hno changed the topic of #linux-sunxi to: /Allwinner/sunxi development discussion - Don't ask to ask. Just ask and wait! - See | | Logs at
ZetaNeta has quit [Ping timeout: 260 seconds]
<libv> hramrach: do you have an example program which would trip up, something that isn't gnome-shell and which would require me to change my current installations extensively?
<libv> hramrach: where did this patch come from?
<libv> ah, found it
<libv> well, i didn't find this exact patch now
<tm512> gonna see if the ubuntu boot partition will work with my rootfs
<libv> hramrach: ah, that code does seem legal
<tm512> doesn't work, same blank screen, although the screen has lit up
<tm512> so there's 0 way to find out what is going on without a serial console?
ZetaNeta has joined #linux-sunxi
<tm512> there is nothing in /var/log/lastlog
ganbold_ has joined #linux-sunxi
vinifr has quit [Quit: Saindo]
auxym has joined #linux-sunxi
popolon has quit [Quit: Quitte]
<auxym> whats the command again to configure kernel options in bsp?
<auxym> i need module snd_usb_audio and apparently dont have it
<Turl> auxym: make linux-config is it?
<auxym> aight, checking it out.
<auxym> btw hi Turl. im back to messing around with linux on my mini-x
<auxym> k, USB Audio/MIDI driver under the ALSA stuff is probably the right thing, and was unchecked
<auxym> so, say I has an SD card i set up with bsp, and only want to change the kernel on it, is that possible?
<tm512> so ubuntu can boot from the kernel I compiled. my LFS setup is broken
<tm512> I've got musl libc, statically linked busybox
<auxym> sorry for repeating myself, but, how do I change only the kernel on an existing bootable sd card?
<tm512> should be able to just overwrite the uImage, then install the new modules in /lib/modules on the rootfs
<auxym> uImage is in rootfs? or on the first uboot partition?
<auxym> sorry new to this
<tm512> uImage is on the boot partition
<tm512> I am new to this too
<auxym> alright, il try that, thanks
<tm512> good luck
<tm512> Once I make mali builtin, and also compile in fbcon, how do I specify that as my console device?
<tm512> instead of ttyS0
egbert has quit [Disconnected by services]
egbert_ has joined #linux-sunxi
<tm512> compiling linux-sunxi, I get:
<tm512> ERROR: "ump_dd_secure_id_get" [drivers/video/sunxi/disp/disp_ump.ko] undefined!
<tm512> ERROR: "ump_dd_handle_create_from_phys_blocks" [drivers/video/sunxi/disp/disp_ump.ko] undefined!
<tm512> is the whole output
<tm512> is that safe to ignore?
ZetaNeta has quit [Ping timeout: 256 seconds]
hipboi has joined #linux-sunxi
michaelwang32 has joined #linux-sunxi
<michaelwang32> is this the allwinner a13 chatroom?
<michaelwang32> why is my allwinner a13 so slow?
<michaelwang32> when i go on the internet?
michaelwang32 has left #linux-sunxi [#linux-sunxi]
<auxym> woo. so kernel compiled and booted, with usb audio module loaded. thanks tm512
<tm512> np
<akaizen> so i just got my A20 HDMI sticks
<akaizen> coming from rockchip development... wheres the flashing tools and info for allwinner?
<akaizen> doh... its all on the wiki
TheSeven has joined #linux-sunxi
[7] has quit [Disconnected by services]
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
FDCX_ has joined #linux-sunxi
FDCX_ has quit [Ping timeout: 260 seconds]
<tm512> LFS system boots
rz2k has joined #linux-sunxi
<mripard> Turl: pong
<tm512> now I gotta figure out how to get a login on busybox instead of it just telling me to hit enter
<diego71> tm512: probably you need to configure a getty process in inittab
<tm512> is it so far though
<tm512> bad picture taken from tablet, but eh
FDCX_ has joined #linux-sunxi
<tm512> I don't like how it is saying "enabling swap space" in the slightest, though
<tm512> telnetd works fine. cool.
FDCX_ has quit [Ping timeout: 246 seconds]
iamfrankenstein has joined #linux-sunxi
techn__ has joined #linux-sunxi
techn_ has quit [Ping timeout: 264 seconds]
techn_ has joined #linux-sunxi
techn__ has quit [Ping timeout: 240 seconds]
massi has joined #linux-sunxi
rellla has joined #linux-sunxi
atiti has quit [Read error: Operation timed out]
Black_Horseman has quit [Quit: Zwi se logou mou!!!]
massi has quit [Quit: Leaving]
rellla has quit [Remote host closed the connection]
massi has joined #linux-sunxi
popolon has joined #linux-sunxi
_enrico_ has joined #linux-sunxi
hipboi has quit [Quit: Leaving]
vaxholm has joined #linux-sunxi
<vaxholm> has anyone figured out what all the differen FE* abbrevs mean? FEL FES FED FET FEX
ZetaNeta has joined #linux-sunxi
<oliv3r> vaxholm: why do you ask?
