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
vinifr has quit [Ping timeout: 260 seconds]
vinifr has joined #linux-sunxi
vinifr has quit [Read error: Connection reset by peer]
_enrico_ has quit [Quit: Bye]
akaizen has quit [Remote host closed the connection]
akaizen has joined #linux-sunxi
akaizen has quit [Remote host closed the connection]
popolon has quit [Quit: Quitte]
akaizen has joined #linux-sunxi
egbert has joined #linux-sunxi
egbert_ has quit [Ping timeout: 264 seconds]
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
Tsvetan has quit [Ping timeout: 268 seconds]
drachenphone2 has joined #linux-sunxi
AndChat-398025 has joined #linux-sunxi
drachenphone has quit [Read error: Connection reset by peer]
drachenphone2 has quit [Ping timeout: 264 seconds]
<Turl> leviathanch: are you around?
<Turl> leviathanch: I just pushed a fresh sunxi-clk based on torvalds/master, it works fine on my A20
TheSeven has quit [Disconnected by services]
[7] has joined #linux-sunxi
deasy has joined #linux-sunxi
deasy has quit [Client Quit]
deasy has joined #linux-sunxi
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
List has quit [Ping timeout: 264 seconds]
ykchavan has joined #linux-sunxi
Black_Horseman has quit [Remote host closed the connection]
ykchavan has quit [Ping timeout: 260 seconds]
tzafrir has quit [Ping timeout: 256 seconds]
mouchon has quit [Ping timeout: 264 seconds]
FDCX_ has joined #linux-sunxi
mouchon has joined #linux-sunxi
gzamboni has joined #linux-sunxi
popolon has joined #linux-sunxi
tzafrir has joined #linux-sunxi
akaizen has quit [Remote host closed the connection]
Quarx has joined #linux-sunxi
akaizen has joined #linux-sunxi
rellla has joined #linux-sunxi
hipboi has joined #linux-sunxi
wingrime has joined #linux-sunxi
ynezz has quit [Read error: Connection reset by peer]
ynezz has joined #linux-sunxi
<leviathanch> Turl: ok
<leviathanch> thx!
ZetaNeta has joined #linux-sunxi
ZetaNeta has joined #linux-sunxi
\\Mr_C\\ has quit []
\\Mr_C\\ has joined #linux-sunxi
ynezz has quit [Read error: Connection reset by peer]
ynezz has joined #linux-sunxi
hramrach has quit [Ping timeout: 240 seconds]
n01 has joined #linux-sunxi
FDCX_ has quit [Ping timeout: 264 seconds]
hramrach has joined #linux-sunxi
fredy has quit [Excess Flood]
fredy has joined #linux-sunxi
<wingrime> #define TVE_03C (0x03c)
<wingrime> thats impressive
<wingrime> for register name
rellla has quit [Remote host closed the connection]
<specing> lol
<specing> But hey, it is not a magic number anymore!
<specing> now its a magic constant!
<wingrime> specing: thats more impressive that realy in user manual
<wingrime> specing: 585 page a20 manual
<specing> allwinner + 600 page manual?
<specing> can't be.
<wingrime> TVE_01C_REG
hipboi has quit [Remote host closed the connection]
ZetaNeta has quit [Ping timeout: 246 seconds]
<lkcl_> ssvb: notify phil hands
<lkcl_> ojn: notify phil hands
<lkcl_> he runs the server
<wingrime> lkcl_: do you finlay get ft5xx ts working?
mcbrick has joined #linux-sunxi
vaxholm has quit [Ping timeout: 264 seconds]
<ssvb> Turl: the comments seem to imply that 400M is a valid mbus clock frequency, that would be surely better than the current default 300M
mcbrick has quit [Remote host closed the connection]
mcbrick has joined #linux-sunxi
<wingrime> ssvb: nice,
ZetaNeta has joined #linux-sunxi
ZetaNeta has quit [Changing host]
ZetaNeta has joined #linux-sunxi
<wingrime> thats stable
<wingrime> > when cpu@1008M,
<wingrime> ssvb: thats for a10 also ?
FDCX_ has joined #linux-sunxi
<ssvb> wingrime: don't know whether it applies to a10, in any case it seems to be a good idea to use PLL5 (DRAMC) also for MBUS
<ssvb> wingrime: Turl overclocked MBUS to 600MHz, and the device did not immediately die for him :) Having MBUS running at the same clock speed as DRAMC would be quite logical
<wingrime> ssvb: good point
<wingrime> ssvb: there is something we can imporve change dram priority?
<wingrime> ssvb: I still have no idea with FE and BE settings differs
<aep> how does the sysfs interface for gpios work? How can i tell if its an in or output?
<aep> i'd expect a 'direction' thing in sysfs, but dont see any
Quarx has quit []
<wingrime> ssvb: witch pll used with mali currently on a20?
<ssvb> wingrime: pll4
<ssvb> wingrime: but we should move mali to pll8
<mnemoc> moin
<wingrime> mnemoc: moin
<wingrime> ssvb: pll8 used?
<mnemoc> gosh... another ton of mails :(
wingrime has quit [Ping timeout: 264 seconds]
hipboi has joined #linux-sunxi
<mnemoc> wb hipboi
wingrime has joined #linux-sunxi
<mnemoc> wingrime: btw, the error posting to dev@linux-sunxi.org was because you CCed wingrime@linux-sunxi.org and the dns hosting who is currently providing the mail forwards doesn't know about such account
deasy has joined #linux-sunxi
<mnemoc> wingrime: oliv3r is going to install a real smtp in the sunxi server as soon as he has time... then devs will be able to send mail from @linux-sunxi.org and @sunxi.org
<mnemoc> and receive mails there too
<mnemoc> (forwards/aliases)
<wingrime> mnemoc: nice
<wingrime> mnemoc: can you give some uncommented old patches from ml for a review
<wingrime> mnemoc: we should somehow deal with amount old stuff
<mnemoc> as I pulled a lot of stuff from hansg's tree I'm not really sure what old commits are still pending. if you can give me a hand identifying the oldest still-pending-but-not-nacked patches it would be great
<mnemoc> now i have a desk and chair ;-)
arokux2 has joined #linux-sunxi
<wingrime> mnemoc: you new home still hostel?
