<kenny>
TheSeven: yes, no output http://pastebin.com/djFQfFeP . I think I'll set up a tftp server and try another kernel on there. I'm not sure if cubian has earlyprintk configured.
<kenny>
in theory should dd-ing a latest u-boot-sunxi actually work though? Or was there other steps I was missing?
naobsd has quit [Quit: Page closed]
<TheSeven>
I *think* that should be sufficient, but I'm mostly dealing with lichee-based nand installs
booss has joined #linux-sunxi
<kenny>
okay, thank you. I might try the previous u-boot nightly as well just to make sure.
heffer has joined #linux-sunxi
<SiCo>
TheSeven: repack require a ramdisk-directory
<TheSeven>
you don't need that part of it
<TheSeven>
just mkbootimg
<TheSeven>
the zip file contains some more
<SiCo>
ok
mosfet_ has quit [Quit: Page closed]
printallthething has joined #linux-sunxi
<SiCo>
TheSeven: mkbootimg require ramdisk, i've touched one file but don't works
<on1x>
ok, theres any updated page in linux-sunxi, any specific git branch ?
<SiCo>
TheSeven: now i do dd if=out.img of=/dev/nandc ?
<SiCo>
TheSeven: sure?
arete74 has quit [Ping timeout: 246 seconds]
arete74 has joined #linux-sunxi
<SiCo>
TheSeven: please don't leave me alone...
arete74 has quit [Ping timeout: 240 seconds]
arete74 has joined #linux-sunxi
<SiCo>
TheSeven: too late, I tried it and it did not work. The cubieboard not start again. A red LED is turned on and shows no signs of life
<TheSeven>
hm... that was supposed to work
<TheSeven>
any uart output?
<TheSeven>
this could be anything from the bootloader not liking the kernel format to the kernel itself being broken
<TheSeven>
hm, or maybe a bad load address
<TheSeven>
you might need to specify --base 0x40000000 in the mkbootimg command line
<SiCo>
yes but now?
<SiCo>
how i can recovery
<TheSeven>
either reflash the old one through FEL mode, or just boot from a card and fix it
<SiCo>
i nedd to reinstall all with phoenixsuite?
<TheSeven>
just writing that partition would be sufficient
<TheSeven>
however it isn't really easy to trick livesuit into doing that
<TheSeven>
easiest way is to boot from the sd card
parabyte has quit [Quit: Leaving]
<SiCo>
ok tomorow i'll try to reflash or using a sd cardfixing the problem
<SiCo>
thank you so much
<SiCo>
bye
SiCo has quit [Quit: Page closed]
<kenny>
u-boot from github.com/cubieplayer/cubian-updates/tree/a20 works fine, nightly doesn't boot, same kernel. Anyone know why?
<Turl>
kenny: different machid or something?
<Turl>
are you using the same environment?
<Turl>
you may want to ask whoever built that uboot
<kenny>
not sure, googling. It's the same environment. I just dd'd the last couple u-boot-sunxi nightly images and I can't get them to boot
<Turl>
build a kernel with DEBUG_LL and EARLY_PRINTK and pass earlyprintk on the bootargs, if it's different machid you'll see an error
<kenny>
I'll try that. I was trying to update u-boot before messing with the kernel, but I'll move to a new kernel build first.
ssvb has quit [Remote host closed the connection]
<Turl>
good night everyone
<kenny>
g'night. Thank for the help
<kenny>
*thanks
<Turl>
you're welcome :)
Gerwin_J has quit [Quit: Gerwin_J]
Gerwin_J has joined #linux-sunxi
Gerwin_J has quit [Quit: Gerwin_J]
Sonic1_ has quit [Ping timeout: 240 seconds]
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
KBme has quit [Ping timeout: 252 seconds]
Gerwin_J has joined #linux-sunxi
hipboi has joined #linux-sunxi
eebrah has quit [Ping timeout: 250 seconds]
eebrah has joined #linux-sunxi
eebrah has quit [Read error: Connection reset by peer]
eebrah has joined #linux-sunxi
eebrah has quit [Ping timeout: 245 seconds]
eebrah has joined #linux-sunxi
Gerwin_J has quit [Quit: Gerwin_J]
eebrah has quit [Read error: Connection reset by peer]
JohnDoe_71Rus has joined #linux-sunxi
eebrah has joined #linux-sunxi
eebrah has quit [Ping timeout: 250 seconds]
eebrah has joined #linux-sunxi
eebrah has quit [Read error: Connection reset by peer]
eebrah has joined #linux-sunxi
<wens>
using SSD with cb2. no high frequency noises :)
TheSeven has quit [Read error: Operation timed out]
TheSeven has joined #linux-sunxi
eebrah has quit [Ping timeout: 245 seconds]
eebrah has joined #linux-sunxi
eebrah has quit [Ping timeout: 272 seconds]
eebrah has joined #linux-sunxi
wolfy has joined #linux-sunxi
tomboy64 has quit [Read error: Connection reset by peer]
tomboy64 has joined #linux-sunxi
ZetaNeta has joined #linux-sunxi
y0g1 has joined #linux-sunxi
ganbold_ has quit [Ping timeout: 245 seconds]
HeHoPMaJIeH has joined #linux-sunxi
Tsvetan has joined #linux-sunxi
BluesBoy has joined #linux-sunxi
rz2k has joined #linux-sunxi
_massi_ has joined #linux-sunxi
cubear has joined #linux-sunxi
<Tsvetan>
hi guys
Black_Horseman has quit [Ping timeout: 252 seconds]
<Tsvetan>
we try to attach 1920x1080 LVDS display to A20 and there is bug in the LCD driver
<Tsvetan>
if the number of LVDS channels is more than 1 the clock is divided by the number of channels and not set properly
<Tsvetan>
and LCD flickering, writting correct values in the registers fix it
<Tsvetan>
but linux reports wrong LCD refresh etc
rellla has joined #linux-sunxi
rellla has quit [Ping timeout: 246 seconds]
rellla has joined #linux-sunxi
Quarx has joined #linux-sunxi
shineworld has joined #linux-sunxi
<JohnDoe_71Rus>
I do not remember who helped me yesterday with wifi. Did the enable of settings. But constantly switched on / off.
