ChanServ changed the topic of #linux-sunxi to: Allwinner/sunxi development discussion - Don't ask to ask. Just ask! - See | | Logs at
notmart has quit [Remote host closed the connection]
mdfe_ has quit [Remote host closed the connection]
christopher has quit [Ping timeout: 264 seconds]
tinti has quit [Quit: Leaving]
ykchavan has quit [Quit: Leaving]
unioah1 has joined #linux-sunxi
BJfreeman has quit [Read error: Connection reset by peer]
_BJfreeman has joined #linux-sunxi
_BJfreeman is now known as BJfreeman
BJfreeman has quit [Quit: had a good time]
[7] has quit [Disconnected by services]
TheSeven has joined #linux-sunxi
\\Mr_C\\ has quit []
ojn has joined #linux-sunxi
sud0x3 has joined #linux-sunxi
sud0x3 has quit [Quit: Leaving]
hramrach_ has quit [Ping timeout: 240 seconds]
hramrach_ has joined #linux-sunxi
Tsvetan has joined #linux-sunxi
shineworld has joined #linux-sunxi
Yaku321 has joined #linux-sunxi
mturquette has quit [Quit: leaving]
mdp has quit [Ping timeout: 264 seconds]
mdp has joined #linux-sunxi
Dreadlish has quit [Read error: Connection reset by peer]
Dreadlish has joined #linux-sunxi
n01|away is now known as n01
shineworld has quit [Quit: Leaving]
eebrah|away is now known as eebrah
lxsameer has joined #linux-sunxi
techn__ has joined #linux-sunxi
techn_ has quit [Ping timeout: 256 seconds]
_BJfreeman has joined #linux-sunxi
_BJfreeman is now known as BJfreeman
<Tsvetan> mnemoc: is there A31 support in linux-sunxi
Tsvetan has left #linux-sunxi ["Leaving"]
luoyi has quit [Ping timeout: 256 seconds]
tzafrir has quit [Read error: Operation timed out]
luoyi has joined #linux-sunxi
hansg has joined #linux-sunxi
luoyi has quit [Ping timeout: 256 seconds]
mdp has quit [Ping timeout: 264 seconds]
mdp has joined #linux-sunxi
hglm has joined #linux-sunxi
BJfreeman has quit [Quit: had a good time]
luoyi has joined #linux-sunxi
<oliv3r> mnemoc: you here?
<oliv3r> anybody that know how the defconfig's work then?
<oliv3r> i modified the arch/arm/configs/sun4i_defconfig, but it doesn't apear to get picked up
<hglm> Did you run 'make defconfig'?
<oliv3r> no
<oliv3r> :p
<oliv3r> i rm -r 'ed my build/sun4i_* directory
<oliv3r> so that should pull in the new defconfig?
<oliv3r> i'm using the bsp btw
<oliv3r> i can manually get it to work; just want to make sure i don't have to submit anything else when changing a defconfig
<hglm> I'm not a kernel hacker, but did compile the kernel.
<oliv3r> oh i've compiled it many times :p i renamed some defconfigs, but it doesn't pick it up when using the bsp seems
<hglm> Default defconfig had a lot of drivers and debugging options that could be disabled.
luoyi has quit [Ping timeout: 256 seconds]
tzafrir has joined #linux-sunxi
<mnemoc> oliv3r: defconfigs aren't a cp of .config, they only include not-default entries
jemk has joined #linux-sunxi
<mnemoc> oliv3r: this old script I used for sanitizing them long ago might give you a better idea about them:
<hramrach_> E200 based reader
jemk has quit [Ping timeout: 252 seconds]
<mnemoc> hramrach_: got hw?
<hramrach_> yes, this one
<hramrach_> not sure it's much use as hw. No docs fr the chip and unknown system
<jinzo> ssvb, is there a place I can track your libhybris work for A31/PVR ?
<hramrach_> so even if you could trace a serial port some how there is no saying if the bootloader or system uses it for anything
<mnemoc> hramrach_: if you get uart we can start playing. we have (incomplete) sun3i support in 3.0 and 2.6.36
<mnemoc> hramrach_: but it looks VERY similar to sun4i
<hramrach_> problem is how to tell what part is uart when you have no means of frobbing it
<mnemoc> from the code the pio controller is identical, and script.bin is loaded in the same address
<mnemoc> so you can see what pins are used/unused and bitbang the others
<hramrach_> so if it can boot from sd card loading a modded u-boot spl might suffice
<mnemoc> yes
<mnemoc> btw, the wiki's img tag supports scaling :p
<hramrach_> yes, it does. I like liiking at full size ;-)
<hramrach_> s/ii/oo/
<mnemoc> *g*
jemk has joined #linux-sunxi
<hramrach_> it says it can boot from mmc but has two controllers
<hramrach_> so you can have non-bootable mmc too I suspect
<mnemoc> doubtful
<mnemoc> all reference designs from allwinner expose the mmc/uart/jtag as external uSD
<hramrach_> should be testable by inserting a SD card with SPL
<mnemoc> just some baremetal bin greeting on the uarts again and again and again....
<hramrach_> good point
<hramrach_> it would *cost* them to design a non-bootable mmc
<mnemoc> exactly
<mnemoc> and would make their own (sw) work harder
luoyi has joined #linux-sunxi
<Turl> theOzzieRat: ping
<tzafrir> Is there any point in running xbmc on a mk802 (linux)? if not: any recommended alternative?
luoyi has quit [Ping timeout: 256 seconds]
<hramrach_> tzafrir it will play some videos. it will probably crash on some, too
jemk has quit [Ping timeout: 252 seconds]
<ganbold> does A31 have driver for its GPU?
<ganbold> or rather does Allwinner provide driver for PowerVR?
<tzafrir> hramrach_, so far I'm rather at loss at even getting there
<tzafrir> e.g. getting the browser to play youtube videos nicely
<hramrach_> tzafrir: actually it plays like 90% videos you can find but when you get one from the remaining 10% it sucks
<tzafrir> hramrach_, what's the difference? What type of videos?
<tzafrir> and so far I couldn't get it to play anything
<tzafrir> (on Linux, that is. Debian Wheezy, specifically)
<tzafrir> s/play anything/play anything smoothly/
<hramrach_> tzafrir: using the first page you should be able to build a working player
<tzafrir> thanks. I'll give it a shot when I get home
<oliv3r> mnemoc: i edited by hand the arch/arm/config/sun4i_deconfig (and friends)
<oliv3r> mnemoc: what else needs to be done to have new defconfig (settings) committed?
<Turl> oliv3r: why by hand?
<mnemoc> oliv3r: first convert that into a .config. then.... 1m
<mnemoc> make savedefconfig
luoyi has joined #linux-sunxi
<oliv3r> Turl: cause i'm renaming things :p
<oliv3r> SUN4I_SOC -> SUNXI_SOC
<mnemoc> oliv3r: make foo_defconfig; (to get .config) ... then make savedefconfig (to get the minimized version)
<oliv3r> mnemoc: ok
<oliv3r> maye you missunderstand mnemoc
<oliv3r> i want to only change whatever we have in git
<mnemoc> oliv3r: and then `mv defconfig arch/arm/configs/foo_defconfig`
<oliv3r> why can't I just do %s/SUN4I_SOC_HDMIAUDIO/SUNXI_SOC_HDMIAUDIO
<mnemoc> you can, sure
<oliv3r> i don't understand why I need that much voodoo magic to change what we have in git
mdfe has joined #linux-sunxi
<mnemoc> if you are only renaming a CONFIG_, set will do
<oliv3r> or rather; if someone pull this change, will they get the changed version
<mnemoc> sed*
<oliv3r> that's what I did
<mnemoc> if you are changing the defconfig, make savedefconfig voodoo is required
<oliv3r> but when I do make linux-config; it doesn't pick it up
<oliv3r> i thought that, KERNEL_CONFIG=sun4i_defconfig forces the loading/building of sun4i_defconfig
<mnemoc> but feedback regarding hramrach_'s defconfig change is needed before any CONFIG_ related change
<oliv3r> can't; i renamed the sun4i audio stuff; sun4i->sunxi needs to happen :p
<mnemoc> oliv3r: the bsp only defconfig-s if there is no .config
<oliv3r> i removed build/sun4i_defconfig -r
<oliv3r> where else is a .config stored?