<vaxholm> oliv3r, looking at what LiveSuit is doing. it would be easier to read the code if that was more clear.
pacopad has joined #linux-sunxi
rellla has joined #linux-sunxi
<oliv3r> wingrime removing the test color bars i would think is a bad idea, since it helps devs see how to use it and how to enable it in a new driver, without having ot look stuff up
<oliv3r> i think 3.4 should also be reference source; since that code won't ever get mainlined in its current form, it's ok to have 'cruft' in comments that explain things possibly
<oliv3r> the sun3i removeal i still think is a good thing, as nobody is working on it, so it can potentially only get in the way. once someone starts on it, most code is probably mergable with sunxi anyway
<oliv3r> libv: starting to work towards fosdem allready? So i should do the same :)
atiti has joined #linux-sunxi
<hno> vaxholm, there is no obvious structure in what they mean. But purpose of each is easily identified.
<hno> but it's not interesting what livesuit is doing. We do not need livesuit at all.
<vaxholm> you use u-boot or something?
<vaxholm> for USB-downloading
<hno> yes.
massi has quit [Quit: Sto andando via]
<vaxholm> hno, do you have any tool to upload u-boot using FEL?
<vaxholm> though, i'm curious about why they've made livesuit so insanely complicated for a simple task...
hramrach has quit [Ping timeout: 240 seconds]
hramrach has joined #linux-sunxi
pacopad has quit [Quit: pacopad]
<steev> hno: naobsd: those work with cb2?
<steev> apparently the mailman attempted to deliver mine yesterday, but, rather than leave it at the office, took it back to the post office
<steev> hopefully it comes out today
<steev> already have the XU and Utilite running Gentoo, need to work on something more difficult ;)
<steev> debated working on a dts for the utilite, but i think there will be others who buy it who will do so
<hno> steev, yes.
<hno> except that we do not yet have a reliable u-boot SPL for NAND. A little undecided on what path to select there.
<hno> and too much other stuff to do at the moment.
<hno> the SPL by yuq should work however.
<oliv3r> hno: what paths are you debating between?
<hno> bit of a headache to decide on the proper path. There is both MTD and Allwinner FTL block driver, and the latter in two versions. Neither can coexists with the other.
<hno> The allwinner FTL is a bit nasty in that it automatically formats the flash without asking.
<hno> plus that I am completely overloaded with other stuff at the moment. Not much time to tinker with these right now.
<hno> should be possible to modify the Allwinner FTL to not automatically format the flash I think.
<oliv3r> i'd drop the allwinner driver at some point imo
<oliv3r> or rather, mv sunxi to sunxi-ftl; and start adding mtd to sunxi; dropping all ftl backwards compatibility
<oliv3r> people who need/want ftl, just stick with the old
<hno> probably, but it's what most are using and for goot reasons as Linux do not have a good replacement.
<hno> s/goot/good/
<hno> patents encumbering development of something that makes sense.
<oliv3r> :s
<libv> oliv3r: well, the organizers are ramping up, the deadline for devroom requests is like tomorrow
<oliv3r> libv: really?
<oliv3r> libv: so I should start writing something today :)
<oliv3r> pff, see, when i brought it up 2 months ago, people where saying 'oh its months away, relax'
<hno> libv / oliv3r , devroom where?
<oliv3r> fosdem
<oliv3r> i wanna do something sunxi related; not sure what yet
<oliv3r> libv suggested a devroom
<libv> oliv3r: file a mainline talk, or try to go onboard with something like an ARM devroom (if that finally has come into existence)
<oliv3r> mainline talk even?
<oliv3r> that sounds a bit big don't ya think?
<oliv3r> hno: was reading that yeah, and it's been up since aug. 6th!
<hno> there is big talks, lightning talks, stands, devrooms, and sharing a devroom with others in similar topic.
tinti has joined #linux-sunxi
<oliv3r> i've been there once in 2010 i think
<oliv3r> or was it 09
<oliv3r> so i really wanna go there
<oliv3r> but a mainline talk about sunxi sounds a bit big :)
<oliv3r> maintrack*
<hno> would be great if there was a arm oriented devroom.
<oliv3r> oh maintrack deadline is oct. 1st; devroom is 15th indeed
<hno> yes.
<oliv3r> well there should be enough to fill 45 minutes in a maintrack room i suppose?