nfet_ has quit [Ping timeout: 260 seconds]
tomboy65 has joined #linux-sunxi
y0g1 has quit [Remote host closed the connection]
<tomboy65>
my endeavours last night with the mali binary driver resulted in the loss of my hdmi module
* tomboy65
sighs
<oliv3r>
morni'n
<oliv3r>
wens: put your ear closer
<oliv3r>
Turl: na, 'real officially' it's with 2, but people so often get it wrong, i never used it as such. But I really really should use my proper name on that stuff
<oliv3r>
perabyte, there is none; zet6221 was RE-ed by wingrime
<oliv3r>
parabyte with the help of some code dump
nfet_ has joined #linux-sunxi
n01 has joined #linux-sunxi
panda84kde has joined #linux-sunxi
n01 has quit [Ping timeout: 252 seconds]
n01 has joined #linux-sunxi
* Montjoie
is going mad, with waitqueue in my crypto driver, cryptsetup benchmark show me x3 perf gain, but in the mean time my benchmarj show me /3 perf gain...
<wens>
oliv3r: pretty sure there isn't :)
<wens>
on the ct, i can here it from arm's length
Fusing has joined #linux-sunxi
deasy has joined #linux-sunxi
<wens>
brcmfmac supports platform callbacks, seems slightly better than rfkill
bfree has quit [Ping timeout: 252 seconds]
arete74_ has joined #linux-sunxi
Wizzup_ has joined #linux-sunxi
speakman has quit [Ping timeout: 248 seconds]
futurisk_ has joined #linux-sunxi
speakman has joined #linux-sunxi
speakman has joined #linux-sunxi
speakman has quit [Changing host]
morfoh_ has joined #linux-sunxi
uwe__ has joined #linux-sunxi
techn_ has joined #linux-sunxi
dwilkins has quit [Ping timeout: 260 seconds]
Wizzup has quit [Ping timeout: 260 seconds]
arete74 has quit [Ping timeout: 260 seconds]
mripard has quit [Ping timeout: 260 seconds]
shineworld has quit [Excess Flood]
morfoh has quit [Ping timeout: 260 seconds]
uwe_ has quit [Ping timeout: 260 seconds]
MadSpark has quit [Ping timeout: 260 seconds]
shineworld has joined #linux-sunxi
MadSpark has joined #linux-sunxi
futurisk has quit [Ping timeout: 260 seconds]
techn__ has quit [Ping timeout: 260 seconds]
mripard has joined #linux-sunxi
shineworld has quit [Quit: Leaving]
shineworld has joined #linux-sunxi
bfree has joined #linux-sunxi
FR^2 has joined #linux-sunxi
maz_ has joined #linux-sunxi
notmart has joined #linux-sunxi
PiyushVerma has joined #linux-sunxi
<oliv3r>
ssvb welcome back! (for a little bit anyway appearantly :p)
<rz2k>
oh god
<rz2k>
dude with @broadcom email at our mailing list
<rz2k>
oliv3r: if you have time - please try to run bluetooth on CT using the brcm_patchram_plus and the .hcd file from Benn, it should work out of the box on low speeds of the UART (like 115200)
<oliv3r>
Tsvetan: oh nice find
<rz2k>
then if you will be still interested you can try to run 300000 and etc, full list of bandwidthes is somewhere in brcm_patchram_plus.c source
<oliv3r>
wens: i'm even getting a noise without anay driver; i should hook up a mic and record it for analsys :p
<oliv3r>
wens: ohh, i can hear it on the CT only with my ear up close
<oliv3r>
rz2k: i have no experience with bluetooth, other then connecting my headphone/keyboard to my phone
<rz2k>
oh
<rz2k>
ok then, I will get there sometime
<rz2k>
it should have no problems, usually, it is pretty standartized
AreaScout has joined #linux-sunxi
boycottg00gle has joined #linux-sunxi
<oliv3r>
yeah but i don't even know how the userspace works :p
maz_ has quit [Ping timeout: 272 seconds]
KBme has joined #linux-sunxi
<tomboy65>
is there any development going on with mali binary drivers? the last commit to mali kernel drivers was 11 months ago.
<tomboy65>
(yeah, i know, it's binary blobs, still ...)
prasannapete has joined #linux-sunxi
tzafrir has quit [Ping timeout: 250 seconds]
<oliv3r>
tomboy65: there won't be any development since its blobs
<oliv3r>
tomboy65: if allwinner releases new bins, those will obviously be commited
enrico_ has joined #linux-sunxi
<wens>
rz2k: i get "writing \ 01 03 0c 00" over and over
<rz2k>
wens: try --use-bandwidth-for-loading or something like that
<rz2k>
--use_baudrate_for_download
<wens>
i did
<rz2k>
check private message
<rz2k>
I posted how it works at bcm4330
<tomboy65>
oliv3r: that was what i was asking, whether any dev is done from the providers (e.g. aw)
<wens>
brcm_patchram_plus just keeps sending hci reset
<wens>
maybe i screwed up my uart config
morfoh_ is now known as morfoh
tiramiseb has joined #linux-sunxi
fredy has quit [Excess Flood]
<tiramiseb>
Hello, I'm trying to install Debian on the NAND of a Mele A2000G. Boot on SD-card is ok but I don't find where is my error when installing on the NAND... I've manually followed the "Pimp my mele" script on guillaumeplayground.net, maybe I'm doing something wrong with the script.bin or the u-boot.bin file.
popolon has joined #linux-sunxi
fredy has joined #linux-sunxi
<oliv3r>
tiramiseb: what is your question?
<TheSeven>
what is that? a10?
<tiramiseb>
Good point :)
<tiramiseb>
Is the a10 yes
<TheSeven>
what happens?