<oliv3r> that's my mistake then :)
<mnemoc> only there
<oliv3r> ok so that's not it then
<mnemoc> exactly, removing that dir you'll get a fresh .config on the next run
<oliv3r> if i do linux-config and go to the audio bit, i see the correct names, just not set as per sun4i_defconfig
<oliv3r> it doesn't though
<oliv3r> i'll rm -r build now
<oliv3r> heck, i'll do make clean
<mnemoc> rm -rf build/ is simpler :p
<oliv3r> yeah
<oliv3r> but i've done that tons of times
<oliv3r> well sun4i_defconfig-linux
<hramrach_> oliv3r: you can rename the configs
<oliv3r> but i had a 3.9 multi-v7 tree there that I didn't wanna nuke yet
<hramrach_> see the 3.0 'prefer new config names' patch
<mnemoc> oliv3r: it's saner to have one bsp per kernel tree
<hramrach_> or just domp the old names and let the config ask the user
<mnemoc> oliv3r: I mean, locally
<mnemoc> oliv3r: to not mixup things
<hramrach_> it should suffice to test defconfigs on standalone kernel build
<hramrach_> when the config changes the android build system should use the new one if you start afresh or pure enough stuff
<ssvb> jinzo: I'm not doing anything A31 related (and don't even have such hardware)
<mnemoc> ssvb: jinzo's question on #olimex was generic, about the performance penalty of using libhybris for the GPU/VPU blob
<ssvb> jinzo: I only hacked some cedarx libve libhybris wrapper for A10 (that's for hardware video decoding, not 3d graphics)
<ssvb> mnemoc: jinzo is better to ask this in #libhybris channel, assuming that A31 has working android drivers for PVR
<mnemoc> good point
<jinzo> ssvb, and mnemoc yeah I misunderstood.
<jinzo> thanks for the anwser.
<Turl> oliv3r: make sure you don't break backwards compat by keeping the old symbol name and depending on the new
<Turl> or selecting it rather
<mnemoc> making the default y might be a better idea
<mnemoc> them*
luoyi has quit [Ping timeout: 256 seconds]
<mnemoc> I mean, m
<mnemoc> so oldconfig will accept the rename blindly, and we won't need it in defconfig
jemk has joined #linux-sunxi
tinti has joined #linux-sunxi
Undertasker has joined #linux-sunxi
\\Mr_C\\ has joined #linux-sunxi
<libv> mali ddk will never be free
<Yaku321> i wonder whats the reason why they are doing that, the chips should sell better when ppl can code more for it
<mnemoc> libv: I only meant that vendors don't have to pay extra to build mali drivers for linux, for pvr they do.
<mnemoc> Yaku321: lawyers
<mnemoc> Yaku321: it's about IP and patents
<mnemoc> and been sued
<hramrach_> Yaku321: the more is known about the chip the more patent trolls sue teh manufacturer
Tartarus_ has joined #linux-sunxi
lxsameer has quit [Quit: Leaving]
_BJfreeman has joined #linux-sunxi
_BJfreeman is now known as BJfreeman
ganbold_ has joined #linux-sunxi
Undertasker has quit [Quit: Leaving.]
Undertasker1 has joined #linux-sunxi
Undertasker1 has quit [Read error: Connection reset by peer]
Undertasker has joined #linux-sunxi
Yaku321 has quit [Disconnected by services]
Yaku-noob has joined #linux-sunxi
Undertasker has quit [Read error: Connection reset by peer]
Undertasker has joined #linux-sunxi
eebrah is now known as eebrah|away
Tartarus_ has quit [Quit: Ex-Chat]
fldc has quit [Ping timeout: 245 seconds]
wingrime has joined #linux-sunxi
<wingrime> hramrach: you finaly buy this E200 ?
fldc has joined #linux-sunxi
sud0x3 has joined #linux-sunxi
<sud0x3> hi there, i just submitted a post to the linux-sunxi message group, but it has not yet appeared, is it awaiting conformation or have i messed up somewhere
<sud0x3> incase it is awaiting moderation, my post was entitled 'meminfo dump and scipt.bin'
<Turl> sud0x3: is it your first email to the list?
<mnemoc> meminfo dump and script.bin for CnM Touchpad 9.7 ?
<mnemoc> it's marked as spam...
<sud0x3> yes i used google groups
<Turl> mnemoc: yay for spam filters :)
<sud0x3> yes thats the one
<mnemoc> Turl: not mine, G's :)
<sud0x3> hmm i hope it isnt spam :)
<sud0x3> i followed the guide here '' and ive not used google groups before so was a little confused
<mnemoc> that page should really include how to turn the dump into something we can commit
<mnemoc> having the dump and the script.bin you are supposed to checkout the sunxi-boards repo, commit the .fex coming from your script.bin. then complete the [dram_para] section using the a10meminfo dump. commit again. and then submit patches instead of a blob and text :p
<sud0x3> to be honest im not sure what your going to be doing with the information, but if there is a way i can better prepare it for you id be happy to do that
<mnemoc> the (verified/completed) dram_para info is need for the bootloader. the rest is to maintain a known to work "database" which later can be used to generate .dtb files
<sud0x3> Im able to do the conversion from bin to fex although you have lost me at 'dram para' but i will do some research and see what i can come up with
<mnemoc> the .fex files has a section called [dram_para], some fields some empty and need to be filled. others need to be verified (basied on the dump from a10meminfo)
<mnemoc> s/some/are/2
<sud0x3> okay thanks :)
<mnemoc> a third commit of the .fex is usually verifying which sections are enabled but should be disabled
<mnemoc> vendors rely in hacked up and customized kernels, we don't
<mnemoc> so a wrongly enabled section in the script file may lead to a kernel panic
<mnemoc> obviusly always testing if the device still works as expected before submiting
<sud0x3> when you say test the, would that be as simple as replacing the sctipt.bin on the partition nanda, or would i need to build a new image a flash it to my tablet to test
<sud0x3> *the script file
<mnemoc> yes, just replace the script.bin in nanda and reboot
<mnemoc> (or uSD card when booting from mmc)
<sud0x3> Okay great, didnt want to have to keep loading a vm to use livesuit to flash
<sud0x3> ive not been able to boot into android from mmc yet but im working on it
rellla has joined #linux-sunxi
ssvb has quit [Quit: Leaving]
mdfe has quit [Read error: Operation timed out]
hglm has quit [Quit: leaving]
<mnemoc> hansg: I love you!
<mnemoc> *cough*
<mnemoc> can anyone can confirm the soc-detect change detects the A10S-olinuxino too?
<mnemoc> strange allwinner's code was explicit about 0 meaning A12 :<
<mnemoc> but f* it
<mnemoc> bbl, coffee
<hramrach_> I bought it a while ago. Got bored of reading e-books atm ;-)
<hramrach_> wingrime:
jemk has quit [Ping timeout: 264 seconds]
paulk-desktop has joined #linux-sunxi
<hramrach_> sud0x3: if you get Android working on mmc don't forget to write an article how to do that ;-)
hansg has quit [Quit: Leaving]
<mnemoc> wiki wiki <3
piyushverma has quit [Quit: Konversation terminated!]
piyushverma has joined #linux-sunxi
Kadosch has joined #linux-sunxi
eebrah_ has joined #linux-sunxi
rellla has quit [Quit: Nettalk6 -]
<drachensun> hramrach_: There is an article about that on the wiki
<drachensun> somebody figured it out for the cubieboard and wrote it up
<drachensun> I managed to follow it and get a working image for my tablets as well but I'm having trouble getting a second type of tablet working with it
jemk has joined #linux-sunxi
mdfe has joined #linux-sunxi
<sud0x3> hramrach: i intend to write articles on any of the work i do with this tablet as there doesnt seem to be much information regarding it on the web, and it will help anyone else tinkering with allwinner devices
<sud0x3> drachensun: I attempted following that guide a few months back but will look into it again
eebrah_ has quit [Ping timeout: 276 seconds]
Kadosch has quit [Read error: Connection reset by peer]
<Turl> mripard: ping
nove has joined #linux-sunxi
sanka has joined #linux-sunxi
rz2k has joined #linux-sunxi
ganbold_ has quit [Remote host closed the connection]
piyushverma has quit [Write error: Broken pipe]
<lkcl> anyone ran into this recently:
<lkcl> make -C board/sunxi/
<lkcl> make[2]: Entering directory `/home/lkcl/src/a10/u-boot-sunxi/board/sunxi'
<lkcl> make[2]: *** No rule to make target `.depend', needed by `libsunxi.o'. Stop.
