wingrime has quit [Read error: Operation timed out]
arokux has quit [Ping timeout: 256 seconds]
BJfreeman has quit [Quit: had a good time]
deasy has quit [Ping timeout: 264 seconds]
focus has quit [Ping timeout: 248 seconds]
Robs58 has quit [Ping timeout: 256 seconds]
deasy has joined #linux-sunxi
focus has joined #linux-sunxi
deasy has quit [Quit: Quitte]
deasy has joined #linux-sunxi
* woprr
is away: Away
<deasy>
wopr in war games :p
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
Turl has quit [Ping timeout: 240 seconds]
Turl has joined #linux-sunxi
Turl has quit [Changing host]
Turl has joined #linux-sunxi
tinti_ has quit [Read error: Operation timed out]
\\Mr_C\\ has quit []
\\Mr_C\\ has joined #linux-sunxi
TheSeven has quit [Disconnected by services]
[7] has joined #linux-sunxi
techn__ has joined #linux-sunxi
techn_ has quit [Ping timeout: 245 seconds]
rz2k has quit []
hurtigbuffer is now known as jelly-home
navlrac has joined #linux-sunxi
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
BJfreeman has quit [Quit: had a good time]
hramrach has quit [Ping timeout: 240 seconds]
jacq|2 has joined #linux-sunxi
hramrach has joined #linux-sunxi
bsdfox_ has joined #linux-sunxi
deasy has quit [Remote host closed the connection]
bsdfox has quit [Ping timeout: 246 seconds]
<mnemoc>
wingrime
mcbrick has quit [Quit: Leaving]
rellla has joined #linux-sunxi
bfree has quit [Remote host closed the connection]
bfree has joined #linux-sunxi
Soul has quit [Ping timeout: 245 seconds]
andoma_ is now known as andoma
<hno>
arokux1, base your work on stage/sunxi-3.4. hans tree is not a sable reference and will be rebased.
<hno>
and participate on the ml. If you find that someone else is working in the same area then collaberate towards a common goal pushing/pulling from each other along the way.
eebrah|away is now known as eebrah
<hno>
<hno> arokux1, base your work on stage/sunxi-3.4. hans tree is not a sable reference and will be rebased.
<hno>
<hno> and participate on the ml. If you find that someone else is working in the same area then collaberate towards a common goal pushing/pulling from each other along the way.
<hno>
<hno> don't base on some other random patchset unless absolutely required for the changes you are making. You never know what will happen with other patches under review.
<hno>
<hno> Did that help?
massi has joined #linux-sunxi
navlrac has quit [Ping timeout: 245 seconds]
panda84kde has joined #linux-sunxi
tzafrir has quit [Ping timeout: 246 seconds]
vicenteH has quit [Ping timeout: 260 seconds]
geecko has joined #linux-sunxi
<geecko>
hi all
<geecko>
got ext4 corruption with the 3.3.0 kernel for A20
<Dreadlish>
how it looks like?
<Dreadlish>
geecko: kernel panics on start?
<geecko>
Dreadlish, nope, but the filesystem was remounted read-only
<Dreadlish>
so maybe fsck should be sufficient to fix it...
<geecko>
Dreadlish, it looks like postgresql database got corrupted... :(
<geecko>
but i don't care about the data
<geecko>
i'd like to know why it happened
arokux has joined #linux-sunxi
<Dreadlish>
brownout
<Dreadlish>
or kernel panic ;d
<geecko>
Dreadlish, no kernel panic, it was still running
<arokux>
hno, yes it helped, thanks hno.
<geecko>
i think the latest linux 3.4 would be much more stable
<Dreadlish>
if filesystem is fucked up, newer kernel would do nothing
n01 has quit [Ping timeout: 264 seconds]
<geecko>
Dreadlish, yes but it wouldn't cause this
n01 has joined #linux-sunxi
<geecko>
Dreadlish, about the brownout, you may be right
<geecko>
i have a lot of sensors plugged in, with two USB ports
bfree has quit [Remote host closed the connection]
Black_Horseman has quit [Quit: Zwi se logou mou!!!]
