deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
cnxsoft has joined #linux-sunxi
protoCall7 has joined #linux-sunxi
Andy-D has quit [Ping timeout: 240 seconds]
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
leviathanch has quit [Quit: No Ping reply in 180 seconds.]
leviathanch has joined #linux-sunxi
protoCall7 has quit [Quit: protoCall7]
xenoxaos has quit [Ping timeout: 258 seconds]
protoCall7 has joined #linux-sunxi
discopig has joined #linux-sunxi
xenoxaos has joined #linux-sunxi
TheSeven has quit [Ping timeout: 265 seconds]
TheSeven has joined #linux-sunxi
akaizen_ has joined #linux-sunxi
akaizen has quit [Read error: Connection reset by peer]
naobsd has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
konradoo77 has joined #linux-sunxi
viccuad has quit [Read error: Connection reset by peer]
konradoo77 has quit [Ping timeout: 256 seconds]
Zboonet has quit [Ping timeout: 245 seconds]
Zboonet has joined #linux-sunxi
Guest82920 has quit [Ping timeout: 258 seconds]
Guest82920 has joined #linux-sunxi
FreezingCold has quit [Ping timeout: 265 seconds]
Zboonet has quit [Quit: Leaving]
Zboonet has joined #linux-sunxi
protoCall7 has quit [Quit: protoCall7]
gianMOD has joined #linux-sunxi
HeHoPMaJIeH has joined #linux-sunxi
protoCall7 has joined #linux-sunxi
domidumont has joined #linux-sunxi
gianMOD has quit [Remote host closed the connection]
cubear has joined #linux-sunxi
selfbg has joined #linux-sunxi
jemk has quit [Ping timeout: 272 seconds]
jemk has joined #linux-sunxi
gianMOD has joined #linux-sunxi
protoCall7 has quit [Quit: protoCall7]
gianMOD has quit [Ping timeout: 240 seconds]
sehraf has joined #linux-sunxi
_massi has joined #linux-sunxi
libcg has joined #linux-sunxi
Skaag has joined #linux-sunxi
Zboonet has quit [Ping timeout: 245 seconds]
Zboonet has joined #linux-sunxi
FR^2 has joined #linux-sunxi
libcg has quit [Ping timeout: 244 seconds]
cnxsoft has quit [Ping timeout: 240 seconds]
cnxsoft1 has joined #linux-sunxi
quitte has joined #linux-sunxi
adj__ has quit [Quit: Leaving]
<vpeter>
hi, I'm looking for a way to write image to SD card in-circuit without removing it from a device and putting in USB writer. Something like this: http://hackaday.com/2014/06/08/the-in-circuit-sd-card-switch/ Is there any commercial device for this purpose?
enrico_ has joined #linux-sunxi
<quitte>
vpeter: switching hi-speed busses is hard. sdio is bidirectional push-pull iirc. I doubt there is a plug in solution. if you fully control the hardware you could switch the devices sdio port to hi-z and afterwards attach an external reader/writer
<vpeter>
I forgot to tell that only one device will use sd card at a time.
<vpeter>
So turning device off, writing image to sd card, and then turning device on.
<vpeter>
Optional step would be to unplug writer from USB bus.
<quitte>
okay. it might be worth a try and it will probably not break. here's what I would do:
<wens>
quitte: really? how come our board schematics have pull-ups
<quitte>
first of all add some diodes to the power supply lines of the sd-card to prevent leaking to the card
<quitte>
wens: I definately had working sdio without pull-up,not even configured in software on stm32. I would add some higgh resistance to any input to make sure it is always in a defined state
<quitte>
vpeter: I meant leaking to the SoC
<quitte>
and then add a plug to the data lines. only attatch the usb adapter when the device is off to your board and remove before powering up
<quitte>
wens: to make consistent what I said: I should add some resistors to make sure there is a defined state on inputs
Renard has joined #linux-sunxi
deasy has joined #linux-sunxi
Black_Horseman has quit [Ping timeout: 245 seconds]
leviathanch2 has joined #linux-sunxi
<quitte>
The hackaday link doesn't provide a lot of information. But my guess is that this only switches SPI and doesn't work with SDIO
leviathanch2 has quit [Ping timeout: 264 seconds]
Renard has quit [Remote host closed the connection]
quitte has quit [Read error: Connection reset by peer]
<wens>
he thinks it really doesn't matter if you get bad performance due to low bus speed
<wens>
btw, someone needs to do the thermal calibration on the A10
<wens>
on sun5i and later, we can always clock ahb from pll6
Seppoz has joined #linux-sunxi
Renard has joined #linux-sunxi
stefan_a has joined #linux-sunxi
<Turl>
wens: I'd be worried about the lower bounds required from some peripherals, not performance
<Turl>
eg this note on the 3.4 cpufreq driver "for usb connecting,* the frequency of ahb must not lower than 60Mhz."