<lkcl> hno: ping? ^
<lkcl> got it. file name wrong
<mnemoc> -j TOOLARGE ?
<mnemoc> ah, ok.
<hramrach_> drachensun: somehow did not find that when searching the wiki, thanks
* hramrach_ needs more SD cards
Tartarus_ has joined #linux-sunxi
Yaku321 has joined #linux-sunxi
Yaku-noob has quit [Ping timeout: 252 seconds]
Tartarus_ has quit [Ping timeout: 260 seconds]
piyushverma has joined #linux-sunxi
<sanka> if I want to build a kernel for an A10s device would you recommend the linux-sunxi or cyanogenmod as a starting point?
<lkcl> mnemoc: any clues on how to use "bootinfo" in sunxi-tools?
mdfe has quit [Remote host closed the connection]
<lkcl> where would i get boot0 and boot1 from?
<sanka> lkcl, from the device?
<lkcl> sanka: yes. under android. no livesuite.
<lkcl> i've run a10-meminfo-static, it's not giving accurate information [that can be dropped into u-boot and fel-boot at least]
<lkcl> sanka: i assume you were asking the question that would mean i would answer "yes from the device" such that you would have some useful further response as to *how* to go about getting that information from "the device"?
<sanka> sorry was afk
<sanka> ehm
<lkcl> ah ok :)
<sanka> I'm remembering how I did it a while back
<sanka> I believe i've dd-ed most of the android partitions
<sanka> and extracted info from there
<lkcl> yep i've grabbed nanda - but it's a FAT partition...
Tartarus_ has joined #linux-sunxi
<lkcl> i need nand_info...
<sanka> nandb and nandc were the boot0 anmd boot1 here
* lkcl goes to ... ah ha!
<lkcl> excellent
<lkcl> that's what i need to know
<tinti> sanka: hi, have you succeeded?
jemk has quit [Quit: bye]
<sanka> what kind of a device are you working on if you don't mind me asking
<mnemoc> boot0/boot1 live outside the logical nand
<mnemoc> first raw block
<lkcl> A20
<lkcl> EOMA68-A20
<sanka> ah fancy new stuff :)
<sanka> tinti, well partly
<lkcl> hmmm...
<lkcl> good god! android has a dd command. wow
<lkcl> i'm amazed
<tinti> :S
<lkcl> hmmm.... no "egon" magic.... hmmm...
<mnemoc> sure it's not from the busybox included by allwinner?
<mnemoc> you won't be able to read boot0/boot1 using the nand driver
<mnemoc> they live outside the logical /dev/nand
<mnemoc> and so, before any nand partition
<lkcl> bugger
<mnemoc> you can try with the mtd-driver from u-boot
<mnemoc> it allows raw reads iirc
<lkcl> mnemoc: this is so i can get the dram timings
<mnemoc> lkcl: usual way is to read the registers of the already initialized dram contorller
<lkcl> a10-meminfo-static?
<mnemoc> yes
<mnemoc> but i don't know if the A20's has the same controller/base/registry
<mnemoc> if not, adapt a10meminfo to fit
<mnemoc> also, can you give me a dump of the sid of your a20? md.l 0x01c23800 4 from u-boot
<mnemoc> github down or it's just me?
<lkcl> mnemoc: i'll drop the u-boot.bin onto my server. ftp to, incoming directory (anon ftp)
<lkcl> gimme 1sec
bsdfox has joined #linux-sunxi
bsdfox has joined #linux-sunxi
<piyushverma> github down :)
<mnemoc> lkcl: i have no use for your u-boot.bin. need some registers from the running SoC :)
<mnemoc> piyushverma: thanks for confirming it :-/
<lkcl> mnemoc: ahh ok
<lkcl> mnemoc: gimme commands to run then!
<mnemoc> lkcl: md.l 0x01c23800 4 :)
<lkcl> ack
<lkcl> md.l
<lkcl> mnemoc: /system/bin/sh: md.l: not found
<lkcl> try again sah.... :)
<mnemoc> from u-boot's console
<lkcl> oh.
<lkcl> ah
<lkcl> right.
<lkcl> 1sec.... if i can stop the boot process...
* mnemoc replaces lkcl's tea with real coffeee
<mnemoc> j/k :p
<lkcl> bootcmd set setargs_nand
<lkcl> Hit any key to stop autoboot: 0
<lkcl> blerk.....
<mnemoc> no timeout? :(
<lkcl> correct.... there's a way to sort that....
<lkcl> hno did it by blowing away /dev/block/nandb :)
<mnemoc> assuming the default env of your u-boot doesn't have the timeout=0 hardcoded
<lkcl> aaaagh! biig mess on-screen
<lkcl> bloody android. no "head" command....
<mnemoc> `busybox head` ?
<lkcl> oo that actually worked
<lkcl> root@teenymac:~/src/a10/sunxi-tools# hexdump -C nandb | more
<lkcl> 00000000 a5 4f 3b ef 62 6f 6f 74 64 65 6c 61 79 3d 30 00 |.O;.bootdelay=0.|
<lkcl> 00000020 72 67 73 5f 6e 61 6e 64 20 62 6f 6f 74 5f 6e 6f |rgs_nand boot_no|
<lkcl> 00000010 62 6f 6f 74 63 6d 64 3d 72 75 6e 20 73 65 74 61 |bootcmd=run seta|
<mnemoc> make a backup (dd) and then zero it
<lkcl> yep.
<lkcl> scarey!
<lkcl> HA!
<lkcl> that did it
<mnemoc> cool :)
<lkcl> 01c23800: 165165c7 80807552 55564848 0842c7fb .eQ.Ru..HHVU..B.
<oliv3r> sorry for goin gawol earlier :p but i'm back now
<ojn> lkcl: yeah, i've done that on a31 too. i can't seem to boot android after though, but that doesn't really concern me
<lkcl> happy?
<oliv3r> oh cool; i'll wikify that immediatly
<oliv3r> what chip is it?
<mnemoc> need the SID (md.l 0x01c23800 4) and the SC (md.l 0x01c00024 1). and then replace the SC with 0xffffffff (mm.l 0x01c00024) and read it again
<lkcl> ojn: what i did was take the first 1024 bytes (dd if=nanda of=nanda.edit bs=1024 count=1)
<mnemoc> oliv3r: A20 :)
<lkcl> then edit it with vi
<lkcl> and then dd it back :)
<oliv3r> Oh!
<lkcl> mnemoc: awww, i booted now!
<oliv3r> ok wikifying it
<mnemoc> lkcl: nanda is just a regular vfat partition
<oliv3r> what kernel are you booting?
<oliv3r> i should backread, but i'm to exited now
<lkcl> ojn. mnemoc: sorry, nandb
<oliv3r> EOMA a20?
<lkcl> oliv3r: yep. 3.3
<mnemoc> lkcl: thanks! can we go for the SC now? .... see instructions 2m ago
<lkcl> mnemoc: ack 1sec
<lkcl> ok stopped
<lkcl> 01c00024: 00000100 ....
<mnemoc> cool. now write 0xffffffff to it and read it again
<lkcl> sun7i#md.l 0x01c00024 1
<lkcl> 01c00024: 00000100 ....
<lkcl> sun7i#mm.l 0x01c00024
<lkcl> 01c00024: 00000100 ? 0xffffffff
<lkcl> 01c00028: 00000000 ? sun7i#md.l 0x01c00024 1
<lkcl> 01c00024: 16518100 ..Q.
<lkcl> happy?