<arokux>
any idea how frequent does gmane sync?
arokux has quit [Remote host closed the connection]
vicenteH has joined #linux-sunxi
qlex has joined #linux-sunxi
* qlex
says hi
<qlex>
i have received my a20 marsboard with android already there, but board seems to be slow in video playback and possibly not outputting 1080, but 720...could any1 help me out ?
<qlex>
is this a good channel to ask about marsboard ?
<qlex>
need to change computers, back in a moment
qlex has left #linux-sunxi [#linux-sunxi]
adrian has quit [Read error: No route to host]
qlex has joined #linux-sunxi
* qlex
is back
<qlex>
can anyone help out with marsboard a20 and android ?
<qlex>
i see that marsboard a20 is similar to cubieboard2 - can i try and use phoenixsuit to put cubieboard android 4.2.2 image onto nand flash of my marsboard a20 ?
<adrian>
hello, i have one A20 olinunxino micro board, i want to enable all 8 UARTS do i put only CONFIG_SERIAL_8250_NR_UARTS=8 or also CONFIG_SERIAL_8250_RUNTIME_UARTS=8 i realy dont want to have the uarts on runtime
FR^2 has joined #linux-sunxi
<rz2k>
qlex: it wont work because marsboard iirc, doesnt have AXP PMIC and thus marsboard will have different .fex file
<rz2k>
comparing to CB2
<qlex>
rz2k: will i ruin my marsboard a20 if i write that image of android onto nand ?
<qlex>
polish distributor of marsboard has just told me on the phone to try cubieboard2 android image
<qlex>
i would like to at least try an image onto sd card
<Dreadlish>
marsboard has a20 and polish distributor?
<rz2k>
qlex: zero ideas, marsboard was always really weird board - mixed feelings about their support for linux-sunxi, strange demands about having own u-boot configuration when they had dram and basic stuff 1:1 to cubieboard's
<Dreadlish>
e, 270zł nie jest źle
n01 has joined #linux-sunxi
<Dreadlish>
whops
<qlex>
rz2k: any other suggestions? cubie2 was more expensive than mars a20 so bought this
<qlex>
but require 10+ units working with a stable android 4+ for next week
<qlex>
should i then switch to cubie2?
<rz2k>
check if marsboard has any official site and check for android image. also get ready to vary your declaration of "stable" :) A20-SDK for android is already around at olimex's website in .torrent (4GB size)
<rz2k>
i bet something will work wrong :p
<rz2k>
so you'd better download that
<arokux1>
adrian, have you seen some some patches flying around for your board on ML, could be useful.
<arokux1>
qlex, Seppoz has marsboard too.
<adrian>
i see some of them but i dont know how to apply
<adrian>
i see some DTS
<arokux1>
adrian, no, what i meant was fex patch
<arokux1>
adrian, A20: Update sys_config FEX for Olimex A20-OLinuXino-MICRO (Rev. C)
<adrian>
fex
<adrian>
i have rev D but is not big difference
<adrian>
the fex is set correctly
<arokux1>
adrian, ok, then
<qlex>
Seppoz: A10 or A20 of marsboard ?
<adrian>
but if i write
<adrian>
CONFIG_SERIAL_8250_NR_UARTS=8
<adrian>
CONFIG_SERIAL_8250_RUNTIME_UARTS=1 to the .config file
<adrian>
the kernel is booting but i have only one /dev/ttyS*
<adrian>
if i put CONFIG_SERIAL_8250_RUNTIME_UARTS=8 is kernel panic
<arokux1>
adrian, you better specify the kernel you're toying with, apart from that I have no idea why it doesn't work.
<adrian>
sunxi 3.3
<adrian>
but i dont remember the branch i used
<arokux1>
adrian, why have you decided to use 3.3, 3.4 seems to be better supported.
<adrian>
i had it working
<qlex>
arokux1: you said Seppoz has marsboard - do you know if its a10 or a20 ?
<adrian>
on 3.4 wwhat branch should i use?
<arokux1>
qlex, A10
<arokux1>
adrian, sunxi-3.4 is itself a branch. you need linux, not androidized , right?
<adrian>
yes linux
<arokux1>
qlex, look for marsboard on ML
<adrian>
i ithing i got the wip
<arokux1>
adrian, do you know what wip means?