<Turl>
wens: so with your minimum of 432000 and the default /4/2 divisor iirc it'd be off already
tomboy65 has quit [Ping timeout: 250 seconds]
indy has joined #linux-sunxi
<Turl>
it'd probably be wise to ask AW what their thoughts are on this
jemk has quit [Ping timeout: 272 seconds]
afaerber_ is now known as afaerber
quitte has quit [Read error: Connection reset by peer]
libcg has joined #linux-sunxi
leviathanch2 has quit [Ping timeout: 264 seconds]
stefan_a has quit [Ping timeout: 246 seconds]
F1skr has joined #linux-sunxi
Andy-D has quit [Ping timeout: 240 seconds]
Netlynx has joined #linux-sunxi
selfbg has quit [Quit: Leaving]
Black_Horseman has joined #linux-sunxi
<wens>
Turl: I'm aware of the requirements on usb
cubear has quit [Quit: Leaving]
viccuad has joined #linux-sunxi
indy has quit [Ping timeout: 265 seconds]
Black_Horseman has quit [Ping timeout: 258 seconds]
<mripard>
wens: Turl: if it's just for USB, we can use the floor/ceiling stuff
<mripard>
from Tomeu
leviathanch2 has joined #linux-sunxi
indy has joined #linux-sunxi
arokux has joined #linux-sunxi
popolon has joined #linux-sunxi
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
Andy-D has joined #linux-sunxi
protoCall7 has joined #linux-sunxi
protoCall7 has quit [Ping timeout: 272 seconds]
<wens>
on sun7i it's /3/2, so 432 MHz gives 144 MHz for AHB
adj_ has joined #linux-sunxi
leviathanch2 has quit [Ping timeout: 258 seconds]
jemk has joined #linux-sunxi
viccuad has quit [Read error: Connection reset by peer]
LinAdmin has joined #linux-sunxi
<LinAdmin>
hello experts ;)
<LinAdmin>
On the linux-sunxi mailing list I have asked a question regarding SATA speed of A20 where I describe that all kernels from 3.4 up to 3.18 that I have built show 4 times higher reading speed from SATA compared to writing.
<LinAdmin>
Imho reading and writing speeds from/to a fast enough 3.5" disk are limited by DMA transfer speed?
<LinAdmin>
Could anybody explain me why one direction is 4 times faster than the opposite??
Atlantic777 has joined #linux-sunxi
<Atlantic777>
I have to make a resistive touch panel work on BPi. I found in A20 manual that there's built in 4 wire RTP controller but I'm not sure which driver I need for it.
cnxsoft1 has quit [Quit: cnxsoft1]
<Atlantic777>
There's touchscreen_sun4i_ts which is renamed to touchscreen_sunxi_ts in some repos.
<libv>
who is forcing you to work with resistive?
<libv>
and why would you want the banana pi in such a case, as opposed to the olimex lime
<Atlantic777>
libv: those are parts which are already bought. I'm trying to help my friend on a project.
imcsk8 has quit [Quit: Reconnecting]
imcsk8 has joined #linux-sunxi
<libv>
well, there are two not so helpful statements here
<libv>
1) very few people here have dealt with touchscreens
quitte has joined #linux-sunxi
<libv>
2) i do not think that much of our hw ever shipped with resistive
<libv>
but send an email to the ml
<libv>
that is much more likely to get an answer
<libv>
as this is definitely not a common question
<libv>
how did you get from banana pi to linux-sunxi btw?
<Atlantic777>
linux-sunxi looks like standard kernel for A20 and similar allwinner's SoC.
<LinAdmin>
@libv: I landed from BananaPI at sunxi because BananPI seems to be glued to the 3.4 kernel ;)
<libv>
yes, but bananapi is a big fraude that really does not have a community or a big support or open source footing
<libv>
and tons of people are very misguidedly buying these devices and then coming to us for help
<NiteHawk>
well, banana pi is essentially an A20 platform - why would sunxi exclude it?