<mnemoc> VERY :)
<lkcl> what's the SID and SC?
<mnemoc> the sid is like an eeprom, the other register tells the sunNi family of the chip and the state of the boot pins
<lkcl> hey i just realised - i could do a boot off of the system now because i can use the u-boot to do manual commands.
<mnemoc> system-control
<lkcl> mnemoc: ahh
<lkcl> anything else? need a wee :)
<mnemoc> lkcl: this means we can detect now what sun[4567]i chip the board is running and act smartly :)
<lkcl> ooOo
<mnemoc> one kernel to rule them all
<lkcl> mwahaha
<oliv3r> lkcl: whats' the name of the board
<lkcl> oliv3r: eoma68-a20
<mnemoc> eoma68-a20
<oliv3r> oh really
<lkcl> back later
<oliv3r> go wee
<mnemoc> oliv3r: VERY early prototype
* rz2k yay for a10s-olinuxino
<oliv3r> should be nearly the same
<oliv3r> mnemoc: have we found any sramc to be idfferent from the SID?
<mnemoc> rz2k: can you test soc-detect + hansg's patch on the a10s-olinuxino?
<rz2k> time to order a20 one
<mnemoc> oliv3r: only in the cases where the sid is blank
<oliv3r> and what was that'8' for again after the 1651, was tthat the bootflag? i forgot
<rz2k> mnemoc: just got it, had no time to run something. will check those patches, probably.
<mnemoc> oliv3r: yes
<oliv3r> mnemoc: obviously, but it's really good to know that it all matches still
<mnemoc> oliv3r: year
<mnemoc> yeah*
<mnemoc> rz2k: please
Yaku321 has quit []
<oliv3r> ok so
<oliv3r> i'll go back read now
<rz2k> i dont even know what u-boot I should run, though :p
<oliv3r> if i say allrady answered stuff, sorry, if i say something stupid, sorry x2
<rz2k> since I have axp152 or something like that
<oliv3r> lkcl: next thing to try btw, stage/3-4
<mnemoc> oliv3r: sorry, the 8 is the chip_id_en bit
<mnemoc> oliv3r: we have a lot to integrate before unning sunxi-3.4 on a20/a31
<oliv3r> just curious how far a10 boots
<oliv3r> erm a20 boots :p should get quite far i'd think
<oliv3r> i thought the 1 was the chip_id_en bit
<oliv3r> ok really backreading now
<mnemoc> 1<<15
<mnemoc> so the 8000 part
<mnemoc> the 1 is the fel button
<lkcl> back
<lkcl> right, mnemoc: what else
<mnemoc> lkcl: i'm done :)
<lkcl> oh ok ha ha
<mnemoc> i only needed to two humble registers
<mnemoc> s/to/those/
<lkcl> ok so what does a10-meminfo-static do
<lkcl> and how do i stuff its info into a dram.c file?
<lkcl> isn't one of the values wrong or something?
<mnemoc> reads /dev/mem and prints the content in known-dramc-struct format
<lkcl> ah
<lkcl> i get this:
<lkcl> dram_clk = 384
<lkcl> dram_chip_density = 4096
<lkcl> dram_type = 3
<lkcl> dram_rank_num = 1
<lkcl> dram_io_width = 16
<lkcl> dram_bus_width = 32
<lkcl> dram_cas = 9
<lkcl> dram_zq = 0x7f
<lkcl> dram_odt_en = 0
<lkcl> dram_tpr0 = 0x42d899b7
<mnemoc> looks perfect
<lkcl> dram_tpr1 = 0xa090
<lkcl> dram_tpr2 = 0x22a00
<lkcl> dram_tpr3 = 0x0
<lkcl> dram_emr1 = 0x4
<lkcl> dram_emr2 = 0x10
<lkcl> dram_emr3 = 0x0
<mnemoc> meaning the A20 share the same dramc as the a10
<lkcl> but there's 4 DRAM chips
<lkcl> 8-bit-wide
_BJfreeman has joined #linux-sunxi
<mnemoc> you can try this dram data with u-boot for a10 and pray
<mnemoc> disable the pmu just in case
<lkcl> ack
BJfreeman has quit [Ping timeout: 256 seconds]
<lkcl> mnemoc: where's a good place to start on disable pmu?
<mnemoc> don't know in sunxi-current, but in sunxi it used to be a define in boards.cfg
<lkcl> ah ok
<lkcl> hmmmm.... am i looking at a SUN4I or a SUN5I? please don't say SUN7I... :)
<mnemoc> sun4i = A10
<mnemoc> the a20 is.... similar
<lkcl> yehhh...
<mnemoc> so starting from a sun4i should be a good starting point. specially considered is confirmed to have the same dramc register
<mnemoc> considering*
<mnemoc> a31 doesn't
<lkcl> a31 has dual-banks 32-bit DDR3
<mnemoc> we will need an a31meminfo :p
<mnemoc> but a10meminfo works great for a10, a13, a10s and now a20 it seems
<techn__> lkcl: mem benchmark? :)
mturquette has joined #linux-sunxi
<mnemoc> wow. this is an unexpected visit
<mnemoc> welcome mturquette :)
<lkcl> techn__: busy. low-level. gotta get a boot running first.
<techn__> lkcl: yeah.. no hurry.. hopefully we get ours soon ;)
<mnemoc> the olinuxino-a20-micro-dev went on sale today, so they are probably sending the freebies these days too
<piyushverma> mnemoc: is mali build available for linux for A20 ?
<piyushverma> either gles2 or x11 ?
jemk has joined #linux-sunxi
n01_ has joined #linux-sunxi
<mnemoc> piyushverma: very unlikely
<mnemoc> piyushverma: we will need to push lkcl and tom to get one
<piyushverma> ok
<piyushverma> do u think other board using A5 with Mali 400 those binay could be helpful here
<piyushverma> like odroid
<piyushverma> sorry A7
<hramrach_> a20 is just too new
<piyushverma> ok
<hramrach_> it will probably have drivers eventually
<piyushverma> A31 too :)
<piyushverma> or that's batter than A20
<hramrach_> but can't tell because next to nobody has any a20 board and available software is unknown
<mnemoc> a linux driver for A31's GPU won't ever happen. PVR charges extra for the linux license and allwinner won't pay it for us
<techn__> piyushverma: no mali on a31
<hramrach_> a31 has different gprahics
<oliv3r> hansg's soc detect fix overlooks one important case; zero sid
<hramrach_> there you go
<piyushverma> right SGX544. I mean ready solution of that platform
<mnemoc> oliv3r: have you seen a sun5i with blank sid yet?
<oliv3r> we have an android on MMC article on wthe wiki; arete74_ wrote it
<piyushverma> May be ask them why so much source issue.
<mnemoc> piyushverma: IP (intelectual poverty) protection and fear to be sued by patent trolls
<lkcl> ahhhh.... does anyone know about this "hostport" thing?
<lkcl> :#define SUNXI_DRAMC_BASE 0X01C01000
<lkcl> mnemoc: ideas? memory-dump of those?
<lkcl> for (i = 0; i < 32; i++)
<lkcl> writel(hpcr_value[i], &dram->hpcr[i]);
<lkcl> sound like a plan?
<mnemoc> lkcl: that's what a10meminfo does
<lkcl> mnemoc: beh?
<lkcl> errrr
<mnemoc> lkcl: printing it later in a human friendly way
<lkcl> so what's this about?
<lkcl> static u32 hpcr_value[32] = {
<lkcl> #ifdef CONFIG_SUN5I
<lkcl> 0, 0, 0, 0,
<lkcl> 0, 0, 0, 0,
<lkcl> 0, 0, 0, 0,
<lkcl> 0, 0, 0, 0,
<lkcl> 0x1031, 0x1031, 0x0735, 0x1035,
<lkcl> 0x1035, 0x0731, 0x1031, 0,
<lkcl> 0x0301, 0x0301, 0x0301, 0x0301,
<lkcl> 0x0301, 0x0301, 0x0301, 0
<lkcl> #endif
<lkcl> as that's the bit i think it's hanging on
<mnemoc> ask hno
<mnemoc> i'm not familiar with that code
<lkcl> mnemoc: ack
<lkcl> hmmmm....
