TheSeven has quit [Read error: Operation timed out]
\\Mr_C\\ has quit []
TheSeven has joined #linux-sunxi
\\Mr_C\\ has joined #linux-sunxi
hipboi has quit [Ping timeout: 246 seconds]
gzamboni_ has joined #linux-sunxi
gzamboni has quit [Ping timeout: 264 seconds]
spenser309 has joined #linux-sunxi
<spenser309>
Hi, what are the various library versions present in the sunxi-mali libraries used for? different boards? different driver versions? etc?
eebrah|away is now known as eebrah
<libv>
spenser309: just use the installation system as described in the readme
<libv>
spenser309: everything else is just fluff for most use-cases
<spenser309>
libv: i'm the author of a patch series to add the sunxi-mali drivers to buildroot and i'm trying to inform the user about which version to use.
<spenser309>
libv: some kind of brief description would be more than adequate but I can't find any documentation for the version.
<libv>
spenser309: r3p0 is the version in most of the kernels.
<libv>
all the others have just been gathered over time, in case other drivers are put into the kernel
<libv>
r3p0 has had some binary fixes to no longer do some stupidity
<spenser309>
libv: so it's related to the version of the driver in the kernel?
<libv>
while others did not have those fixes
<libv>
yes
<libv>
arm does not maintain any semblance of a stable kernel abi, let alone api
<spenser309>
libv: i'll make a note of this the user and select r3p0 by default.
<libv>
we keep those other drivers around because we have those drivers, and its better to have some backup plan
<libv>
spenser309: just always enable r3p0
<libv>
if you are using the standard sunxi kernel that is
<libv>
let it depend on that
<libv>
the sunxi-mali buildsystem is meant to run on the booted target
<libv>
and it autodetects the version of the kernel driver
<spenser309>
libv: I have it working in a cross compile environment by specfying all the options it autodetects.libUMP
<libv>
yes, that's part of why those options are there
<spenser309>
libv: thanks for your help!
atiti has quit [Ping timeout: 248 seconds]
Offshore has joined #linux-sunxi
\\Mr_C\\ has quit [Excess Flood]
\\Mr_C\\ has joined #linux-sunxi
rellla has joined #linux-sunxi
dragonn has joined #linux-sunxi
n01 has joined #linux-sunxi
<oliv3r>
Turl: zero docs, only code
<oliv3r>
u-boot is most revealing if anything
<oliv3r>
check my a20 branch on github
<oliv3r>
cmake is the devil's spawn, it's a horrible mess
<oliv3r>
and i think the default prefix should be /usr/local iirc
<oliv3r>
Turl: sprunge has color highlighting for diff? how awesome is that!
<theOzzieRat>
where is the most appropriate place to instantiate i2c devices?
_BJFreeman has joined #linux-sunxi
BJfreeman is now known as Guest19073
_BJFreeman is now known as BJfreeman
<oliv3r>
ssvb: the network bugs; hans has a wemac patchset on the ML not yet pushed to linux-sunxi afaik
Guest19073 has quit [Ping timeout: 252 seconds]
<oliv3r>
which reminds me, i have a wemac rename patchset pending to finish and submit
<ssvb>
oliv3r: if something is not broken, don't fix it :)
<oliv3r>
if it's not broken, improve it
<oliv3r>
tss tss ssvb :)
bamvor has quit [Quit: Leaving]
vicenteH has quit [Ping timeout: 240 seconds]
bamvor has joined #linux-sunxi
FR^2 has joined #linux-sunxi
* Black_Horseman
is away: Min enoxleite to nekro8afti... krataei ftyari!!!
BJfreeman has quit [Quit: had a good time]
<oliv3r>
mnemoc: ping
<oliv3r>
can anybod confirm/deny that sun3i uses the same emac driver? According to the Makefile it does, so it could be a change, i don't see who authored the patch, mnemoc commited it however
<oliv3r>
btw, how do we feel about dropping sun3i support from 3.4? Nobody has hardware, code can't be tested etc
<oliv3r>
i'll post it to the ML I guess
n01 has quit [Read error: Operation timed out]
n01 has joined #linux-sunxi
rellla has quit [Ping timeout: 268 seconds]
notmart has joined #linux-sunxi
rellla has joined #linux-sunxi
\\Mr_C\\ has quit []
\\Mr_C\\ has joined #linux-sunxi
hglm has joined #linux-sunxi
hipboi has joined #linux-sunxi
vicenteH has joined #linux-sunxi
tzafrir has joined #linux-sunxi
<mripard>
oliv3r: ping?
cajg has quit [Ping timeout: 248 seconds]
dlan has quit [Ping timeout: 240 seconds]
dlan has joined #linux-sunxi
<oliv3r>
mripard: pong
dlan has quit [Ping timeout: 240 seconds]
dlan has joined #linux-sunxi
<mripard>
oliv3r: following back on the sid discussion, where is your sysfs file created ?
<mripard>
(ie, what find -name eeprom under sysfs return?)
<mripard>
(and what file do you use to test your driver?)
bsdfox_ has quit [Ping timeout: 276 seconds]
lunatik has joined #linux-sunxi
<oliv3r>
erm
<oliv3r>
if you give me 30-45 minutes; i'll tell you; I am in the middle of adding the regular attributes next to the binary ones (dual sysfs) as greg requested to test things, but from my bare memory, /sys/bus/platform/devices/sunxi_sid/eeprom
<oliv3r>
devices or drivers not sure on that one
<PiyushVerma>
ssvb: I did native build qt5. but it's unable to run gles demo program like hellogl_es2. Can't find EGLConfig returning null config.
cajg has joined #linux-sunxi
<oliv3r>
typo of the day: git stash poop
bsdfox_ has joined #linux-sunxi
<hglm>
PiyushVerma: Is hellogl_es2 a qt5 application? Does the "test" program from the sunxi-mali repository work?
