ChanServ changed the topic of #linux-sunxi to: Allwinner/sunxi development discussion - Don't ask to ask. Just ask! - See http://linux-sunxi.org | https://github.com/linux-sunxi/ | Logs at http://irclog.whitequark.org/linux-sunxi
calris has joined #linux-sunxi
bfree has quit [Ping timeout: 252 seconds]
bfree_ has joined #linux-sunxi
Guest99624 has quit [Ping timeout: 260 seconds]
calris_ has quit [Quit: Page closed]
Guest99624 has joined #linux-sunxi
Dave77 has quit []
egbert has quit [Disconnected by services]
egbert_ has joined #linux-sunxi
bremen has joined #linux-sunxi
<bremen> hello
<bremen> what's the latest on the nand mtd driver? I was wondering if it was possible to put linux onto the flash memory replacing android..
drachensun has joined #linux-sunxi
<bremen> drachensun: hi.. any idea what's the latest on the nand mtd driver?
<drachensun> I dont know, I boot Linux from flash with the u-boot that is already installed
<drachensun> I just change the uboot env on nandb, the kernel on nandc and then replace largest Android partition with my Linux rootfs
<bremen> and that works just fine??
<bremen> what about wear levelling??
<drachensun> its got a few complications
<drachensun> you have to use the allwinner memory hack in the kernel
<drachensun> mnemoc was telling me the lichee branch in uboot will do the same thing but not need the hack, I haven't had a chance to try it yet
<bremen> ATAG_ MEM -> y in the kernel config IIRC
<drachensun> yup
<drachensun> but thats only if you have a 1G ram device
calris_ has joined #linux-sunxi
<drachensun> I guess that is the really the only issue I have had with it so far
<bremen> So you've not been having any problems as such .. that sounds pretty good... and plus the FS on the nand hasnt ruined any of your nand chips
<drachensun> not that I've seen
<drachensun> I've had one running for 3 months or so on it
<bremen> what the heck... I was told that without wear levelling (which the nand mtd driver would provide) Linux running off of it would damage the nand .. :9
<bremen> :(
<drachensun> I don't know, it might eventually
<bremen> drachensun: Darn.. I'm gonna try running it off of the nand... you did not use part-probe eh? you've kept the partitions intact.. erm.. is it too much to ask for a quick and dirty script to do that?
<drachensun> In fact, I've got a link all ready
<drachensun> I'm off for now, let me know how it goes
<bremen> drachensun: Thanks :)
calris_ has quit [Quit: Page closed]
bremen has quit [Ping timeout: 245 seconds]
ZaEarl has quit [Ping timeout: 264 seconds]
jquip has joined #linux-sunxi
<jquip> Hallo all :)
mdfe has joined #linux-sunxi
jquip has quit [Ping timeout: 245 seconds]
mdfe has quit [Ping timeout: 258 seconds]
carlo_ is now known as n01
rellla has joined #linux-sunxi
shineworld has joined #linux-sunxi
Dave77 has joined #linux-sunxi
<mnemoc> hno: ping
hansg has joined #linux-sunxi
shineworld has quit [Quit: Leaving]
n01 has quit [Remote host closed the connection]
mdfe has joined #linux-sunxi
n01 has joined #linux-sunxi
arete74 has quit [Ping timeout: 246 seconds]
arete74 has joined #linux-sunxi
<theOzzieRat> mnemoc, did I submit my latest correctly?
<mnemoc> theOzzieRat: by format, it's perfect. haven't looked at the changes yet :p
<mnemoc> lichee/a10-dev and lichee/a13-dev in my github updated up to feb 2013!
<mnemoc> the sanitization was taking too long, so you'll have to live with the sick "real" lichee branch for now
<mnemoc> lichee/a10s-dev updated as well
<n01> mnemoc, sorry for the newbby question, but what is exactly the lichee branch and what is the difference with sunxi-3.0/3.4?