<LinAdmin>
@libv what do you mean by "a big fraude" ?
<libv>
so the nasty answer would be "surely the magnificent banana pi community should help you here"
<LinAdmin>
@libv: ???
<libv>
LinAdmin: they tried their absolute best to get the maximum coverage and sales by leeching the rpi design and name (badly), and then using the work from our community
<libv>
NiteHawk: exclude is a big word
<libv>
NiteHawk: but we do get an awful lot of people in here asking bpi questions, questions which, if the bpi was not such a fraude, the bpi community and/or the company behind it should be able to answer
<NiteHawk>
i agree that they're riding piggyback on the successs of raspi and are not as "open" / helpful as their advertising / labelling suggests
<LinAdmin>
@libv: sorry for not knowing, but what is "rpi design"?
<libv>
LinAdmin: raspberry pi?
<libv>
the layout, from afar, is meant to be the same
<libv>
but everything is off by a few mm
<Atlantic777>
That's the GPL spirit, isn't it? If you can do better, fork it.
<libv>
Atlantic777: to some extent.
<LinAdmin>
@libv: I see, but they did a step that the raspberry manufacturers should/could have done long ago?
<libv>
Atlantic777: but people come in here and expect us to be responsible
afaerber_ has joined #linux-sunxi
<Atlantic777>
libv: nope, GPL excludes responsibility of any kind. :)
<libv>
Atlantic777: ok, do you want to see how long that will last?
<Atlantic777>
And linux-sunxi leached linux source. And that's good, that's how it should be.
<NiteHawk>
okay, but wouldn't that mean you'd end up with olimex, cubieboard2, banana, .... micro ecosystems (communities)? they're sparse enough on their own compared to the raspberry, so it might be more sensible to go "a20 of the world unite" ;)
<LinAdmin>
@libv: Imho it is Allwinner which should have given more support for their A20 chips
<libv>
LinAdmin: oh, definitely
<libv>
NiteHawk: banana pi is all over the market right now
<libv>
and they actively try to portray themselves as a competitor to rpi, in every way
<libv>
which is a big lie
<NiteHawk>
well, sure. it's nice value for money (in terms of hardware capababilities) - i got one myself, too :P
viccuad has joined #linux-sunxi
<LinAdmin>
@libv: I hope you are aware of the fact that BananaPI now is exclusively sold by Sinovoip Ltd. in China? LeMaker can not deliver any product at the momen!
afaerber has quit [Ping timeout: 240 seconds]
<libv>
we here get the people who did not research things and ask all sorts of questions which one would've expected that oh so magnificent bpi community to handle
<libv>
at least with cubieboard and olimex, linux-sunxi was not used parasitically
<Atlantic777>
I came here with the question "is there a driver for touch panel controller in A20". Let's forget BPi.
<libv>
Atlantic777: oh, i am not complaining about your specific question
<libv>
just that 90% of new people in here ask about bpi
<NiteHawk>
that's a valid point for sure. but i agree it's probably not so much lemaker / sinovoip, but a lack of proper documentation from allwinner
<libv>
and some of the questions really are pretty bad
<LinAdmin>
I came here with the question regarding "SATA speed" - Let's forget BPi.
<libv>
"i got this rootfs from rpi, and then got a kernel from bpi, and now it won't boot. linux-sunxi must help me now."
<libv>
LinAdmin: again, that is something for the ml
<LinAdmin>
@libv: the ml did not give any solution
<libv>
LinAdmin: then why did you think that you would get an answer on irc?
paulk-collins has joined #linux-sunxi
<LinAdmin>
@libv because the explanantions for the ml give the link to the irc channel :O
<libv>
LinAdmin: email and irc are two totally different mediums
<NiteHawk>
libv: you'll encounter that sort of attitude with almost any hardware - people buy 'gadget' hardware, and then fail on the basics. and the right answer in _those_ cases is "go figure".
<LinAdmin>
@libv: that is clear, but imho some expert in the chat might know that very technical problem of SATA-DMA????
<libv>
LinAdmin: "some" "might"
<libv>
those are the two keywords in that sentence
<libv>
LinAdmin: and why do you think that people here don't read email?
