thought ours ML will be a good pre-mainline mailing list check
<mripard> Turl: well, I don't see anywhere in its mail or IRC ping a mention that it's not usable/has some drawbacks
<ZetaNeta> Turl, You mean the IBM?
<arokux> mripard, do not be angry plz, I'll fix things ppl told me sofar and will post patches to proper mainline lists.
<ZetaNeta> When they done "Dual CPU" thing. That allowed you quickly switch to a lightweight less battery intensive linux
<arokux> mripard, and next patches will go to mainline lists too.
docice has quit [Quit: docice]
<arokux> mripard, or do you want to give me some comment before I sent patches to mainline lists?
<mripard> arokux: I'm not angry at you specifically, it's just a pattern that arose recently, and you're just the last example of it.
<deasy> arokux, thank you for code :p
<arokux> mripard, yes, I remember you complaining. I should say I'm a bit too shy to send to mainline lists directly :)
<mripard> arokux: and since you're the only one that really has a clue about what you've been doing
<deasy> you are shy to get a flame :p
<mripard> the only in depth and meaningful review you'll get will be by the USB maintainers
<deasy> Linus is goind to eat you hehehe
<deasy> going*
<mripard> deasy: honestly, this is just a tale to frighten young kernel developpers before they get to sleep
<arokux> mripard, ok, patches will be at the proper mailing lists tonight.
<mripard> arokux: other than that, wingrime's right, use your real name
<mripard> make it through checkpatch.pl
<Turl> ZetaNeta: no, I mean some dude that made like an arm board you put inside a certain kind of notebook and made it bootable from that small board or normal computer
<arokux> mripard, there was P.S. - I was going to.
<ZetaNeta> Turl, Canonical?
<mripard> and give a reference to the name of the patch serie you're using
<mripard> arokux: PS?
<Turl> ZetaNeta: no, it was not corporate
<Turl> ZetaNeta: can't find it now
<arokux> mripard, post scriptum in my e-mail.
<mripard> ah, right, sorry then :)
<ZetaNeta> Turl,Artem Kyznecov?
<arokux> mripard, no problem, nobody noticed it :D
<Turl> ZetaNeta: looks like a cracker? dunno :p
<Turl> ZetaNeta: found it! http://cuppcomputing.com/?p=501
<ZetaNeta> Close to what IBM did
<ZetaNeta> Exept, they have done it built in :D
naobsd has joined #linux-sunxi
<tgaz> is there a list of a20 coprocessors and their numbers anywhere?
<arokux> tgaz, our main wiki page?
<tgaz> arokux: i can't find the number assignments. (like p15 is vmmu)
<arokux> tgaz, what are those numbers p15?
<Turl> tgaz: look for cortex-A7 ones
<Turl> tgaz: they should be standarized
<tgaz> arokux: the number used in the machine code to address the different subsystems
<tgaz> Turl: will do. thanks.
<Turl> tgaz: I bet you can download a manual from ARM's site with them
ganbold__ has joined #linux-sunxi
ganbold_ has quit [Ping timeout: 245 seconds]
tzafrir_laptop has quit [Ping timeout: 264 seconds]
<tgaz> (cp0-cp7 implementation defined. cp10-11 FPU. cp14 debug. cp15 system control. it seems.)
<ZetaNeta> Is there any other EOMA boards exept the one oliv3r linked to
<arokux> ZetaNeta, which one did oliv3r show to you?
<ZetaNeta> The A10 one
<arokux> ZetaNeta, you mean the cpu card or the board you plug it into?
<arokux> ZetaNeta, ok, this is the only one so far.
ykchavan has joined #linux-sunxi
<ZetaNeta> :P
<ZetaNeta> Well, monopoly is also good
<ZetaNeta> Well, its even better.....
<arokux> ZetaNeta, you are the guy that wants to create a netbook?
<ZetaNeta> Yup
<ZetaNeta> I wonder, what are the sizes of the Card, and the sizes of the motherboard you plug it in
rellla has quit [Quit: Nettalk6 - www.ntalk.de]
<arokux> ZetaNeta, haven't you seen this? http://rhombus-tech.net/community_ideas/kde_tablet/news/
<ZetaNeta> otnx
<soldoKyn> ssvb: after the 2 tweaks (fb0_framebuffer_num, sunxi_fb_mem_reserve) now my Xorg.log seems ok http://pastebin.com/uPrfiBJq . The results from glmark2 are http://pastebin.com/A2KxcXVb (a 800x600 window session and a 1280x720 fullscreen one). Are this results as expected or maybe too low (your results on Mele A2000 much higher)?
<soldoKyn> Also I noticed from glmark2 log some errors...also these are expected or I might have done something wrong?
<libv> mapbuffer is an implementation thing
<libv> vertex texture fetch is a hw thing, it is not manadatory by the spec
<libv> the mapbuffer issue should disappear with the free driver, at one point
<libv> the other is never going to get fixed
<ZetaNeta> arokux, Does it have nand on it? Or SD card is the only memory?
<arokux> ZetaNeta, i'm not its designer, lkcl_ is.
<ZetaNeta> :D
<soldoKyn> libv: ok thanks! do you think that my glmark2 results are good?
<arokux> ZetaNeta, so when should we expect the netbook?