<oliv3r> about sunxi? :)
<libv> oliv3r: yes, there is, and it would be good for the project
<oliv3r> libv: ok; then i'll write a proposal this weekend and register for a maintrack room soon
<oliv3r> before oct. 1st :)
<hno> architecture, mali, cedarx, mainlining, etc etc. No issue to fill up 45 min.
<oliv3r> i imagine i'd be done talking in 10 minutes :p
<naobsd> does anyone try 3.4 kernel from SDK2.0 for A20?
<hno> haven't tried it. Only looked and saw too many blobs where I did not want to see them...
n01 has joined #linux-sunxi
<oliv3r> naobsd: wouldn't really care for running it, only to compare against it
<n01> yo 0
vinifr has joined #linux-sunxi
massi has joined #linux-sunxi
jemk has joined #linux-sunxi
deasy has joined #linux-sunxi
hipboi has joined #linux-sunxi
<naobsd> hmm
wingrime has joined #linux-sunxi
<oliv3r> right, going to the 'serverroom' to sort some shit; bbl
<hramrach> libv: technically any program that does a X11 call before passing the display to Mali would trip this. Any call that allocates data, that is. Most libX11 data structures have a mutex in them when the flag set with XInitThreads is set and null otherwise.
tinti has quit [Remote host closed the connection]
<hramrach> I don't have a program ready, sorry
<libv> that's fine
<libv> hramrach: this patched binary of yours, this was a direct diff of the objdump, right?
<hramrach> yes, it does not show where the branch jumps to
<hramrach> the backtrace shows XGetDefault so that call should trigger the crash if done pre/post EGL init
<libv> it jumps to a call which is already in the loaders symbol table
<hramrach> I have a bsdiff as well but no way to print that
<libv> so it jumps straight to the XInitThreads ptr
<hramrach> yes, that's how I expect it to work but can't see it form the disassembly
<libv> there is an extra level of indirection there, something which objdump does not or cannot easily resolve for you
<libv> hramrach: so this is what you use?
<libv> and it solves the problem for you?
<hramrach> yes, I run the patched binary. Wnated to test if the patch works and it did
<libv> ok, good, i will hexedit the same fix in, and commit it with this russian as the author
<libv> i will then go and see if i can implement the same for r3p2-01rel2
<hramrach> thanks
<hramrach> I can give you the bsdiff which hexedits it for you on r3p0
<libv> that also would be good
<hramrach> you cannot pastebin a bsdiff :s
<libv> hramrach: thanks, stored, will integrate this in a few h
<hramrach> thanks
wingrime has quit [Ping timeout: 260 seconds]
stekern has quit [Ping timeout: 245 seconds]
hipboi has quit [Remote host closed the connection]
wingrime has joined #linux-sunxi
<wingrime> oliv3r: ping
Soru has quit [Ping timeout: 260 seconds]
<wingrime> libv: thanks for rewiev, but better some one commets it two _weaks_ ago )
<wingrime> mnemoc: I have strange problem with
stekern has joined #linux-sunxi
<oliv3r> wingrime: pong
<wingrime> oliv3r: some new guy trying merge cedar, can you rewiev
<oliv3r> wingrime: i heard some noise, i'll see if i can find time, very busy with a few things atm
<wingrime> oliv3r: bad
<oliv3r> wingrime: the patch that bad?
<wingrime> oliv3r: bad you have no time
<oliv3r> wingrime: i know! but i am very busy at work these days :S
<oliv3r> need to earn money to pay for my home :)
<wingrime> oliv3r: patch also without name , and singned-off
<oliv3r> so useless :)
<wingrime> oliv3r: v2 needed
<oliv3r> hmm, private inbox: 43 unread; list inbox, 200 unread :(
<oliv3r> ohh patrick's patches?
rellla has quit [Quit: No Ping reply in 180 seconds.]
massi has quit [Quit: Sto andando via]
rellla has joined #linux-sunxi
<wingrime> oliv3r: yes
<atiti> w000t
<atiti> theres a vdpau-sunxi
<atiti> aweeesoomee
<oliv3r> erm no there isn't
<oliv3r> there's heavy development towards something like that
<oliv3r> but it's not a 'drop in replacement' of cedarx
<atiti> well it seems promising none-the-less
ZetaNeta has quit [Remote host closed the connection]
<wingrime> atiti: at least better than blob
<atiti> so no blob is involved?
<wingrime> atiti: thats was general idea make it
<atiti> nice
<vinifr> (OFF) Do you guys have seen the Stanford online courses?
<Turl> vinifr: I probably have, as well as all the others :p
<Turl> vinifr: I signed up on coursera for two courses, but I always forget to follow through :(
<vinifr> nice!
<vinifr> the courses are very good, high level
<Turl> vinifr: did you take one?