<adrian>
work in progress
<adrian>
i think
<arokux1>
adrian, yep, so no reason to pick those. they are used by hackers to add/fix stuff
<arokux1>
wigyori, that's not the full error message
<wigyori>
o/ :)
n01 has joined #linux-sunxi
<mnemoc>
wingrime: sun7i not enabling plat_sunxi ?
<hramrach>
mnemoc: no.
<hramrach>
I wrote a patch for that but hansg put it on top rather than fixing up the initial sun7i import and rebasing the sun7i patches on top of that
<hramrach>
hence you have like 70 commits which build with f19 config but if you enable some random config option that requires a plat header you have to symlink it into sun7i :/
<wingrime>
mnemoc: also sun7i_defconfig fails with PM build
<wingrime>
mnemoc: I have fixed this but hsang not applied
<hramrach>
he used another patch
<mnemoc>
:(
<hramrach>
also problem with the se missing headers is that the thing appaerntly builds even if you don't have the header in hte commit
* qlex
signs off but will return tomorrow
<qlex>
bye
<arokux1>
qlex, byu
<arokux1>
bye*
<hramrach>
so hansg can't see the build errors in his tree. would need a fresh checkout
qlex has left #linux-sunxi [#linux-sunxi]
<hramrach>
I had a missing header too and the kernel built fine
n01 has quit [Read error: Operation timed out]
vicenteH has quit [Ping timeout: 245 seconds]
<mnemoc>
hramrach: sun7i must enable plat, not symlink
<arokux1>
hramrach, one could find untracked files with: git ls-files --other --exclude-standard
eebrah|away is now known as eebrah
eebrah is now known as eebrah|away
n01 has joined #linux-sunxi
ganbold_ has joined #linux-sunxi
<arokux1>
nice reviewer I've got
eebrah|away has quit [Ping timeout: 245 seconds]
eebrah has joined #linux-sunxi
eebrah is now known as Guest18487
Guest18487 is now known as eebrah_
eebrah_ is now known as ibrah
ibrah is now known as eebrah_
buZz__ is now known as buZz
bfree has quit [Remote host closed the connection]
bfree has joined #linux-sunxi
hipboi_ has quit [Ping timeout: 268 seconds]
damo22 has joined #linux-sunxi
vinifr has joined #linux-sunxi
<vinifr>
arokux1, hi
<arokux1>
vinifr, hi
<vinifr>
do you use kernel with DTB appended or kernel with DTB not appended?
<arokux1>
vinifr, I fatload kernel and dtb separately. what's your problem?
<vinifr>
i can not creat initramfs with 'mkinitramfs' for sunxi-mainline
hipboi_ has joined #linux-sunxi
<arokux1>
vinifr, why do you need that? have you got *.cpio from Turl?
<vinifr>
can I replace 'fatload mmc 0 0x45000000 uInitrd' by 'fatload mmc 0 0x45000000 *.cpio'
<arokux1>
vinifr, you need to pack *.cpio into uImage, there is a option for this in kernel config, please refer to the wiki page on Linux mainline how to or smth
<vinifr>
i pack .cpio into uImage, but I'll test when at job
<arokux1>
vinifr, what?
adrian has joined #linux-sunxi
<vinifr>
I already did it
<wingrime>
mnemoc: leave stage broken not good idea
<arokux1>
vinifr, ok, so now you fatload uImage and dtb, is it working?
<vinifr>
arokux1, i can not test right now :)
<arokux1>
vinifr, so why were you asking questions in the first place? :p
<vinifr>
I was wondering how you 'I fatload kernel and dtb separately'
<arokux1>
vinifr, ok, so now you know. when you'll test it, tomorrow? (i've promised to update wiki with instructions, but forgot to do so, will do it tonight)
<vinifr>
I will test soon
<wingrime>
ssvb: ping
<ssvb>
wingrime: pong
vinifr has quit [Quit: Saindo]
<wingrime>
ssvb: how do you think HW RNG will have some performane impact with common usecase?
<ssvb>
what are you going to use it for?
bsdfox has joined #linux-sunxi
<mnemoc>
wingrime: sure it's not, please send fixes and then poke me with the subject of the mail
<wingrime>
mnemoc: I thinked thay you know best way fix it (not #ifdef way)
pacopad has joined #linux-sunxi
<arokux1>
ssvb, what are you working on, btw?