<lkcl> hprrrrr ooaarrrrr
<lkcl> 01c01250: 00000301 00000301 00000301 00000301 ................
<lkcl> 01c01260: 00000301 00000301 00000301 00000301 ................
<lkcl> 01c01290: 00001031 00001031 00000735 00001035 1...1...5...5...
<lkcl> 01c01270: 00000000 00000000 00000000 00000000 ................
<lkcl> 01c01280: 00000000 00000000 00000000 00000000 ................
<lkcl> 01c012a0: 00001035 00000731 00001031 00000735 5...1...1...5...
<lkcl> 01c012b0: 00001035 00001031 00000731 00001035 5...1...1...5...
<lkcl> 01c012c0: 00000001 00001031 00000000 00001031 ....1.......1...
<lkcl> 01c012d0: 00000000 00000000 00000000 00000000 ................
<lkcl> ooaarrrrrr
<lkcl> ooo, different!!
<oliv3r> what PMU do we have? axp209?
<lkcl> err... errr.... how do i get CONFIG_BLAHBOARD?
<lkcl> oliv3r: yep
<lkcl> identical to eoma68-a10 except CPU and RAM
<mnemoc> lkcl: for now you can just alter SUN4I's data
<mnemoc> lkcl: but eventually you'll need a proper board
<oliv3r> i'd pretend the A20 is an A10 for early experiemnts
<lkcl> mnemoc: euuw, euuuw!
* lkcl throws up
<mnemoc> well... go ahead and introduce sun7i :)
<mnemoc> boards.cfg does the core magic
<oliv3r> (still backreading it seems, you guys talk a lot :p)
<oliv3r> mnemoc: doesn't matter if we've seen it; we see only very little SID's so far. There's always (a minor) chance that that happens
<mnemoc> oliv3r: true, and that would lead to detect sun5i/a12
<lkcl> oliv3r: you're kidding! this is normal!
<mnemoc> BUT the sunNi part is the most important bit. and eventually a a12-so-claimed person will arrive and get us more data to disambiguate
<oliv3r> lol ok back-read!
<oliv3r> but yeah, SID has the really big advantage, that we can probe it, without writing to any registers
<oliv3r> so its much safer imo
<oliv3r> while sramc is more accurate
<lkcl> interesting. it's "getting ramsize" that fails!
<oliv3r> anyway, i'd build it as if it where an A10 u-boot; get the right mem-timings in; boot it :)
<oliv3r> should work
<oliv3r> dram_chip_density = 4096 should be memsize
<mnemoc> oliv3r: soc-detect only reads it after knowing which sunNi we are talking about
<mnemoc> and the SC/CHIP_ID method is confirmed to work on sun[4567]i. hramrach will have to tell us if it works on sun3i too
simosx has joined #linux-sunxi
simosx has joined #linux-sunxi
simosx has quit [Changing host]
<mnemoc> i bet it does too
<hramrach_> heh, did not get to even boot anything on that board too
<hramrach_> but where did all those people with f20 go?
<oliv3r> well the only 'bad' thing about sramc is, that you have to write to a register
<oliv3r> mnemoc: ideally; we put the SRAMC data into -> serial
<oliv3r> :p
<oliv3r> or something in /proc
<lkcl> errr dribble, dribble wibble...
<oliv3r> 'chip name' 'something'
_BJfreeman has quit [Read error: Connection reset by peer]
_BJfreeman has joined #linux-sunxi
<mnemoc> oliv3r: yes, eventually we can use the stuff printed by soc-detect in cpuinfo
<oliv3r> nonono
<oliv3r> ignore that
<oliv3r> that's better :)
<oliv3r> this file you need to modify and put in what mem10dumpinfo said
<lkcl> oliv3r: am ahead of you :)
<oliv3r> lkcl: you pasted the dram_clk settings earlier
<oliv3r> one thing that's important you forgot
<oliv3r> i think you have to manually calculate that or sumat
<oliv3r> anyway, it's only the 'size' parameter :)
<oliv3r> which kinda matches your error
<oliv3r> i'm so jealous right now :p
<lkcl> oo no crashy-crash!
<techn__> mnemoc: how you changed encodings ?
<lkcl> ok now it crashy-crash when writing to 0x4a000000 to drop u-boot in :)
<oliv3r> lkcl: what's that patch for?
<lkcl> oliv3r: eoma68-a20
<oliv3r> we only change sunxi/dram_eoma68-a20.c
<oliv3r> that's the file you copy/paste add
<mnemoc> techn__: iconv
<oliv3r> that should be all the chagnes required to u-boot
<lkcl> dram settings are still f*****d.
<oliv3r> because in that list you forgot the 'size' :)
<lkcl> oliv3r: should doesn't quite cut it.
<lkcl> oliv3r: where?
<oliv3r> and normally, all memories we've seen so far are near identical; only clock often different
<oliv3r> i didn't see .size = ; in your mem10info dump
<oliv3r> dram.c MIGHT need a modification of course for you
<lkcl> oliv3r: mmm.... in the script.fex file they don't set the size
<mnemoc> a10meminfo doesn't include size because it's not in the register
<mnemoc> you have to fill it yourself in the struct
<lkcl> dram_baseaddr = 0x40000000
<lkcl> dram_clk = 384
<lkcl> dram_type = 3
<lkcl> dram_rank_num = -1
<lkcl> dram_chip_density = -1
<lkcl> dram_io_width = -1
<lkcl> dram_bus_width = -1
<lkcl> dram_cas = 9
<lkcl> dram_zq = 0x7f
<lkcl> dram_odt_en = 0
<lkcl> dram_size = -1
<lkcl> dram_tpr0 = 0x42d899b7
<lkcl> dram_tpr1 = 0xa090
<lkcl> dram_tpr2 = 0x22a00
<lkcl> dram_tpr3 = 0x0
<mnemoc> that's totally f*ed up
<lkcl> dram_tpr4 = 0x0
<lkcl> dram_tpr5 = 0x0
<lkcl> dram_emr1 = 0x4
<lkcl> dram_emr2 = 0x10
<lkcl> dram_emr3 = 0x0
<lkcl> looovelyyy....
<lkcl> yeah wits-tech rush-job....
<mnemoc> your previous a10meminfo dump had meaningful data
<lkcl> yeah
<lkcl> that's wits-tech script.bin
<mnemoc> lkcl: please commit the stock .fex into sunxi-boards, and then replace all empty and -1 fields in [dram_para] with the data from a10meminfo
<mnemoc> livesuit scans your memory and fills the gaps. we can't do the same
<mnemoc> so we need real values
<oliv3r> also submit raw nanda dump for:
paulk-desktop has quit [Quit: Ex-Chat]
<lkcl> mnemoc: sunxi-boards?
<lkcl> mnemoc: i don't have an account on github, and won't be creating one.
<oliv3r> well if you can host the raw nanda.img; i'll upload it for yo
<lkcl> oliv3r: ack. it'll be at (ftp) in the incoming directory in about 1 minute
<mnemoc> lkcl: clone it, commit the stock sys_config/a20/eoma68-a20.fex, edit [dram_para], commit again, submit the patches
<oliv3r> lkcl: yeah, script.fex doesn't have proper mem info. it never does; livesuit 'injects' it when you flash it into boot0/1 dynamically somehow
<mnemoc> lkcl: after commiting git-format-patch will generate .patch files you can attach on an email or paste service
<oliv3r> we don't know
<oliv3r> that's why we have a10meminfo; it extracts the info from the registers
<lkcl> ack. i love git-format-patch
<oliv3r> ?
rellla has joined #linux-sunxi
<lkcl> oliv3r: nooo, i said "" not "" - that's a whopping 30% extra characters to type!
<lkcl> (dwarfed by the number just typed now... oh well... :)
<lkcl> mnemoc: livesuite scans my memory - holy fuck!
<lkcl> waaah, i am being scanned! bzzzt, epic fail on the proprietary applications thing....
<mnemoc> well... it's very practical for their usual customer
<mnemoc> less things to f* up
<mnemoc> they probe the dram of each unit during flashing, and write proper data for boot0 to use on each boot
<lkcl> i was just thinking of how useful it would be for livesuite to probe a customer's memory....