<Turl> nice
<Turl> part 2 looks even more interesting :)
<rah> mripard: ping?
<wingrime> libv: we need fb_draw_colorbar ?, it can be done by userspace
<libv> wingrime: simply deleting stuff != cleanup.
rellla has quit [Read error: Connection reset by peer]
<wingrime> libv: that can not be in kernel
<libv> wingrime: upstream is lightyears away
<wingrime> libv: and what is cleanup for you?
<libv> structure.
<libv> wingrime: monkeys follow preprocessor directives, coders find structure where there isn't any
<wingrime> libv: I not follow preprocessor directives here, I just see testcode that draws colorbar , afaik you can write it in userspace
<wingrime> libv: disp have 2-3 api layers , change structure involv massive patches, and may unkilled bugs
<wingrime> libv: I can't remove one layer with small patches
<Turl> oliv3r: ping
akaizen_ has joined #linux-sunxi
akaizen has quit [Ping timeout: 256 seconds]
vinifr_ has joined #linux-sunxi
vinifr has quit [Ping timeout: 260 seconds]
popolon has quit [Ping timeout: 240 seconds]
\\Mr_C\\ has quit []
vinifr_ has quit [Read error: Connection reset by peer]
soul has joined #linux-sunxi
soul is now known as Guest95337
\\Mr_C\\ has joined #linux-sunxi
stekern has quit [Ping timeout: 260 seconds]
stekern has joined #linux-sunxi
Soru_ has joined #linux-sunxi
Guest95337 has quit [Read error: No route to host]
Soru__ has joined #linux-sunxi
Soru_ has quit [Ping timeout: 256 seconds]
jemk has quit [Ping timeout: 260 seconds]
Soru has joined #linux-sunxi
Soru__ has quit [Ping timeout: 260 seconds]
vinifr has joined #linux-sunxi
popolon has joined #linux-sunxi
<atsampson> hno: have you tried u-boot-sunxi HEAD on an A10 device lately?
<atsampson> for me, the "sunxi: set arch timer CNTFRQ" commit breaks it -- it doesn't get into the kernel (irritatingly, I've got no serial port on this device so I can't see what it's saying otherwise...)
<atsampson> (that commit's 26771a8188c88dd3da27e32a118126cceba7c7a9; the previous one fe0cd13e20a42a037064a8a89384e96f1a05c087 works fine)
<oliv3r> Turl: pong
<Turl> oliv3r: I've got two q's for you :p
<Turl> oliv3r: 1) can you confirm the dram on A20 is clocked from PLL6 instead of PLL5 like on A1X?
<oliv3r> wingrime: not entirly true, yes you can write colorbars in userspace, but what if you are writing a driver and are not even close to that, you only want to make the output stage work; you need that then ;)
<oliv3r> Turl: i belive there was a change there, i'd have to consult my u-boot patches though
<oliv3r> i did clean everything up with macro's so should be good
jemk has joined #linux-sunxi
n01 has quit [Ping timeout: 264 seconds]
<oliv3r> there you mean?
<oliv3r> i do see that sun7i and sun4i use different PLL's
<oliv3r> the only thing that I find really strange, is that just above, only PLL5 gets configured
<oliv3r> and PLL6 is just used 'as is'
<oliv3r> hno ^
<wingrime> libv: [PATCH] sunxi:disp: Use msleep_interruptible instead LCD_delay_ms
vicenteH has joined #linux-sunxi
<Turl> oliv3r: 2nd question is, any idea what's the "DLL" AHB gate?
<Turl> it's not described in docs
<wingrime> Turl: DLL - some part of DRAMC
FDCX_ has joined #linux-sunxi
<wingrime> Turl: actualy Delay Locked Loop
<Turl> oliv3r: I discovered all this as I tested PLL6 on A20 and my board died as it was gating PLL6 :)
<oliv3r> ah, why where you messing with PLL6? :p
<oliv3r> well hno can answer you better what the dll is, but there's quite some stuff in dram.c mentioning it, dll0, dllx; DLLCR etc
<Turl> oliv3r: PLL6 is needed for SATA
<oliv3r> on A20 aswell?
<oliv3r> oh whre you gonna try the sata patch? :)
<oliv3r> anyway, the setup of PLL5 and PLL6 will need to be investigated aswell
n01 has joined #linux-sunxi
<oliv3r> so there could still be rudimentary mistakes there
<oliv3r> Turl: what does the a20 sdk think?
<Turl> oliv3r: about what?
<Turl> oliv3r: what sata patch? :p
rellla has joined #linux-sunxi
<oliv3r> ahci patch
<oliv3r> :(
<oliv3r> that one :p
<wingrime> oliv3r: have time for new reviews?