<PiyushVerma>
yes it's qt5 app
<PiyushVerma>
yes basic test in repository works
<hglm>
PiyushVerma: I am not very familiar with qt5, does it use compositing with GLES?
<PiyushVerma>
hglm: ok
<PiyushVerma>
many time qt have platform specific patch
<PiyushVerma>
may be the way qt5 initialise egl is not exactly the mali works
<PiyushVerma>
ssvb have already build qt5 so I ask him he may have some patch or comment
<hglm>
PiysuhVerma: Could be, maybe the framebuffer/configurations details that Qt5 wants are not available in the Mali driver.
<PiyushVerma>
ok any suggestion to fix ?
<hglm>
PiyushVerma: But in any case I have distinct impression that Mali400 was never designed to do compositing or blending well. In a 3D rendering engine anything related to brending/transparency is extremely slow on Mali400.
<libv>
?
<libv>
hglm: mali400 on a10(s)/a10 just has a a single slow fragment shader
<PiyushVerma>
hdlm: yes I realise that it's limitation of pixel fill rate. But I think in case of DRI if compositor disable then it will directly render to FB right
<libv>
hglm: things are slightly different with 4 fragment shaders clocked twice as fast.
<PiyushVerma>
hglm : sorry typo
<libv>
(like ony exynos)
<PiyushVerma>
I test on other x86 platform with xface if I enable composit manager it take 40% cpu just in basic cube render. and if I disable it take 2 %
<hglm>
libv: I see, it was just that in my experience with multi-pass rendering with Mali400 (with additive blending) it was slow.
<PiyushVerma>
which let me think that if compositor disable it directly render to FB
<FR^2>
I'm still trying to fixate my cubieboard2's mac address. I used bin2fex on /boot/script.bin, added two lines to the result: "[dynamic]" followed by "MAC = "5a805f56ea95"", used fex2bin and moved the output back to /boot/script.bin - still, on every reboot I get a new mac address. What am I doing wrong?
<PiyushVerma>
FR^2: i prefer to set mac address in boot script after installation it's more simple
<FR^2>
PiyushVerma: Ah? How do I do that?
<PiyushVerma>
ifconfig eth0 hw macadd
<FR^2>
but still the script.bin way should work, shouldn't it?
<PiyushVerma>
FR^2 must should work. But I never use that. we use our own flash script which set macid after installing all
<FR^2>
k
<PiyushVerma>
libv: does mali a20 lib available for linux now ?
<libv>
PiyushVerma: have you tried the standard libs?
<hramrach__>
the r3p0 ones?
<PiyushVerma>
will that run on Cortex A7 ?
<hramrach__>
the 3.3 kernel shipped with the board has r3p2 interface but did not find the lib for that
<hramrach__>
but maybe r3p0 would work with hansg's 3.4 kernel
<libv>
try throwing in r3p0 kernel side
<libv>
or grab the odroid binaries
<hramrach__>
odroid
<PiyushVerma>
libv : aah odroid could be good a idea. Do u think lima ready to use in commercial product ?
<hramrach__>
hardkernel blocks me from accessing their web site, heh
<libv>
PiyushVerma: lima is not a proper graphics driver at all atm.
<hramrach__>
PiyushVerma: technically it's illegal to use the odroid libs on other board
<libv>
PiyushVerma: and are you really intending to use a20 and mali in a commercial product?
<PiyushVerma>
libv: yes we already have Digital Signage Solution running on A10
<libv>
germans would call that "sportlich"
<PiyushVerma>
now want to do A31 but linux side cedar and gpu driver issue
<hramrach__>
you will probably have issues on a20 too
<hramrach__>
it's too new
<PiyushVerma>
yes
<PiyushVerma>
but a20 cubi board is there so may be some support from there for lib
<hramrach__>
and not sure the r3p0 driver will work with mp
<hramrach__>
but you can be the first to try ;-)
<PiyushVerma>
A31 don't have mali
<hramrach__>
yeah, no driver for that going to be released, ever
<PiyushVerma>
yes may be some day libhybris
<libv>
heh
<libv>
PiyushVerma: also, be aware that the legal status of our own mali binaries is pretty grey
<libv>
PiyushVerma: and i trust that your customers are being appropriately notified about the use of GPLed software on the solutions you sell
<PiyushVerma>
libv: yes
<hramrach__>
tecnically they need not if you like rent the devices and not sel them
<hramrach__>
that's one of the reasons for gpl3
hipboi has quit [Ping timeout: 248 seconds]
hipboi has joined #linux-sunxi
<hramrach__>
hmm, battery dead
* hramrach__
going offline
<PiyushVerma>
hramrach__: ok most I try to use is LGPL or bsd
<PiyushVerma>
hramrach__ : see u soon
<libv>
PiyushVerma: ship with a linux kernel, and you have to actively let your customers know that parts of the software shipped is GPLed, and that all the rights of the GPL are being upheld
<PiyushVerma>
libv: that's usualy in software licance like android or apple so we do same
<libv>
PiyushVerma: ok
bsdfox_ has quit [Ping timeout: 240 seconds]
bsdfox_ has joined #linux-sunxi
<lunatik>
guys, can someone tell me why internal nand in unavailable when booting from sd card? am I doing something wrong?
<lunatik>
I've moved the stock rom to sd card and now the kernel only sees /dev/block/mmcblk*, but there are no /dev/block/nand*
hipboi has quit [Ping timeout: 240 seconds]
paulk-desktop has joined #linux-sunxi
paulk-desktop has quit [Remote host closed the connection]
wingrime has joined #linux-sunxi
<wingrime>
oliv3r: ping
bsdfox_ has quit [Ping timeout: 240 seconds]
notmart has quit [Quit: notmart terminated!]
bsdfox_ has joined #linux-sunxi
<wingrime>
ssvb: ping?
n01 has quit [Ping timeout: 264 seconds]
<oliv3r>
wingrime: pong
<wingrime>
oliv3r: cubie arrived!