<NiteHawk>
but if i have specific problems related to the hardware ('generic' A20) - say i need more details on the FEL boot process, i'd think this is the right place to come / ask
<Atlantic777>
I don't like when conversations goes this way... Let's do something more productive.
<LinAdmin>
@atlantic777 fully agreed
<libv>
irc is temporarily limited, email is not.
<libv>
yes, use the mediums productively, and tap out an email.
<LinAdmin>
@libv: methinks that questions of a newbee of the ml are not read by experts?
diego71_ has joined #linux-sunxi
<libv>
LinAdmin: depends on how the question is asked
<libv>
the same is absolutely true for irc
<libv>
actually worse
<libv>
on irc you're more likely to get larted, on email people just stay silent
<LinAdmin>
@libv: what do you mean by "how"?
<libv>
i am not going to write a big explanation on how to ask good questions
<LinAdmin>
@Atlantic777: What do you call "a similar kernel/driver stack"?
<LinAdmin>
@adj_: Have used that one too.
<Atlantic777>
LinAdmin: the same kernel release.
<Atlantic777>
And with the same configuration, tweaked only in places where it needs to so it can run on the other platform.
<adj_>
tkaiser reported less performance when using ethernet and sata simultaneously
<Atlantic777>
IRQ collisions?
<LinAdmin>
@Atlantic777: I have mainline kernel 3.18 on X86 and on A20.
<Atlantic777>
LinAdmin: ou, x86 is so ackward and different... I mean another ARM platform.
<LinAdmin>
@Atlantic: I do not have other Arm hardware with SATA.
<LinAdmin>
@Atlantic777: When DMA transfer between disk and RAM occurs, then there are only very few other interrupts. Do you think those could interfere?
<Atlantic777>
What tkaiser reported makes sense. Similar things happen with RPi when you use USB and ethernet because ethernet controller is also connected to usb if iirc.
<adj_>
rpi ethernet is usb2ethernet in the usb hub
<Atlantic777>
that ^
<LinAdmin>
@Atlantic777: I do not have RPi, but I guess that these limitations happen in both directions the same way?
<adj_>
A20 has native sata, although i don't know what resources are shared inside SOC
<LinAdmin>
What surprises with A20: Why is reading so blindly fast and only writing 4 times slower??
<Atlantic777>
and this is what wanted to say, adj_
<LinAdmin>
Imho the same DMA engine is active in both directions?
<Atlantic777>
LinAdmin: are you sure it's not the medium and slave controller?
<Atlantic777>
The problem doesn't exist if you use the drive on a PC?
<adj_>
"taiser: The funny thing is: When measured individually neither network nor storage performas that bad. Only when used concurrently in a NAS setup especially write performance nearly stalls: http://kaiser-edv.de/tmp/mZz3Ek/"
<Atlantic777>
Hm, I'm late for an exam. -.- See you later!
<adj_>
he is using olinuxino lime2 (A20), so the board design is different
<LinAdmin>
@adj_: tkaiser is wrong. Writing also stalls when no network traffic runs.
<Atlantic777>
happy hacking :)
<LinAdmin>
Atlantic777: thx, take it easy
<adj_>
He is not wrong, with his board and his workload " When measured individually neither network nor storage performas that bad. Only when used concurrently"
<adj_>
maybe it's different in your case
<LinAdmin>
@adj_: he did not give me clear numbers and I still do have some doubts
<LinAdmin>
Detail question: RAM reading and writing speeds in A20 must be the same?
Froolap has joined #linux-sunxi
libcg has quit [Ping timeout: 258 seconds]
<adj_>
LinAdmin, obviously A20 is not a high performance SATA implementation, as it doesn't saturate even SATA I with a SATA III SSD fast device
<LinAdmin>
@adj_: for sure, but this does not explain 4 times slower writing to disk
<LinAdmin>
@adj_: Detail question: RAM reading and writing speeds in A20 must be the same?
<adj_>
haven't tested
<LinAdmin>
@adj_: is it reasonable that I do test that?
<adj_>
ram speed is orders of magnitude faster than a hdd, don't think that it is related
<LinAdmin>
@adj_: yes, but in case there is a miraculous slow down of factor 4, I check the unprobable
<adj_>
are you testing sequential write speed?
domidumont has quit [Read error: Connection reset by peer]
<adj_>
are you sure that the disk cache is enabled
<LinAdmin>
@adj_: yes, sequential read and write. How would disk cache become disabled?