<mnemoc> n01: lichee branches are allwinner's
<mnemoc> sunxi are community driven
<n01> ohhh, ok, thank you :)
<mnemoc> main problem with lichee/* is that they keep a separated branch per chip, while we need to unify everything is a single mach-
<mnemoc> merging can't be done blindly
<n01> goot starting point anyway I guess
<mnemoc> the other problem is that they work on windows, and have no sense of "style"
<mnemoc> n01: yes, specially because they have documentation and info we don't
<theOzzieRat> mnemoc, I was just making sure I submitted correctly. The problem with patches to SPI is 99.9% of ppl will not have any interest
<mnemoc> so they can fix things we don't
<mdfe> hi all, I have some trouble with breaking bcm4330.ko modules build of latest sunxi-3.4 kernel. I took sun4i_defconfig and called make oldconfig before. any sugguestions?
<mdfe> In case I disable bcm4330 the build went fine
<mnemoc> mdfe: that's a gcc/ld error...
<mdfe> bleh
<mnemoc> mdfe: do you get it every time? or haven't tried?
<mnemoc> mali also gives "random" gcc crashes :|
<mdfe> I get this issue every time I tried to build this module
<mdfe> try times in a row
<mdfe> and only this module is affected
<mnemoc> there might be something in your env interfering with that hackish Makefile
<mnemoc> like using the wrong CC
<mdfe> I will increase the verbosity
<mnemoc> and drop parallelism :)
<mdfe> good point
<mdfe> I will try next
<mnemoc> I build -j16, but with V=1 that's a bad idea
Dave77 has quit []
<mdfe> mnemoc: sadly -j1 does not help but the log shows now the build breaks at linking time
<mdfe> I will try to build-in this stuff next
<mnemoc> mdfe: and the right gcc/ld is been used?
<mnemoc> for all involved files I mean
<mdfe> dunno
<mdfe> its mer
<mnemoc> I mean, when you look at the verbose log
<mnemoc> do you see all gcc and ld calls using the right toolchain?
<Turl> mnemoc: ?
<mdfe> the build is done via scratchbox2 inside mer OBS
<mnemoc> Turl: some clock related info allwinner just gave to tsvetan (olimex) to fix some problem they are having
<mdfe> gcc 4.6.4 (should be linaro)
<mnemoc> in that case you should see arm-linux-gnueabihf-gcc-4.6 and arm-linux-gnueabihf-ld
<mnemoc> instead of gcc and ld
<mdfe> Its a fake native build via qemu
<mnemoc> that error you get usually means you are trying to link incompabitible bins...
<mdfe> /opt/cross/lib/gcc/armv7hl-meego-linux-gnueabi/4.6.4/include
<mnemoc> uhm
<mnemoc> need to leave :< back in 30m
<Turl> mnemoc: what was/is the problem?
<mnemoc> no idea
<mnemoc> but those two posts/replies count as usual documentation about the clocks I think
<mnemoc> s/usual/interesting/
<mnemoc> bbl for real now
hramrach has quit [Remote host closed the connection]
hramrach has joined #linux-sunxi
rellla has quit [Ping timeout: 246 seconds]
rellla has joined #linux-sunxi
mdfe has quit [Remote host closed the connection]
rellla has quit [Ping timeout: 264 seconds]
rellla has joined #linux-sunxi
egbert_ is now known as egbert
hipboi has joined #linux-sunxi
theOzzieRat has quit [Ping timeout: 256 seconds]
theOzzieRat has joined #linux-sunxi
hipboi_ has joined #linux-sunxi
hipboi has quit [Ping timeout: 264 seconds]
hipboi_ has quit [Client Quit]
torqu3e has quit [Ping timeout: 245 seconds]
torqu3e has joined #linux-sunxi
rz2k has joined #linux-sunxi
paulk-desktop has joined #linux-sunxi
plan_b has joined #linux-sunxi
<mnemoc> oliv3r: when you have a minute, can you test http://sprunge.us/iejN on top of your sunxi-media-create changes?