<wingrime>
oliv3r: can you introduce current state of support
n01 has joined #linux-sunxi
<wingrime>
ssvb: ping
<oliv3r>
wingrime: OMG AWESOME! :D
<oliv3r>
i bet you are one happy camper now :)
<oliv3r>
wingrime: but support is shaky at best :)
<wingrime>
oliv3r: yeax
<oliv3r>
do a dance!
<oliv3r>
anyway, first thing to know: Make sure you have a good power supply
<wingrime>
oliv3r: it have linaro on nand
<oliv3r>
yep :)
<oliv3r>
A20 is a little more power hungry it seems
<oliv3r>
could also depend on board though; so if you have really strange issues, PSU could too weak
<oliv3r>
did you get UART and uSD adapter?
<wingrime>
oliv3r: stock have glitches with video , it notmal ?
<wingrime>
oliv3r: yes
<oliv3r>
cool; you can use uSD for work on tablet ;)
<wingrime>
oliv3r: where I can see uSD pinout
<oliv3r>
stock can be glitchy, a10 was glitchy with android
<pitillo>
Hello. I'm trying to cross-build a kernel for cubieboard2 but I'm not sure which config is needed. Can someone give me a hand? (I'm using sources from https://github.com/cubieboard2/linux-sunxi, is this right?)
<oliv3r>
wingrime: nothing yet, mripard wanted to start on it last weekend, but he's been busy. I think by working with a31 he has some idea's/conecpts ready for a20, so should be really soon
<oliv3r>
pitillo: that's not ours, that's from #cubieboard i think; ours is at github.com/linux-sunxi
<wingrime>
oliv3r: what working with 3.4 ?
<oliv3r>
so it could be very well the stock 3.3 kernel?
<oliv3r>
wingrime: hansg's wip/a20 tree is your best bet
<pitillo>
ummm interesting oliv3r. I've used that to cross-build an A10 kernel for the cubieboard. Can it be used too with A20? if so, which config is needed?
<oliv3r>
wingrime: that's the 3.3 code dump we received from AW a while ago
<oliv3r>
it should have all the latest cedarX stuff
<oliv3r>
(kernel bits)
<pitillo>
ah ok oliv3r, thank you very much. I'll continue reading about A20 and its kernel support currently. I'm a bit lost with cubieboard source too because I don't know which kernel config is needed (no sun7i in defconfigs...)
<wingrime>
oliv3r: what's problem with dram?
<oliv3r>
wingrime: u-boot?
<oliv3r>
wingrime: or with the kernel
<oliv3r>
pitillo: basically; we haven't done much work for sun7i yet
<wingrime>
oliv3r: it preformance problem with new mem controller?
<oliv3r>
pitillo: but allwinners 3.3 tree (link to wingrime above) is what allwinner gave us that should run on stock nand
<oliv3r>
pitillo: so should come with a config.
<oliv3r>
Turl: ping
<pitillo>
ummmm really interesting. Taking note and checking. Thank you very much oliv3r
<oliv3r>
wingrime: yeah, I think ssvb said the dram controller is the same, defaults are set a little more conservativly, but the big issue is, that a7 is slower
<wingrime>
oliv3r: you have hdmi working
<oliv3r>
pitillo: I don't know what cubieboard has in their github, so can't comment on that ;) it is most likly a fork of either our kernel, or amery's
<oliv3r>
wingrime: i never used video output; only uart
<oliv3r>
wingrime: and I've only done u-boot work for a20
<wingrime>
oliv3r: kernel still not boots?
<oliv3r>
wingrime: which kernel? stock should :p
<wingrime>
oliv3r: ours 3.4
dragonn has quit [Remote host closed the connection]
* Black_Horseman
is back (gone 03:13:45)
<FR^2>
sensational news! ;)
<oliv3r>
wingrime: i think hansg has it working, so should boot
<oliv3r>
jwrdgoede is hansg, try building his latest kernel, it's either wip/a20 or for-amery branch
<wingrime>
oliv3r: It very hard debug kernel early without uart working
<oliv3r>
uart works for 3.4
<oliv3r>
sata i think; ethernet
<hglm>
AFAIK jwrdegoede/sunxi-3.4-for-amery is more up to date then jwrdegoede/sunxi-3.4-a20-wip, but there might be configuration issue with USB that prevents a clean compile
<oliv3r>
sound
<oliv3r>
hglm: yeah disable USB is easiest fix :p
<oliv3r>
that's what I did :)
<wingrime>
oliv3r: I about problems with kernel before uart prints can be used
<oliv3r>
wingrime: with a20-uboot you have uart output; and if you enable earlyprintk you should get that without problem too
<wingrime>
oliv3r: I talking about JTAG
<oliv3r>
ahh, i odn't know if jtag works
<oliv3r>
only that uart works
<oliv3r>
i haven't used jtag myself on these boards
<oliv3r>
BUT the uSD jtag should work just fine
<wingrime>
oliv3r: I should know witch jtag table required
<wingrime>
oliv3r: also witch default uart speed
<wingrime>
oliv3r: ok, I just conneced and powered using uart
Xaros has joined #linux-sunxi
Black_Horseman has quit [Ping timeout: 264 seconds]
Xaros has quit [Client Quit]
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
<wingrime>
oliv3r: nice, emacs leds not show any sign of life
<wingrime>
oliv3r: it reay require reboot for iyt
<wingrime>
?
<oliv3r>
wingrime: emac should work just fine
<oliv3r>
plug lan in, go
<oliv3r>
leds are controlled by the RTL PHY
<oliv3r>
wingrime: do not connect 3.3V power from uart!