<mnemoc>
wingrime: I don't have time at the moment to look into it.... barely to apply pre-reviewed patches
<wingrime>
ssvb: rng driver are realy easy
<wingrime>
ssvb: but It have strong connection with other crypher
<damo22>
err this looks like good stuff, what do you guys reckon
<swdwdw>
width 640, height 480 set video layer param Open video display layer failed config! layer open hdl:0,ret:-1 w:640 h:480 Open video display layer failed render_render!
<swdwdw>
the tablets of allwinner soc is even less people to use in china
damo22 has left #linux-sunxi [#linux-sunxi]
<wingrime>
swdwdw: maybe, but allwinner linux better than rockchip
tat has joined #linux-sunxi
<wingrime>
swdwdw: community matter
<tat>
is it possibill to recreate a .config from /proc/kallsyms
<wingrime>
tat: not nessesory
<wingrime>
tat: you always have /proc/config.gz
<tat>
wingrime: not allways
* wingrime
mostly
<tat>
but isn't any thing that is configured into the kernel somhow reflected in exported symbols ?
<wingrime>
tat: If you not dissabled
<tat>
wingrime: it wasn't me
<wingrime>
tat: can you explain?
<tat>
wingrime: it is a chinese tablet with a sun5i kernel ( android) i need ftdi driver, so i don't wan't to mak my own kernel, just the ftdi driver, since i don't have the exact config some stuff in the kernel is shifted and i can't resolve all symbols
<wingrime>
tat: you can try build 3.0.8+ from crane branch
<wingrime>
tat: in some cases
<tat>
why from crane branch ?
<wingrime>
tat: module will not work with linux-sunxi
<wingrime>
tat: we have other version
<wingrime>
tat: in case you want build kernel for stock kernel
<swdwdw>
are you a chinese ?is so give me your qq number ,I give you the driver
<wingrime>
swdwdw: I am a russian and liver in Russia
<swdwdw>
tat :are you a chinese ?is so give me your qq number ,I give you the driver
<swdwdw>
tat :are you a chinese ?if so give me your qq number ,I give you the driver
<tat>
swdwdw: i am not chinese, but do i need to be chinese to get a gpl licensed driver ???
<Turl>
tat: uploading stuff from china is often hard
<Turl>
big firewall doing random stuff :)
<swdwdw>
you know this
<swdwdw>
?
<tat>
well i am not so interested in binary driver, rather in a working source tree and a .config that works for my device
<Turl>
swdwdw: I think ubuntu one works good from china, you can try
<tat>
so what is about the crane brunch and the other version ( version of what exactly )
<wingrime>
tat: crane is allwinner stock branch
<tat>
i have the right kernel revision, the same compiler, if i have the config i should be able to build a clean driver that can resolve the right addresses
<Turl>
tat: to build module without rebuilding kernel you often need exact same config, kernel source and compiler
<pitillo>
wingrime: thank you very much for your help on friday. Cubieboard is booting right since uboot and sunxi-spl were rebuilt using cubieboard config (instead of mele_a1000). Now we have the release ready for it. Thank you very much to all people working on these devices and their documentation.
<swdwdw>
which source code are you need ftdi-elan or serial-ftdi
<swdwdw>
?
<Turl>
tat: if you do not have config you can try with "sun4i_crane_defconfig" if tablet
<wingrime>
tat: I normaly use linux on sun5i, you better try linux on mmc
panda84kde has quit [Quit: Konversation terminated!]
<wingrime>
Turl: tat use sun5i
<tat>
one more question, where would i find that branch ?
<arokux>
Turl, why would you need a full config to build a single module?
<Turl>
then a13_nuclear_defconfig or such
<Turl>
arokux: because you build the full kernel then take whatever module you need? :p
<arokux>
Turl, mm.. if we have stock src, then we have headers?