<ZetaNeta> arokux, Step 1: Get 2000 euros, which i should get in the end of the year
<ZetaNeta> Step 2: Invest into a 3d printer
<ZetaNeta> Step 3: Finish the design of the netbook
<ZetaNeta> With all the cool features
<arokux> ZetaNeta, why not use online 3D printing service, just to check your idea works. for mass production you can then buy a 3D printer.
<ZetaNeta> Step 4: Realize that this is out of expected price for netbook.
<ZetaNeta> Step 5: Remove half of the features
<ZetaNeta> So.... You probably understood
<ZetaNeta> "It will take a while"
<ssvb> soldoKyn: it gets synchronized with monitor vertical refresh, you can change option "SwapbuffersWait" to "false" in /etc/X11/xorg.conf to run at full speed
FDCX_ has joined #linux-sunxi
<ssvb> soldoKyn: though it's a bit strange that some of the tests are showing FPS as 53 or 55, they all should normally be close to 60
<arokux> ZetaNeta, do you have some design already?
<ZetaNeta> arokux, I got a long long list of ideas to implement
<ZetaNeta> i wonder if its easy to have a Win8 on arm
<ZetaNeta> Becouse if it will support windows. Sells will be much higher
<ZetaNeta> If it will support winCE/win8
<ZetaNeta> Also, i think about doing a kickstarter thing. Becouse there are quite a bunch of guys who "Reinvent LiveUSB drives", and actualy get money there.
<ZetaNeta> And that may "make my dreams come true", and allow me to order "more details", thus making the device cheaper to produce
<ZetaNeta> + A flow of customers!
<ZetaNeta> + Unspent money.......
<ssvb> libv, soldoKyn: nice, running strings on the new mali x11 blobs seems to shows some use of DRI2SwapInterval, which means that probably benchmark programs can be made to run fast even without xorg.conf tweaks
<libv> ssvb: me no likey
<libv> i got the horse spinning btw
<ssvb> older r3p0 mali blobs did not have any references to DRI2SwapInterval
<ZetaNeta> arokux, How do you think, how long a "x86 laptop" sized battery will keep the arm
<ZetaNeta> day, 2? Month?
<arokux> ZetaNeta, just do some calculation
* ZetaNeta can.....
<ZetaNeta> but
* ZetaNeta is lazy
<buZz> yes math
atiti has quit [Ping timeout: 276 seconds]
<buZz> do it
<ssvb> soldoKyn: what is your memory clock frequency? also please let me know the glmark2-es2 score with SwapbuffersWait=false
<ssvb> soldoKyn: some days ago I have added a "troubleshooting" section in the readme at https://github.com/ssvb/xf86-video-sunxifb (which should cover the basic performance tuning)
<ssvb> soldoKyn: btw, what kind of x11 desktop environment are you using?
<ssvb> soldoKyn: I'm just afraid that the 53 fps problem might be caused by compositing
ganbold__ has quit [Remote host closed the connection]
<ssvb> soldoKyn: or the xorg config might be actually in /usr/share/X11/xorg.conf.d
<arokux> how is this neat command called that looks for maintainers for the patch?
<soldoKyn> ssvb: with swapbufferswait set to false the results are closer to yours on a10/a13... 113 for 800x600 window and 73 in 1280x720 fullscreen and like mentioned in your readme file some individual test run with more than 200 fps...
<soldoKyn> ssvb: I'm using xfce4 with openbox...
<soldoKyn> ssvb: where I can get the memory clock freq? from dmesg?
<ssvb> soldoKyn: the results look ok for r3p0 mali drivers, r3p2-01rel2 drivers are going to be faster
<ssvb> soldoKyn: for getting the memory clock speed you can use https://github.com/maxnet/a10-meminfo
TheSeven has quit [Ping timeout: 264 seconds]
TheSeven has joined #linux-sunxi
n01 has quit [Remote host closed the connection]
<soldoKyn> ssvb: meminfo http://pastebin.com/MCvFuvEF
iamfrankenstein has quit [Ping timeout: 268 seconds]
iamfrankenstein has joined #linux-sunxi
<ssvb> soldoKyn: you have rather conservative memory settings (384mhz clock frequency, cas=9), which explains the glmark score not going up to 160
<arokux> Turl, how do I need to credit allwinner code?
<arokux> Turl, This code is based on the sources provided by .....
<arokux> hno, mnemoc, maybe you have ideas?
<Turl> arokux: you need to keep the original copyright notice
<Turl> you can prepend "Based on blah.c, " if you want
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<arokux> Turl, this one: http://sprunge.us/YGEf ?
<Turl> arokux: it'd end up looking something like this I guess http://sprunge.us/EJIQ
<soldoKyn> ssvb: I'm happy that the glmark2 performance are good although limited by memory speed!! Thank you for you precious suggestions and for the time spent!
<soldoKyn> libv: thank you too!!
<arokux> Turl, thanks a lot! do you think saying "Based on sunxi-ehci.c" is enough? ppl would ask themselves where this sunxi-ehci.c is... maybe Based should be really omitted?