<oliv3r>
only tx; rx and GND
<oliv3r>
never use power
<oliv3r>
things will break!
<oliv3r>
wingrime: don't use power from UART!
<oliv3r>
i guess if you ONLY use uart power, it won't break i guess
<oliv3r>
uart is as always, 115200
<oliv3r>
screen /dev/ttyUSB0 115200
<wingrime>
oliv3r: whats with emac
<oliv3r>
EMAC is controller via the RTL PHY on the PCB, the leds etc
<oliv3r>
a20 soc has only ethernet MAC (Media ACces), and connects to a PHY via mii
<wingrime>
[ 197.949952] wemac wemac.0: eth0: link down
<oliv3r>
your cable ok? and your switch?
<oliv3r>
i haven't used ethernet on mine i don't thihnk
<oliv3r>
and on my cubie 1.0 it's broken :)
<oliv3r>
don't be supprised if the driver is broken; there's several fixes in hansg's a20 3.4 tree
notmart has joined #linux-sunxi
notmart has quit [Changing host]
notmart has joined #linux-sunxi
gzamboni has joined #linux-sunxi
gzamboni_ has quit [Ping timeout: 240 seconds]
<oliv3r>
Turl: ping
<wingrime>
oliv3r: where hansg's repo?
<oliv3r>
i pasted it above :p
<hglm>
oliv3r: I got the hansg's latest branch (sunxi-3.4-for-amery) to compile with USB support, by selecting the right USB options (not as modules) in the right order.
<wingrime>
hglm: witch hw work/not work?
<hglm>
wingrime: I don't know the exact status of the A20 support.
<hglm>
You have to select:
<hglm>
drivers -> USB -> USB Gadget Support -> USB Peripheral Controller (X)) -> SoftWinner SUNXI USB Peripheral Controller (X)
<hglm>
drivers -> USB -> SUNXI USB2.0 Dual Role Controller support -> USB0 Controller Support -> otg support
<oliv3r>
hglm: the patch that's supposedly lingering on the mailing list; doe sit fix that?
<oliv3r>
wingrime: if you backread hansg's A20 mails on the ML; he says what the current support is
<oliv3r>
i have high doubts if disp fits the bill
<hglm>
oliv3r: I don't know. Which patch do you mean?
Black_Horseman has quit [Remote host closed the connection]
<hglm>
The sunxi-3.4-for-amery branch seems to run OK on my A10 and also fixes the wlan not working problem I had with the current stage/sunxi-3.4.
<oliv3r>
Check Jussi Kivilinna's reply on the same thread; i haven't looked to be fair
<hglm>
oliv3r: I didn't try that patch, the branch compiles cleanly with the right USB options.
<oliv3r>
well if that patch fixes the 'wrong options' behaviour, it should be pulled back in ;)
<oliv3r>
but having a compiler errors from bad config options is bad imo
<hglm>
hglm: That's true, it needs to be fixed properly.
<hglm>
Talking to myself there :)
<oliv3r>
lol
<oliv3r>
not sure if I have time to look at it right now
<oliv3r>
so if anybody has a few :p shouldn't be a huge thing
<hglm>
oliv3r: It may a little more complicated if the USB driver has been unified (from being completely seperate for sun4i, sun5i etc before).
<oliv3r>
but i'm not sure if the .show is called at all
_BJFreeman has joined #linux-sunxi
<oliv3r>
all other uses of attributes are device_attr; class_atrr; bus_attr, it's very rare to use raw attributes (i think that's what kobj_attribute is?)
<oliv3r>
mripard: i was hacking in the textual version
<oliv3r>
mripard: but wasn't sure if i could use kobj_attribute so was busy looking that up
<oliv3r>
for the path it doesn' tmatter
<Turl>
ssvb: if you can give that a try on a real world setting it'd be great
<ssvb>
Turl: thanks, I'll try
<mripard>
oliv3r: I'm more concerned about *where* that file might be, and why you have those two different pdev pointers
<ssvb>
Turl: though looks like it is including multiple unrelated changes
<ssvb>
Turl: I hope you know what you are doing :)
<Turl>
ssvb: well my mele seems to be happy :)
rellla has quit [Ping timeout: 240 seconds]
rellla3 has joined #linux-sunxi
<Turl>
ssvb: there's a register cache, some bug fixes, and the timeout thing is inspired on what we have in mainline now
<ssvb>
Turl: so was mine until hansg broke it :)
jemk has joined #linux-sunxi
<Turl>
ssvb: btw, I discovered that conservative governor helps with the net performance a bit too
<oliv3r>
booting
<wingrime>
oliv3r: yes accroding cubie1 schenematic it realy connected to 3.3 PWM before , others schnematics feeds using parazite (power regulators will produce normal voltage if you send 5v not 3.3
<ssvb>
Turl: I'm always using the 'performance' governor
<oliv3r>
but i mailed him 2 pull requests so hopefully he will push those before you pull
<jemk>
wingrime: oh, bad. it neads libavformat/codec, do you have it installed?
<oliv3r>
mnemoc: just out of curiosity; when will you be able to test etc again?
<mnemoc>
oliv3r: hope to solve my connection problems this monday (moving out of the hostel finally)
<mnemoc>
oliv3r: after monday
<oliv3r>
nice one
<ssvb>
oliv3r: the current stage/sunxi-3.4 is not so nice, it introduces at least two troubles for me (and who knows, maybe there are more?)
<Turl>
nice, no spamusers in the last two days
tinti has joined #linux-sunxi
<oliv3r>
ssvb: hansg may have fixed all of those
tinti has quit [Changing host]
tinti has joined #linux-sunxi
<wingrime>
jemk: ffmpef not enought ?