<wingrime>
arokux: also, version string
<Turl>
arokux: you need to build it to get headers :p
<tat>
arokux: if the config differs kernel symbols can be shifted and your module has the wrong address, the kernel loader can't handle that in linux
<arokux>
tat, I thought the symbols are resolved at module loading
<tat>
but not dynamicaly
<pitillo>
now is time to finish some server tasks and soon we'll put hands on the cubieboard2 :D
<tat>
if your module has a different address to find the symbol in memmory, then the kernel loader won't accept that, ( sadly windows can handle that kind of bloke )
<arokux>
pitillo, what are you working on, btw?
<swdwdw>
is there anyone who have port the xmbc android to sun-5i platform ?
<pitillo>
arokux: we are building optimized CRUX-ARM releases for the cubieboard, next step will be cubieboard2 (far from all your development work)
<arokux>
pitillo, interesting though
<wingrime>
pitillo: interesting, but you must have already noticed you can easly bring it not only for cb but many other a10 hw, only replace SPL and script
mcbrick has joined #linux-sunxi
<wingrime>
pitillo: like hasng fedora
<pitillo>
wingrime: yeah, we were noticed that people were running on the hackberry A10
<pitillo>
they were using our generic release, giving support only for vfp (v1) but now, as we have hardware, we can support an optimized release for it :)
<tat>
swdwdw: reallocation ( but still not dynamically )
<tat>
arokux: that was
rellla has joined #linux-sunxi
<tat>
well i am still on the lookout for that allwiner crane allwiner stock branch, havn't found it yet, mabe someone can tip me off
<arokux>
tat, well, maybe *this* is working, but something else not.
<wingrime>
pitillo: it little problematic make 100% generic image , SPL related on DRAM settings (how much mem, and how wide bus to it) , script related with screen settings (HDMI or VGA or LVDS or RGB) in this case it important only on tables
<tat>
arokux: that is from my dmesg when i insmod the module "usbserial: Unknown symbol usb_alloc_urb (err -22)"
<tat>
but when i look at /proc/kallsysms there is "c0375794 T usb_alloc_urb"
<wingrime>
pitillo: for make new spl (that we not support ) user need dump registers on living android
jemk has joined #linux-sunxi
<wingrime>
pitillo: same for script with tablets
rellla has quit [Remote host closed the connection]
<pitillo>
wingrime: that depends if I'm right. Currently we are running a generic image as rootfs (not .img to dd directly to a device). There is always a dependencie for device's bootloader and kernel. May be my lack of knowledge (and language) makes me missunderstand your sentence.
<wingrime>
pitillo: so I think best way supply /boot/ partiniton separately
<arokux>
tat, this is puzzling.
<pitillo>
wingrime: well, really we are working directly on the rootfs, boot is directly managed by the user and per device
<arokux>
hno, an out-of-kernel built module should work just fine, provided the ABI hasn't changed
<hno>
arokux, depends.. the version checks is quite strict.
<arokux>
hno, I haven't figured out what "version checks" are, though
<hno>
There is two modes. Version based and symbol based.
vrga has joined #linux-sunxi
vrga has left #linux-sunxi [#linux-sunxi]
<hno>
arokux, see Documentation/kbuild/modules.txt for some details on the symbol based approach.
<arokux>
hno, symbol based is interesting, because it the one that failed, although there was such symbol in kallsyms, maybe with different ABI?
tzafrir has quit [Read error: Operation timed out]
rellla has joined #linux-sunxi
<hno>
arokux, very likely.
<hno>
very many CONFIG_* variables changes module ABI in various parts of the kernel.
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
eebrah is now known as eebrah|away
<arokux>
hno, oh, ok.
<arokux>
hno, so the config is indeed useful.
<arokux>
hno, this make modularity kind of limited. one can develop modules only for special version and special config.
rellla has quit [Remote host closed the connection]
<hno>
arokux, yes. But does not limit modularity as such, only trying to cross modules built for different kernels.
<hno>
loading a module built for another kernel than the current running is not supported.
<wingrime>
arokux: it realy problem try use linux-sunxi on it
<wingrime>
arokux: we have almost all driver works
<arokux>
hno, this limits the ability to supply binary drivers in form of modules...
<hno>
When you build the kernel it's trivial to prepare an "Kernel module SDK" tree with the headers needed for building out-of-tree modules. But that requires that the one that builds the kenel wants you to be able to build modules.