<oliv3r> we guess, that it uploads some program, that either reads SPD? or does some timing tests
* lkcl is wondering when the penny is going to drop for mnemoc...
<mnemoc> lkcl: doh. missed the joke :[
<oliv3r> though, technically; not quite sure how they do it
<lkcl> :)
<oliv3r> PC memory, has SPD that is factory programmed
<oliv3r> i don't think we have spd ON the actuall memory chips
<oliv3r> since it's a single smalle eprom on the memory module
<oliv3r> so i doubt SPD is 'the way'
<hramrach_> does it still work when the flash is empty?
<hramrach_> you could read it from the existing boot0
<oliv3r> doubt they do it that way
<oliv3r> if you completly BROKED your nand
<oliv3r> borked
<hramrach_> you cannot overwrite it normally
<oliv3r> we're talking about broken
<oliv3r> also, that means that they need to program each board with a specific boot0
<oliv3r> actually
<hramrach_> each batch
<oliv3r> yeah doubtfull though
<oliv3r> memory detecting sounds more logical (though I dont' know how)
<hramrach_> you can't really detect memory
<oliv3r> that's what I thought
<oliv3r> but you can't assume boot0 will always be 'ok'
<oliv3r> i guess AW does :p
<hramrach_> you can detect layout - eg. holes if you map everything and bits are missing
<hramrach_> but not timings
<oliv3r> well you 'could' but it would be slow
<oliv3r> do some timing tests
<oliv3r> actually; i doubt they do that; cause tom would have known about that trick
<hramrach_> you can't really do that
<hramrach_> I could run memtest on a x86 box all day but the moment I started DMA intensive data transfet it would crash
<hramrach_> underclocking the ram helped so it was timing issue
<hramrach_> and try to detect thet reliably ;P
<oliv3r> nobody said Allwinners methods are reliable :p
<oliv3r> but how would tom build his boards?
<oliv3r> i mean, the flash comes empty from the factor
<oliv3r> he uses livesuit to flash the initial image
<hramrach_> it can possibly use the included timings when the board has none
<hramrach_> included in image
<hramrach_> also the timings are random
<oliv3r> no i mean
<oliv3r> tom buys a tray of A10 chips
<oliv3r> he buys a nother tray of flash chips
<hramrach_> why does cubie run at 480 and mele at 360?
<oliv3r> he solders everything together
<oliv3r> and uses livesuit
<mnemoc> hramrach_: tom claims better PCB design ;-)
<hramrach_> yes, I know, you must get the initial timings there
<hramrach_> but the livesuit may do something different when the boot0 is empty
<oliv3r> well tom would have told us a little on how he does that
<oliv3r> or he only changed the clocks and the rest is still 'stock'
<hramrach_> but you cannot detect PCB sedign with livesuit :p
<oliv3r> but memory timings! :p
<oliv3r> well they do 'something'
<oliv3r> so its still quite interesting what they do
<oliv3r> and why has lkcl gone cold :(
<mnemoc> hramrach_: i believe livesuit only probes certain things, and only if they are NULL or -1 in the script
<hramrach_> I guess you could probe ram size quite reliably but not the parameters
<oliv3r> that i find plausible
<oliv3r> mnemoc: ok back to bsp issue i have
<oliv3r> mnemoc: i've done make clean
<hramrach_> so if you are careful in picing chips you can make boards with different ram size but same image
<oliv3r> i do make; yet it does not use arch/arm/configs/sun4i_defconfig
<lkcl> oliv3r: because i'm busy staring at source code
<oliv3r> so, where does the default config come from :)
<mnemoc> oliv3r: check
<oliv3r> mnemoc: sun4_defconfig
<oliv3r> which reminds me, gota submit a patch on that
<oliv3r> lkcl: did you change the 'size' parameter?
<mnemoc> rm build/sun4i_defconfig-linux/.config
<hramrach_> btw old PC bios did probe ram size by trying to write every page or so ;-)
<oliv3r> mnemoc: rm -r build
<lkcl> oliv3r: it's set to 1024.
<oliv3r> lkcl: ok :(
<lkcl> hmmm, where's the source for that a10-meminfo-static program?
<mnemoc> github
<lkcl> ack
<lkcl> got it
<oliv3r> lkcl: official name of the eoma; in caps or not? e.g. 'EOMA86-A20'?
<oliv3r> and how did you compress that, it's called tgz but it's only a tar?
<mripard> Turl: pong
awafaa has quit [Ping timeout: 240 seconds]
<oliv3r> mripard: i've been busy so hadn't had time to resubmit patch! sorry :(
<techn__> mnemoc: check allwinner-tools ? :)
<mripard> oliv3r: that's fine, don't worry :)
<oliv3r> i'm slacking! :(
<oliv3r> i just have to send-email it
<mnemoc> techn__: a1a386bb8aeb2a4a98d3fb3dbb817d5a4ffa4b80 is horrible
<mripard> then just send it and you'll be done with it
<mnemoc> techn__: you commited modified bin files
<mnemoc> techn__: you should really revert that
<techn__> mnemoc: rebase?
<mnemoc> git revert a1a386bb8aeb2a4a98d3fb3dbb817d5a4ffa4b80
<oliv3r> mripard: been busy :p
<oliv3r> mripard: tonight
<lkcl> mnemoc: - a20-meminfo.txt and meminfo.c
<mnemoc> spaces...
<oliv3r> lkcl: ^
<lkcl> spaces?
<mnemoc> lkcl: where are the .patch files for sunxi-bards? :)
<lkcl> mnemoc: busy. working shit out.
<oliv3r> lol
<oliv3r> lkcl: so officailyl its all caps? EOMA68-A20? yeah?
<mnemoc> lkcl: how does your meminfo.c differ from maxnet's?
<mnemoc> oliv3r: can you push your sound backport set in a branch I can pull?
<oliv3r> mnemoc: yeah
<mnemoc> chinese comments are patch friendly :(
<oliv3r> i MAY have fucked up the comments
<oliv3r> i'll push it to my github
<oliv3r> once lkcl answers the official naming :)
`26 has joined #linux-sunxi
<lkcl> mnemoc: extra debug info
<lkcl> dcr = 0x0
<lkcl> iocr = 0x0
<lkcl> csr = 0x0
<lkcl> drr = 0x0
<lkcl> tpr0 = 0x0
<lkcl> tpr1 = 0x0
<lkcl> tpr2 = 0x0
<lkcl> gdllcr = 0x0
<lkcl> rslr0 = 0x0
<lkcl> rslr1 = 0x0
<lkcl> rdgr0 = 0x0
<lkcl> rdgr1 = 0x0
<lkcl> odtcr = 0x0
<lkcl> dtr0 = 0x0
<techn__> mnemoc: I updated submodules
<lkcl> moo???
<lkcl> worra?
<lkcl> naaaah
<oliv3r> ouch
<oliv3r> so lkcl EOMA86-A20; all caps?
<mnemoc> techn__: allwinner-tools has submodules??
<techn__> no
<techn__> sunxi-bsp
<lkcl> oliv3r: yep
<oliv3r> ty; committing
<oliv3r> pushed
<oliv3r> just need some stock fex files etc; but this is good
<mnemoc> techn__: thought it was already at the head of sunxi-3.0
<mnemoc> techn__: we should probably make branches, for both 3.x, both stages and -next
<techn__> mnemoc: yeah.. only submodule which I didn't update was kernel :p
<techn__> mnemoc: I was thinking to create branch for sunxi-next
<techn__> which has u-boot-current and sunxi-net.. and full felboot support
<oliv3r> mnemoc: all on github now
<mnemoc> oliv3r: pull url?
<oliv3r> makin' me work even harder eh
<mnemoc> techn__: i think we should use u-boot-current on the bsp branches for stage/sunxi-3.x too
<lkcl> wadderfuck?