<oliv3r>
i doubt it :p but it's possible
<wingrime>
*ffmpeg
<jemk>
wingrime: dont know, on arch ffmpeg contains libav*, so it should be ok
<mnemoc>
oliv3r: i sub-rented an apartment for 2 month, but there i'll have a proper desk to plug boards, code, test, ....
<jemk>
wingrime: any compiler error messages for me?
<mnemoc>
here in the hostel it's more.... complicated
<ssvb>
oliv3r: well, it is possible, but wouldn't it be a good idea to submit a clean patch set instead of breaking/fixing stuff randomly in the middle of it?
<oliv3r>
mnemoc: awesome; nice to be out of the hostel too
<oliv3r>
ssvb: well I think the big issue is, that stage/sunx-3.4 is VERY outdated, wasn't it?
<wingrime>
jemk: wait, need connect all
<mnemoc>
let's fix it first. tag. stable-jump. tag.
<oliv3r>
mnemoc: i went over some linux-sunxi repositories and tried closing issues as best as I could; havne't touched the linux one, too many issues there :p
<Turl>
ssvb: btw, if you still have perf issues, can you test commenting lines 77-79 on my patch? (the if(!in_suspend...) thing, leave just the udelay)
<ssvb>
Turl: maybe, let me check if the latest pull request is any better first
<wingrime>
main.c:287:21: error: ‘AV_CODEC_ID_MPEG1VIDEO’ undeclared (first use in this function)
BJfreeman has quit [Read error: Connection reset by peer]
<wingrime>
jemk: ^
<jemk>
wingrime: oops, try to add an #include <libavcodec/avcodec.h> to main.c
n01 has quit [Ping timeout: 246 seconds]
<ssvb>
Turl: hmm, the last pull request from hansg does not compile
<mnemoc>
*g*
<Turl>
ssvb: o.O
<Turl>
it did for me, except for a usb issue I think, which was due to what I had selected on oldconfig :p
<oliv3r>
ssvb: it's USBC
<oliv3r>
the default config uses a combination nobody has used before and fails
<oliv3r>
it's on the ML
<ssvb>
yep, but it is still annoying
<oliv3r>
highly
<oliv3r>
i saw a few mails about it yet, but i still have 100 unread to go through ;)
<oliv3r>
busy replying to silly linux-sunxi mails from people who have little clue :(
<ssvb>
which patch would be the best to address this issue?
<hglm>
ssvb: See the thread on the mailing list. You have to set the USB config precisely in the right way.
<oliv3r>
set the right options in .config
<wingrime>
jemk: SAME
<mnemoc>
i saw a comment from techn about a missing { in the last pull from hansg on the usbc code
<mnemoc>
on github
<ssvb>
somebody needs to submit a patch for the default config or something
<Turl>
oliv3r: I just enable _EHCI and _OHCI ones for sunxi and it Just Works (TM) for me
<hglm>
Yeah the default config should be fixed so that usb0 is OTG (host + device)
<jemk>
wingrime: could debian really have very old ffmpeg, there it was called CODEC_ID_... (without AV)
<wingrime>
jemk: I use debian SID
<wingrime>
jemk: I not sure I use ffmpeg (not fork)
<wingrime>
jemk: but I have ffmpeg installed
<mnemoc>
sid should have libav, not old-ffmpeg
<oliv3r>
Turl: disabled otg?
<Turl>
oliv3r: I dunno, but I have a mele, so I don't use it :)
<oliv3r>
Turl: then that's why it works, if you enable otg the wrong way, you get issues :)
<oliv3r>
it's an otg only issue
<wingrime>
mnemoc: strange dev war
<mnemoc>
wingrime: yes
<hglm>
oliv3r: well, the defconfig by default selects host-only and that also doesn't compile.
<oliv3r>
major netowrk regression, yay i'm almost up to speed on my mails
<oliv3r>
libav fs old-ffmpeg; yeah it's confusing and it's like '*G*'
<wingrime>
jemk: I remove AV_
<wingrime>
jemk: but than
<wingrime>
main.c:253:17: error: storage size of ‘video_codec’ isn’t known
<jemk>
wingrime: seems like my code is somehow incompatible to your libav*
<wingrime>
jemk: I bild it after two fixes
<wingrime>
1) replace video_codec type to int
<wingrime>
2) comment out printf with codec name due:
<wingrime>
main.c:(.text.startup+0x2da): undefined reference to `avcodec_get_name'
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
<wingrime>
jemk: Futurama not plays((
<jemk>
wingrime: bad, here it played. any errors or just no picture?
<wingrime>
jemk: not errors, some lines on screen nothing more
<wingrime>
jemk: you can comparea my Futurama trace with mpeg_test trace
<jemk>
wingrime: when i traced they matched, if you give me a trace from your non working ill check
<oliv3r>
is 'patrick wood' in here?
mdp has quit [Ping timeout: 248 seconds]
<jemk>
I think i need to add more error checking or testers won't have much fun
mdp has joined #linux-sunxi
<wingrime>
jemk: strange , first time I tested , it not worked, now works
<wingrime>
jemk: also it interfere with blacklight
<wingrime>
jemk: when backlight turned on after of screen was distored
<wingrime>
jemk: gotcha!
<wingrime>
jemk: If you will play futurama, than press ^C in a middle , than If you try run mpeg-test again you will see lines on sceen
<wingrime>
jemk: looks like you not fully configure disp
<wingrime>
jemk: lines looks like as invalid layer format
n01 has joined #linux-sunxi
<wingrime>
jemk: find something?
eebrah is now known as eebrah|away
<jemk>
wingrime: if you ^C a programm using disp, then disp will be broken till reboot/module reload
<jemk>
wingrime: if disp isn't closed properly it breaks
<wingrime>
jemk: funny
<jemk>
wingrime: maybe i could add an signal handler and catch ^c, but it would be better to fix disp sometime
<wingrime>
jemk: better find whats wrong with disp
<wingrime>
jemk: it looks like wrong layer format
<jemk>
wingrime: wrong is that it doesn't close layers a program opened when the program gets killed
<jemk>
wingrime: but i don't know how this could be fixed
bsdfox_ has quit [Read error: Connection reset by peer]
vicenteH has joined #linux-sunxi
<wingrime>
jemk: do you have any idea how cedar do a-v sync ?