<arokux>
wingrime, I'm just curios. it's not sunxi specific here
<hno>
arokux, binary drivers are not supported, period.
<arokux>
:)
<hno>
accepted by gentlemen agreement, but not allowed.
<arokux>
if binary are not supported, then it makes sense
<hno>
there is ways around it for those who persist in providing binary modules, see for example what NVidia does with their graphics driver.
<wingrime>
arokux: it simple way avoid kernel blobness
bsdfox has joined #linux-sunxi
<arokux>
wingrime, yes, MODULE_LICENCE too.
bsdfox has quit [Changing host]
bsdfox has joined #linux-sunxi
<wingrime>
DKMS
<arokux>
nice reviewer I've got
<hno>
DKMS is a solution to how to hande out-of-tree modules in a reasonably user friendly manner. Actually unrelated to binary modules, but as those providing binary modules have to do it out-of-tree....
<hno>
The "binary modules not allowed" thing is about two aspects
<hno>
- Software freedom
<hno>
- Software support
<hno>
The kernel is GPLv2 licensed, with no amendments. Loading a module into the kernel is runtime linking the module with the kernel and using GPL licensed internals of the kernel.
<hno>
strictly speaking GPLv2 do not regulate what is done runtime, but still.. The module interface is NOT seen as a natural boundary for the kernel. The syscall interface for most parts are however.
<hno>
And software support is simply that free software developers have no chance to know what that binary module you loaded is doing to their code, so most developers won't look at problem reports if the kernel have any binary-only modules loaded.
<wingrime>
hno: but still , ko is ELF, it notmaly must support such things like relocs and export/import tables
<hno>
wingrime, yes it does.
<hno>
and with MODVERSIONS enabled those export/import tables are carefully versioned to avoid mistakes of loading a module built from an incompatible kernel source.
<wingrime>
hno: also module signatures and hashes coming
<hno>
+ that he kernel version as such is also verified.
<hno>
yes, but that is another topic.
<wingrime>
hno: also loading non-tree ko makes problem for things like rootkits
<hno>
and a quite contraversional one.
<wingrime>
hno: also in winnt kernel there such thing like Request Level
<hno>
rootkits can be embedded in-tree as well.. orthogonal issue. But module signatires is relevant there.
<wingrime>
hno: have linux someyhing simular for modules
<hno>
I am not familiar with winnt Request level.
<hno>
but winnt is a different kernel using a different architecture. was originally a microkernel architecture, which allows kernel to be divided in multiple separate processes.
<wingrime>
hno: Easy to explan, when you module get access by callback or something also you, get some Request Level (IRQL). you can't run some functions with that level otherwice you will get bluescreen ( IRQL must be Less or equal)
<wingrime>
hno: it like do somesing funny in IRQ context
deasy has joined #linux-sunxi
<deasy>
samknows box is a real pain...decade of try and it never works
<hno>
Linux also have such rules obviosly. But differently from winnt.
<wingrime>
hno: also , form it level depens can be you code be interrupted by other sings
<wingrime>
*code
<wingrime>
hno: for example, code with level HIer than MemoryManager can have problems with swapped pages
jlj_ has joined #linux-sunxi
<hno>
Very many Linux drivers runs in fully interruptible and scheduled state today.
<wingrime>
hno: when you try access to somesing in SWAP, you get crash, becose MM can't interrupt your code and load page from disk
<hno>
but Linux kernel do not support paged kernel memory.
<wingrime>
hno: its not about code only
<wingrime>
hno: any data
<wingrime>
hno: also I have saw scheduled while atomic very often
<hno>
yes, linux kernel do not support paged kernel memory, neither code or data.
<wingrime>
hno: when debug some SLOW framebuffer
<hno>
scheduled while atomic is bugs. Gets logged to bring attention to the fact that the driver is not acting wel.
<wingrime>
hno: expectly funny that bug also want print and makes boot on it endless
<wingrime>
hno: when I writed IR driver, I get problem, that I can't use printk for debug it
<wingrime>
hno: you will simply lose data, becose printk performance costly
<hno>
wingrime, printk is fine, but you likely do not want them logged to serial console..