<ssvb> soldoKyn: there is one more thing to check, you can try "performance" cpufreq govornor instead of "ondemand" (with the gpu benchmark, the cpu is used only sporadically and probably sits on a low clock frequency)
<Turl> arokux: yeah, you can omit that I suppose
<mripard> arokux: I usually just say "Based on code " and the allwinner copyrightt
<ssvb> soldoKyn: this may potentially improve the glmark score a bit
<mripard> it's enough
sioxs has joined #linux-sunxi
<arokux> thanks mripard
<sioxs> hi everyone! just got a Android A10-20 v4.2.2 with sun7i ARM7 and have a few issues I can't seem to solve any help would be appreciated setting up the box
<sioxs> it seems the device has group permissions set and I can't figure out how to add the xbmc user to the media_rw group
<buZz> edit /etc/group ?
<sioxs> any device that gets attached can be read but not written to how do i change this behavior?
<sioxs> there is no /etc/group file in this box
<buZz> well then no clue
<sioxs> is it lust a matter of adding the file and adding the content?
<buZz> no!
<buZz> are you even running linux?
<sioxs> I'm not familiar with this platform but know linux somewhat
<sioxs> yes the Android is a linux kernel v 3.3
<buZz> ahhh android
<buZz> i think the /etc folder is in /system somewhere
<buZz> not 100% sure
<buZz> and usually you are not allowed to write to it
<buZz> also, adding users to groups shouldnt be needed on android
<sioxs> a teminal only uses sh but busybox is installed and chown or the like doesn't write the changes as the permissions are set ----rwxr-x
<Mithrian> Anyone having trouble with USB0 in host mode after reboot?
<Mithrian> Seems like it won't detect the device
<Mithrian> you have to remove it and replug it
<sioxs> I looked in the system folder there is a link to the root dir but no group file and i can;t find where the permissions are set for users
<sioxs> yes
<sioxs> tried almost everything including reformatting no luck
<sioxs> every device gets mounted with uid=1000 gid=1023
<techn_> ssvb: It works atleast 340Mhz .. I left there yesterday.. since I was wondering why I got only 1700 points on antutu
<techn_> ssvb: so there might be other things wrong
<sioxs> but users are set uid=1000 gid=1015
<techn_> (mbus??)
<sioxs> ?
<ssvb> techn_: interesting, is it a worse score than with lower mali clock speed?
<techn_> ssvb: I havent benchmarked that yet
<techn_> I'll try to get highest clock first
<ssvb> techn_: the slower speed is actually one of the possible symptoms of gpu failures (the picture freezes for a while, and you get "Mali PP: Hard reset of core Mali_PP0" in dmesg)
<sioxs> buZz I have root access but it still doesn't let me have write access to mounted devices - is there a way to simply add the user to the group?
<sioxs> I know but when things arn't working ssh or installing a terminal usually can solve the problem
<buZz> i have never ever fixed a problem on android with doing anything in terminal
<buZz> nor is this #android-sunxi
FR^2 has quit [Quit: Connection reset by peer]
<sioxs> Sorry didn't know there was a IRC fro #android-sunxi
<buZz> i dont either :)
<sioxs> thanks BuZz
<buZz> cause i dont run android on sunxi
<buZz> nor does anyone here i guess
<sioxs> this is a big problem for me too not too much around on the net for it - but thought being linux someone might know how I can get around it
_BJFreeman has joined #linux-sunxi
<buZz> there is #android
<buZz> and #android-dev
<sioxs> Isn't there a generic bysybox command to add a user to a group without adduser or addgroup? (not available command)
<sioxs> Thanks again I'll check it out
_BJFreeman is now known as BJfreeman
<sioxs> Cya and play safe
<buZz> have fun :)
<hno> arokux, I do not think we should require an initramfs to boot on most devices. Our defconfig is to make life convenient for us and the majority of our users.
panda84kde has quit [Quit: Konversation terminated!]
atiti has joined #linux-sunxi
_enrico_ has quit [Quit: Bye]
tzafrir has joined #linux-sunxi
<oliv3r> The popular Cyanogen Mod distribution of the Android Open Source Project dropped a bombshell today: the founding members have formed a corporation (currently with a team of seventeen hackers) to work on the project with the founder of Boost Mobile as the CEO. Quoting the announcement:
massi_ has quit [Quit: Sto andando via]
arokux2 has joined #linux-sunxi
<arokux2> hno, ok, i'm good.
Guest3689 has joined #linux-sunxi
<ssvb> soldoKyn: the performance difference between "ondemand" and "performance" cpufreq governor is actually quite measurable in glmark2-es2 (209 fps vs. 243 fps with mali r3p2-01rel2 overclocked to 408mhz)
BJfreeman has quit [Ping timeout: 240 seconds]
<ssvb> soldoKyn: err, not fps but glmark2-es2 score for running in x11 with 800x600 window
ykchavan has quit [Ping timeout: 248 seconds]
ykchavan has joined #linux-sunxi
sioxs has left #linux-sunxi [#linux-sunxi]
jemk has joined #linux-sunxi
<arokux2> Turl, can I ignore: WARNING: line over 80 characters ?
<Turl> arokux2: it depends
<arokux2> Turl, ok)
<Turl> arokux2: if it's a long string/err message, it's ok to go over 80
<Turl> to maintain grepability
<Turl> but as a rule of thumb, try not to go over 80
eebrah is now known as eebrah|doze
Guest3689 has quit [Ping timeout: 245 seconds]
<oliv3r> Turl: the whole 80 line limint is redicilous
<oliv3r> it should be a soft limit if anything, or atleast for strings
<oliv3r> Turl: so CM is going comercial eh
<oliv3r> did you know about that?