<wingrime> or better apartment
<arokux2> aep, which kernel?
<aep> arokux2: 3.3.0
<arokux2> aep, ok, no idea then, but I know how to do it in mainline, not sure it will apply.
<aep> yeah i think this uses the mainline gpio driver
<aep> or something
<aep> at least its not the SUNXI_UGLY_GPIO thingy
<mnemoc> wingrime: better is subjective, but it's a different apartment... on some things better (desk/chair/washing machine) on others worse.
<mnemoc> we have a gpiolib driver using the same script.bin section, the ugly should be removed
<arokux2> aep, in the mainline, first you need is to enable sysfs gpio access in the kernel config
<aep> jo its there
<aep> i can read the pin, but its missing something to set it as ouput
<aep> err other way round
<aep> i can set the pin but not read it
<mnemoc> wingrime: but only for 3 weeks...
<aep> since there's no 'direction' file
<wingrime> mnemoc: but whats next?
<aep> i'll try the 3.4 kernel
<wingrime> mnemoc: after 3 weeks?
<mnemoc> uncertain
<wingrime> mnemoc: thats sucks, realy
<mnemoc> hopefully something more permanent that *I* can furniture. but still uncertain
auxym has quit [Ping timeout: 264 seconds]
<wingrime> mnemoc: how many times you life such way?
<mnemoc> almost all this year
<n01> is it that hard to find a good apartment in berlin?
<mnemoc> i only arrived to berlin in july. but for real renting you need to show 3 payrollls. i only have 1 atm
<n01> ohh yeah, that sucks
<mnemoc> for days you have hostels, for longer you have WG (shared apartments, on which you have to be approved by all the residents and can take ages) or renting the 1-room apartment of someone while he/she/it goes on vacations
fredy has quit [Excess Flood]
<mnemoc> I do the last
<mnemoc> cheaper, faster, but... only a couple of weeks
auxym has joined #linux-sunxi
fredy has joined #linux-sunxi
<arokux2> aep, sorry, that is all I know
<aep> :)
jukivili has quit [Ping timeout: 246 seconds]
jukivili has joined #linux-sunxi
<wingrime> mnemoc: russia somehow different , may places where you can rent some place without offical documents. people not do offical rent agreement, thats for tax avoidment
<mnemoc> wingrime: thing with berlin is that renting is very cheap, and lot of people looking, so owners choose the "most reliable"
<mnemoc> asking for payrolls is their way to know if the person can pay
<wingrime> mnemoc: if that people-people agreement you can always axplian situation and be understanded
<wingrime> mnemoc: pre-pay can be helpfull too
rz2k has joined #linux-sunxi
<mnemoc> that's how subrenting works. prepaid + large warranty for possible damages in cash
atiti has joined #linux-sunxi
ZetaNeta has quit [Ping timeout: 246 seconds]
auxym has quit [Ping timeout: 264 seconds]
auxym has joined #linux-sunxi
vinifr has joined #linux-sunxi
auxym has quit [Ping timeout: 240 seconds]
auxym has joined #linux-sunxi
<wingrime> techn__: ping
<n01> mnemoc: where are the latest sources leaked for A20? let's see if I can start mainlining something again
Black_Horseman has joined #linux-sunxi
vicenteH has quit [Read error: Connection reset by peer]
vicenteH has joined #linux-sunxi
<vinifr> mripard, ping
vicenteH has quit [Read error: Connection reset by peer]
vicenteH has joined #linux-sunxi
n01 has quit [Quit: leaving]
<vinifr> what does Free Electrons use for Source Navigation?
<mripard> lxr, that's mainly why we host an instance :)
<mnemoc> mripard: does lxr use git (tags/branches) directly these days or you need an extracted tree of each?
<wingrime> anyone here tryed to use Tv out?
<mripard> I don't really know, I didn't actually set it up
<mripard> but I think we're still using different trees for each version
FDCX_ has quit [Remote host closed the connection]
FDCX_ has joined #linux-sunxi
vaxholm has joined #linux-sunxi
<aep> darn, how does this machine id stuff work?
<aep> Error: unrecognized/unsupported machine ID (r1 = 0x00000f35).
<wingrime> aep: uboot sipply it to linux kernel
<aep> yeah, i know, but why does my self built kernel not have that one
<wingrime> aep: kernel have MACHINE_START macros for each supported
<lkcl_> wingrime: yes i did.
<wingrime> lkcl_: wow, how
<lkcl_> http://git.rhombus-tech.net/?p=linux.git;a=tree;h=refs/heads/lkcl-3.3-a20;hb=refs/heads/lkcl-3.3-a20
<lkcl_> i had to use.... don't freak out too much.... polling of the I2C bus (!!)
<lkcl_> on a timer, because the INT line isn't activating, and even if it was it's not connected to an EINT-capable GPIO... whooops..
<lkcl_> wingrime: it turned out that there was a short-circuit on the hand-assembled PCB: I2C clock was connected to RGB/TTL clock. This Is Bad :)
<lkcl_> the 2nd PCB arrived yesterday: powered that up and i2cdetect found it straight away.
<wingrime> lkcl_: how much revisions board have ?
<wingrime> lkcl_: is 2nd?
<wingrime> lkcl_: mass production with autoplacment I hope solve most issues with PCB
<aep> how do i choose the correct machine id in the kernel config? i set system type to "A20" but that doesnt appear to be it :/
hipboi has quit [Quit: Leaving]
<rz2k> aep: if you want to run something on a20 - please use our https://github.com/linux-sunxi/u-boot-sunxi and kernel https://github.com/linux-sunxi/linux-sunxi/tree/stage/sunxi-3.4
<aep> i kinda do
<rz2k> they both have everything already set up
<rz2k> including machine id mess from allwinners 3.3 kernel
<aep> apparantly the machine type is 0x10bb. i dunno why archarm uses 0f35 :/
<rz2k> what archarm has to do with kernel and bootloader?
<rz2k> you downlaoded premade image?