<tiramiseb>
Nothing displayed on the screen, system not reachable on the network
<tiramiseb>
when booting on the sdcard, I can connect with SSH within 15 or 30 secs
Fusing has quit [Ping timeout: 250 seconds]
<tiramiseb>
I've partitioned the nand with /dev/nanda 16M, /dev/nandb the rest. I've put the Debian files hierarchy in nandb
<TheSeven>
ok, so you don't have a serial port for debugging?
<TheSeven>
if you have a different a10 board, you'll probably have to replace the fex file on /dev/nanda
<tiramiseb>
on /dev/nanda the majority of files are from the previous sytem (android in this case)
<tiramiseb>
I've only changed the "linux" directory
<tiramiseb>
is this wrong ?
<TheSeven>
you'll have to replace uboot inside that linux directory, not sure if anything else as well
<tiramiseb>
I'm not sure if the uboot I have is correct
<TheSeven>
I prefer just wiping that crap away and putting there what I need (see the image that I linked above)
eebrah has quit [Quit: Lost terminal]
<tiramiseb>
I've tried with the one from guillaume, the one pat wood has link in guillaume's blog comments, I also have cross-compiled it from the lichee-dev branch...
<tiramiseb>
no luck with these...
adb_ has joined #linux-sunxi
<TheSeven>
my recommended approach is:
<TheSeven>
2. reboot
<TheSeven>
3. create ext4 filesystem on /dev/nandb
<TheSeven>
1. flash my image to the nand
<TheSeven>
4. copy a working rootfs there
<TheSeven>
5. copy your kernel to /boot/uImage inside that rootfs
<TheSeven>
6. replace the fex in /dev/nanda
<TheSeven>
7. reboot and remove the sd card
<TheSeven>
(with "the fex" I mean script.bin)
<TheSeven>
alternative approach: get a uart cable, check what's wrong, and fix it
<tiramiseb>
There is no serial port on the A2000G. at least no apparent one... I'd rather not open it to check if there is one hidden in there
<tiramiseb>
I'm flashing your image
kanedafp has joined #linux-sunxi
Fusing has joined #linux-sunxi
<tiramiseb>
TheSeven : unfortunately, I get the same result : blank screen and unreachable on the network
<oliv3r>
tiramiseb: it should have a serial port; all mele's do
<oliv3r>
some need some soldering
HeHoPMaJIeH has quit [Quit: Leaving]
<tiramiseb>
so I should open it and check on the board...
<tiramiseb>
ok, I'll first try with the (somewhat old) headless "Debian wheezy with nand install" image from Guillaume and check if the NAND install works this way and if I can get a clean system with a no-neadless kernel. then if I don't get any result, I'll open the mele...
jinzo has joined #linux-sunxi
<TheSeven>
tiramiseb: do you have a 3.3v usb-serial cable?
<tiramiseb>
the only usb-serial stuff I have is an adapter which adds a serial port on /dev/ttyUSB0, on another computer...
<TheSeven>
you'll need such a one with 3.3V TTL output
<TheSeven>
(or build a simple level shifter with 2 transistors and a few resistors)
<tiramiseb>
ouch
<tiramiseb>
I'm more a system&network guy than an electronic guy :p
<tiramiseb>
I nearly never play with electronic stuff
<oliv3r>
mripard: more free-electron folks involved! :p
paulk-collins has joined #linux-sunxi
<tiramiseb>
TheSeven: I'll first check if I can run Guillaume's old Debian image on the NAND, change the kernel and upgrade the system. Then if it doesn't work I'll try finding the serial port and debugging the boot process...
<oliv3r>
paulk-collins: hey paul!
<paulk-collins>
hi oliv3r !
<paulk-collins>
oliv3r, guess what? I received a ZaTab recenty!
<paulk-collins>
(first gen)
<oliv3r>
paulk-collins: that's awesome! perfect for allwinner hacking :D
<oliv3r>
paulk-collins: little update, remember how atsampso1 was planning on doing replicant/CM support for allwinner chips?
<oliv3r>
paulk-collins: he kinda kdidn't liket he UI so quit that effort :(
<paulk-collins>
mhh yes, I haven't heard about it in a while
<oliv3r>
while i'm no expert (and dn't have time atm) it shouldn't be a lot of work though, i can't imagine it is
<oliv3r>
ignoring things that require cedarX for now
<oliv3r>
besides Mali, everything we need is available in the 3.4 kernle
<paulk-collins>
right
<plaes>
paulk-collins: could you make some pics from innards of the Zatab?
<paulk-collins>
I was thinking of writing an allwinnerd which would be daemon that'd set things up right for android reading the script.bin
<oliv3r>
yeah i don't think we have those on our wiki yet
<paulk-collins>
for instance linking the right wifi module in place
<n01>
oliv3r: i see your name with two "l" on sunxi ml (gmail web)
<oliv3r>
paulk-collins: why? i mean, allwinners android images deal with script.bin allready (the kernel does all the work)
<oliv3r>
n01: yeah :p
<oliv3r>
n01: that is my true and real name as my mother gave it to me
<paulk-collins>
oliv3r, well Android has to load the wifi driver module at a fixed location
<oliv3r>
n01: i decided to start signing my mainline work with my real proper name
<paulk-collins>
and there are different wifi chips
<n01>
hooo ... lol, ok
<paulk-collins>
I know that for most of the rest the kernel does it
<oliv3r>
paulk-collins: how does replicant/android deal with USB wifi (that doesn't need special GPIO treatment)
<paulk-collins>
oliv3r, well it's not supported at all
<paulk-collins>
but it could be
<oliv3r>
paulk-collins: so if I have some random android box, and plug in a USB wifi; it won't pick it up?
<paulk-collins>
nope, it won't
<paulk-collins>
a lot of things are hardcoded in Android
<TheSeven>
paulk-collins: USB wifi just works fine here...