ganbold__ has joined #linux-sunxi
shineworld has joined #linux-sunxi
dl9pf has joined #linux-sunxi
ZaEarl has joined #linux-sunxi
<hramrach> calris: finding and reading a file on fat does not require much memory
<hramrach> if you can live with restarting the process from scratch for every file you read
<hramrach> ext* might be different
<libv> dl9pf just made some opensuse rpms for sunxi-mali
<hramrach> cool
* libv is now working through firststeps for an opensuse, to give those packages a whirl
gzamboni has quit [Read error: Operation timed out]
wingrime has joined #linux-sunxi
<mnemoc> if there is a maintainer I would love to have a yum repo at packages.linux-sunxi.org
<hramrach> whatis the mouse text entry app?
<libv> mnemoc: a word in your ear :)
<hramrach> I always forget the name
<mnemoc> :)
<hramrach> dasher I guess
gzamboni has joined #linux-sunxi
<Turl> mripard: we got feedback on the uart patches :)
<hramrach> hmm, it does not work :<
<hramrach> mnemoc: is there streaming parser for fex?
<mripard> Turl: yes, I saw :)
<mripard> I'll modify it the way he wants, and resend them
<hramrach> or do you read whole file an thenn fish out the data
<Turl> mripard: I modified it already
<Turl> mripard: http://sprunge.us/DaRe
<mripard> Turl: oh, really?
<mripard> great
<Turl> mripard: if it looks good to you feel free to send it to the list, I've got some errands to run :)
<mripard> yes, I'll send it
<mripard> don't know when though
<mripard> so, feel free to send it as well when you have some time if I didn't send it already
<libv> hansg: that ba10 tv box that you added fex files for, is it thing documented on the wiki?
<libv> hansg: is it the net'sPC 2?
rz2k has quit [Ping timeout: 256 seconds]
<mnemoc> hramrach: the .bin is not streamable, the blob is separated in 3 parts. first the index of sections, then the index of properties, and then all the values
<hramrach> but if you want only one section you can toss the section index
<mnemoc> the name of the section goes there
<hramrach> read and toss
plan_b has quit [Quit: plan_b]
<mnemoc> wrote that to clean aw's sys_config.c
<libv> i decided to add some "how to get this device up with sunxi" to the a7hd page
<libv> my guess is that users will first try to find whether their specific device is supported, and then like to jump into getting it up from there
<mnemoc> that could be a template I guess
<mnemoc> hramrach: but all drivers using plat/sys_config.h can be switched to plat/script.h for .bin parsing. it should clean up a lot the driver initialization
<hansg> (mine is black, but that is the basic model)
<hramrach> so the properties can be technically anywhere but the prop index for a section is continuous
<libv> hansg: can you spend some time taking photos and filling out a wiki page for it?
<mnemoc> hramrach: yes
<libv> i think that getting the device specific pages filled out better is the key to getting more people to use sunxi
<hramrach> not so bad
<mnemoc> libv: absolutely. I only meant that as there will be duplicated content/hints it might be better to implement some parts as templates. and these templates can also help to categorize the pages
<hansg> libv I agree, and I'll put this on my todo. but I'm not sure when I'll get around to actually doing this.
<libv> mnemoc: yeah, but templates are not for me atm :)
<mnemoc> :)
<libv> oliv3r is the wizard there iirc :)
<mnemoc> libv: make the content thinking in reusage and I can turn it into a template
<hramrach> yes, I have a tegra tablet which I don't use because it has no debug facilities and there are no deailed instructions on getting ut running
gzamboni has quit [Read error: Operation timed out]
<hramrach> I could flash random kernels on it but that is going to get boring quickly with no output unless it works perfectly
<libv> hansg: yeah, i just found a tv box in sunxi-boards, which might have been the one i found on pollin.de (which i am still trying to convince egbert of buying :)) and no information for it on our wiki :)
<libv> and no, it is not the same one :)
<hansg> libv, yeah a wiki with photos is probably the way to go to allow people to identify which device they actually have, I called this one the ba10 because that is what is says on the pcb, the case is not marked in any way
<libv> mnemoc: what would need to be changed? no more direct mention of hyundai a7hd?