<Turl> oliv3r: I don't have any spare HDDs :)
<Turl> :(*
iamfrankenstein has quit [Ping timeout: 260 seconds]
<oliv3r> wingrime: for some reason USB stopt working on my server, kinda important that I fix that first ;)
<oliv3r> but i have no clue why it's gone :S
<wingrime> oliv3r: your home server?
<oliv3r> aye
<wingrime> oliv3r: just reboot
<Turl> oliv3r: sunxi? :D
<oliv3r> lol no asus :)
<oliv3r> wingrime: i actually even disconnected power
<oliv3r> it appears as if the bus isn't being powered at all anymore
<wingrime> oliv3r: look carefully to wires
<oliv3r> i removed a device, but it still thinks it's there
<oliv3r> pretty sure it worked 10 minutes ago
<wingrime> oliv3r: check power wires, maybe you kill "fuse"
<vaxholm> naobsd, thanks.
atiti has quit [Ping timeout: 264 seconds]
<Turl> oliv3r: intel controller or external one?
<Turl> (or are you an AMD guy? :P)
<oliv3r> amd ohci controller + ehci controler
<oliv3r> lspci shows them
<oliv3r> lsusb shows nothing
<oliv3r> well it shows 2 hubs
<oliv3r> Bus 001 Device 001: ID 1d6b:0002 Linux Foundation 2.0 root hub
<oliv3r> is the only thing i get
<oliv3r> x2
<oliv3r> but i should have tons more
<oliv3r> all the 1.1 root hubs are gone
<oliv3r> maybe that's whats wrong, all my devices are very likly usb 1.1; printer, cardreader, UPS etc
n01 has quit [Read error: Operation timed out]
<wingrime> oliv3r: no, it 2.0 but low speed
<oliv3r> wingrime: 2.0 sits 'ontopof' 1.1 afaik
<oliv3r> | | < > OHCI support for PCI-bus USB controllers | |
<oliv3r> that's a new option!
<oliv3r> arokux: ^ is that because of platform vs pci USB host controller?
<rah> hno: ping?
vinifr has quit [Ping timeout: 256 seconds]
<vaxholm> back from reality. so... u-boot-sunxi. which branch should i use for a cb2?
rellla has quit [Remote host closed the connection]
<oliv3r> that was it
<oliv3r> usb 1.1 driver missing for pci
drachenphone has joined #linux-sunxi
<libv> wingrime: don't harass people
<oliv3r> libv: he means well :)
<Turl> oliv3r: can you recommend some household material that's nonconductive and doesn't have a tendency to attract static?
<atsampson> olive oil?
<wingrime> Turl: antistatic cases from comuter equipment
<wingrime> Turl: pakets form motherboard, etc
<atsampson> no, those are conductive; that's how they work
<oliv3r> Turl: household, that leaves out anti-static bags :p
<oliv3r> Turl: paper.
<atsampson> if you're after something to put a board on, I'd use a towel or something
<Turl> atsampson: yes, that's what I'm needing
<Turl> this cubie had no case
<atsampson> mine's just taped to a bit of card...
<oliv3r> mine just lays on the flat table surface
<oliv3r> though I did move the case from my cb1 to cb2
<oliv3r> i put it either on anti static bags, or just on wood/desk
<oliv3r> not too worried
<atsampson> not a very good photo, but
<oliv3r> atsampson: what's yoru CM status atm?
<atsampson> has the current device tree
<atsampson> graphics, wireless, rotation sensor, SD card, sound all work
<atsampson> the camera initialises now but doesn't actually produce any pictures
<atsampson> this is with a 3.0 kernel, since the prebuilt gcc that configuration uses segfaults while compiling linux-sunxi 3.4 ;)
<oliv3r> you can't use external compiler with CM?
<atsampson> you probably can; I've not looked at it yet though
<atsampson> there's a configuration thingy to select which toolchain to use for the kernel
<atsampson> in terms of changes to other stuff: there's a pull request to fix the vold config in device_allwinner_common...
<atsampson> ... and a hand-hack to build/core/Makefile to pass --configid=$(modelidcfg) to ota_from_target_script
<wingrime> atsampson: two cubie?
<atsampson> which I suspect should actually be fixed by creating the script it's trying to install, although it's not immediately obvious
<atsampson> wingrime: Cubieboard 1 on the left, BeagleBone Black on the right
<wingrime> atsampson: cool
<wingrime> atsampson: they run somethign
<wingrime> ?
<buZz> uboot
<buZz> in endless loop
<buZz> :P
<atsampson> both running Debian -- I use them for build-testing stuff on ARM
<Turl> oliv3r: if you're bored, can you help me find what is currently using pll5 on A20?