<libv> upstream adheres to it
<oliv3r> libv: linus disagrees with it :)
<Turl> oliv3r: it is kind of soft, but don't go over 80 on all lines :)
<oliv3r> it should be 80 limit, 90 softlimit, 100 hard limit
<libv> oliv3r: but he is just the tip of the iceberg
<oliv3r> Turl: well for checkpatch 80 is a hard limit
<Turl> oliv3r: you use a 100x24 term? :)
<Turl> I use a 80x24 one
<oliv3r> i have a 1920x1200 monitor
wingrime has joined #linux-sunxi
fredy has quit [Excess Flood]
<oliv3r> my term is randomly 101x37 atm :)
<Turl> I have a 1080p one :)
<Turl> heh
<oliv3r> i have 2x 1920x1200 :)
<wingrime> oliv3r: find more annoning bugs
<Turl> oliv3r: re. cm, no I didn't know until today :)
rz2k has joined #linux-sunxi
<oliv3r> at work i have 1x 1920x1200 1x ... 27" dunno res; but ?x1440
fredy has joined #linux-sunxi
<oliv3r> wingrime: don't worry too much about 3.4; mainline is more important ;)
<wingrime> oliv3r: if, screen of by xorg timeout, It not restore backlight
<libv> my terminals are all 80x25
<wingrime> *off
<libv> ah, not true, but at least they all are 80 wide
<libv> anyway, stick to the rules if you want to get upstream
<oliv3r> 80 is way tyo small for me
<oliv3r> oh i do
<oliv3r> but for strings/comments 80 can be really short
<wingrime> libv: better see my patches and I contine hacking tvenc
<Turl> you can go over 80 for strings
<Turl> to keep them grep'able
<oliv3r> Turl: but i don't :p
<wingrime> oliv3r: also a20 can't shutodwin
<libv> i only break it for big tables, like the board enable table in flashrom
<oliv3r> Turl: i make them real short; id on't split it either
<wingrime> *shutdown
<libv> wingrime: are you going to review my lima patches?
<Turl> libv: kernel patches?
<libv> no
<wingrime> libv: I can see , but not shure my optinion was usefull
guilherme_ has joined #linux-sunxi
<wingrime> *sure
<oliv3r> wingrime: you can always try :)
<libv> in any case, wingrime should learn to collect related patches, and to be patient
<oliv3r> patience is a hard thing; i know it very well myself
<wingrime> damn noizy kbd ...
TheViking has joined #linux-sunxi
boycottg00gle has joined #linux-sunxi
<wingrime> libv: I not perfectly understand witch patches you like / not like , so I afraid hack it more
<wingrime> oliv3r: If you have cb2 nearly you can try trigger some nice bug, that I fixed, and send to mainline
<wingrime> oliv3r: some review
<oliv3r> wingrime: what ML did you send your patch to?
<oliv3r> care to link it?
<wingrime> oliv3r: see my pcmaudio patch
<oliv3r> i glanced at it
<wingrime> oliv3r: try play with mplayer any music to analog out
<oliv3r> is that a bug with x86 too?
<wingrime> oliv3r: and try pause it with *space*
<wingrime> oliv3r: no
<wingrime> oliv3r: on cb2 only
<wingrime> oliv3r: I mean any sun7i
<boycottg00gle> how do i disable frame buffer console blank?
<oliv3r> so how is that a mainline patch?
<boycottg00gle> (sunxi stage-3.4)
<boycottg00gle> and why is the lcd backlight on when in console blank?
<wingrime> libv: you looks talkative today... you finished glmark support ?
<oliv3r> wingrime: i can't find your mainline patch
<wingrime> oliv3r: I not talked about maninline
<oliv3r> ohhhhh
<oliv3r> in that case, i think i saw
<oliv3r> i have no clue what it does though
<wingrime> oliv3r: what we have no in mainline usefullt
<wingrime> ?
Mithrian has quit [Quit: Leaving]
<oliv3r> wingrime: you don't know what driver to make for mainline
<wingrime> oliv3r: no, make priority order
<oliv3r> touchscreen drivers :)
<oliv3r> well i'm not sure if i'm qualified to make a list
<oliv3r> SmartCardReader sounds interesting, but we need hardware for that
jemk has quit [Quit: bye]
jemk_ has joined #linux-sunxi
<wingrime> oliv3r: do you saw radeon feature matrix?
jemk_ is now known as jemk
<oliv3r> wingrime: mem2mem driver ;)
<oliv3r> wingrime: years ago
<oliv3r> wingrime: well sometimes i look at it
<wingrime> oliv3r: I want something in our wiki
<oliv3r> wingrime: MTD driver
<oliv3r> TV Decoder!
<wingrime> oliv3r: without hw tunner
<wingrime> oliv3r: I have no idea how connect it
Mithrian has joined #linux-sunxi
<wingrime> oliv3r: begin make new kms driver?
<arokux2> mripard, if I select "Kernel low-level debugging functions" the kernel will hang after "Starting kernel...." no matter which sunXi UART I pick 0 or 1.