<mnemoc> libv: templates can have arguments, and we can dump the page name from a macro
<libv> hansg: ah, as in, a wiki page with a set of smaller frontal photos which the device name underneath and linking to the device specific page?
<libv> s/which/with/
oliv3r has quit [Read error: Operation timed out]
<mnemoc> pictures of teh PCB are important, and if you have the uart pads or other identified magic pads, they are great for the device wiki page too
<libv> mnemoc: for the specific device page, yes
<libv> mnemoc: but i think hans wants to make identification easier
gzamboni has joined #linux-sunxi
<mnemoc> per-feature categories can help there too
<hramrach> well, mabe try something like tablet list and stick list
<mnemoc> we have that
<libv> yeah
<mnemoc> categories are lists
<mnemoc> and you can also add content (as it were a page)
<hramrach> when the photo is generic iPad clone it is not so much help but some devices are distinguished
<libv> hramrach: it will help people to narrow things down to only a few devices
<libv> visually
<mnemoc> awesome ida
<mnemoc> idea*
<hramrach> well, the category list is of known devices, not necessarily supported
<mnemoc> if the reference image is named just like the page of the device we can probably tweak mediawiki to do it magically
<hramrach> can you have photos in categry list or does it require manually created pages?
<libv> mnemoc: is there a way to have such a page autogenerated, when there is like a special 256x256 picture attached to each device page?
<hramrach> magic ftw
<mnemoc> i've seen wikis (in wikia) using pictures as category links
<mnemoc> by default it's only the page name. of pages tagged with that [[Category:Foo]]
<mnemoc> i suppose it's one of the many mediawiki extensions
<hramrach> it sucks that you cannot do subcategories
<hramrach> sw/distro/debian
<mnemoc> hramrach: of course you can categorize categories
<mnemoc> creating a graph of them
<hramrach> but that does not make subcategories
<hramrach> that just adds the cat page innto a cat
<mnemoc> the best way to have pages in multiple categories in simple steps is to tweak the infobox to add all the needed categories based on the given args
<hramrach> maybe the cat listing can do tree if cat is added? there is depth that does not seem to do anything
<mnemoc> there is a category tree thing
<mnemoc> we use in the frontpage to list the device tree
<mnemoc> see "Devices" at the bottom. that's 100% generated
<hramrach> cool
<hramrach> did not know about that bit
<libv> whatever the solution will be, adding "Identifying your device" as the first big thing users see, is going to really simplify things for users
<mnemoc> +1
<hramrach> but can you make like expanded tree?
<libv> if we then add this "images, bsp, hwpack, full build" list to each page, users know exactly what to do
<mnemoc> i think there is an argument to make it expanded by default
<hramrach> oh, it's JS-expandable clicking on tiny [+]
<hramrach> can you weed out the translations somehow?
<libv> mnemoc: a second link after "identifying your device" could be "buying a device", with the community friendly boards, a "caveat emptor" section, and then some features/formfactor based lists :)
<hramrach> users should get the page in their language hopefully or pick when they arrive at a page
<hramrach> but having 10 translations in device list ..
shineworld has quit [Ping timeout: 246 seconds]
<libv> translations are a waste of time and a maintenance nightmare imho
<libv> we are not wikipedia
<hramrach> they are good but should not get in the way
<libv> no, only a handful of initial user pages should be translated, at most.
<libv> anything else is just stupid
<hramrach> i don't care about french translations so long as they do not clutter the category list
<hramrach> but they do
<libv> we really need a wiki nazi, someone who can think structuredly
<hramrach> if people want to translate let them
<hramrach> but remove from category
oliv3r has joined #linux-sunxi
<libv> too much to do on sunxi, and too little good people to do it
<mnemoc> our wiki nazi just arrived
<mnemoc> he silently just volunteered himself
<mnemoc> categories and translation subpage don't fit well. the only way for that is to have separated per-language wikis. which imo it's not good for us
shineworld has joined #linux-sunxi
<bfree_> btw for sunxi-mali on debian you can rebuild the ubuntu libdri2 package untouched ... though I guess without the rest packaged one more dirty make install doesn't really matter :-/ I got it working (test running) yesterday after failing the first time round (I think that was just because Make.config didn't see libX11.so, as the libX11.so symlink to libX11.so.6 comes from libx11-dev which isn't mentioned or really needed)
bfree_ is now known as bfree
<hramrach> mnemoc: can you rename Hackberry A10/fr to Hackberry/fr ?