<jemk>
wingrime: That's not job of cedar. cedar only decodes single frames. syncing, reordering and so on needs to be done by software i think
<wingrime>
jemk: not fully as we have hardware interface for it
<wingrime>
jemk: not sure it used , but we have AV-timer for it
<jemk>
wingrime: that's only a timer that is running with typical audio/video freqs
<wingrime>
jemk: also we have strange AVS2 ioctls
<wingrime>
jemk: and strange sub-engine called avs2
<wingrime>
*avs
<wingrime>
*avc
<wingrime>
It may be encoder?
<wingrime>
jemk: I will make unification for cedarx driver in kernel for single one file for a13 and a10 and will make cleanup
<jemk>
wingrime: avc is the h264 encoder
<wingrime>
jemk: there is any sample for it?
<wingrime>
jemk: or something we can trace?
<jemk>
wingrime: some android libs somewhere on github, but i couldn't try to run them
notmart has joined #linux-sunxi
notmart has quit [Changing host]
notmart has joined #linux-sunxi
<wingrime>
jemk: what plans are you have ?
<wingrime>
jemk: ^
n01 has quit [Ping timeout: 264 seconds]
n01 has joined #linux-sunxi
notmart has quit [Ping timeout: 264 seconds]
<jemk>
wingrime: sry, bbl. only mpeg engine for now, maybe h264 dec next, but i wont do encoding in near future
jemk has quit [Quit: bye]
n01 has quit [Ping timeout: 240 seconds]
n01 has joined #linux-sunxi
<wingrime>
mnemoc: ping
<wingrime>
ssvb: ping
<wingrime>
hno: ping
<ssvb>
wingrime: pong
<wingrime>
ssvb: we have issue with disp, if not normaly close it layer get strange config
<wingrime>
ssvb: do you saw something simular ?
<Turl>
ssvb: any luck with emac?
<ssvb>
wingrime: I guess we need to be sure to disable colorkey before closing, because it is messing the layers stacking order a bit
<wingrime>
ssvb: also there issue with blacklight
<ssvb>
Turl: well, disabling that hansg's patch is necessary in all cases (even with the latest pull request)
<ssvb>
Turl: your patch still seems to be excessive, but I can run more tests with it
<ssvb>
Turl: wemac is weird :)
<Turl>
ssvb: it'd be great if you could, esp. the register read caching bit
<ssvb>
Turl: can we find somebody with CB1 for also testing WEMAC there?
<wingrime>
ssvb: yeah, wemac is very strange , it use dma callbacks for TX (dma callbacks OMG in IRQ context)
<ssvb>
wingrime: if we have sleeps called from the IRQ context, it might be not the brightest idea
lunatik has quit [Ping timeout: 250 seconds]
<wingrime>
ssvb: do tx in irq context too
<wingrime>
ssvb: also I get cubie2
<ssvb>
congrats!
<wingrime>
ssvb: build in linaro image have no free space at all ((
<ssvb>
wingrime: just install something on the sd card :)
<ssvb>
wingrime: it must be way easier now than we had it a few weeks ago
<wingrime>
ssvb: debootstrap take a loog time with mmc
<ssvb>
wingrime: use nfs root :)
<wingrime>
ssvb: ince idea
* ssvb
wants to find some compatriots for nfs root wemac testing
<wingrime>
ssvb: uboot can use emac for boot kernl?
<ssvb>
hmm, don't know
<ssvb>
I'm currently booting the kernel from mmc
<wingrime>
ssvb: about cedar, It wery nice have jemk, and nove on board , It realy surpricely fast for me have working mpeg example
<wingrime>
ssvb: are you tested it ?
<ssvb>
wingrime: not yet, but if it works, then we can make it play nice with VDPAU
<ssvb>
wingrime: it's a bit more difficult than XV, but still quite doable
FR^2 has quit [Quit: Connection reset by peer]
<wingrime>
ssvb: its still have no any sync , it do Frame after Enter press
<wingrime>
ssvb: this is PoC at all
<ssvb>
wingrime: this does not matter much
<ssvb>
wingrime: as long as we can kick it to get frames one after another, it's good enough
<wingrime>
ssvb: problem, that mpeg less interseting than h264, but I can't get all register names so easy as I get form mpeg
<wingrime>
ssvb: mpeg work quite good with software decoding
<ssvb>
wingrime: yes, mpeg is not quite interesting from the practical point of view
<wingrime>
ssvb: mpeg can be used at all, for example I can write modules for mplayer to do it
<wingrime>
ssvb: jpeg can be used only for android
<ssvb>
wingrime: the needed modules are already written for mplayer ;)
<ssvb>
wingrime: they are called vdpau and vaapi
<wingrime>
ssvb: vdpau can be used if we will have normal disp DRM kernel module
atiti has joined #linux-sunxi
<ssvb>
wingrime: vdpau can be used if xorg ddx gets supports for it
<ssvb>
wingrime: the drm kernel module is just one of many possible internal implementation details
<ssvb>
wingrime: and not the easiest one for the sunxi 3.4 branch
<wingrime>
ssvb: what currently works on a20 with 3.4 branch?