<wingrime>
hno: it not cached?
<hno>
it's direct.
<wingrime>
hno: also, question how kprint act with serial, when I printk in IRQcontext
<woprr>
lo
<hno>
if loglevel is above the console limit then it only goes into the klog ringbuffer.
<wingrime>
hno: it usualy need irq for ask from hw
<woprr>
on porting the TSC driver, in which header do we have script_item_u type def or whats our version?
<hno>
woprr, not sure. let me check.
<woprr>
grep is running... but youre faster ;)
<hno>
mach/sys_config.h
deasy has quit [Ping timeout: 264 seconds]
<hno>
a grep of the kernel tree is only some seconds on this old laptop.
<hno>
woprr, ls -l ./arch/arm/mach-sun7i/include/mach/sys_config.h
<hno>
was added as a glue to support sun7i drivers during transition, but wrongly done so don't work well.
<woprr>
haha a link ./arch/arm/mach-sun7i/include/mach/sys_config.h -> ../plat/sys_config.h
<hno>
you should use plat/sys_config.h
<arokux>
wingrime, hno interesting read guys.
<woprr>
ok, and which is the corresponding type for script_item_u ?
<hno>
woprr, there is none.
<hno>
you need to use the type you are asking for..
<woprr>
the I must rewrite the funtions somehow
<woprr>
thx
<hno>
yes a little bit to use gpio type when asking for gpio etc. don't know what they smoked when designing the new interface but it is not an improvement.
<hno>
old interface was bad, this is....
<hno>
woprr, might be a good opportunity to port the driver to mainline instead :)
<woprr>
script_get_pio_list() needs to read from script.bin, right?
<arokux>
vinifr, double check you did everything as described. if it is still not working, then ask.
<hno>
right... need to fix that in u-boot still..
<vinifr>
hno, what branch?
<vinifr>
arokux, ok
<hno>
vinifr, what?
<arokux>
hno, there is 128MiB limit for the kernel to be happy, remember?
<hno>
wasn't hat 16MB?
<hno>
yes I rememebr.
<arokux>
hno, no, 128, it is because of some automatic guessing of zeroaddr (smth like that)
<vinifr>
uboot/sunxi or uboot/sunxi-current
<hno>
vinifr, for what?
<vinifr>
loading .dtb and kernel mainline
<hno>
vinifr, both works.
<hno>
but if you are doing development then please also help testing u-boot current...
<vinifr>
i am using current
<hno>
arokux, I am not sure that applies to fdt placement, but yes kernel load address (what is recorded in the uImage) needs to be in first 128MB.
<vinifr>
arokux, ok. it works!!! :)
<hno>
never seen any other than 0x40008000 used as load address for the kernel on sunxi.
<arokux>
hno, now I see what you mean. I think for the fdt the constraint was it shouldNOT be in HIGHMEM, what ever HIGHMEM is.
<arokux>
vinifr, ;)
<hno>
HIGHMEM is whatever memory the kernel could not map directly.
<vinifr>
Turl, ping
<hno>
which should be none on the available sunxi devices.
<arokux>
hno, mripard_ said there could be an issue on A31s
\\Mr_C\\ has joined #linux-sunxi
<arokux>
(or A31?)
<hno>
I guess there could be an issue on devices with more than 1GB of memory with kernel in 3G/1G user/kernel split.
<vinifr>
Turl, worked well, thank you
<arokux>
vinifr, what are your intentions with mainline kernel?
<arokux>
hno, no idea. sorry.
notmart has quit [Ping timeout: 264 seconds]
<vinifr>
arokux, i wanna test maxine's i2c driver and my lradc driver
<arokux>
vinifr, ok, you'll then continue to hack?
<vinifr>
after other drivers also
<hno>
vinifr, nice
jinzo has quit [Read error: Operation timed out]
mcbrick has quit [Quit: Leaving]
<vinifr>
hno, is it possible to place mmc driver into mainline?
<arokux>
vinifr, I recommend you turn on low level debugging
<vinifr>
no change
<vinifr>
arokux, good idea
<hno>
vinifr, it uses DMA and is not quite up to mainline standards on style and register access.. plus a whopping 0 pages documentation of the registers.
tat has quit [Read error: Connection reset by peer]
<vinifr>
:0
<Turl>
vinifr: :) you're welcome
<arokux>
Turl, why is mturquette so silent?