<aep> i just left the uboot env thingy from the arch install
<aep> yeah
<rz2k> update the u-boot
<aep> wouldnt have expected them to be incompatible
<aep> yeah
<rz2k> after u-boot update and latest kernel from stage/sunxi-3.4 everything will work. archarm probably has old version of u-boot before a20 support was added
<aep> yeah
<aep> well its booting now but i lost LVDS. i guess i'll try updating uboot
<aep> and i still have no idea how to set a gpio as input
<aep> there's nothing new, even with kernel 3.4
pacopad has joined #linux-sunxi
<WarheadsSE> aep: if there is an issue, you can let us know, you know. rz2k: i didn't make that image in particular, but I can address issues like this when they are raised.
<pacopad> Hi All , i don't find how to compile the sunxi_mem driver
<pacopad> in which kernel part can i find it ?
<aep> WarheadsSE: sure. got a bugtracker?
<WarheadsSE> rz2k: and yeah, I would bet that the u-boot is older, since we had it early
<Turl> pacopad: what do you mean by sunxi_mem?
<WarheadsSE> aep: well, we have a github :P https://github.com/archlinuxarm/PKGBUILDs/issues
<aep> WarheadsSE: btw, that uboot patch in the pkgbuild doesnt apply anymore. do i need it?
<pacopad> "/dev/sunxi_mem"
<aep> something with changing path where the script.bin is loaded
<Turl> pacopad: looks like drivers/char/sunxi_mem
<pacopad> it was i thought but there is no sunxi_mem in the 3.4 kernel tree
<WarheadsSE> aep: well, if its on the wrong githash, or even source tree, how would it matter :P
<aep> WarheadsSE: err, i updated it locally, so i can send a merge request when this stuff boots
<aep> instead of compiling everything outside of archarm
<WarheadsSE> Yeah, pull request is great :)
<aep> trying to boot without the patch now///
<WarheadsSE> Just match up the expected defaults for uEnv.txt/script.bin location to be sure it is compatible for other users, and we're all good.
<aep> that didnt work
<aep> spl: not an uImage at 80
<aep> ### ERROR ### Please RESET the board ###
<wingrime> aep: mmc can do it somethimes
<WarheadsSE> mm, are you doing eMMC or SD boot?
<WarheadsSE> I would assume SD, and sometime it gets timing issues like that
<aep> thats supposed to be mmc
<aep> err sd card
<aep> just happened after flashing the new uboot image. pretty sure something is wrong with it
<lkcl_> wingrime: 2nd revision. fucking well stopped working and didn't tell me they'd stopped, so there are still components on the bottom side of the PCB, and connectors in the wrong fucking places.
<aep> WarheadsSE: oh yeah... i see. the install file of uboot flashes it to /dev/mmcblk0 that doesnt look like an sd card...
<WarheadsSE> mmc = sd
<aep> hurm
<WarheadsSE> eMMC is just attached the same way
<WarheadsSE> The question is which one comes up as mmcblk0
<aep> aye
<rah> when I run "./build.sh pack" in a lichee/ SDK directory, I get an image whose bootfs contains a file named "linux" instead of a directory named "linux" with files in it
<aep> looks like there's android preinstalled on the internal chip, so that answers the question
<rah> I have the strace output of someone else's execution of fsbuild, which is executed by the pack script: http://pastebin.com/SaFMdk0U
<rah> it differs from mine: http://pastebin.com/TtmcxTZT
<rah> whereas the other person's calls getdent() on linux/, my fsbuild calls read()
<rah> does anyone have any idea why my fsbuild might be running differently to the other person's?
<rah> I've tried fsbuild binaries from two different SDKs
<rah> I've tried different sys_config.fex files
<rah> I've even tried different underlying filesystem types
<rah> I've installed the same distribution and version as the other person
<rah> my fsbuild still calls read() on the linux directory
<rah> this is driving me nuts
<rah> can anyone help?
<aep> weird. the sunxi-spl.bin produced by my own uboot build from https://github.com/linux-sunxi/u-boot-sunxi is broken. the one from archarm works fine
<techn__> wingrime: pong
<aep> do i need to update that part?
<rz2k> rah: poke techn__
* rah pokes techn__
<rah> rz2k: is this like a ritual where I poke people, drink some concoction, do some weird dance and then fsbuild works?
<rz2k> rah: no, this ritual is called "only techn__ here is known android user"
<rah> rz2k: err.. this isn't an android build
<rz2k> rah: also he did the sunxi-bsp
<rz2k> i bet he is your guy
<rz2k> for any bsp problems
<wingrime> techn__: can you make small review, I sended patch that makes emac builds by default as module
<rah> rz2k: what is "bsp"?
<wingrime> and many other patches
<wingrime> in plans
<rah> rz2k: thanks but neither of those links explain what it is :-)
<WarheadsSE> Board Support Package
<rah> and what is that?
<WarheadsSE> basically , here, this is the base set of packages and code that make this board function.
<rah> packages of what? what kind of code? which boards?
<rz2k> rah: it creates hwpacks - hardware specific packages that you can flash to sd among with your rootfs
<rz2k> this is created for ease of maintaining everything
<rah> I still don't understand
<rah> *shrug*
<rz2k> rah: you hit ./bsp SDCARD=/dev/sdc ROOTFS=debian.tar.gz, wait 50 minutes and have latest kernel, u-boot and drivers already installed with your root filesystem on your /dev/sdc, then you eject /dev/sdc and smash it in your device and have everything working
<rz2k> out of the box
<aep> actually... what's the typical device node name for the NAND? i dont have anything except the mmc
n01 has joined #linux-sunxi
* rz2k cant explain further
<rah> I see
vicenteH has quit [Read error: Connection reset by peer]
<rah> does this work with A31 systems?
<rz2k> does linux-sunxi work with A31 systems?
<rah> does "./bsp SDCARD=/dev/sdc ROOTFS=debian.tar.gz" work with A31 systems?
<rah> or particularly, does "smash it in your device" with with A31 systems?
<rz2k> nope
<rah> they boot differently to A10/A13
<rah> :-/
vicenteH has joined #linux-sunxi
* rah pokes techn__ again
<rz2k> where did you get a31 sdk?
<rah> it came from someone else
<rah> I think they got it from Allwinner
<arokux2> rz2k, do you want a31 sdk?
<popolon> aep, isn't it : /dev/mmcblk0p1 ?