<atsampson> there's a Raspberry Pi not in the picture doing temperature/light level logging too... (it's too slow for build testing!)
<buZz> light level logging?
<buZz> how did you implement that?
<wingrime> atsampson: cb1 fast for building?
<atsampson> exactly the same way as for temperature, but with a photocell instead of a temperature sensor ;)
<oliv3r> Turl: a little busy, but i can help think and search a little
<atsampson> wingrime: substantially faster than the RPi, which has *veeery* slow IO
<Turl> oliv3r: maybe DDR is fed from PLL5 and "MBUS" whatever it is uses PLL6?
<hno> oliv3r, The DLLs are delay locked loops adjusting the timinig for the DDR signals. Don't know the full detials.
<hno> atsampson, thanks. I'll double check that change tomorrow.
<hno> rah, better to ask what you need to ask than trying to ping. I am here at random times.
<rah> hno: my machine says "[ 0.816] open img file c:\linux\u-boot.bin failed"
<hno> but always listening..
<rah> hno: and yet the pack script works without error
<Turl> rah: does your nanda have a linux/u-boot.bin file?
<rah> there's a u-boot.bin in bootfs/linux
<rah> Turl: what do you mean by "nanda"?
<buZz> atsampson: hah ok, how did you calibrate the light sensor? :)
<rah> Turl: this isn't an android image
<hno> rah, can't help you much with the pack script. Not doing livesuit images. But there is others here that are familiar with it.
<atsampson> buZz: I didn't -- I don't think it's accurate enough to be worth it ;)
<atsampson> buZz: although I do have a streetlight outside my window that provides a convenient reference level at some times of day
<rah> Turl: if that's what you mean?
<buZz> ah nice
<rah> hno: such as?
<buZz> i use a local university's watt/m² data ;)
<hno> rah, just ask and the'll respond in due time.
<rah> ok
<rah> my machine says:
<hno> nanda is the first partition in your flash, where bootfs goes.
<rah> [ 0.816] open img file c:\linux\u-boot.bin failed
<rah> how do I determine the contents of the first partition from a LiveSuit image?
<hno> you can unpack it and then look what is in the fat filesystem. Or of it's flashed just press 1 (or is it 2, never remember) during boot and you will see it as an USB drive.
<rah> should the usb cable be plugged in before, or after powering on and pressing 1?
<hno> doesn't matter. You need to press the key while boot1 starts.
<rah> well this is interesting
<rah> what is a directory named "linux" under pack/out/bootfs is a file on the device
<hno> that won't work...
<hno> what do the file contain?
<rah> I'm just trying to find out
<rah> less just shows ^@
<hno> hexdump -C
<hno> but if less says ^@ then it's likely just zeroes.
<rah> all zeros
<hno> so check your pack recipe again..
<hno> and maybe try another version of the pack scripts.
<rah> I have
<rah> I've tried two different versions
<rah> both give the same results
<rah> well, the same issue booting at least
auxym has quit [Read error: Connection reset by peer]
<drachensun> rah: can you paste me the output on when pack runs?
<Turl> rah: first partition on nand, as it looks you're using allwinner's boot0/1
<rah> aye, will do
<rah> drachensun: when I get there
<drachensun> ok
<Turl> ssvb: ping
<Turl> ssvb: do you have an A20?
<Turl> ssvb: I did a small change on a clock and I'm getting 1GB/s extra neon fill :)
<ssvb> Turl: yes, I have cubieboard2
<Turl> ssvb: same here
<ssvb> Turl: how fast is the total neon fill performance now?
http_pot_ has joined #linux-sunxi
<Turl> ssvb: 3086.5MB/s
<ssvb> Turl: very nice
<Turl> ssvb: full results
<atsampson> you're sure you didn't change the clock you're using to do the performance measurement? ;-)
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
techn__ has joined #linux-sunxi
techn_ has quit [Ping timeout: 264 seconds]
<wingrime> Turl: what are you changed?
<Turl> wingrime: mbus clock
<wingrime> Turl: how much
<Turl> wingrime: double of what uboot sets now
<ssvb> Turl: didn't we already tweak the MBUS clock earlier?