<adj_>
do you mount with a parameter that always syncs?
<adj_>
tons or reasons are more probable than ram speed
<LinAdmin>
@adj_: no special mount params given, but I will check that
<adj_>
what filesystem are you using?
<adj_>
are you testing writing to the disk device?
<LinAdmin>
ext3, ext4, jounals disabled
<LinAdmin>
the slow down is also visible when writing to the raw disk, when writing to a partion or to a file
MackBoy has quit [Remote host closed the connection]
<adj_>
try with a ssd if you can, most people report ~40MB/s writing to a ssd
<LinAdmin>
@adj_: I saw these values and 40JB/s on a fast SSD is proof of the slowdown
MackBoy has joined #linux-sunxi
<adj_>
you are reporting 25MB/s, not so different
<LinAdmin>
Yes - in both cases writing is much slower than reading
<LinAdmin>
my best guess: either the setup of dma engine is faulty or it is a problem in A20 chip itself
<adj_>
A20 is a SOC for tablets and media centers, doubt allwinner is too worried about only 40MB/s write speed trough sata
<LinAdmin>
@adj_: fully agreed
HeHoPMaJIeH has quit [Quit: Konversation terminated!]
<LinAdmin>
@adj_: thanks for your hints, I will try with a SDD, benchmark DMA and RAM and see for IRQ conflicts. bye
<Turl>
if you pick the wrong one you could hose your PC's OS
_massi has quit [Quit: Leaving]
formruga has quit [Quit: Konversation terminated!]
<EL3>
okey now I get sfdisk: kunne ikke åbne /dev/sdb til læsning. In englisg thats could not open for read
<EL3>
or something
<Turl>
are you running it as root?
popolon has quit [Ping timeout: 258 seconds]
<EL3>
okey that helped, but disk is in use, I guess I just unmount it
<EL3>
hmm then it cannot open for read again
<EL3>
ofcousre because it is unmounted
<Turl>
I bet you ejected it completely instead of unmounting it
<EL3>
oh yea probably.
<gbisson>
hi, would any of you have some advice on where to get a linux kernel for the f1e200 processor? on the wiki it is clear that linux-sunxi community does not support it but maybe some of you would have some pointers?
<EL3>
hmm it says device is in use
<EL3>
okey worked now, I used gparted to unmount it
<Turl>
gbisson: that's sun3i isn't it?
<gbisson>
Turl: wiki says that yes
<Turl>
gbisson: the sun4i code drops had some sun3i drivers floating around, but as far as I know there's nothing complete for sun3i
<Turl>
maybe wingrime knows something I don't
<Turl>
gbisson: I believe AW didn't ship linux on their sun3i devices, they used their own OS or something
<gbisson>
Turl: okay thanks! I'm confused with the naming to be honest
<EL3>
Okey now I am stuck at the next command.. --> cat <<EOT | sfdisk --in-order -L -uM /dev/sdb
<gbisson>
Turl: the linux-sunxi mentions sun3i F20 support but I'm not sure how different is the F20 from the E200
<EL3>
sfdisk: kunne ikke åbne /dev/sdb til skrivning - english could not open for writing
<Turl>
el3: not running as root?
<EL3>
sry
<EL3>
got it again
<EL3>
sudo cat <<EOT | sudo sfdisk --in-order -L -uM /dev/sdb... did not realize I should put sudo in 2 places
<Turl>
gbisson: there's not that many people with sun3i hw and info is rather scarce
<Turl>
el3: you can live without the sudo on cat if you want to use just one :)
<Turl>
gbisson: what device is this btw?
<EL3>
:) I try to get more advanced linux user. Kinda noob at the moment
<gbisson>
Turl: custom design from a customer that wants Linux support
<Turl>
el3: practice makes perfect :)
<Turl>
gbisson: so it's 2014 hardware as to say, cool
<gbisson>
Turl: I don't know much right now but I wanted to have the basics covered and it turned out to be pretty hard to get a kernel
<Turl>
gbisson: you should be able to get an SDK from allwinner if you ask and are buying them the chips
<EL3>
These next commands confuse me --> mkfs.vfat ${card}${p}1. What should I put in there ? mkfs.vfat /dev/sdb${p}1
<gbisson>
Turl: what about the sunxi-3.0 branch, it has a sun3i defconfig, do you know if that would be a good starting poitn?