<aep> nop thats the mmc
<popolon> oh, yes, start by mmc
<aep> p1 being partion 1 ;)
<rz2k> rah: you want to talk with allwinner about problems with their sdk's, also they released new version of A20-SDK with 3.4 kernel, maybe they did another one for A31?
<popolon> blk0 = block (device) 0 probably
<aep> i hope its /dev/nand and i dont brick the thing by writing to it
<popolon> :)
<popolon> First question, where is uBoot ?
<arokux2> rz2k, guys from pengpod shared a31 kernel with us
<popolon> if you erase uBoot, don't know how you will boot
<rah> this BSP can allegedly create LiveSuit images
<WarheadsSE> popolon: the allwinner chips generally will boot from eMMC/NAND or SD. Not really brickable without physical damage
* rah installs
<techn__> wingrime: yeah.. that looked good
<popolon> uBoot is inside the LiveSuit image, isn't it ?
<aep> WarheadsSE: phew, thanks
<rah> techn__: hello?
<aep> WarheadsSE: and SD takes priority i assume?
<techn__> rah: hello
<aep> so if i erase the nand, i can just write the sd from a pc
<techn__> rah: current community a31 support is currently very poor.. I'm not sure if you can even boot community kernel with a31 :/
<techn__> no one have had interest to put effort for it (since lack of mali)
<rah> techn__: my problem is with fsbuild
<rah> which, as I understand it, isn't chip-specific
<techn__> yeah.. it's common.. but sadly closed source
<rah> I've managed to get it to build correctly
<popolon> firefoxOS sadly choose a31 for their first device ... another bad choice for them
<rah> I'm not sure how
<rah> but it's in the directory which I moved onto an ext3 filesystem from a reiserfs filesystem
<rah> techn__: do you know whether fsbuild has problems being run on reiserfs filesystems?
<techn__> I might not have support for reseirfs
<rah> hmm
<rah> techn__: why would it need access to the filesystem the bootfs files are on?
<rah> doesn't it just read() them?
<wingrime> techn__: thanks
vicenteH has quit [Read error: Connection reset by peer]
vicenteH has joined #linux-sunxi
<techn__> rah: oh.. problem might occur if directory is in different directory hierarchy
<techn__> rah: those image packing tools have really bad bugs.. and no tips of what has failed :(
<rah> techn__: sorry, too much ambiguity in there; I don't understand :-)
<rah> techn__: problem might occur if which directory is in a different hierarchy to which other directory?
<techn__> rah: you told that you moved rootfs directory to other place?
<rah> techn__: yes
<rah> techn__: no
<rah> techn__: I initially moved the SDK build directory to another FS, but the build failed so I unpacked a fresh SDK onto the FS
<rah> (and built that)
<rah> and it works
<rah> the SDKs that have been freshly unpacked and built on my normal FS don't work
<techn__> ok.. there is symbolic links
<techn__> make sure that they point to correct place
<rah> techn__: symbolic links where?
vinifr has quit [Ping timeout: 256 seconds]
<rah> that's not much help
auxym has quit [Read error: Connection reset by peer]
<rah> I know how to find symlinks
<rah> techn__: why are you suggesting checkout symbolic links?
<rah> techn__: I'm not using the SDK tree I moved
vicenteH has quit [Read error: No route to host]
auxym has joined #linux-sunxi
FDCX_ has quit [Remote host closed the connection]
<techn__> rah: It just came into my mind that broken symbolic link might show like you described
massi has joined #linux-sunxi
guizamboni has joined #linux-sunxi
<wingrime> oliv3r: ping
ganbold_ has quit [Remote host closed the connection]
<aep> is there any way to get an even when an interrupt changes?
<aep> erf
<aep> when a gpio input changes
Soru has quit [Read error: Connection reset by peer]
<wingrime> Turl: ping
<Turl> wingrime: pong
<Turl> aep: if the pin supports external interrupts and you mux it as such, possibly
<wingrime> Turl: have you any a20 device with linux-sunxi kernel nearly to test fix for one annonyng bug
massi has quit [Remote host closed the connection]
<aep> Turl: what do you mean by "mux it as such" ?
<aep> something in the fex file?
<Turl> aep: pins are defined with a string like
<Turl> port:<port><mux feature><pullup/down><drive capability><output level>
<Turl> on fex
<aep> i set it as input in the fex thing, but the gpio drivers apparantly doesnt support the edge stuff
Soru has joined #linux-sunxi
<aep> gpio_pin_3 = port:PE04<0><default><default><default>
vinifr has joined #linux-sunxi
<Turl> aep: that pin for example doesn't support external interrupts
<aep> where can you see that?
Black_Horseman has quit [Remote host closed the connection]
hark has quit [Ping timeout: 248 seconds]
<Turl> aep: see for example PH00, mux 6 column
Amitabha has quit [Ping timeout: 248 seconds]
<aep> ooh
hark has joined #linux-sunxi
<aep> i wonder if setting one of them to 6 will actually make it work with gpio-sunxi.c
<aep> nope :(
<Turl> aep: are you using one of the listed pins with EINT?
<aep> yeah, PH14
<aep> gpio_pin_2 = port:PH14<6><default><default><default>
<aep> but i get no /sys/class/gpio/gpio2_ph14/edge
<aep> and poll on /sys/class/gpio/gpio2_ph14/value never returns
<Turl> aep: is it an A10 device?
<Turl> or something else?
<aep> A20 cubieboard 2
<aep> i get the correct value on /sys/class/gpio/gpio2_ph14/value
<aep> just polling doesnt work
<aep> maybe there's another interface for interrupts?
<Turl> aep: can you check /proc/interrupts and see if the interrupts actually trigger?
<aep> dunno how to read this http://npaste.de/p/C53l
<Turl> aep: I don't see the gpio driver on there, can you paste your gpio_para fex section?
<aep> yeah. and none of the numbers increase when i change the input level, except rhe obvious ones like serial port and eth
Amitabha has joined #linux-sunxi
<Turl> aep: so /sys/class/gpio/gpio*ph14/edge does not exist?
<aep> nope
wingrime_ has joined #linux-sunxi
<wingrime_> ssvb: wtf http://paste.debian.net/40773/
<Turl> aep: can you add some debugging printk's on drivers/gpio/gpio-sunxi.c ?