<oliv3r> wingrime: a LOT of work; but VERY needed
<techn_> ssvb: 456MHz crashed
<oliv3r> video driver probably requires more then 1 person to work on
<wingrime> oliv3r: yes
<oliv3r> wingrime: i did like your video out u-boot proposoal
<wingrime> oliv3r: but I alread see many people who want do sata
<oliv3r> yeah because it's very easy
<wingrime> oliv3r: but no one display driver
<wingrime> oliv3r: but I need wait Turl
<oliv3r> axp might be easier then disp
<wingrime> oliv3r: not sure
<wingrime> oliv3r: gpio, leds, adc, irq contr, regulator driver,
<wingrime> oliv3r: input driver
<oliv3r> we should make a list on wiki, based on memory map
<oliv3r> and then make a priotization order
<oliv3r> tv-decoder has the advantage that its' input
<oliv3r> adc is interesting, we have the LRADC on ML right now
<Turl> nobody wants to do dma? :)
<wingrime> Turl: I can, but without users it useless
<wingrime> Turl: we furstly must have some testable dma users, than do
<oliv3r> Turl: DMA is being done by matt
<oliv3r> i agree with alexsey here
<oliv3r> er wingrime
<oliv3r> have drivers that also work in pio mode; then add dma mode
<ssvb> techn_: hmm, 456MHz is a really high clock speed for mali (A10 and A20 already crashed at 416MHz for me)
<wingrime> oliv3r: firslty we must have pio users , than do dma
<wingrime> oliv3r: can you try make color-full table with legend
<wingrime> oliv3r: about SoC patrts we support in mainline and linux-sunxi
<wingrime> oliv3r: like radeon matrix
<ssvb> techn_: looks like 381MHz might be indeed the realistic clock frequency limit for A13, at least A13 does not seem to be worse than A10/A20, thanks for testing
<Turl> oliv3r: we have emac :)
<wingrime> ssvb: how can you explain diferent clock max with same techprocess?
<Turl> wingrime: luck, not all chips are created equal
<techn_> ssvb: 432Mhz was ok.. but there is no great performance difference between 180Mhz -> 432Mhz :(
<wingrime> Turl: no, cpu have 1080 Ghz normaly
<wingrime> why mali can not?
<wingrime> err
<wingrime> MHz
<Turl> wingrime: heat design?
<wingrime> Turl: no
<oliv3r> Turl: maxime said emac needs a cleanup!
<oliv3r> and we need GMAC!
<oliv3r> wingrime: GEMAC
<wingrime> Turl: i think only about conveer longness
<ssvb> wingrime: I'm not trying to explain anything, I'm collecting data right now to make conclusions later :)
<wingrime> ssvb: thats depend on IC
<ssvb> wingrime: somebody decided that it was a good idea to run mali at 180mhz in a13, I just want to confirm that the chip really can work at a much higher clock frequency
<wingrime> ssvb: marketing
<ssvb> techn_: maybe the memory is way too slow and is becoming the bottleneck, or maybe your benchmark is not a pure gpu test
TheViking has quit [Ping timeout: 250 seconds]
<wingrime> ssvb: a13 cheaper than a10, so a13 must not interfere with more expensive a10, thats means it must be slower
<techn_> that gives boost from 1600points to 1800points
<wingrime> ssvb: if you make a13 too powerfull, you never sell a10
<ssvb> wingrime: I find it hard to believe, most likely it was decided that the low resolution screens do not need fast mali and decided to save battery by running mali at the lower speed
<jelly-home> wingrime: or dissipation numbers
<techn_> ssvb: yes.. tablet runs hotter when clock is higher :p
<jelly-home> imagine that!
<techn_> ssvb: is there clock throlling in r3p2 driver?
<techn_> or does it work
<rz2k> techn_: we dont have mali_utilization_handler implemented
<ssvb> techn_: the odroid sources have some sort of dvfs for mali
<rz2k> mali can report utilization status to function, which should scale freq. accordingly
<techn_> rz2k: nice :)
mcbrick has joined #linux-sunxi
<wingrime> ssvb: most problematic I think, aw use old 55nm still but rockchip use 28nm
<wingrime> ssvb: overheat and , bad tablet battery time
<rz2k> have no idea if we can do that on sunxi, switching pll on the go. exynos has a dedicated video PLL (called VPLL :p) for this purposes
<ssvb> rz2k: switching pll dividers is no good?
<wingrime> hipboi: AW still contine use 55nm ?
<arokux2> oliv3r, are you up to test usb host now? ;)
<ssvb> rz2k: a20 has a dedicated extra pll for mali, but unfortunately not the earlier sunxi chips
<wingrime> ssvb: aw enginers mosly correct use single pll for mali and cedar
<oliv3r> arokux2: my boards are packed up at this moment
<rz2k> needs test, that was all I wanted to say. our clk code might not be ready for this sort of switching right in the middle of something.
<arokux2> oliv3r, so then later tonight?
<wingrime> ssvb: if you want save chip space
<ssvb> rz2k: actually I was a bit horrified when I looked at the allwinner pll frequency switching code, each pll multiplier is implemented as a bitfield and they are updated separately!
rz2k has quit [Read error: Connection reset by peer]
rz2k has joined #linux-sunxi
<ssvb> why on earth could not they just do a proper atomic update of the whole 32-bit pll configuration register in one go?