<Turl>
gbisson: that's based on a sun4i code drop, and I don't think anyone ever tested the sun3i stuff
<Turl>
but sure, better than nothing I guess
<Turl>
el3: /dev/sdb1
<EL3>
okey thanks
<gbisson>
Turl: arf, ok thanks Turl! that was very helpful
<EL3>
But not under the Build, should I enter those commands too
<EL3>
make -j4 ARCH=arm CROSS_COMPILE=arm-linux-gnueabihf- uImage modules ??
<EL3>
It says: no rule for making uImage. Stop
quitte has quit [Read error: Connection reset by peer]
<EL3>
u-boot-tools is installed. I run the command : make -j4 ARCH=arm CROSS_COMPILE=arm-linux-gnueabihf- uImage modules . And I get --> make: *** No rule to make target `uImage'. Stop.
konradoo77 has quit [Ping timeout: 258 seconds]
Guest82920 has quit [Remote host closed the connection]
<ssvb>
el3: does running "mkimage --help" work (to check whether you have really installed u-boot-tools)?
jemk has joined #linux-sunxi
domidumont has quit [Ping timeout: 264 seconds]
<EL3>
ssvb: sudo make -j4 ARCH=arm CROSS_COMPILE=arm-linux-gnueabihf- uImagae worked now. I suppose I was in wring dir as you mentioned
<EL3>
andmake ARCH=arm CROSS_COMPILE=arm-linux-gnueabihf- INSTALL_MOD_PATH=output modules_install also worked
<EL3>
ssvb, Now I shall create the boot.cmd.. It says I shall tyrpe this -->setenv bootargs console=ttyS0,115200 root=/dev/mmcblk0p2 rootwait panic=10 ${extra}
<EL3>
what do I put in $ {extra}
<ssvb>
el3: whatever you want, maybe even nothing
<EL3>
ssvb, okey.
<EL3>
hmm setenv commando not found
ricardocrudo has quit [Remote host closed the connection]
tomcheng76 has quit [Read error: Connection reset by peer]
<EL3>
Okey now when I do: setenv bootargs console=ttyS0,115200 root=/dev/mmcblk0p2 rootwait panic=10, I get too many arguments
tomcheng76 has joined #linux-sunxi
cubear has quit [Quit: Leaving]
wingrime has quit [Ping timeout: 245 seconds]
<linkmauve1>
I’m trying to setup a GLES context with libMali, so I do eglGetDisplay() on an X11 Display, which returns 1 as the EGLDisplay, and then when I call eglInitialize() with it I don’t get GL_TRUE.
<linkmauve1>
What could I do to get it to work?
<linkmauve1>
Ah btw, something strange is that I have both card0 and card1 in /dev/dri/
<linkmauve1>
And AFAIK there is only one Mali on my A10.
<linkmauve1>
… and now that I changed the permissions on /dev/dri/card*, I do get that context!
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
<linkmauve1>
Also, there doesn’t seem to be any vsync, and tearing is terrible.
wingrime has joined #linux-sunxi
bonbons has quit [Quit: Leaving]
<ssvb>
linkmauve1: looks like either the ddx driver is not using disp layers, or maybe your xorg.conf file is disabling vsync
<ssvb>
linkmauve1: please check /var/log/Xorg.0.log (maybe pastebin it somewhere)
nicksydney has quit [Quit: No Ping reply in 180 seconds.]
<ssvb>
linkmauve1: [ 8933.256] (II) FBTURBO(0): tear-free zero-copy double buffering needs more video memory
<linkmauve1>
Heh. ^^
<linkmauve1>
Thank you.
<linkmauve1>
I’ll change my uEnv.txt accordingly.
<linkmauve1>
It’s the sunxi_ve_mem_reserve?
<ssvb>
linkmauve1: it's explained in Xorg.0.log
<linkmauve1>
Ok, sorry.
<ssvb>
linkmauve1: sunxi_fb_mem_reserve is the right variable for uEnv.txt
<linkmauve1>
I already had it set to 30.
nicksydney has quit [Quit: No Ping reply in 180 seconds.]
<linkmauve1>
(For a 1280×720@60 display.)
nicksydney has joined #linux-sunxi
<ssvb>
linkmauve1: yeah, unfortunately another modification needs to be done in script.bin (fex)
<ssvb>
linkmauve1: without that tweak, sunxi_fb_mem_reserve=30 just reserves extra memory but does not really use it
<linkmauve1>
Heh. ^^
<linkmauve1>
So with fb0_framebuffer_num=3 that should be ok?