<paulk-collins>
the wifi module path is
<paulk-collins>
TheSeven, in Android?
<TheSeven>
yes
<paulk-collins>
TheSeven, while the device has an embedded wifi chip?
<TheSeven>
no, it doesn't have an embedded one
<paulk-collins>
ok
<TheSeven>
that would be pointless, as android only supports just one chip at a time anyway
<paulk-collins>
also I suppose you took the dongle that is supposed to work with the board?
<paulk-collins>
because if you use another one, I doubt it'll work
<paulk-collins>
because the wifi module path is hardcoded
<paulk-collins>
at least usually it is
<paulk-collins>
but that can be worked around
<TheSeven>
well you do of course need to have the right kernel modules loaded and firmwares available
<paulk-collins>
ah ok
<paulk-collins>
yes sure if you load the module manually
<paulk-collins>
it works
<paulk-collins>
right
<JohnDoe_71Rus>
oliv3r: you need kernel module to wifi
<TheSeven>
or just compile it in
<oliv3r>
JohnDoe_71Rus: i'm obviously assumeing that linux will work just fine with the wifi dongle
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
<TheSeven>
no real need to keep it as a module anyway
<paulk-collins>
TheSeven, mhh I never tried to have it builtin
<paulk-collins>
I wonder if it's properly turned off then
<oliv3r>
so what if i have a box, plugin 2 wifi usb dongles (that are fully supported by the kernel midn you)
<oliv3r>
android will only work with the first one (hardcoded path thing)
<paulk-collins>
TheSeven, that's worth a try, thanks for the idea
<oliv3r>
but if i open a terminal, i could in theory enable the second one etc
<TheSeven>
(or just load all the modules in init.rc)
<TheSeven>
you could possibly even load them on demand, check what allwinner did with usbmodeswitch for some 3g dongles
<JohnDoe_71Rus>
In Linux more space, more supported hardware. androdi to build only the modules needed.
<paulk-collins>
well I know for sure it handles only one at a time
<TheSeven>
that must be using some trigger when a new device is plugged in
<oliv3r>
so you can't turn it off with built in :)
<paulk-collins>
yes that's my concern
<paulk-collins>
I'll make a quick try on my laptop
<oliv3r>
so android doesn't do rfkill
<paulk-collins>
brb
<paulk-collins>
no I don't think it does
shineworld has quit [Quit: Leaving]
<paulk-collins>
mhh
<paulk-collins>
I should ask GNUtoo, he probably knows that
<oliv3r>
i don't see him (or TJ) in replicant; they still around?
<paulk-collins>
yes GNUtoo is stil around, but he's taking a break these days
<paulk-collins>
though we still meet in real life
<paulk-collins>
for TJ, I don't know how far he went the GPS things
<paulk-collins>
I have an university teacher helping me with GPS btw
<oliv3r>
ah cool
<oliv3r>
so you picking up where TJ left off?
<paulk-collins>
I'm rather working on the GSD4t, the one in the Galaxy S2
<paulk-collins>
He was working on the other one
rz2k has quit []
fredy has quit [Excess Flood]
fredy has joined #linux-sunxi
<oliv3r>
the S3 one
<oliv3r>
ok cool
<oliv3r>
well my GF has the S2
<oliv3r>
and i strongly feel on getting the S3
<paulk-collins>
it's a nice phone, yes
<oliv3r>
as it's a bit better; and just as well supported i think
<oliv3r>
and getting an S2 new is almost impossible; and used S2, you have to get really lucky to get a nice one that's not messed up (internally (flash) and externally (glass)
<oliv3r>
but I need TJ's work! :p
<tiramiseb>
TheSeven: ok, the result with Guillaume's image is worse : it doesn't see /dev/nand and it doesn't want to boot on another kernel. I don't like the way it's going. I will maybe try to open the Mele later.
<tiramiseb>
I will come back here another day if I need more help
<tiramiseb>
thanks for the suggestions, thanks oliv3r too for your comments
<tiramiseb>
bye
tiramiseb has left #linux-sunxi [#linux-sunxi]
<paulk-collins>
oliv3r, well honestly, I don't have a lot of hope for GPS…
<paulk-collins>
ah btw, I think the Fairphone will have working GPS with Replicant
<paulk-collins>
we'll see that soon
enrico_ has quit [Remote host closed the connection]
HeHoPMaJIeH has joined #linux-sunxi
HeHoPMaJIeH has joined #linux-sunxi
HeHoPMaJIeH has quit [Changing host]
uwe__ is now known as uwe_
prasannapete has quit [Quit: prasannapete]
<oliv3r>
paulk-collins: yeah im' not gonna go the fairphone way :(
<oliv3r>
paulk-collins: seperation with the modem is bad, and Mediatek 6589 appears to be a powerVR part?
boycottg00gle has quit [Ping timeout: 260 seconds]
<oliv3r>
paulk-collins: so the argument is then modem seperation + powerVR vs gps
<oliv3r>
i looked it up on the wiki; and the fairphone is appearantly using the 6589
<oliv3r>
So no thank you for now
<slapin>
oliv3r: what is so ad about separation with modem?
<slapin>
I mean what is so *bad* about modem separation...
<oliv3r>
slapin: if the modem has direct access to RAM, GPS, the modem basically is a small computer inside your SoC
<oliv3r>
if you have the source for that, it doesn't matter
<oliv3r>
but usually, these radio(-firmwares) are blobs
<oliv3r>
now if mediatek is forced by chinese/usa government to spy on its users
<oliv3r>
they can use the gps data, and send it out over the air, without the kernel/linux/android even knowing it is happening
<slapin>
oliv3r: so there is no actual separation, it is just damn part of SoC, and even probably in software using some soundish uart or i2s to communicate to radio...
<oliv3r>
if the modem is tightly integrated, it could even enable your microphone and transmit taht data
<oliv3r>
slapin: if they are properly seperated, yes :)
cubear has quit [Quit: Leaving]
<slapin>
oliv3r: for me it looks more like marriage, not separation...