<rz2k> its allwinner
<rz2k> :p
<wingrime> ssvb: thats need additional hw logic, like shadow regs
<wingrime> jemk: ping
<boycottg00gle> hmm ioctl fd FBIOBLANK 0 should do it
notmart has quit [Quit: notmart terminated!]
<jemk> wingrime: pong
<ssvb> boycottg00gle: you can also add "consoleblank=0" option to kernel cmdline to prevent framebuffer console blanking
<wingrime> jemk: some guy from ML had strange issure with vdpau
<wingrime> jemk: do you readed that?
<boycottg00gle> ssvb: ha - thanks! will try
<Turl> ssvb: don't they? I think they read the value to a temp file, update each multiplier and the apply it
<Turl> s/file/variable/
<jemk> wingrime: im trying to read it, but i didn't have time the last days, so i missed much
<jemk> wingrime: maybe you want to give me a short summary
<wingrime> jemk: than wait some time
<wingrime> jemk: I find like
<ZetaNeta> Is there a easy way to implement a "Vector Hypertext" multibootload on A* boards? (Android/Debian)
fredy has quit [Excess Flood]
<mripard> arokux2: what hardware are you running it on?
<arokux2> mripard, Mele A1000
<arokux2> mripard, sun4i
<mripard> A10 then ?
<mripard> ok
<mripard> then it's UART0
<mripard> UART1 is only for A13
<mripard> have you passed earlyprintk in the boot arguments?
<arokux2> mripard, mm.. not sure. should I?
<ssvb> Turl: here they set the pll4 clock frequency, one multiplier at a time - https://github.com/linux-sunxi/linux-sunxi/blob/sunxi-v3.4.61-r0/arch/arm/mach-sun7i/clock/sys_clk.c#L656
<ssvb> Turl: and here they define a big volatile structure for accessing the registers - https://github.com/linux-sunxi/linux-sunxi/blob/sunxi-v3.4.61-r0/arch/arm/mach-sun7i/include/mach/ccmu.h#L540
fredy has joined #linux-sunxi
<mripard> arokux2: yes
<arokux2> mripard, ok, i'll try later and let you know.
<ZetaNeta> Is there a "advanced" bootloader for arm?
<ZetaNeta> I mean the one which support a normal multiboot, like most x86 loaders
<Turl> ssvb: wonder if gcc optimizes it to be a single write
<wingrime> jemk: short enought
<wingrime> Turl: no, writel have volitale
<wingrime> Turl: volitale should not be optimized
<wingrime> *volatile
<Turl> wingrime: but it is not pointer, it's volatile structure
<Turl> never seen it in use, I don't know how that behaves
<Turl> ZetaNeta: uboot is scriptable, make your own :)
<ZetaNeta> Turl, I mean.... Like if there any user friendly multiboot. That i can put on a final device
<ZetaNeta> Not even like text grub multiboot
<ZetaNeta> But more like simple Apple multiboot loader
<ZetaNeta> Anything like that?
<Turl> dunno what apple multiboot is
<Turl> ZetaNeta: you can use kexecboot I guess
<Turl> or Petitboot
<ZetaNeta> Greyish background, icons of availible OS. Thats apple stuff
<jemk> wingrime: the strange vdpauinfo is normal, i didn't implement the query functions yet, so he gets uninitialized random garbage.
<Turl> ZetaNeta: check those two out then
<ZetaNeta> Ok
<Turl> ZetaNeta: AW bootloaders kind of do that too I think
<Turl> they can show an icon, dunno if you can choose the OS though
<jemk> wingrime: that it doesn't run... bad, i'll have to add some informative error messages to know what happens there. is on my todo list.
<ZetaNeta> i mean like, something what will work nicely on a Allwinner netbook with TS.
<ZetaNeta> And yet, dont scare the "girls" away
<jemk> wingrime: and the cache problems, i might have an idea: https://github.com/willswang/libcedarx/blob/master/libcedarx.c#L206
<ZetaNeta> Well, i can put a text like thing. But......
<jemk> wingrime: maybe it only needs to be implemented
<ZetaNeta> Turl, kexec thing is genius!
<wingrime> ssvb: ....
<ssvb> jemk: lol
focus has joined #linux-sunxi
<ssvb> jemk: but this should affect vlc only, right?
<jemk> ssvb: all players using libcedarx, don't know if there are others
<ZetaNeta> is there any rules about "TouchScreens" in eoma68?
<ZetaNeta> oliv3r,
<ZetaNeta> Can i choose a single TS, that will work with all eoma68, becouse its in off. spec of it?
<ZetaNeta> Or there is nothing official about TS
* ZetaNeta reads about I2C....