<hramrach> oh, there is redirect
<hramrach> no need I guess
oliv3r has quit [Ping timeout: 252 seconds]
<hramrach> the spanish translation of cubieboard page is sooo lame
mdfe has joined #linux-sunxi
<hramrach> deletionistic clique exists?
<hramrach> yes, every cubieboard translation is at best a snapshot if previous version of the english page, at worst stub
<hramrach> with the exception of Chinese perhaps
oliv3r has joined #linux-sunxi
<hramrach> bfree: I expect it to be buildable all right
<hramrach> but no debian repo for it
<hramrach> also they have maybe some ubuntu-specific patches or older snapshot
oliv3r has quit [Ping timeout: 240 seconds]
oliv3r has joined #linux-sunxi
<hramrach> bfree: you have any idea how would packaging of sunxi-mali look?
<hramrach> I have some bin-only packages but could not make them replace each other
rz2k has joined #linux-sunxi
<hramrach> ther is glx-alternatives but not covering gles
<bfree> hramrach: nope, I'm not very motivated to look at non-free stuff so I've not looked/thought about it much :-/ not really even sure what your are pointing at as a problem vis-a-vis gles-alternatives
<hramrach> there are fb and x11 separate libs
<hramrach> it's somewhat worth packaging now since we have patches for the libs
<hramrach> mdfe: did you try to push the destdir patch upstream?
<mdfe> not yet
<hramrach> that would certainly help
<mdfe> luc ask me as well
<mdfe> I will do so
<hramrach> thanks
<libv> mdfe: yeah, throw me the patch and i'll push it up straight away
<mdfe> please give me few minutes
<libv> oh, no haste there from my side :)
<hramrach> bfree: do you know some package hosting for Debian like Ubuntu has on LP?
<libv> btw, the image does not seem to want to boot, i'll spend some time after cooking to find out what exactly is wrong there
<zumbi> hramrach: what are you after?
<zumbi> hramrach: there is mentors, but linux-sunxi.org has it's own space for debian-ish stuff
<hramrach> no build server I expect
<zumbi> no, just build locally and upload
<hramrach> one good feature of LP PPAs is clean-build checking
<zumbi> I am working on a local build server, maybe I can share it once it's up
<zumbi> there is build.opensuse.org as well
<hramrach> for RPMs I guess
<zumbi> hramrach: what do you want to build
<zumbi> hramrach: also supports deb
<hramrach> nothing at this moment
<bfree> hehe I've just been using pbuilder native on a cubie ;)
<zumbi> hramrach: but planning on building?
<bfree> I was trying to check there if build.opensuse.org does debian armhf (and if so what releases it covers) ... couldn't see an obvious answer yet
<zumbi> bfree: where you the guy trying to build linux kernel deb packages?
<hramrach> could try building debs for the 3d libs
<zumbi> bfree: OBS site I think does cross builds or emulated,
<hramrach> could do that as well I guess
<bfree> zumbi: yep ... I have some kernel and u-boot packages ... and a horrible rough sunxi-tools now locally
<hramrach> just need to stuff in some disk
<zumbi> bfree / hramrach: try to get your packages done, once I get my stuff up, I'll include them to do clean builds and place them into public repo
<zumbi> please ping me with your dsc files, and I'll take them
<bfree> zumbi: gimme a minute and I'll update my little repo on linux-sunxi.org so you can puke over sunxi-tools in particular :-p
<zumbi> hehe..