<aep> yeah, like where?
<Turl> aep: add some on sunxi_gpio_eint_probe to make sure it enters the a20 case
<aep> yo
<aep> oooh
* aep has a feeling
<Turl> aep: also on sunxi_gpio_irq_init and sunxi_gpio_to_irq
<ssvb> wingrime: you have no g2d, ioctls just fail
<Turl> wingrime: nice optimization ;)
<wingrime_> ssvb: yes, thats a20, g2d still not builded in, but code must be more correctly
<ssvb> wingrime: patches are welcome
<wingrime_> ssvb: consider that like bug report
<Turl> wingrime, ssvb: did you get a chance to play with mbus?
<ssvb> Turl: yes
<aep> Turl: i think its this: "[*] enable sunxi_is_foo() even for other mach-sunNi"
vaxholm has quit [Quit: Leaving]
<wingrime_> ssvb: I already sended today some patches to linux-sunxi
<aep> in the sunxi_gpio_eint_probe code it will just use the first one i guess
vaxholm has joined #linux-sunxi
<Turl> ssvb: and? :)
<ssvb> Turl: tried 600mhz mbus just for lulz, the system booted but deadlocked when running glmark2-es2
<ssvb> Turl: 400mhz and 480mhz mbus seems to work without any visible problems
<Turl> ssvb: how can you make a 400mhz mbus?
<Turl> pll6 runs at 600, and you only have dividers
<ssvb> Turl: 1200MHz / 3
<Turl> so you upped the pll6 clock
<Turl> using PLL5 would be a good compromise I think
<Turl> and frees up PLL6 for sata only
<ssvb> Turl: reg_val = (0x1U<<31) | (0x1<<24) | (0x2<<0) | (0x0<<16); /* 400MHz */
<ssvb> Turl: PLL6*2 = 1200, N=1, M=3
<Turl> ssvb: that's gate, pll6, /3 indeed
<Turl> ssvb: if pll5 with no divisions works for everyone, it could provide a nice perf boost while remaining rather conservative
<wingrime_> Turl: are you thinked about cpufreq?
<wingrime_> Turl: how it can be downclocked
<Turl> wingrime_: mbus cpufreq?
<wingrime_> Turl: yes
<wingrime_> Turl: you decided witch plls will be used for cpufreq>
<wingrime_> ?
<Turl> wingrime_: it could be scaled with frequency I suppose, but I wonder how much power would that actually save
<Turl> wingrime_: cpufreq controls pll1, axi, ahb, apb0
<wingrime_> Turl: but do you not thinked that cedar and mali and dram may be downclocked too
<Turl> wingrime_: everything can be downclocked to become a 8086 competitor if you wish :)
<ssvb> Turl: A10 seems to have no register for MBUS configuration, and MBUS in A13 is configured as PLL5/2 (half of the memory clock speed)
<Turl> ssvb: yes, looks like mbus is sun5i+ (and I've implemented it as such)
<wingrime_> ssvb: are you sure? thats can be undocumented
<Turl> ssvb: I only own an A10S, haven't played with it yet
<Turl> ssvb: if PLL5 /1 works it could be nice
<aep> Turl: sunxi_gpio_eint_probe doesnt have a case for a20, i think that means none of the pins are accepted as interrupt, are they?
<aep> it falls through else and sets gpio_eint_data = {-1, -1, -1, -1}
wingrime_ has quit [Read error: Connection reset by peer]
<ssvb> wingrime_: the memory location where MBUS configuration register is supposed to be just reads as 0 on Allwinner-A10
<Turl> aep: first one has a sunxi_is_a20 on my copy
<Turl> what tree are you using?
<Turl> ssvb: uboot writes it unconditionally
<aep> should i be using something else?
<Turl> aep: yeah but which branch?
<aep> sunxi-3.4
<Turl> aep: maybe it gets lost in the mem bus
<ssvb> Turl: yes, but we just read it back as 0
<Turl> aep: try stage/sunxi-3.4
<aep> ooh
<aep> thanks
<arokux2> Turl, around? small question. if I get a clock with devm_get_clk and enable it should I then disable it or this is done by devm_* framework?
<wingrime> ssvb: thats sad, but, don't forget about gate on and other placement
<wingrime> ssvb: realy sad
<Turl> arokux2: you should disable it, devm will only handle the clk_put for you
<arokux2> Turl, thanks
<ssvb> wingrime: first I tried to hack u-boot to modify mbus settings on a10, but it had no effect, then I tried to read this value back and seems like this register just does not exist
<ssvb> wingrime: maybe a10 has mbus, but uses some sort of default clock for it (which we can't change)
<wingrime> ssvb: I can have very simple nonconfigurable dividor
<wingrime> *It
<wingrime> *divider
<ssvb> wingrime: yeah, maybe something like pll5/2
<wingrime> ssvb: for sure we can always ask benn
<ssvb> wingrime: that's a good idea
<vinifr> Does someone already analyzed sun4i-ts.c?
<vaxholm> FEL -> u-boot. what's the reason for having the SPL code? why is the two-step booting needed?
<wingrime> vinifr: are you talking about termo sensor and resistive ts part?
<vaxholm> hmm. is that purely to initialize the DRAM maybe...
<vinifr> wingrime, touch pad driver
<n01> Turl: for the RTC I need to be sure that I'm using the external osc. Can I do thot with clock framework?
<vinifr> it seems that there are many unneeded lines
pacopad has quit [Read error: Operation timed out]
pacopad has joined #linux-sunxi
AndChat-398025 has quit [Ping timeout: 264 seconds]
drachensun has quit [Ping timeout: 245 seconds]
<wingrime> vinifr: you realy have hw that use it?
<vinifr> unfortunately I have no
<wingrime> vinifr: than why you so interested in it?
<Turl> n01: which external osc?
<vinifr> wingrime, I'm trying to do tp driver for mainline
Seppoz has quit [Ping timeout: 264 seconds]
<wingrime> vinifr: without any hw thats difficult///
<n01> 32k osc
<vinifr> wingrime, well, you have right, I will first try to exploit the ADC
<vinifr> ... but I will...
<Turl> n01: what reg is that?