<wingrime>
hasng's
<ssvb>
wingrime: as for the mainline support, be prepared for lots of bikeshedding and many people demanding contradicting things :)
jemk has joined #linux-sunxi
<ssvb>
wingrime: because this is a relatively new area and nobody really knows how to do it 'right'
<paulk-desktop>
I understand the sunxi-3.4 branch is based upon the android-3.4 kernel? Did you try running it with Android?
<ssvb>
wingrime: I have not tried the 3.4 branch on a20 yet
<wingrime>
ssvb: I think that sunxi currently good choose for 'protected from hw backdoors' there less closed-source stuff
<wingrime>
ssvb: wifi only over usb and have no access to main memory
<wingrime>
ssvb: we have no 'satelite core' that runs closed RTOS
<wingrime>
ssvb: also we have no MCU in south bridge that have own tcp-stack and access to main-memory and ethernet conntroller on PC
<wingrime>
ssvb: maybe a10 next candidate for RMS notebook
rellla has joined #linux-sunxi
rellla has quit [Remote host closed the connection]
jemk has quit [Remote host closed the connection]
<ssvb>
wingrime, jemk: about the mpeg1/mpeg2 videos decoding, I remember that I recommended you to check it after jpeg precisely because it's a natural next step to gradually increase the complexity and use it as a kind of practice
<ssvb>
wingrime: also if we know everything about decoding mpeg2, we can already start working on the vdpau wrapper
<ssvb>
wingrime: while at the same time you can continue reverse engineering more practically useful and more complex codecs
<ssvb>
wingrime: hope this makes sense
<bfree>
ssvb: "compatriots for nfs root wemac testing"? I've been doing nbd root for a little while (3.3 on A20, 3.4 and debian's 3.10 on a10). is it specifically nfs you are interested in or ???
<ssvb>
bfree: just general ethernet performance
<ssvb>
bfree: nbd is naturally also interesting
<ssvb>
bfree: what kind of a10 device do you have?
<bfree>
just a cubie (and a cubie2)
<ssvb>
have you checked the current stage/sunxi-3.4 branch?
<bfree>
nope, I'd kinda been waiting for some good sounding a20 potential before trying to build newer kernel packages for 3.4 (and prefer the main branch for packages I share obviously)
<wingrime>
ssvb: we know mostly for mpeg12 but mpeg engine can handle more
<wingrime>
ssvb: mpeg4 ms-mpeg vp6 xdiv and divx
<ssvb>
wingrime: I think mpeg1 and mpeg2 are good enough for getting started with the higher level video acceleration wrappers
<wingrime>
ssvb: but I can't say 100% know becose some bits are unknown and if you looks at code if full with unknown action according traces
<ssvb>
wingrime: mpeg4/divx could be probably skipped because CPU is still good enough at decoding these formats and they don't add much testing/verification value over mpeg2, but I personally don't mind either way :)
<wingrime>
ssvb: according you ideas we must begin work on h264 soon as possible
<wingrime>
ssvb: but difficulty grows eponential
<wingrime>
*exponential
<wingrime>
ssvb: and more pushing that all mainline - one more side of it
<wingrime>
MT6592 - 8xcortex a7 with 1,7-2 ghz
<Turl>
wingrime: yes, you can tftp boot kernel from uboot
<wingrime>
Turl: good
<Turl>
ssvb: I have a CB1 but it's set up to boot mainline over tftp, if nobody else can in a reasonable timeframe, I can reconfigure it and give it a try
* Turl
should get more SD cards
<wingrime>
Turl: I also need more one sd for cb2
pirea has joined #linux-sunxi
<wingrime>
Turl: can it 64 ot 128 Gb?
<pirea>
wingrime hy
<pirea>
i don't know how to use ioctl :|
<Turl>
wingrime: I don't know
<pirea>
wingrime what value must have first argumet of ioctl function
<pirea>
?
<wingrime>
pirea: file handle
<Turl>
pirea: run "man 3 ioctl"
<pirea>
wingrime file handle means?
<Turl>
pirea: return value of open()
<wingrime>
pirea: ^
<Turl>
int fd = open("/dev/...", O_...); ioctl(fd, THING, ...);
pirea has quit [Read error: Connection reset by peer]
pirea has joined #linux-sunxi
<wingrime>
mripard: ping
ibrah has joined #linux-sunxi
pirea has quit [Quit: Leaving]
<oliv3r>
wingrime: a13 has special "image enhancer" to deal with that i think
<oliv3r>
(flickering of backlight)
<oliv3r>
wingrime: i Know there's an 'av sync' timer and maybe more; outside of cedarx
<oliv3r>
since i think cedar doesn't do audio at all
<wingrime>
oliv3r: blacklight glitch you can test with mpeg-test on a10 (I tested on a10)
<oliv3r>
Turl: ssvb I think emac will see some fixes when we port 3.3
<oliv3r>
ssvb: i can't test emac on cb1; my phy i sbroken :(
shineworld has joined #linux-sunxi
<ssvb>
oliv3r: too bad :( how did it end this way?
<wingrime>
oliv3r: replace, phy have no many outputs, you can remove it with hot-solder and knife,than solder new
BJfreeman has quit [Ping timeout: 252 seconds]
<oliv3r>
wingrime: yeah uboot can boot via emac
<oliv3r>
you can tftp the kernel
<hglm>
ssvb: I have been trying the new sunxifb, it seems to work pretty well with mp4 playback in vlc, but I am running my screen in scaler mode, wasn't that supposed to be a problem or is that only devices like A13?
<wingrime>
oliv3r: I want have more fun with mainline using x-modem with u-boot and send kernel over uart
<ssvb>
hglm: A13 has only one scaler, so the scaler mode should be a big problem there (no scaler available to the actual video layer)
<ssvb>
hglm: also I believe scaler mode has some problems with tearing (will try to verify it now)
<wingrime>
oliv3r: but perhaps I need wait mripard while he bering basic support
<hglm>
ssvb: Looks smooth to me on my A10, no tearing with limited testing and scaler mode is enabled on the screen.