<Turl> ssvb: possibly, I dunno
<wingrime> Turl: nice result,
<Turl> switching mbus to PLL5 (480Mhz) yields slightly slower results than the tweaked one over PLL6
<Turl> so it might be 600Mhz and the A20 documents may be correct in saying it's "PLL6*2"
<Turl> A13 docs say "PLL6"
<wingrime> oliv3r: wow drivers/video/sunxi/lcd/lcd_bak/ have strange drivers that do i2c and spl by bit-bang
<Turl> ssvb: experimenting with that and the display engine might be good
<wingrime> Turl: also you tryed change dram host priorites
<ssvb> Turl: I think mbus clock was increased from 240mhz to 300mhz earlier -
<ssvb> Turl: but if A20 can handle 600mhz mbus, then it's definitely good
rellla has joined #linux-sunxi
<Turl> ssvb: I'm running mainline headless, maybe one of the other modules craps out with a higher mbus
vinifr has joined #linux-sunxi
<wingrime> ssvb: sun7i cedarx driver also raise freq upper limit
<Turl> ssvb: you can reconfigure mbus from inside linux with no issues, so it should be easy to test, not needing to recompile uboot
<rellla> hno: ping
<wingrime> Turl: you can't test stability so easy
<Turl> wingrime: I know
<ssvb> Turl: how did 480mhz mbus perform (with PLL5)?
<ssvb> Turl: was it faster than the current default?
<Turl> ssvb: yes
<Turl> somewhere in the middle, so I'm inclined to think it's 600 and not 300, and the A13 docs are wrong
<techn__> Turl: strange that C fill is better than neon
<ssvb> Turl: according to the comments in the boot0 code, the current default is 300mhz (not 150mhz) -
<Turl> ssvb: that confirms source is PLL6*2 then
<Turl> techn__: dunno if it affects C fill, but I think this binary is C-A8 optimized
<ssvb> Turl: C fill is using uint64_t pointers for writing to memory (which should map to STRD instructions), so it's not too shabby either
<ssvb> Turl: maybe that's some subtle timing issue
<techn__> Turl: maybe latest compiler optimizes it better than hard coded neon function
<ssvb> techn__: the compiler has no clue about these things, that's mostly a random effect
<Turl> techn__: this is 201305 codesourcery, armel build, mtuned to cortex-a8
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<Turl> (I built it as part of buildroot)
<ssvb> techn__: it could even be that waiting a few cycles between writes could result in better memory bus utilization, or something like this
<ojn> Does anyone here have a copy of the a31 branch from That git server looks completely broken:
<ojn> error: Unable to find 4d6537b9e473af95baef30c2cdf232b5253e0640 under
<ojn> that's the object that's supposed to be at the head of the a31 branch
<Turl> lkcl_: ^
<Turl> ojn: nice to see you playing with an A31 btw :)
<ojn> Turl: Hooked up the eval board and wired it up to the auto-boot system I have. Need to get gmac going so I can at least nfs root it though
<Turl> ojn: if you just need something to take a look at and not that specific release, amery's github has a bunch of A31 code
<ojn> is there a copy of those branches on github? That'd be ideal.
<ojn> url?
<Turl> I don't know if they're the same
<ojn> thanks, that should be close enough
<rah> is charging controlled by software on Allwinner tablets?
<wingrime> rah: no
<wingrime> rah: axp209 ic
<rah> my tablet is in an unbootable state; if I plug it in, will it charge?
<wingrime> rah: yes
<rah> wingrime: this is an A31
<wingrime> rah: than other axp
<rah> it has an AXP221
<rah> I'll presume it will charge
rz2k has quit [Read error: Connection reset by peer]
<hno> atsampson, u-boot A10/A13 bootup fixed.
<hno> rellla, just ask what you need to ask, don't ping. I am here at random times but always listening.
<rellla> hno: i think my problem has been fixed since 2 minutes. let me see if a10 boots with current HEAD
<rellla> and sorry for pinging ;)
<atsampson> hno: yep, that works -- thanks very much!
jemk has quit [Ping timeout: 264 seconds]
rellla has quit [Remote host closed the connection]
<popolon> I see here, there is already code for Cortex-A15 virtualisation on KVM (so should work with A7) for at least exynos and versatile :
<popolon> si is it already supported on A20 please ?
<popolon> s/si/so/
rellla has joined #linux-sunxi
<rellla> hno: as atsampson said, a10 boot is fixed here, too. thanks
<hno> rellla, you see. much smoother just asking instead of pinging.
<rellla> :p
<rellla> though no network access. but thats sth for tomorrow.
<Turl> wingrime: A20 user manual calls cedar "Phoenix 3.0"
rellla has quit [Remote host closed the connection]
<wingrime> Turl: )
wingrime has quit [Ping timeout: 264 seconds]
iamfrankenstein has joined #linux-sunxi
iamfrankenstein has quit [Excess Flood]
List has joined #linux-sunxi
<rah> my LiveSuit images end up with two files, linux and os_show, instead of directories:
<rah> any idea why I end up with files instead of directories?