<wingrime> vinifr: temp sensor
<wingrime> thats can be easy checked and debuged
<wingrime> vinifr: but if you talking about simple resistive TS
<wingrime> vinifr: thats very cheap and can be easy connected with 4-wires
<n01> Turl: LOSC_CTRL_REG
<vinifr> yeah, TP is a tochscreen + ADC controller for 4-wires TS
<n01> 0x01c20d00
drachenphone has joined #linux-sunxi
<wingrime> vinifr: begin with temp sensor
drachensun has joined #linux-sunxi
drachenphone2 has joined #linux-sunxi
<Turl> n01: I think it should be the bootloaders' responsibility to set that
<Turl> n01: does uboot set it?
<n01> Turl: probably, I didn't check. So do you suggest to skip that checking?
<vinifr> wingrime, I want that the driver supports ADC by IIO framework and touchscreen by input framework
drachenphone has quit [Ping timeout: 246 seconds]
<Turl> n01: the internal osc is 32k+-20% I think I saw somewhere
<n01> Turl: probably for RTC the 20% is not enough
<wingrime> vinifr: nice
<n01> so it needs the external osc
<Turl> hno: does uboot currently configure this?
<n01> Turl: otherwise how can I check the mutex for internal/external osc? (clock tree diagram in the manual)
atiti has quit [Read error: Operation timed out]
auxym has quit [Quit: Lost terminal]
<vinifr> wingrime, but I do not want to use sun4i-ts as a model, I want to make a new driver
iamfrankenstein has joined #linux-sunxi
<Turl> n01: you mean mux? snooping at the register
auxym has joined #linux-sunxi
<Turl> n01: there was a discussion about this on linux-arm, an AW engineer commented on this
<Turl> I dunno why but his replies are not archived anywhere, you'll have to read from the quoted text :(
<Turl> probably because he sent HTML email
<n01> Turl: it says that it is enabled by bootloader
atiti has joined #linux-sunxi
<rah> for future reference
<rah> apparently the fsbuild program can't cope with being run on a reiserfs filesystem
<rah> I presume it will b0rk on anything other than extX
<hno> Turl, configure what?
<rah> s/b0rk/bork/
<hno> Turl, 32K clock source? Configuring that requires knowledge if there is an external 32K oscillator. Most boards do not have one.
FDCX_ has joined #linux-sunxi
<specing> yeah, 32K crystal is 0.25$, too expensive.
focus_it has joined #linux-sunxi
<focus_it> Hi, trying to compile sunxi v3.4 for EOMA-68 A20 based CPU, but it hangs after boot - UART log file here http://www.gplsquared.com/eoma_boot/sunxi_3.4_eoma_boot1.txt
<Turl> hno: should it be configured with an uboot flag then?
<hno> Turl, what is wrong with the driver configuring it?
<wingrime> focus_it: ...use our uboot
<Turl> hno: it is board dependent and not really part of the rtc
<Turl> hno: read the discussion I linked earlier
<hno> Turl, can you link it again.
<focus_it> wingrime: I built the uboot from following http://linux-sunxi.org/FirstSteps
<focus_it> wingrime: nope I lied! :( - it seems I followed the instructions at http://rhombus-tech.net/allwinner/a20/boot/
<wingrime> focus_it: are you realy booting on mmc?
<focus_it> wingime: yes booting off mmc
<Turl> hno: read the quoted text as the replies from the AW engineer do not seem to be archived anywhere
<wingrime> focus_it: according your log
<wingrime> focus_it: you use aw nand boor
<wingrime> *boot
<wingrime> hno: ^
<focus_it> wingrime: can you show me tell me which line that is by describing it
<hno> wingrime, EOMA68-A20 is different...
<wingrime> hno: whats with it?
<hno> MMC0 is not easily available, so normal booting is via nand loading our u-boot.
<hno> which then loads kernel etc from MMC3.
<wingrime> hno: facepalm
<hno> indeed.
<focus_it> I agree with hno - nand has been set up to start booting off the mmc
<hno> which also means all hardware is set up by AW bootloaders, not u-boot.
<wingrime> hno: thats imposible place our spl on nand only?
<focus_it> chaining the boot from nand to mmc works with another image I produce - so I know it can work
<wingrime> focus_it: what a you trying to do? where kernel placed? mmc0 , mmc3 or nand?
<hno> wingrime, just need to ic up the driver so access to the boot area is easier..
<focus_it> using that bootable uSD image, I replace uImage and lib/modules from compiling sunxi 3.4
<focus_it> wingrime: the kernel is in mmc3
<focus_it> EOMA has space problem that let got the mmc3 pins wired to the outside world instead of mmc0
<wingrime> lkcl_: if you have no space on eoma , why no just use spi-nor instead nand?
<wingrime> lkcl_: and use spi-nor for boot strap only
<focus_it> wingrime: there is space, there routing of wires become issue in small dense boards
<wingrime> focus_it: in you log kernel just hung?
<focus_it> just hangs - yes
<wingrime> focus_it: are you not sure thats hw problem?
<wingrime> *not hw problem
<wingrime> thats very early hung
<focus_it> wingrime: it boots fine with another image I prepare using cubie's a20 SD card image
<focus_it> don't believe its hardware
<wingrime> focus_it: so you replace only uImage , uboot still same with last check?
<focus_it> i replace uImage and lib/modules from compiling sunxi 3.4
<focus_it> uboot not replaced
pacopad has quit [Quit: pacopad]
<focus_it> There is boot log from working linaro image and a different kernel here : http://www.gplsquared.com/eoma_boot/linaro_log.txt
<focus_it> (I want to get it to boot with sunxi 3.4 linux however and reduce duplication of efforts)
<n01> hno: Turl: so in the RTC do I just assume that the osc has been already configured by u-boot?
<wingrime> hno: thats very early hung
<wingrime> [ 0.000000] aw_clksrc_init(120) must be next
<wingrime> focus_it: are you have Architect Timer enabled in current config?
<focus_it> i used eoma set up done by lkcl - here http://rhombus-tech.net/allwinner/a20/boot/eoma68-a20-config
<focus_it> there is a lot of difference with sunx v3.4 so I just 'gave up' trying to understand the differences
<wingrime> focus_it: try our default config for sun5i
<wingrime> err
<wingrime> sun7i
<wingrime> cp arch/arm/configs/sun7i_defconfig .conig
<hno> sun7i_defconfig is quite crappy in sunxi-3.4.