<paulk-collins>
oliv3r, yes I know -- though many people want it, so we'll likely port to it
<paulk-collins>
we know it's not the best target out there
<oliv3r>
slapin: a proper marriage, has a uart/spi/i2c link for control/data; i2s or the like for audio (voice) for where that still is done the old fashioned way
<oliv3r>
paulk-collins: playing devils advocate here, won't that go against the project?
<oliv3r>
paulk-collins: I mean, how different will support for that phone be compared to CM?
<oliv3r>
CM - 3D blob?
<slapin>
oliv3r: for properly separated I mean communicating via UART for commands and i2s/analog for voice, SoC or not...
<oliv3r>
slapin: that's a proper marriage
<paulk-collins>
oliv3r, first of all, we cannot know whether the phones we support do have modem separation
<oliv3r>
slapin: these integrated modems tend to have access to NAND, RAM, GPS, Audio chips
<oliv3r>
wens: heh, i was sitting on the throne today; thinking once things settle down for me a bit and the other bits i'm working on are done and merged; audio looks like a nice bigger, yet doable, challagne
<oliv3r>
wens: i was guessing by that time, DMA might be done too
<oliv3r>
wens: but how about this; nobody is working on DMA right now :)
<oliv3r>
wens: i think matt abandonend it :(
<oliv3r>
mripard: ^
ZetaNeta has quit [Ping timeout: 246 seconds]
Black_Horseman has quit [Remote host closed the connection]
zeRez has joined #linux-sunxi
rz2k has quit [Read error: Connection reset by peer]
<PiyushVerma>
any idea how to disable consol screen off. In 3.4.x it turning consol off after 10 min so nothing displayed on screen. where as no such behaviour in 3.0.x
<oliv3r>
afaik that's controlled by userspace
<PiyushVerma>
userspace is same all rootfs same just update kernel and got 2 issue
<PiyushVerma>
one of usb and one of display off
<PiyushVerma>
even if I restore kernel then problem gone
apo has quit [Remote host closed the connection]
apo has joined #linux-sunxi
<torbenh3>
PiyushVerma: either setterm or pass consoleblank=0 kernel param
<PiyushVerma>
torbenh3: thanks going to try
<Montjoie>
oliv3r, you said nobody works to mainline DMA ?
wingrime has joined #linux-sunxi
<oliv3r>
Montjoie: i well
<slapin>
oliv3r: ??!!
<Montjoie>
arg I need DMA to mainline for my driver
<n01>
oliv3r: actually could be a way to improve cooperation among devs
<oliv3r>
n01: well us devs usually know where to find eachothers repo's or ask here
<n01>
also true
<libv>
ssvb: wouldn't that just affect one client?
<oliv3r>
i do not object if someone wants to add a link to my repo obviously
<ssvb>
libv: well, not calling XInitThreads and then doing multiuthreaded stuff is wrong, that's one of the bugs in the mali blob
<libv>
i believe that someone gave me a binary diff for that many many moons ago, i doubt that i ever got round to applying it
<oliv3r>
n01: i'm just worried about 'joe user' or 'Muhammed 'kang' packager' starting to package stuff from dev tree's and then users show up 'oh i have this and that kernel with this and that driver patch'
<mripard>
oliv3r: yep
<mripard>
oliv3r: it was matt's plan all along as well.
<ssvb>
libv: based on the symptoms description from notmart, the problems are happening on the client side (EGL bailing out)
<n01>
hahaha
<oliv3r>
mripard: i'm slowly untangeling and working out something that's like sdhci for ahci. It's hard, since i'm not really sure what i'm doing about 60% of the time :p
<oliv3r>
mripard: what happened to matt?
<libv>
ssvb: again, why would it work after the mali-libs test, and not when ran on its own?
<n01>
btw I bought plane tickets for fosdem
<wens>
does Turl have a repository somewhere?
<mripard>
oliv3r: actually, all of this code should be in sunxi-devel.
<notmart>
oliv3r: ahah, yeah, that was pretty much me :p (about joe user starting to package things and trying to have the complete puzzle)
<mripard>
oliv3r: and I have no idea what happened to him
<libv>
ssvb: why would xinitthreads a lasting effect on the xserver?
<libv>
+have
<libv>
ssvb: if that were the case, wouldn't this be a big security issue?
<mripard>
oliv3r: if you have no idea of what you are doing, then step back, stop doing code, try to understand how it should work, and start work on it again.
<Montjoie>
mripard, If you have work on SPI, could you explain me the DMA magic value 0x07070707 used for DMA config please?
<mripard>
coding by copy/pasting is actually a bad idea
<libv>
and why would one not just set X up so that it would not be harmed?
<mripard>
like you already figured out by the AHCI thing I think.
<mripard>
Montjoie: hmmm, I worked on SPI without DMA
<Montjoie>
arghhh
<mripard>
and I'll need a bit more context than tha.
<libv>
ssvb: it also does not explain why the test would fix it, as the test does not run xinitthreads
<ssvb>
libv: don't know, something like non-initialized memory buffers related to xlib which persist across multiple application runs? but the whole issue that notmart is observing seems to be rather unusual
<Montjoie>
I have asked here several time and nobody seems to know how to get good values for DMA config
<libv>
ssvb: my money is on some fbturbo interaction
<libv>
xinitthreads cannot logically be the cause with those symptoms
<libv>
ssvb: did notmart get round to running es_tri?
<notmart>
libv: yes, didn't work as well tough
<libv>
as this would get a minimally sized xtrace log
<oliv3r>
mripard: well it's a lot that intermingles and you have to understand a few frameworks; but i'm getting there, and I do step back, read some more lxr, try to understand etc, hence the slow progress :p
<libv>
notmart: that's good news
<libv>
that way the xtraces of the mali-libs test and es_tri can be compared
<libv>
ssvb: the fact that the very first render happens on the _real_ fb as well as in a window...