<rah> this is in the bootfs
<hno> a guess is whatever pack is using for generating the fat filesystem image is borked.
aep has joined #linux-sunxi
<aep> hi, how would i expect the backlight to be exposed to userspace when enabling it in this fex stuff? Can't find anything in /sys ?
<rah> hno: I've written to the exposed USB disk
<rah> hno: I created a removed the linux file
<rah> hno: and the os_show file
<rah> I created a directory named "linux" and copied u-boot.bin into it
<rah> I unmounted the USB drive, pulled the cable out and turned on the machine
<rah> s/turned on/turned off/
<rah> when I turned it on and booted, it still says "open img file c:\linux\u-boot.bin failed"
<rah> when I mount the "nanda" partition file system again, the linux file has reappeared, as well as the "os_show" file
<hno> rah, always have issues when I try writing to the FAT that way.
<rah> :_/
<hno> it is writeable, but seems to be aggressively cached with writeback somewhere.
<rah> (this has put my nose out of joint)
<rah> what a pain
<rah> especially considering I get the same problem from more than one pack script
<atsampson> rah: are you particularly desperate to build LiveSuit images, rather than (say) CyanogenMod .zip files?
<rah> atsampson: what would I do with cyanogenmod .zip files?
<atsampson> well, I don't know; what are you doing with images at the moment?
<rah> atsampson: at the moment I'm trying to find a method of booting a kernel on my machine
<atsampson> if you've got a CM recovery image, it can read .zip files off an SD card or USB, and they can do arbitrary stuff to the filesystems...
<atsampson> hmm -- couldn't you just boot off an SD card, then?
<rah> well, let me rephrase
<rah> I'm trying to find a method of installing a bootable kernel on my machine's nand
<rah> which I can then use in future
<atsampson> ah, there are some instructions for doing that on the cubieboard wiki -- gimme a minute...
<atsampson> (starting from the "The next step is to get the Debian system onto the internal NAND." bit)
<rah> these instructions amount to "boot from SD, hack up the NAND"
<atsampson> yes, essentially
<atsampson> the advantage being that if you fail to hack up the NAND correctly, you can still boot from SD to fix it
<rah> this does not lend itself to automation
<atsampson> what sort of automation are you after? have you got a large number of boards to do this with, or...
<rah> the kind of automation that already exists :-)
<TheSeven> what's the goal?
<rah> to port Ubuntu Touch
<TheSeven> oh, interesting.
<TheSeven> with or without android below it?
<rah> without
<TheSeven> so you basically want to write a pre-built linux image to the nand, without booting from SD?
auxym has joined #linux-sunxi
<rah> I want to get something working so I can start building and testing images
<rah> at the moment, I'm stuck.. at boot1
<TheSeven> on sd or nand?
<rah> nand
<TheSeven> are you using my nand image?
<rah> what do you mean?
<rah> I don't know
<rah> I'm guessing not
<TheSeven> so right now you're booting from SD and trying to write a working nand image from there?
<TheSeven> if so, do you have a /dev/nand device node? or just /dev/nanda etc.?
<rah> I'm not booting from SD
<rah> I'm writing images using LiveSuit
<TheSeven> urgh.
<rah> ?
* TheSeven doesn't like livesuit
<TheSeven> but, tbh, I've never successfully used it (yet)
<TheSeven> I just booted from SD once to write something sane to the nand
<TheSeven> (on my cubieboard, I guess you're dealing with some A10 tablet here?)
<rah> it's an A31 tablet
* TheSeven can't help much then
<vinifr> a patch for sunxi should be against which branch?
deasy has joined #linux-sunxi
<rah> I think I may have found the problem
<rah> $ du -sb out/bootfs out/bootloader.fex
<rah> 5976754 out/bootfs
<rah> 1981440 out/bootloader.fex
<Turl> vinifr: a patch for what?
<vinifr> linux-sunxi
FDCX_ has quit [Ping timeout: 264 seconds]
http_pot_ has quit [Quit: Sto andando via]
<Turl> vinifr: stage/sunxi-3.4
<vinifr> ok
List has left #linux-sunxi [#linux-sunxi]
List has joined #linux-sunxi
<vinifr> Turl, there several code style issues in sun4i-ts, should I fix the problems of style from snippet changed?
<Turl> vinifr: if you want to fix code style do it on a separate patch
<Turl> do not mix style changes and code changes
<vinifr> hm, alright :)
<vinifr> Turl, really get very messy
<popolon> is it possible to write the nand after booting from the sd ?
<popolon> you just need to format the device isn't it ?
<popolon> or at least the ext partition
akaizen_ has quit [Remote host closed the connection]
akaizen has joined #linux-sunxi
akaizen has quit [Remote host closed the connection]
akaizen has joined #linux-sunxi
gzamboni has quit [Read error: Connection reset by peer]