<focus_it> ok i try it - originally i tried make ARCH=arm CROSS_COMPILE=arm-linux-gnueabihf- sun4i_defconfig
<focus_it> it takes 40 min to compile and another half hour to jig everything up
<wingrime> focus_it: add -j4
<wingrime> if you have more cores
<focus_it> netbook with ssd :)
<wingrime> i3?
<focus_it> atom dual core 1.6GHz
<wingrime> thats explain everything
<hno> oh, almost as fast as my 5 year old laptop.
<aep> what? it takes 30 minutes to compile it on an A20 oO
<Turl> n01: I guess you can just ignore it for now; we can always revisit this if it turns out to be troublesome
<hno> Turl, what parts other than the RTC uses the 32K clock?
<focus_it> (i have plenty other PCs - this work i keep in netbook to carry between home and work)
<Turl> hno: many parts can use it
<hno> Turl, yes many parts can, but which ones do?
<hno> I know the CPU can, but you don't normally do that because it does not make sense other than in the standby code.
<Turl> hno: buses on suspend
<Turl> and CPU of course
<hno> and those do not really care what 32K source is used, just some that ticks at a low frequency.
<hno> RTC cares about having a good 32K clock source.
<Turl> indeed
<Turl> I dunno if timers use it
<Turl> they are probably hooked to the 24M osc though
Seppoz has joined #linux-sunxi
<hno> Heh... A31 internal LOSC is +/- 30% accuracy.
<hno> Turl, most if not all timers can be using LOSC but normally use 24M. But TIMER3 seems to be LOSC only.
vinifr has quit [Ping timeout: 240 seconds]
<wingrime> Speedy Boot
<wingrime> 5s
<wingrime> thats famous Super Suspend
guizamboni has quit [Remote host closed the connection]
iamfrankenstein has quit [Ping timeout: 256 seconds]
wingrime has quit [Ping timeout: 263 seconds]
guilherme has joined #linux-sunxi
guilherme is now known as Guest32283
<vaxholm> hno, re the CB2 LED defs. what is "current u-boot" in this context? i assume it's not branch sunxi, then.
<vaxholm> would it be the denx master?
<vaxholm> nvm. "Merge and reformat boards.cfg and MAINTAINERS" :)
<hno> vaxholm, it is the sunxi branch.
<hno> just pushed some fixes to that merge.
<vaxholm> right. will do
<lkcl_> wingrime: we need to sell boards, before making major redesigns. if we'd used the same x16 RAM ICs as on the cubieboard there would be space
<lkcl_> but.... sell boards first, redesign later!
<lkcl_> focus_it: i've just moved that config to eoma68-a20-3.3-config
<lkcl_> and have just tested sunxi-3.4 - everything i tested worked (see post 3mins ago on list)
<lkcl_> focus_it: also, use ccache for goodness sake! :)
<arokux2> focus_it, hi Joe
<focus_it> hi arokux2
iamfrankenstein has joined #linux-sunxi
iamfrankenstein has quit [Excess Flood]
<arokux2> focus_it, I've thought you are using your working hours only:) solved your problem?
<focus_it> 24/7 to get this done
<focus_it> still attacking one point at a time
<focus_it> work has more tools to be productive
<focus_it> 5GHz scope for starters
<arokux2> focus_it, you could setup ssh/sshfs to compile the kernel at home
<arokux2> focus_it, at work*
<focus_it> work internet has problems and isn't allowing home>work connections
<focus_it> the netbook is very handy - it goes with me everywhere and easy to back up as needed
<arokux2> focus_it, so does it still hang during boot? lkcl_ got it working.
<focus_it> still compiling - will have answer in about 1 hour
<arokux2> focus_it, alright.
<lkcl_> arokux: and posted the config, so you should be good.
<lkcl_> focus_it: joe i did the conversion from the eoma68-a20-3.3 config over to 3.4 and added in the extra config options needed which aren't present by just doing a make menuconfig
<lkcl_> some of the config options have changed
<lkcl_> changed names that is
<arokux2> lkcl_, me? I was always good here. amazing community. *you* guys were missing it. but now you are here too, so I'm excited to see fruits of this coop
<focus_it> lkcl_: I'm compiling with arch/arm/configs/sun7i_defconfig
<lkcl_> CONFIG_AW_AHCI has become CONFIG_SW_AHCI and weird shit like that
<lkcl_> focus_it: that'll do... but it's got extra stuff in that you won't need, and will take a hell of a long time, and then you'll still have to reconfigure it after compiling.
<arokux2> lkcl_, have you enabled ethernet btw? it is called differently too, but you've sent packets....
<lkcl_> the default kernel sun7i_defconfig kernel doesn't have USB-OTG for example
<lkcl_> whereas the one i just tested an hour ago - and posted on the rhombus-tech wiki - does.
<hno> yes, sunxi-3.4 sun7i_defconfig is a bit messed up at the moment.
<focus_it> lkcl_: after this compile finish and test, i try the one you posted
<lkcl_> arokux2: i've always been here, right from the beginning - just in the background until we've had working hardware
<hno> needs a major cleanup to more closely match sun4i_defconfig
tzafrir has quit [Ping timeout: 245 seconds]
<lkcl_> focus_it: nods
<lkcl_> arokux2: yes i enabled it - it ain't having none of it though. but it's a hardware issue - i tested with known-good u-boot (the one i used a couple of weeks ago) and that's having none of it, too.
<lkcl_> so it's almost definitely _not_ software.
* lkcl_ spotted some crap around the ethernet pins.... brb... scalpel. me. scarey. *lol*
<arokux2> lkcl_, what about sata? first you found out the pins should be switched, but in your e-mail you've said it works with *unmodified* cpu card.
iamfrankenstein has joined #linux-sunxi
iamfrankenstein has quit [Excess Flood]
rz2k has quit []
<hno> vaxholm, led2 merged.
<lkcl_> arokux: yes that's right. unmodified CPU Card. no need to replace the capacitors on the SATA lines with larger ones.
n01 has quit [Remote host closed the connection]
<vaxholm> hno, thanks
<arokux2> hm.. but focus_it found out it works with switched pins only?