<libv>
ssvb: that's a really big clue that this is something to do with the disp layer handling
<libv>
ssvb: in fbturbo
<ssvb>
libv: yes, it's also possible that something is wrong with ump buffers there
<libv>
ssvb: if so, that issue cannot exist on the client side
<libv>
as this is some persistent state
<mripard>
oliv3r: if you don't feel up to the task, you can always let Thomsa do the hard work :)
HeHoPMaJIeH has joined #linux-sunxi
HeHoPMaJIeH has joined #linux-sunxi
HeHoPMaJIeH has quit [Changing host]
<gzamboni>
mripard, did you got spi without dma working on mainline ?
<oliv3r>
mripard: IF i copy pasted, it wouldn't have gone wrong :p but those where magic values, and i rewrote the magic value wrong :p
<libv>
plucking dri2 calls out of xtrace might reveal what it is
kanedafp has quit [Ping timeout: 250 seconds]
<oliv3r>
mripard: it's very interesting to do; and once he sees what i'm doing he'll probably take it over :D but i wanna atleast be able to do the start
<ssvb>
libv: yes, that's what I would do myself if I had this problem reproducible on my system
<wens>
oliv3r: some parts are easier to understand (clocks, pinctrl, gpio), some harder (audio, dma?), but all are a lot of fun
<libv>
yup, the main developer is also the packager apparently
<ssvb>
notmart: if you are going to package it, a better name is x11trace :) as explained in the "why the name is stupid" section there
<notmart>
good to know
<notmart>
and yeah, will probably end up packaging it
<ssvb>
libv: as for the render on the _real_ fb as well as in a window, this might be not easy to achieve without some "help" from the kernel messing up ump buffers...
<ssvb>
notmart: can you identify the exact commit your linux-sunxi kernel has been built from?
zeRez has quit [Quit: zeRez]
<notmart>
ssvb: should be the last commit of sunxi-3.4 branch, so a7350cb6a9ec1aae510e26cdc730f05f12e13f9f
<ssvb>
ok, it should not make any difference for mali
<notmart>
ok
<notmart>
aanyways.. what i should do with xtrace?
<ssvb>
just run your x11 application via it and watch the log
<ssvb>
similar to strace and friends
<notmart>
ok, so running the test app, right?
<ssvb>
right
<ssvb>
if it fails to run, also try adding "--nocopyauthentication, -nDo not copy credentials" option
m1r has joined #linux-sunxi
m1r has quit [Client Quit]
m1r has joined #linux-sunxi
<PiyushVerma>
torbenh3: Perfect it work thanks again
<Montjoie>
mripard, if you start working on DMA, please tell me, I greatly look forward on it for enhancing my crypto driver
<notmart>
ok, this is ridiculous..
m1r1 has joined #linux-sunxi
<notmart>
i did rebuild the kernel, same revision, with different config options (changes that doesn't have anything to do, i enabled the nand flash drivers) and on this one opengl now just works
* notmart
retries with the sd with the older kernel
Quarx has quit []
<notmart>
one problem persists, that any opengl application is shown for the first time on the display no matter what the virtual terminal is
<notmart>
so on a terminal as well as on x11
m1r has quit [Ping timeout: 246 seconds]
<ssvb>
notmart: something might be wrong with ump buffer secure id numbers, the mali userspace blob expects secure id 1 and 2 to be assigned to the framebuffer, and we also have one extra ump buffer with secure id 3
<wens>
Montjoie: can the crypto engine work without DMA?
<ssvb>
notmart: these id numbers are assigned sequentially, and if something has tried to allocate ump buffers before starting xorg, then they may be off (framebuffer gets secure id numbers higher than 2)
<Montjoie>
wens, yes it can work without it, and for dm-crypt for example DMA is useless
wolfy has quit [Quit: Paradoxul homosexualilor este ca, desi nu reusesc sa se reproduca, sunt in fiecare an din ce in ce mai multi. (Tristan Bernard)]
<Montjoie>
I need info on DMA only for optimization
<ssvb>
notmart: I guess I need to add more safety checks and some extra logging to verify the sanity of secure id numbers assignment
<wens>
Montjoie: dm-crypt doesn't use DMA?
<notmart>
ssvb: i would be happy to test :)
<wens>
mripard: i finished the stmmac rewrite. should i send it off to relevant lists/people for review?
<Montjoie>
wens, dm-crypt crypt with buffer of 512bytes, too small for having any performance increase with DMA vs SS CPU mode
<ssvb>
Montjoie: ouch, this sounds painful, does it have any advantage over purely software encryption without dma?
<ssvb>
Montjoie: do you have some benchmark numbers in cpu cycles (or nanoseconds) comparing different methods?
<Montjoie>
ssvb, yes very painful, you will see benchmark in my patch submission soonly
<Montjoie>
for a short teaser, raw peformance of SS seems good, but real life benchmark show that SS is useless for dm-crypt
<Montjoie>
and it made me cry
<torbenh3>
Montjoie: even with 8k blocks ?
<torbenh3>
Montjoie: also keep in mind, that having free cpu time, is more important than raw throughput.
<Montjoie>
any number I give to cryptseyup, dm-crypt use 512bytes because my hardrive have 512 bytes sector
<Montjoie>
but I need to investigate more
<ssvb>
torbenh3: there will be no free cpu without using dma for dm-crypt
Gerwin_J has joined #linux-sunxi
<Montjoie>
for the moment a dd of 250M on a dm-crypt relying on tmpfs is writed in 23s without SS and in 33s with SS (37 with DMA)
<ssvb>
Montjoie: is this on A10 or A20 hardware?