<ZetaNeta> Question closed
pacopad has joined #linux-sunxi
<ssvb> jemk: I thought that willswang implemented his own high level wrapper for cedar, xbmc should be using libcedarv from https://github.com/linux-sunxi/cedarx-libs
<ssvb> jemk: (which is also a big mess)
<ssvb> jemk: but there are multiple implementations, and I have no idea which one is actually used
<jemk> ssvb: i don't know if this really is the problem, can't test it, a10 magically works without it
<ssvb> jemk: :)
<jemk> ssvb: but i doubt this is the only problem, as this function only gets called some times at video start, not after each new bitstream data
ykchavan has quit [Quit: Leaving]
rz2k has quit [Read error: Connection reset by peer]
rz2k has joined #linux-sunxi
<ssvb> jemk: I'm more worried that Paullo612 seems to have used the cache flush ioctl quite heavily around the place with no luck and still has to use uncached mapping - http://irclog.whitequark.org/linux-sunxi/2013-09-17#4982131;
<ssvb> jemk: I guess I just need to reproduce the problem on my A20 (but different things are piling up)
<arokux2> Turl, how is this done shorter normally?
<arokux2> Turl, static int sunxi_ehci_init(struct platform_device *pdev, struct sunxi_ehci_hcd *sunxi_ehci)
<mripard> arokux2: put the "struct sunxi_ehci_hcd *sunxi_ehci)" on a new line
<mripard> indented to the opening brace of the above line
<arokux2> mripard, thanks
<arokux2> mripard, should it be intended so the next line just just after an opening brace on the previous line?
<mripard> hmm, I don't understand your question
<mripard> if that clarifies
<arokux2> mripard, is it allowed to add spaces for intending?
Mithrian has quit [Quit: Leaving]
<arokux2> mripard, (because otherwise I wont match parameters on the previous line so nicely as you did)
boycottg00gle has quit [Remote host closed the connection]
n01 has joined #linux-sunxi
<Turl> arokux2: yes, tabs to get near, and fill with spaces to match the first char after (
<arokux2> Turl, thanks
<Turl> arokux2: spaces should always be < 8 then :)
<arokux2> :)
<ZetaNeta> does anyone know keyboards that are "to be embedded">
<ZetaNeta> ?
<ZetaNeta> I mean which are not made specialy for something, but are made for the laptop to be made for them?
<ZetaNeta> Better, if netbook sized
<Turl> ZetaNeta: look for laptop replacement keyboard on aliexpress
<ZetaNeta> I thinked about it
<ZetaNeta> But, they are more problematic
<ZetaNeta> 1. Most of the time, its not a standard interface
<ZetaNeta> 2. When laptop gets old......
<ZetaNeta> 3. Etc stuff
dwilkins has quit [Ping timeout: 245 seconds]
pacopad has quit [Quit: pacopad]
<Turl> I don't think such a thing as a "generic laptop keyboard" exists, because there's no "generic laptops"
dwilkins has joined #linux-sunxi
<Turl> a product line from a OEM usually has all the same keyboard though, so they tend to last in time
<rz2k> ZetaNeta: you should look at industrial market, I cant remember right now, but I did search for such products for one of the products at my job
<rz2k> there are industrial keyboards for kiosks and other embedded applications
<rz2k> ready to be embedded to your design
<arokux2> Turl, how about a small driver that dumps some registers, like hipboi did for sunxi-3.4? sunxi-dbg.
<ZetaNeta> jelly-home, I got them sold near me. (In brive-la-something)
<arokux2> Turl, actually it is not sunxi specific
<Turl> arokux2: that's what /dev/mem is for
<Turl> arokux2: just use devmem2 or busybox devmem
<ZetaNeta> jelly-home, So like, i can experiment
<ZetaNeta> But... Why you need a A20 device like this?
<ZetaNeta> (And whats not less important, how much are you willing to pay for this?)
<arokux2> Turl, I've asked several times how I can do it in user space!!!!
<arokux2> libv, ^
<arokux2> Turl, THANKS!
Black_Horseman has joined #linux-sunxi
Black_Horseman has quit [Changing host]
Black_Horseman has joined #linux-sunxi
<ZetaNeta> rz2k, Couldnt find any which may fit
<libv> arokux2: "22:30 < libv> arokux: C code to map the relevant part of /dev/mem" on september the 3rd
<jemk> ssvb: did you have any thoughts on dri for vdpau, or busy with mali?
<libv> arokux2: it's just a handful of lines and a nobrainer
<libv> and you can then build all sorts of functionality on top
<arokux2> libv, if you want a remote user do dump registers for you, would you send him a small *.c and ask him to compile it?
<ZetaNeta> Is it hard to make a keyboard by self?
<ZetaNeta> A laptop one
<arokux2> ZetaNeta, we here do software, you know...
<ZetaNeta> D:
<libv> arokux2: yup
<ZetaNeta> I cant return to ##hardware, after the last stupid question i asked
n01 has quit [Read error: Connection reset by peer]
<ssvb> jemk: yes, I got sidetracked with mali and some other things, hopefully this is going to be resolved soon
n01 has joined #linux-sunxi
<ssvb> arokux2: or you can give the remote user a shell script, which invokes devmem2
<arokux2> ssvb, thanks for the tip
jemk has quit [Ping timeout: 260 seconds]
<ZetaNeta> Can i like do a "large amount" of cheap arm laptops for 1000 euros,
<ZetaNeta> ?
<ZetaNeta> Like order to make them in china?