<ssvb>
yes
<linkmauve1>
It was currently set to 2.
<ssvb>
we should just do the necessary changes in the sunxi-3.4 kernel, this mess involving 'sunxi_fb_mem_reserve' and fb0_framebuffer_num just does not make much sense
<ssvb>
adj_: maybe I'm pinging you too early, but did you have time to test that sd card image on your olinuxino a10s?
<adj_>
i just arrived at home
<adj_>
i write as soon as i test
<ssvb>
adj_: thanks!
<ssvb>
this should not take much time, just insert this sd card, plug a hdmi cable, power on the device and check
<ssvb>
1) if you see a text menu on your monitor
<ssvb>
2) if you can use this menu (have some means of handling input)
<EL3>
okey. I still dont reallyknow what I am doing, but I came this far. :) ill try that thank you
<ssvb>
el3: and feel free to update the linux-sunxi wiki to fix the broken rootfs links if you confirm that this one works :-)
nicksydney has quit [Quit: No Ping reply in 180 seconds.]
nicksydney has joined #linux-sunxi
<EL3>
okey
<ssvb>
linkmauve1: how do we know which fex file is up to date?
<linkmauve1>
Alright, I don’t have that message anymore when I start Xorg, but the tearing still looks identical.
<linkmauve1>
ssvb, I just diff’d the provided fex file and the disassembled script.bin.
<linkmauve1>
Since the provided script.bin made the display work and the provided fex didn’t, I guess the script.bin is the correct one.
<linkmauve1>
But this is something I should look into a bit more.
<ssvb>
linkmauve1: oh, how nice of them :-) https://github.com/linux-sunxi/sunxi-boards is the "official" repository for the fex files, but there may be different revisions of it floating around
<ssvb>
linkmauve1: hmm, but having tearing is not nice
<linkmauve1>
Yet another one, different from the two I currently have, yay!
<linkmauve1>
Actually I don’t intend to use this board as a gaming console, just as a GLES-only computer on which to test my stuff.
<linkmauve1>
So having tearing is not the worst thing.
<ssvb>
linkmauve1: you have "[ 8933.209] (**) FBTURBO(0): Option "SwapbuffersWait" "true"" in the Xorg.0.log, which means that it at least should try to do vsync
<linkmauve1>
(Of course it’d be better without it. ^^)
<linkmauve1>
Is there an environment variable I could set to force it to vsync, like on mesa?
<linkmauve1>
Btw, the tearing is something I had never seen, it’s quite horizontal, but like a hand-drawn line.
<ssvb>
linkmauve1: oh, what, does the "tearing" look like the whole screen is shaking vertically?
<linkmauve1>
No, it’s localized to the OpenGL window.
<linkmauve1>
(A 512×512 window at the center of the screen.)
<linkmauve1>
I am on HDMI, if that could change anything.
<adj_>
ssvb, it depends on the speed of the modem downloading the image
<ssvb>
adj_: that image is only ~7MB, downloading should not be a problem unless you are using an old dialup modem :-)
<ssvb>
linkmauve1: eglSwapInterval is not supported by the mali drivers at all
<linkmauve1>
Ok, that’s one less worry. ^^
<linkmauve1>
It doesn’t error anything though when I call it.
<ssvb>
linkmauve1: you have an a10 lime board with a 16-bit dram bus width, which is not great, but supposedly 1280x720 should be ok (1920x1080 could be a big problem)
Andy-D has quit [Ping timeout: 272 seconds]
<ssvb>
linkmauve1: can you try to also enable "scaler_mode" for fb0 in the fex file?
<linkmauve1>
Alright, rebooting.
<linkmauve1>
No, it didn’t change anything wrt vsync.
<linkmauve1>
Btw, my method of testing is to call glClearColor() with a random color, then glClear() then SDL_GL_SwapWindow() for 120 times, which should take exactly two seconds but currently takes less than one second.
<ssvb>
adj_: thanks for trying, do you have a serial console?
<adj_>
i have a usb serial cable
<linkmauve1>
This is using SDL2 master, compiled on the board itself with Mesa’s headers and libMali’s everything else, without any other modification.
<adj_>
i'm going to investigate
<linkmauve1>
ssvb, I don’t notice any tearing here in gles-rgb-cycle-demo.c.