<focus_it> that be on meb board
<focus_it> the pins come out correctly from the eoma
<arokux2> lkcl_, also, we had public ssh access to the EOMA, so ethernet issue is yours only?
<arokux2> focus_it, ah, ok!
<vaxholm> if i want fancy graphics on my cb2, what's the currently best supported env? android/xorg/fb ?
<libv> what is fancy graphics?
<mnemoc> spining cubes are fancy
<libv> vaxholm: would a banksy style stencil satisfy your needs?
<libv> it's not too fancy, but pretty hip
<vaxholm> libv, i'm more of a spinning cube than art guy ;)
<libv> vaxholm: why do you want a spinning cube?
<vaxholm> libv: i can live without the cube. i'd like something to use for an HTPC, maybe even be able to play quake3.
<vaxholm> so video and possibly opengl
<libv> just install a linux and follow the Binary_Drivers wiki page
<libv> if you want to play android games, then install android
<mnemoc> the cedarx RE project is pretty advanced these days
<libv> the mali RE project is currently stuck on the main developer trying to figure out why his jellyfish is all arty
<mnemoc> uh
<specing> haha
<specing> libv: is your jellu missing any triangles?
<libv> nope, all triangles are there
<libv> but it looks like i am tripping on lsd, but then confined to just one window
<vaxholm> how much of the mali(400) is understood?
<libv> vaxholm: libv.livejournal.com
vinifr has joined #linux-sunxi
akaizen has quit [Remote host closed the connection]
akaizen has joined #linux-sunxi
<aep> Turl: gpio event finally works. thanks :)
<lkcl_> arokux2: powered up the spare board (early one, wire running from PWRFBOUT to the connector... *sigh*) and ethernet works fine
<arokux2> lkcl_, congrats!
<arokux2> lkcl_, btw, remember the commit of the working sunxi-3.4 kernel, so we can bisect in case of regressions later.
<arokux2> lkcl_, sunxi-3.4 is developing at a quick pace. yes, you might be not interested in recompiling it every day, but still, if smb comes and wants a *working* kernel for eoma saying sunxi-3.4 won't suffice probably.
auxym has quit [Quit: Lost terminal]
<lkcl_> arokux2: ack.
<lkcl_> btw arokux2: i *have* been at this for 10+ years: http://hands.com/~lkcl/himalaya/
Black_Horseman has joined #linux-sunxi
<arokux2> lkcl_, what is himalaya?
<lkcl_> arokux2: it was only HTC's 2nd ever product. the first one was the wallaby.
<lkcl_> back then there were only about 4 to 6 of us regularly on #htc-linux
<arokux2> lkcl_, so you were mainlining it?
<lkcl_> arokux2: russell rejected anything that was more than 2 levels deep. between us we'd managed to reverse-engineer over *twelve* HTC products, and it would have meant dropping *250* files into the top-level ARM directory for the intel pxa processor
<lkcl_> at that point i went "yeah this is really worth bothering with" and unfortunately many people were then forced to duplicate the drivers because they didn't know that xanadux.sf.net existed.
<lkcl_> 8 years later and russell's _still_ overloaded....
<arokux2> lkcl_, I see. sunxi will get mainlined this time :)
<arokux2> lkcl_, btw I wanted to ask you about this router thingy
<lkcl_> arokux2: i mention this because when you say to me "it'll be a lot of effort to get 3.3 up and running" for example, you're forgetting that i really _do_ know what i'm doing, in a weird-and-random sort of way. in this case i decided i couldn't be bothered to forward-port the AHCI driver :)
auxym has joined #linux-sunxi
<lkcl_> i got better things to do
<lkcl_> ya. router thingy. yes.
<lkcl_> the primary driver on that is a potential customer. only like qty 100 a month (and it would _only_ be 100/month) but it'd be a steady 100/month so nothing to be sneezed at
<lkcl_> so. yes. router thingy. oh. a PM. right. just spotted that :)
auxym has quit [Client Quit]
<Turl> aep: you're welcome
akaizen has quit [Ping timeout: 246 seconds]
akaizen has joined #linux-sunxi
tzafrir has joined #linux-sunxi
vaxholm has quit [Ping timeout: 240 seconds]
<popolon> Hi
<popolon> I tried to update wiki
<popolon> and was blocked by Spam protection filter
<popolon> This is probably caused by a link to a blacklisted external site.
<popolon> :)
<arokux2> Turl, ^
<popolon> didn't added link
<popolon> there is a git clone line in modified text
<popolon> (make this more wiki and more readable
<Turl> popolon: try again please
<popolon> Turl, thanks
<popolon> ok, work fine :)
<popolon> thank you
FDCX_ has quit [Ping timeout: 260 seconds]
auxym has joined #linux-sunxi
<arokux2> what is the best allwinner tablet right now?
Black_Horseman has quit [Remote host closed the connection]
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
Black_Horseman has quit [Read error: Connection reset by peer]
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
arokux2 has quit [Remote host closed the connection]
<focus_it> arokuz: I recompiled sunxi using 7i and put it into the linaro distro - got much further ahead with EOMA - sata now working although it does not pick up ethernet and HDMI
<focus_it> the boot log notes are here http://www.gplsquared.com/eoma_boot/linaro_log2.txt
<focus_it> HDMI monitor is detected when its turned on, but no video - not sure why
<focus_it> even a command line would be fine
<focus_it> ethernet also not detected - not sure why
<focus_it> the ethernet and hdmi are known to work from the on board flash with android that can boot and operate the ethernet and hdmi
<Turl> focus_it: I see no mention of the eth driver on there, make sure it's enabled
<Turl> focus_it: disp is printing a lot of 'Lcd in danger', sounds like your fex config is incorrect
<focus_it> the lcd setting is for luke's lvds screen which I do not have attached
<focus_it> how do i enable eth driver?
<focus_it> sorry to ask
<Turl> what branch are you using?
<focus_it> sunx 3.4
<Turl> stage/sunxi-3.4 or sunxi-3.4?
<focus_it> git checkout origin/sunxi-3.4
<Turl> focus_it: then make sure you've got CONFIG_SUNXI_EMAC enabled on your kernel config