<wingrime>
ssvb: you are still here?
<ssvb>
wingrime: yes
<wingrime>
ssvb: about flush cache in dma driver , I stull not sure about we need it
<wingrime>
ssvb: normaly io mem non cachable
<wingrime>
ssvb: Witch means we can flush cache only in specific cases
<wingrime>
ssvb: any idea?
<ssvb>
wingrime: it really depends on the memory mapping, if mapping is writecombine, then cache flushes are not necessary
<ssvb>
wingrime: but you may need a barrier
<oliv3r>
mpeg4/divx should be secondary targets; jpeg, mpeg1/2 and h264 are primary imo
<wingrime>
ssvb: cedar normaly not use cache flush, but new cedar kernel driver add ioctl for it, but I have newer seen that was used
jemk has joined #linux-sunxi
<oliv3r>
ssvb: i have no clue, it just stopped working
<oliv3r>
ssvb: could be a power thing, but i get nothing from over mii, lights don't even go on
<wingrime>
ssvb: in dma case, it realy not nessesory when you send somethig to IOMEM , and I think that normaly driver must allocate buffer with such flag
<oliv3r>
ssvb: if i poke some pads with my scope i sometimes get some leds to blink shortly, but that's it
<oliv3r>
wingrime: but where do I get new one
<ssvb>
hglm: indeed, looks like there are no problems with tearing in scaler mode
<wingrime>
oliv3r: you can possible kill some power regulator, you firstly need multimiter and check all voltages
<ssvb>
hglm: but I remember you said something about sunxi_disp_vsync_demo not working properly some time ago
<wingrime>
oliv3r: check all regulators for poweroutput
<wingrime>
oliv3r: you must check 1.5, 3.3 ,5 , 2.5 power sources
<oliv3r>
wingrime: i boot my kernels via sd; but tftp is really good alternative
<wingrime>
oliv3r: x-modem with uart
<wingrime>
oliv3r: u-boot can push kernel over uart
<wingrime>
ssvb: how much unnesesory flushes can influence to performance?
<hglm>
ssvb: Yeah that was in the console, it didn't quite look right (tearing). I'll test that again.
<wingrime>
ssvb: how much unnesesory flushes can influence to performance?
<ssvb>
wingrime: cache flushes are definitely not free, and cost maybe like half of memcpy performance (but need to measure it again on a10 to be sure)
<oliv3r>
wingrime: yeah for a20 we best wait for maxi
<ssvb>
wingrime: still uncached reads are a lot more expensive
<wingrime>
ssvb: now every (mostly) dma transefer on sunxi do flush at end . so you mean, it can cost more than simply memcpy without DMA
<ssvb>
wingrime: that's unlikely, memcpy with non-cached source buffer may be really slow unless it is using NEON or VFP
<oliv3r>
wingrime: i checked voltage with osciloscope, it looked fine, even mii messages seemed fine; i compared it with datasheet and traced out tons of pins; all 'looked' fine (from soc to phy + power)
<ssvb>
wingrime: and the use of NEON or VFP in the kernel is prohibited (or at least highly non-recommended)
<oliv3r>
wingrime: uart is slow :p
<oliv3r>
wingrime: but i should setup a tftp setup; much easier, copy file; push
<oliv3r>
then again, on cubie 1 i can't; broken phy :p
<ssvb>
wingrime: you can probably try to run some benchmarks yourself for different types of dma transfer handling (with and without cache flushes)
<oliv3r>
yay i've backread :p
<hglm>
ssvb: I tried the tearing test videos, although not 100% smooth (which may be related to my LCD), there is no tearing apparent, while there is tearing when not using XV for video output. The sunxi_disp_vsync_demo also, while not 100% smooth, doens't seem to be showing tearing.
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
<ssvb>
hglm: good, thanks for testing
<wingrime>
oliv3r: a20 captable uboot in your wip/a20 ?
<oliv3r>
yeah
<wingrime>
I will try
<wingrime>
with mmc
ibrah has quit [Ping timeout: 264 seconds]
<wingrime>
olvi3r: wow, a10 kernel boots that hung , its more than I expect form it
<wingrime>
oliv3r: it have same machine number?
<wingrime>
oliv3r: it hung at:<6>Calibrating delay loop...
jemk has quit [Quit: bye]
<oliv3r>
wingrime: someone mentioned exactly same error message today
wingrime has quit [Ping timeout: 240 seconds]
rellla has joined #linux-sunxi
soul has joined #linux-sunxi
shineworld has quit [Read error: Connection reset by peer]
rellla has quit [Remote host closed the connection]
paulk-desktop has quit [Quit: Ex-Chat]
atiti has quit [Ping timeout: 240 seconds]
soul has quit [Read error: Connection reset by peer]
soul has joined #linux-sunxi
soul has quit [Read error: Connection reset by peer]
Soru has joined #linux-sunxi
atiti has joined #linux-sunxi
utzig has joined #linux-sunxi
Offshore has quit [Ping timeout: 276 seconds]
<oliv3r>
wingrime: check mailing list; fedora 19 for a20 from hans
<Turl>
mripard: ping
tinti has quit [Read error: Connection reset by peer]
rz2k has quit []
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
<Turl>
oliv3r: ping
<hno>
oliv3r, merged your wip/a20 branch, and updated to 2013.07-rc3.
<hno>
that is to my wip/a20 branch..
<hno>
and updated sunxi-current.
<hno>
Hmm... github seems to hung on the sunxi repo.
<hno>
and pushed wip/a20 to sunxi repo.
\\Mr_C\\ has quit []
<hno>
oliv3r, would be good if dramc magics patch could be split to sunxi-current. Almost impossible to review a20 changes now.