<hno>
vinifr, the only documentation of the MMC controller is a copy of the text of the maketing brief for the MMC controller IP they are using. Likely not their own, but knowing Allwinner it's likely significantly mutilated.
<hno>
+ the available driver sources.
<vinifr>
thanks to all, i have to go
<arokux>
vinifr, that was fast :)
<vinifr>
arokux, i am in my job, i go to home :)
<lkcl_>
arokux: ping. sorry (checked the irc logs) - to answer your question
<lkcl_>
the wiki is yellow because a friend of mine advised that it's a good colour for making sure you finish off the details of a project :)
vinifr has quit [Quit: Saindo]
<lkcl_>
once the first cards and products are out, i'll pick another colour
<arokux>
lkcl_, yes, please, it is very uncomfortable for me to read
<arokux>
lkcl_, I wonder if its ok for you? :)
<lkcl_>
arokux: i'll find another colour. in the meantime i'll tone it down a bit
<lkcl_>
then consider if you need to ask that question :)
<arokux>
lkcl_, ok :0
<lkcl_>
*lol*
<lkcl_>
i'll make it.... aalllmost white ok?
<jlj_>
lkcl_: I visited that page without javascript on. lol
<hno>
oh, there is a new MMC driver from AW as well..
<arokux>
lkcl_, ok, thank that you care
<arokux>
hno, a new? where have you have it from?
<hno>
It's in out tree alreayd. drivers/mmc/host/sunxi-mci.[ch]
<arokux>
hno, which branch?
<hno>
stage/sunxi-3.4
<hno>
got pulled in as part of the A20 merge.
<arokux>
hno, what is "A20 merge"?
<arokux>
hno, does AW drop the code to us from time to time?
<hno>
the job Hans did at bringing in changes from the A20 SDK source drop into our 3.4 based tree.
<hno>
AW do very rarley send patches. Only 1.5GB tarballs of Android + kernel + u-boot and a bit more.
<lkcl_>
hno: allwinner are supposed to be getting me the 3.4 code (kernel, u-boot) to put into git.rhombus-tech.net
<arokux>
hno, lkcl_ do I understand correctly that the whole allwinner success in europe (eoma, olimex, cubie, hackberry, mk8XX etc.) is due to linux-sunxi.org and rhombus-tech?
<hno>
arokux, yes, all starting with Allwinner releasing mostly complete GPL code to lkcl i initial request.
<hno>
lkcl_, The rumored A20 SDK 2.0?
<arokux>
I wonder what percentage that might be of their overall sales. having said that, there is enormous amount of publicity they've got.
<hno>
of sales it's barely noticeable.
<hno>
completely dwarfed by the tablet market.
<hno>
but marketing quite significant percentage.
<lkcl_>
arokux: yes. they're on farnell, now, and so on.
<lkcl_>
hno: yes.
<arokux>
lkcl_, so allwinners most recent kernel is 3.4?
<lkcl_>
arokux: i believe so
<arokux>
lkcl_, last question: how do you think, if mainlined, will AW support mainline or continue to hack on its own?
<hno>
the bigger question is what improvements have they made compared to earlier release, and how to identify those.. dealing with source drops is rather time consuming.
<arokux>
hno, if we have all the (untouched) code from before it shouldn't be so difficult, if one is familiar with the codebase.
<hno>
diffing two different trees with different and unknown origin is not that easy..
<arokux>
ok
<hno>
AW trees is based on Android kernel + some bits from arm, mainline etc, + their own additions.
naobsd has quit [Quit: Page closed]
ganbold_ has quit [Remote host closed the connection]
Soru has quit [Ping timeout: 276 seconds]
mripard_ has quit [Ping timeout: 248 seconds]
mripard has joined #linux-sunxi
<Turl>
arokux: because he is afk? :) I dunno. You can always email him
<arokux>
Turl, afk?
<Turl>
away from keyboard
<arokux>
Turl, I doubt he'll have time to read an e-mail from a random person.
<Turl>
arokux: most people do when it's directed at them