<lkcl> DRAM:dramc_init pre setup clock
<lkcl> dramc_init pre disable pad power
<lkcl> dramc_init pre set drive
<lkcl> dramc_init clock off
<lkcl> dramc_init pre ddr3 reset
<lkcl> dramc_init clock pre select dram controller 1
<lkcl> dramc_init clock pre itm disable
<lkcl> dramc_init clock pre enable dll0
<lkcl> dramc_init configure external dram
<lkcl> dramc_init external dram reg written
<lkcl> dramc_init clock on
<lkcl> dramc_init wait for ccr
<lkcl> dramc_init enable dllx
<lkcl> dramc_init set impedance
<lkcl> dramc_init set iocr
<lkcl> dramc_init set refresh period
<lkcl> dramc_init set timings
<lkcl> dramc_init set mode
<lkcl> dramc_init set dqs
<lkcl> dramc_init reset dram
<lkcl> dramc_init scan read pipe
<lkcl> dramc_init read pipe: 0
<lkcl> dramc_init configure hostport
<techn__> floood
<lkcl> dramc_init configured hostport, getting ramsize
<lkcl> dramc_init got ramsize 1073741824
<lkcl> dcr = 0x0
<lkcl> iocr = 0x0
<lkcl> csr = 0x0
<lkcl> drr = 0x0
<lkcl> tpr0 = 0x0
<mnemoc> techn__: it seems it will be long
<lkcl> tpr1 = 0x0
<lkcl> tpr2 = 0x0
<lkcl> moo?
<lkcl> wadderfuckisthosevalues doing at zero??
<lkcl> i just wrote the fucking things, why are they zero??
<mnemoc> lkcl: <--- my personal favorite paste service
<lkcl> mnemoc: ack
<lkcl> mnemoc: oooo, awesome
<hno> Oh, playing with an A20 board?
_BJfreeman has quit [Quit: had a good time]
rellla has quit [Quit: Nettalk6 -]
<lkcl> hno: yep
<mnemoc> oliv3r: thanks :)
<lkcl> and finding that reading from SUNXI_DRAM_BASE returns all zeros!
<lkcl> which is awesome!
<lkcl> hno: any bright ideas?
<mnemoc> oliv3r: and the rename? :p
<oliv3r> mnemoc: i hope that's workable; i can do github pull requests
<oliv3r> only done that for 3.4
<oliv3r> can squash them all to one if needed
<mnemoc> oliv3r: `git pull git:// wip/hansg-sound-fix-backport` is the real men's way
<mnemoc> oliv3r: not a "github pull request"
<hno> lkcl, I have not looked at the A20 yet. But a good ide to see if the basics of u-boot works is to soft-fel-boot our u-boot.bin. Press 2 on the console when Allwinner boot1 sarts, then upload & exe u-boot.bin using fel.
<oliv3r> mnemoc: as you please :p
<lkcl> hno: well i'm doing fel-boot.bin
<mnemoc> oliv3r: anyway, need you to send it to the ML first... for the records
<oliv3r> lkcl: might have to try that a few times, sometimes it boots so fast ...; booting to fel happens around 0.240 ..seconds?
<oliv3r> mnemoc: will do
<hno> fel-boot.bin is a bit more sensitive. Do you get the u-boot SPL banner?
<lkcl> hno: in theory i could knock out the ram setup from u-boot.
<lkcl> yes.
<lkcl> i've added a ton of debug info.
<lkcl> writel(0x16237495, &dram->csel);
<lkcl> printf("post select dram, csel = 0x%x\n", readl(&dram->drr));
<lkcl> on every line.
<lkcl> oliv3r: i'm holding down the boot fel thing. oh right - this is for boot0?
<lkcl> riight.
<lkcl> ok, actually that would work.
<oliv3r> sending in a few minutes mnemoc
* lkcl has an idea
<mnemoc> oliv3r: thanks :)
<lkcl> ok it's in FEL mode...
<hno> You can try dumping boot1 header. But meminfo should give the same.
<hno> have you got an hpcr table for A20?
<lkcl> hno: oh of course!
<lkcl> hno: yes.
<lkcl> - look for the patch there
<lkcl> right. where is boot1 dropped into RAM? i can read it with fel read can't i?
<oliv3r> should work yeah
<oliv3r> dump the BROM while your add it :)
<lkcl> oliv3r: you know the memory address?
<lkcl> :)
<oliv3r> hno can do magic with it in his github
<lkcl> gimme some addresses and sizes, i'll run them
<lkcl> The size of Boot1 is 0x00068000.
<lkcl> The file stored in 0x00000000 of block 2 is perfect.
<lkcl> block 2.... that's 0x2000 isn't it?
<hno> lkcl, you can dump it from nand u-boot. See Adding new board section.
mdfe has joined #linux-sunxi
<oliv3r> 0xffff0000 - 0xffff7fff
<hno> no, block 2 is block #2 of the NAND chip.
<oliv3r> ^ should be BRO
<lkcl> ack
<lkcl> ok
<lkcl> hno: ack
sud0x3 has quit [Remote host closed the connection]
<oliv3r> hno: could you put that in your github?
ykchavan has joined #linux-sunxi
<`26> mnemoc: pardon my ignorance of the u-boot development process, but I just downloaded and unpacked the debian source package of u-boot version 2013.01.01 and there is no mention in it of sunxi or cubieboard
<lkcl> U-Boot 2013.04-06935-g5ca78a1-dirty (May 30 2013 - 21:34:16) Allwinner Technoloy
<lkcl> CPU: SUNXI Family
<lkcl> Board: EOMA68_A20
<lkcl> I2C: ready
<lkcl> DRAM: 1 GiB
<lkcl> MMC: SUNXI SD/MMC: 0
<lkcl> woo!
<oliv3r> what did you change?
<oliv3r> `26: i'm note ven sure we have anything upstream yet
<lkcl> oliv3r: hno's idea to run from boot0
<mnemoc> `26: branch sunxi-current
<mnemoc> `26: hno is the maintainer. I don't know the state of it's mainlining
<mnemoc> `26: anyhow, 2013.01.01 is old
<hno> mnemoc, progressing slowly, but will get there.
<lkcl> reading script.bin
<lkcl> Error reading cluster
<lkcl> ** Unable to read file script.bin **
<oliv3r> lkcl: using fel-boot you mean?
<lkcl> barf :)
<oliv3r> lkcl: well evertyhting is setup properly i think
<`26> hno: is upstream usable yet?
<lkcl> oliv3r: yes. start from android (boot0, boot1), press "2" within 0.24 seconds
<hno> `26, there is nothing upstream yet
<lkcl> then use fel-boot to upload the compiled u-boot.
<`26> right.
<lkcl> soooOooo..... the allwinner boot0 is doing something that fex-boot.bin isn't
<hno> lkcl, yes.
<lkcl> hno: looovelyyyy....
<lkcl> buuut, i have a procedure....
<mnemoc> `26: branch sunxi-current is based on 2013.04
<hno> sunxi-current is tracking upstream fairly closely.
<`26> right, I'll diff against that then
<oliv3r> i'm SO jealous of lkcl irigt now
<oliv3r> gets to have all the fun
<oliv3r> :(
<lkcl> hno: MMC: SUNXI SD/MMC: 0
<hno> lkcl, yes?
<lkcl> that means sd/mmc 0 is recognised, right?
<oliv3r> aye
<hno> yes
<lkcl> ok.
<oliv3r> try fatls mmc 0
<oliv3r> or ext4ls mmc 0
<hno> but you might need "mmc rescan" when booting from boot1.
<oliv3r> (assuming ... wel you know)
<lkcl> ahhh that did it
<hno> haven't looked into why yet.
<lkcl> mmc rescan did the trick
<oliv3r> yeah a10 needs that too
<oliv3r> so that's 'normal'
<oliv3r> probably beacuse nand boot0 doesn't touch mmc?
<hno> yes, all of them needs it when booting via boot1.
simosx has quit [Quit: Αποχώρησε]
<lkcl> sun4i#fatload mmc 0 0x43000000 script.bin
<lkcl> reading script.bin
<lkcl> Error reading cluster
<oliv3r> boot1 i meant
<lkcl> unf.