<ssvb>
adj_: the USB serial cable is fine, this SD card image is supposed to be using serial console too and there might be some useful debugging output
<linkmauve1>
Ah!
<linkmauve1>
I can reproduce if I pass 1 as its first argument!
<linkmauve1>
With 42 it’s fine though.
<linkmauve1>
16 has some tearing at the very bottom of the window, 20 has none.
<linkmauve1>
That tearing can do some pretty nice drawings. ^^
konradoo77 has joined #linux-sunxi
<linkmauve1>
But I can confirm it’s present on gles-rgb-cycle-demo.c too if msleep_time is low enough.
<ssvb>
ok, that's good to know
<linkmauve1>
Oh, I had an issue with apitrace which was trying to load libGL.so, apparently it’s due to libepoxy.
<linkmauve1>
It works fine with your demo.
LinAdmin has quit [Quit: LinAdmin]
<ssvb>
linkmauve1: hmm, was it using libGL from Mesa when it had tearing?
<linkmauve1>
No, I haven’t installed Mesa at all on this board.
<adj_>
ssvb, possible problem sdcard, i try with other one
<linkmauve1>
Oh btw, “[EGL-X11] [2254] DETECTED ONLY ONE FRAMEBUFFER - FORCING A RESIZE”
<ssvb>
adj_: well, either that, or it could be indeed a problem with this sd card image
<ssvb>
adj_: , I don't have a10s hardware myself to test and debug it :-(
FreezingCold has quit [Ping timeout: 255 seconds]
konradoo77 has quit [Ping timeout: 240 seconds]
<ssvb>
linkmauve1: that's a normal spam from the mali blob
<linkmauve1>
Ok.
<ssvb>
linkmauve1: do you have any other windows overlapping the test window?
<linkmauve1>
No.
<ssvb>
linkmauve1: or actually a compositing window manager?
<linkmauve1>
No, just dwm.
<linkmauve1>
Which is the most minimalist one I’ve found.
EL3 has quit [Ping timeout: 264 seconds]
<ssvb>
linkmauve1: this is all strange, I can't reproduce tearing with "gles-rgb-cycle-demo 1" here, the screen blinking is rather uniform for me
<ssvb>
linkmauve1: or window blinking in the windowed mode
<ssvb>
linkmauve1: then run the sunxi_disp_vsync_demo from the test directory
naobsd has quit [Quit: Page closed]
<ssvb>
linkmauve1: sorry for asking you to test so many things, I just want to be sure that the kernel itself is not fubared by some third-party patches
<linkmauve1>
No problem, I’m here for that. :)
<linkmauve1>
I think I have every dependency, but “./configure: line 18555: syntax error near unexpected token `RANDR,'” “./configure: line 18555: `XORG_DRIVER_CHECK_EXT(RANDR, randrproto)'”
<linkmauve1>
This is on an ArchLinuxARM chroot, so maybe I’m missing something.
<linkmauve1>
Ah yes, xorg-server-devel.
<linkmauve1>
Which is installing mesa. :/
<ssvb>
chroot? what is your primary system on this device?
<ssvb>
adj_: do you have some other allwinner devices (a10 or a20)?
<adj_>
a13
<linkmauve1>
./sunxi_g2d_bench has a lot of tearind though.
<adj_>
olinuxino a13
<adj_>
ssvb, can be that is trying to boot from board flash?
<ssvb>
adj_: sd card is the primary boot media on a10/a10s/a13/a20, it should have higher priority than nand
<ssvb>
adj_: a13 has no hdmi, so it's not very useful
<ssvb>
linkmauve1: yeah, the bench program does not care about vsync
<linkmauve1>
So the kernel is ok, but not the Mali integration with it?
<ssvb>
linkmauve1: right, and I'm out of ideas at the moment
<linkmauve1>
:)
<linkmauve1>
Thank you for your help anyway, if I find something I’ll tell you.
<ssvb>
linkmauve1: other than trying some good known image, which I'm not ready to provide offhand :)
<ssvb>
linkmauve1: ok, thanks
<linkmauve1>
I’ll test some other image another day, for now I’m going to try to see why my game engine is displaying a black screen. ^^
<linkmauve1>
On Mesa/i965 it’s working fine in GLES mode, but I wouldn’t exclude some extension making it work although it shouldn’t in a pure GLESv2 environment.