<Montjoie>
A20
<Montjoie>
but perhaps my test with dd is bad because iozone give better results for SS
tzafrir has joined #linux-sunxi
<ssvb>
and A10 even has Cortex-A8 CPU core, which handles crypto stuff much better than Cortex-A7
<oliv3r>
it starts to appear strongly that A10 may be (much) faster depending on workloads then A20
<oliv3r>
even though A20 is dualcore
<oliv3r>
i hope that performance 'gap' will be bridged with A60 (big.LITTLE 2.2)
<Montjoie>
but I keep hope, each day I found some optimization
<oliv3r>
power efficiency of a7; performance of a15
<oliv3r>
and in a year or so, I wouldn't be supprised if the kernel is smart enough to utilize all 4 cores
<wens>
i recall samsung committed big.LITTLE support?
<oliv3r>
yeah i think we are slowly heading that way
<oliv3r>
that the scheduler takes account that there's big.LITTLE
<mripard>
gzamboni: it probes, it doesn't work yet
<mripard>
wens: if you feel like it, pes, sure :)
<mripard>
Montjoie: sure, i will
zeRez has joined #linux-sunxi
zeRez has quit [Client Quit]
zeRez has joined #linux-sunxi
<wens>
anyone besides mripard want to be cc'ed?
JohnDoe_71Rus has joined #linux-sunxi
<Turl>
arokux1: ping
<arokux1>
рш
<arokux1>
hi
<arokux1>
Turl:
m1r1 has quit [Ping timeout: 250 seconds]
<Turl>
arokux1: I got a nice ct delivered today :)
<arokux1>
Turl: ... it took long.
<Turl>
arokux1: but it arrived, didn't it? :)
<Turl>
I got way more cables this time
<arokux1>
:)
<arokux1>
Turl: but no UART
<Turl>
even an otg one
<Turl>
arokux1: do you know how to install the dissipator thing?
<arokux1>
Turl: there is a video on youtube, but I haven't installed it
<wens>
dissapator?
<Turl>
wens: heatsink
<wens>
Turl: doesn't it have adhesive on the under side? just peel off the backing and stick it on the A20?
<Turl>
yeah I just did it
<Turl>
the cb2 one brought a thermal paste tube though
<keebler>
focus: /msg
<Turl>
how do you stack the screw thingies?
<Turl>
I always mess up the heights
HeHoPMaJIeH has quit [Remote host closed the connection]
<focus>
keebler: not understand
<focus>
Turl: they supply excess screw thingies me thinks
<Turl>
focus: I put, from bottom to top, short, layer, long, layer, middle, ct, middle, layer, short
<focus>
if you put a sata disk, you need it from the way it is shown in the photos, but when I look at it closely, i don't need it if i put the sata on the same side as underside of cubie - save one layer of plastic as well
<Turl>
there is 4 nuts I don't know where to put
<focus>
photos somewhere...
<focus>
i put the smallest of the hex nuts as feet, and the m3 nuts at the top, that leave me 4 spare big hex nuts and one layer of plastic free
panda84kde has quit [Quit: Konversation terminated!]
arokux has joined #linux-sunxi
jinzo has quit [Ping timeout: 240 seconds]
arokux1 has quit [Read error: Connection reset by peer]
FR^2 has quit [Quit: Connection reset by peer]
<wens>
Turl: this is how i assembled it, except for the HDD. i had to buy extra screws
n01 has quit [Ping timeout: 250 seconds]
kivutar has quit [Ping timeout: 264 seconds]
<Turl>
wens: assembling it like on the picture, I've now got 4 extra 'feet' pieces
<Turl>
which are probably useful to attach an hdd
_massi_ has quit [Quit: Leaving]
<wens>
Turl: no extra nuts?
<Turl>
wens: I only got 4 nuts, which I've used at the top
<Turl>
8 short, 8 medium and 4 large of the other things
adb_ has joined #linux-sunxi
<wens>
I got 8 nuts, 4 short, 8 medium, and 4 large spacers
<wens>
so looks like you got the correct package :)
xeros has quit [Ping timeout: 245 seconds]
wingrime has quit [Ping timeout: 264 seconds]
popolon has quit [Quit: Quitte]
<hno>
Heh.. Cubieboard. Some of the screws that came with my Cubieboard2 clearly can not have passed any quality check. Some of them was completely blank with no screw thread in the hole.
<hno>
spacers is the correct term for those pieces i suppose.
<slapin>
hno: this is open source! you can make your own threads, the ones you like most!
<slapin>
hno: btw, long time no see!
<hno>
slapin, I know, but no, Cubieboard is not open hardware so halts a little there.
<Turl>
hno: at least you got the screws :) my cb2 is caseless
<hno>
always been lurking here. just not speaking up very often.
<hno>
An a little sorrow on the Cubieboard side.. my CB1 is ill.
<KBme>
hno, it's as open as any other hardware, isn't it?
<hno>
a bit troublesome as it's been my A10 reference.
<hno>
KBme, no.
<hno>
KBme, it's somewhat more open than random chineese tablets, but about the same. There is others that are very open on their hardware (i.e. Olimex)
<TheSeven>
well, the cubieboard does at least have schematics, unlike most of those tablets. but that's about it.
xeros has joined #linux-sunxi
<oliv3r>
u-boot ML is speaking of using fdt_high
<oliv3r>
Turl: finally! :D
<oliv3r>
Turl: don't bother with the heatsink; i doubt it is needed
<mnemoc>
moin
<Turl>
oliv3r: I sticked it anyway :p can't hurt can it?
<Turl>
hi mnemoc
<mnemoc>
hi Turl
Sonic1_ has joined #linux-sunxi
Sonic1_ has quit [Remote host closed the connection]
<oliv3r>
Turl: depends
<oliv3r>
if the thermal pad is crap
<oliv3r>
it may isolate the chip isntead :)
<oliv3r>
insulate*
<oliv3r>
hno: what's wrong with your A10? my PHY is still broken! (or again, probably bad power lines somewhere)
<oliv3r>
Turl: it looks better nakid :)
<oliv3r>
Turl: now go dl our u-boot and test2 gb :)
<oliv3r>
Turl: when I want to use clocks based on name, what's the normal procedure
<oliv3r>
ask the dt-node what the clock names are, and then pass that to devm_get_clk?