<lkcl> but fatls works
<oliv3r> maybe fat broken :p
<oliv3r> fsck :)
<lkcl> sun4i#fatload mmc 0 0x48000000 uimage
<lkcl> reading uimage
<lkcl> Error reading cluster
<lkcl> barf
<hno> lkcl, load what you need via fel.
<lkcl> hno: ack.
<lkcl> ehmm...ehmehmehm.... can u-boot go into fel mode?
<hno> no, but you can type "reset" and hold down 2.
sanka has quit [Quit: Leaving]
<lkcl> *sigh* ok
<lkcl> ack
<lkcl> hno: you see the fel dump @ 0xffff0000->ffff7fff?
<hno> yes. already snarfed it.
<mnemoc> hno: is `md.b 0x42400000 0x2084` the real thing or a left over on ram?
<hno> left overs.
<oliv3r> md.b probably doesn't work reliably though; due to allignment issues?
<hno> it works fine from DRAM.
<hno> but not from most of the I/O space.
<lkcl> Error: unrecognized/unsupported machine ID (r1 = 0x00001008).
<lkcl> Available machine support:
<lkcl> 00000f35 sun7i
<lkcl> ID (hex) NAME
<lkcl> oh fer fuck's sake :)
<hno> lkcl, convert the hexdup to binary, then run bootinfo on it to print the dram info used by the bootloaders.
<oliv3r> mnemoc: sent
<hno> lkcl, you can set the machine id in uboot environment to temporarily override what is built into u.boot
<lkcl> hno: 00004000 06 00 00 ea 65 47 4f 4e 2e 42 52 4d 20 00 00 00 |....eGON.BRM ...|
<lkcl> 00004010 31 31 30 30 31 31 30 30 31 36 35 31 00 00 00 00 |110011001651....|
<lkcl> offset 0x4000 into that memory-dump?
<lkcl> hno: ack on the machine id
sud0x3 has joined #linux-sunxi
<mnemoc> hramrach_: no one ever wants to comment on your defconfig patchset :(
jemk has quit [Quit: night]
<lkcl> ./bootinfo < boot2
<lkcl> 4: Magic : eGON.BRM
<lkcl> c: Length : 32
<lkcl> 10: BOOT ver : 1100
<lkcl> 14: eGON ver : 1100
<lkcl> 18: Chip? : 1651
<hno> lkcl, no I meant the output of the md.b given in the sunxi u-boot wiki page.
<hno> and quoted by mnemoc a moment ago.
<oliv3r> looks like a new version no?
<lkcl> hno: ack
<oliv3r> mnemoc: i'll review the defconfig patchset
<hno> oliv3r, sure, and very expected that A20 have an updated BROM.
<oliv3r> maybe, i didn't think they'd change it
<oliv3r> a1* all had the same version :p
<oliv3r> i wonder how mcuh really changed though
<hno> no, there is quite substantial differences between even A10 and A13. Haven't looked at A12s yet but that should be very similar to A13.
<lkcl> logging... logging in minicom.....
<hno> lkcl, use screen or something else that works...
<lkcl> capture :)
<oliv3r> i prefer screen for console work very strongly
<oliv3r> screen /dev/ttyUSB0 115200
<oliv3r> and it stays connected until you unplug it
<lkcl> what's the opposite of hexdump??? never needed it!
<oliv3r> even survives standby on the pc
<oliv3r> bin2hex?
<oliv3r> hex2bin?
<oliv3r> good question
<mnemoc> od ?
<lkcl> ack
<oliv3r> mnemoc: if i can find them
rz2k has quit []
<oliv3r> mnemoc: [PATCH 3.0 v3 0/3] defconfig patches for 3.0?
<oliv3r> mnemoc: apply my deconfig patch before hramrach_
<lkcl> hno: got it.
<lkcl> where am i sending this? email to you....
<mnemoc> oliv3r: not really fair :) he's been updating this patchset for months
<oliv3r> mine first, no conflict
tzafrir has quit [Ping timeout: 252 seconds]
<oliv3r> his first -> conflict
<mnemoc> oliv3r: :)
<oliv3r> also, mine is only for 3.4
<oliv3r> his 3.0
<mnemoc> send sent both
<mnemoc> and need your rename for 3.0 too...
<oliv3r> i don't have one
<oliv3r> i'll do it after hramrach_ 's patch
<oliv3r> cause i'll fix the new files then
<oliv3r> i don't think i can cherry pick this patch; i'll have to redo it manually
n01_ has quit [Ping timeout: 276 seconds]
<oliv3r> mnemoc: i don't have the mails anymore; i'll check google's archive
<oliv3r> only have the sumary mail
<mnemoc> deleted the .patch files??
<oliv3r> hramrach_'s mails
<mnemoc> without feedback/acks I can't apply them anyway :(
`26 has quit []
<mnemoc> it's not something I can impose
<oliv3r> can you 'stack' config SUN4I_NANDFLASH; config SUN5I_NANDFLASH; config SUNXI_NANDFLASH; ?
<oliv3r> e.g. where ; is an enter
<oliv3r> is it like a switch/case fallthrough thing?
<mnemoc> those are used for compatibility
<mnemoc> old CONFIG forces the new CONFIG
<oliv3r> gah, now I have to reply; cause I don't like the PLAT_SUNXI thing; mainline isn't going to PLAT, it's staying with MACH; arch/<arch>/play_ is going away
<mnemoc> but at this point I think we should just make better defaults and let the new CONFIG_ leak in
<oliv3r> anyhow
<mnemoc> oliv3r: we have plat in 3.[04] because it's the way we have of unifying
<oliv3r> fine :p
<oliv3r> but it'll go away again eventually :p
<mnemoc> sure
<oliv3r> anyway, can I do a 'fallthrough' with the config?
<oliv3r> e.g. keep the ol sun4i and have the new sunxi too?
<mnemoc> just see PLAT_SUNXI as it was a ARCH_SUNXI
<mnemoc> with the name that the 3.[04] infrastructure allows
<oliv3r> +CONFIG_
<mnemoc> a config without descript is "hidden"
<mnemoc> but it's selects are honored
<oliv3r> explain more
<oliv3r> will it help when using make oldconfig?
<mnemoc> yes
<mnemoc> if SUN4I_SOC_HDMIAUDIO selects SUNXI_SOC_HDMIAUDIO, it will take the same value by "magic"
<oliv3r> in defconfig files?!
<mnemoc> defconfigs should only tell not-default things
<mnemoc> that's what `make savedefconfig` is for
nove has quit [Quit: nove]
<oliv3r> so in Kconfig we need both old + new
<mnemoc> for an smooth migration, yes
<oliv3r> ok rebasing
<mnemoc> at least during some "releases"
<mnemoc> 3.4 people is brave, 3.0 people isn't
<oliv3r> so only have config SUN4I_oldcrap followed by an empty tristate above it
<mnemoc> so we need more backward compaiiblit in 3.0
<oliv3r> wanna keep 3.4 then as it is?
<oliv3r> or want a oldcrap included patch :)
<mnemoc> 3.0 :)
<oliv3r> i'll do that too :p but for 3.4
<oliv3r> i'll do 3.0 after hramrach's patch
<oliv3r> i'll review it :p
<mnemoc> hramrach also included a 3.4 patchset
<oliv3r> really? didin't see it
<oliv3r> i'll review that too; and even do a update patch; if you apply mine first :p
<oliv3r> or
<oliv3r> hmm
<mnemoc> 6 mails total
<oliv3r> do his first, and i'll resend mine
<oliv3r> people can review it meantime
sud0x3 has quit [Ping timeout: 240 seconds]
<mnemoc> forwarding you the patchset, as .eml attachments
<oliv3r> send them to my -list mail
<mnemoc> that I did
<oliv3r> i'll review them in bed later :p
<mnemoc> good night
awafaa_ has joined #linux-sunxi
<oliv3r> nn
wingrime has quit [Ping timeout: 264 seconds]
sud0x3 has joined #linux-sunxi
tzafrir has joined #linux-sunxi
tzafrir has quit [Read error: No route to host]
tzafrir has joined #linux-sunxi
mdfe has quit [Remote host closed the connection]
sud0x3 has quit [Quit: Leaving]
bsdfox has quit [Ping timeout: 252 seconds]