<zumbi> I have just ordered my 19" rack to put all the stuff in there
<zumbi> so it'll take a while to get it all up, so no hurries
<zumbi> it would be good to know where your packages are, so I can track them
<bfree> what are you going to put in there?
<hramrach> 3 cdozen odroids on a shelf ;-)
<hramrach> s/c//
<zumbi> hehe.. all devices I have collected over the years
<zumbi> still got mx51 cluster we did for bootstrapping armhf :P
<zumbi> bfree: awesome, thanks
torqu3e has quit [Ping timeout: 245 seconds]
torqu3e has joined #linux-sunxi
hansg has quit [Remote host closed the connection]
<wingrime> mnemoc: can you cherry pick i2c fixes
<mnemoc> wingrime: unless I seriously broke something and needs urgent fixing, all changes go via the ML
<mnemoc> and there I need to give people some time to react and comment
rellla has quit [Quit: Nettalk6 - www.ntalk.de]
mdfe has quit [Read error: Connection reset by peer]
<mnemoc> wingrime: oh, that's from the -dev
<mnemoc> wingrime: cherry-picked to both stage branches
mdfe has joined #linux-sunxi
plan_b has joined #linux-sunxi
carlo_ has joined #linux-sunxi
wingrime has quit [Ping timeout: 256 seconds]
shineworld is now known as silverio
silverio is now known as shineworld
n01 has quit [Disconnected by services]
carlo_ is now known as n01
carlo_ has joined #linux-sunxi
n01 has quit [Remote host closed the connection]
mdfe has quit [Remote host closed the connection]
plan_b has quit [Quit: plan_b]
rz2k has quit []
vinifm has joined #linux-sunxi
vinifm has quit [Quit: Saindo]
shineworld has quit [Quit: Leaving]
<hramrach> wtf, sw_hcd_suspend only gets device, not state
<hramrach> techn_: suspending the console does not avoid the fbcon oops: http://hpaste.org/84125
<Turl> mripard: :) I was going to reply to your eth patch
<Turl> mripard: check wemac.c too, it says allwinner
<techn_> hramrach: you could widen console locking
<techn_> hramrach: I cannot reproduce that oops
<hramrach> just susp/res
<techn_> for me it doesnt crash
<hramrach> with X running if that matters
<Turl> mripard: also, if you send a v3 of the uart series, please pick "ARM: sunxi: cubieboard: Add UART muxing" so it goes in together
<techn_> I'm not running X
<hramrach> for me it crashes sometimes
<mripard> Turl: actually, it still sits in my inbox
<hramrach> thought the disp suspend helps but may be irrelevant
<Turl> v4*
<mripard> so don't worry, it will get in :)
<Turl> ok
<hramrach> oh, so it crashes when disp is already suspended during suspend
<hramrach> eg xset dpms force off
torqu3e has quit [Ping timeout: 245 seconds]
torqu3e has joined #linux-sunxi
<hramrach> ugh
<hramrach> hcd registers struct platform_driver.driver.pm.suspend but disp registers platform_driver.suspend
<hramrach> the latter gets state, the earlier does not
<hramrach> reverted the disp part - no change
<hramrach> techn_: I have no idea what else I would lock
<hramrach> it's not like there is much more to do
<hramrach> and generally the disp suspend seems to have no effect
<techn_> I checked other similar drivers
<hramrach> so it should lock in hte notify?
<techn_> or maybe we should suspend console while we are changing modeset
<techn_> suspend fb
<hramrach> I put in some debug prints to see where it crashes
<hramrach> it crashes way after disp resume finishes
<hramrach> so not much related
<hramrach> but also wakes up system immediately
<hramrach> mhh, it resumes some clock which should keep down
<hramrach> well, it's just broken
<hramrach> I can connect and disconnect the hdmi in dpms off and no reading is done
<hramrach> but on resume from suspend you get hdmi reading while in dpms off and crash
Guest99624 has quit [Remote host closed the connection]
<libv> pfff booting suse on this thing is rougher than expected...
paulk-desktop has quit [Quit: Ex-Chat]
<libv> the hyundai definitely is not the device for that...