<oliv3r>
or just use the string ind evm_get_clk
<hno>
oliv3r, it's quite ill. Can run u-boot but any attempts to run anything serious under linux and it locks up hard.
<Turl>
oliv3r: I think you can use devm_clk_get(dev, "name")
<oliv3r>
i had a bad SD card the other day; everything worked, except loading the kernel/dt into ram from the sd card
<Turl>
oliv3r: I need to buy another SD to test :p
<hno>
oliv3r, it boots fine.
<oliv3r>
Turl: i should have sent you 2 sd cards on apostcard, but i ran out lol; i have them scattered and don't know where they all are
<oliv3r>
Turl: well if i want to be 'generic' etc, could i get of_get_property("clk-name"); and pass that to devm_clk_get
<oliv3r>
or rather, ist hat the recommended way
<oliv3r>
and how would I deal with having multiple clocks to get, just parse strings?
<Turl>
oliv3r: if you need to get clks from dt just use of_clk_get
<oliv3r>
but what if i don't know how many clocks there are to get
<techn_>
Turl: How to handle pll selection when display mode changes dynamicly?
<techn_>
Turl: should device tree give all available options and then code should somehow search best match?
<Turl>
oliv3r: there's a function to count how many elements were given to a property
<oliv3r>
Turl: and i can use that with of_clk_get?
<KBme>
hno, well, i'm not sure that makes a wholme lot of difference as long as many chips on these devices are closed anyways.
<oliv3r>
Turl: in any case, it is possible to have a driver start/load/get all clocks without knowing which ones are available from the dt
<Turl>
oliv3r: yeah, you can then loop for(i=0;i<nr;i++) of_clk_get(..) or sth
<oliv3r>
Turl: good; i'll play with that
<oliv3r>
Turl: and then store those in an array of clocks or something (for suspend/resume etc when all clocks need to be changed again)
<Turl>
techn_: you can set rate dynamically
<Turl>
techn_: I'm not sure if I understood your question though
<Turl>
oliv3r: yeah I suppose
<Turl>
oliv3r: but usually the driver knows how many clocks there are
<Turl>
dunno what you're writing
n01 has joined #linux-sunxi
FR^2 has joined #linux-sunxi
<hno>
KBme, it does make a difference, just as open source makes a difference.
<oliv3r>
Turl: unless you have a generic platform driver and want to have much shared, minimally individual stuff :)
<oliv3r>
Turl: like ahci_platform :)
kivutar has joined #linux-sunxi
<KBme>
hno, none of those olimex boards look opensource to me
<oliv3r>
Turl: so with of_clk_get(); you need to give an index; so you blindly assume that clk 0 and clk 1 are the clocks you want, even if they are 'reversed' (you could of course use name based clks)
<KBme>
oh, don't get me wrong, I would be so thrilled by an opensource arm board, but there is none
aseigo has quit [Remote host closed the connection]
aseigo has joined #linux-sunxi
n01 has quit [Ping timeout: 260 seconds]
n01 has joined #linux-sunxi
m1r has joined #linux-sunxi
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
deasy has joined #linux-sunxi
deasy has quit [Remote host closed the connection]
deasy has joined #linux-sunxi
n01 has quit [Read error: Connection reset by peer]
n01 has joined #linux-sunxi
tomboy65 has joined #linux-sunxi
tomboy64 has quit [Read error: Connection reset by peer]
tomboy64 has joined #linux-sunxi
tomboy65 has quit [Write error: Connection reset by peer]
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
deasy has joined #linux-sunxi
ZetaNeta has quit [Ping timeout: 260 seconds]
AreaScout has quit []
aesok has joined #linux-sunxi
<eagles0513875>
hey all
<eagles0513875>
hey oliv3r
m1r has quit [Ping timeout: 240 seconds]
notmart has quit [Quit: notmart terminated!]
adb_ has quit [Ping timeout: 252 seconds]
wingrime has joined #linux-sunxi
rz2k has joined #linux-sunxi
<libv>
ssvb: wow.
<libv>
ssvb: i think aseigo should go use another SoC.
shineworld has joined #linux-sunxi
<libv>
and irc meetings is not going to make the sunxi documentation or code any better
<libv>
also, aren't the mer guys still using the work mdfe did back in march/april?
<libv>
like kernel build, mali packaging and image creation...
<libv>
that was all done on a hyundia a7hd
<libv>
_my_ hyundai a7hd
rz2k has quit []
n01 has quit [Ping timeout: 260 seconds]
shineworld has left #linux-sunxi ["Leaving"]
n01 has joined #linux-sunxi
netlynx has quit [Quit: Leaving]
rah has quit [Ping timeout: 245 seconds]
ssvb has quit [Read error: Operation timed out]
rah has joined #linux-sunxi
ssvb has joined #linux-sunxi
wingrime has quit [Ping timeout: 250 seconds]
zeRez has quit [Quit: zeRez]
<libv>
ssvb: btw, a week or two ago, i ran into a (mental mostly) wall with dri2
ZetaNeta has joined #linux-sunxi
ZetaNeta has quit [Changing host]
ZetaNeta has joined #linux-sunxi
<libv>
ssvb: just like i wasn't getting any swapbuffer events, i am not getting an invalidatebuffer events
<libv>
i had just hooked up some sane FB (tile scanning pattern) resizing code, there was nothing i could hook it up to that would actually make resizing work
ssvb has quit [Ping timeout: 260 seconds]
<libv>
and if i then tried to ... hrm...
ssvb has joined #linux-sunxi
<speakman>
Are there any way to control gpio through /dev? Going through /sys will require root privileges.
<WarheadsSE>
most methods of doing that require sudo or root..
Turl has quit [Ping timeout: 245 seconds]
ssvb has quit [Ping timeout: 240 seconds]
ssvb has joined #linux-sunxi
discopig has quit [Ping timeout: 240 seconds]
arokux2 has quit [Remote host closed the connection]