<ZetaNeta> Because i suddenly choose in the side of china power, instead of "at home" 3d printing
<arokux2> ZetaNeta, I'm sorry, but you really spam here. (check out armdevices.net) and ask further questions in #arm-netbook
<arokux2> Turl, can you test usb host for me now/later tonight?
deasy has joined #linux-sunxi
wingrime has quit [Ping timeout: 248 seconds]
<oliv3r> ssvb: in u-boot they actually do it a bit better :)
<arokux2> deasy, around?
rz2k has quit []
<Turl> ZetaNeta: I doubt you'd even start covering factory tooling with that amount
<Turl> arokux2: on a CB2 I can
<arokux2> Turl, is PLL6 added on A20?
<ZetaNeta> Turl, Well.... What about 2000?
<arokux2> ZetaNeta, do not spam plz
<ZetaNeta> arokux2, Who says about spam?
<arokux2> ZetaNeta, do you have allwinner based board?
<ZetaNeta> Yes
<Turl> arokux2: yes
<ZetaNeta> 2 of em
<arokux2> ZetaNeta, which one?
<ZetaNeta> inet9something
<ZetaNeta> and another which i need to figure out
<arokux2> ZetaNeta, what? what is that?
<ZetaNeta> inet97f-ii
<Turl> ZetaNeta: dunno, ask in the other channel better :p but even if you don't consider tooling it'd be a 20 device run
<ZetaNeta> and another is a A13 based one.
<ZetaNeta> Turl, Ok.
sioxs has joined #linux-sunxi
<Turl> arokux2: what do you want me to test?
<arokux2> Turl, usb host, but wait, i'll add bindings to a20 first.....
<Turl> ok
<Turl> arokux2: add bindings for all of them except A31
<arokux2> Turl, all of them = dtsi?
<Turl> arokux2: there's part of the binding on the dtsi, yes, and the muxing goes on the board dts
<arokux2> Turl, so you tell me I should add ehci bindings to all dtsi and all dts for all except of A31 based?
<Turl> arokux2: if they have usb ports, sure :)
<Turl> if they don't, don't :)
<oliv3r> ZetaNeta: the whole problem with touchscreen, well the i2c ones, is that you need many drivers to support all; all need to be matching the display etc, touchscreen, while nice, is hard to support that early in the boot stage. with keys it should be doable
<arokux2> Turl, ok, i'll add to A20 only, i do not know the other hardware really.
<deasy> arokux2, yes
<arokux2> deasy, do you want to help by testing things?
<deasy> maybe :p
<ZetaNeta> oliv3r, ok
<arokux2> deasy, you have cb2, right?
arokux2 has quit [Remote host closed the connection]
ZetaNeta has quit [Ping timeout: 248 seconds]
n01 has quit [Quit: Lost terminal]
<deasy> arokux, nop
arokux2 has joined #linux-sunxi
<arokux2> deasy, what then?
<deasy> the
<deasy> 1
<arokux2> deasy, cb1?
<deasy> yep
<arokux2> perfect.
<arokux2> deasy, so you know how to compile kernel and stuff? did you compile mainline?
<deasy> i have never do it on cubieboard
<deasy> but i know how to compile kernel
<arokux2> deasy, ok, you don't see to be enthusiastic enough )
<deasy> seems*
<deasy> i'm a bit tired
<deasy> say what to do and i do it
ZetaNeta has joined #linux-sunxi
<arokux2> deasy, sorry, it is a long list of things. I've thought you did it once.
<deasy> tomorrow :p
<deasy> i need to sleep
<deasy> prepare it
<Turl> arokux2: ping? :p
<arokux2> Turl, yes
<deasy> oh oh :o sound of mice
<deasy> mouse*
<Turl> arokux2: ETA for node added? :P
<arokux2> Turl, no idea, maybe not tonight even, I've decided to do thorough testing by myself first, but smth wrong with my rootfs......
<Turl> arokux2: ah, ok then :)
<arokux2> Turl, I cannot login into alip rootfs, do you know what can be the problem?
<Turl> dunno, what pw have you tried?
<arokux2> Turl, seems like it creates login on another console...
<Turl> arokux2: you can always boot passing init=/bin/sh and skip all the things
<arokux2> Turl, it worked but bash behaves ugly.. no arrow keys and backspace
<deasy> gn
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<arokux2> Turl, any idea what could be the problem: http://sprunge.us/XjbX ?
<Turl> bad inittab?
<arokux2> Turl, any other headless rootfs you can recommend? buildroot works but most of the command are stripped off. there is not libusb -v, for example.
techn_ has quit [Read error: Connection reset by peer]
techn_ has joined #linux-sunxi
arokux2 has quit [Read error: Connection reset by peer]
arokux2 has joined #linux-sunxi
<focus> hi, I'm trying to get LCD operational on EOMA-68 A20 http://www.gplsquared.com/eoma_boot/eoma_boot.html#lcd_lashup
<focus> the script.fex file I follow to set up a lcd http://www.gplsquared.com/eoma_boot/script.fex
<focus> KD43G18-40NB-A11
<focus> Lot of LCD in danger messages
<focus> any clue would be helpful
arokux2 has quit [Remote host closed the connection]
<Turl> arokux: Arch Linux ARM
focus has quit [Quit: Leaving]
naobsd has quit [Quit: Page closed]
ZetaNeta has quit [Ping timeout: 248 seconds]
arokux2 has joined #linux-sunxi
popolon has quit [Quit: Quitte]
tinti_ has quit [Read error: Connection reset by peer]
<arokux2> Turl, arch linux did the trick, my distribution actually :)