<adj_>
the hdmi blinks, changes resolution and puts text incorrectly most of the time
<adj_>
hdmi linux boot messages say "missing enviroment variable" and something about retrieving pxelinux/...
<adj_>
i can't read it because it blinks too fast
<adj_>
i hope it helps you the output i pasted in pastebin
<adj_>
ask if you want some more testing
<adj_>
ssvb: what i pasted expires in 1 hour
naobsd has quit [Quit: Page closed]
<ssvb>
adj_: ok, thanks a lot! the console output looks garbled, but it gave away a useful hint - "File not found /sun4i-a10s ..."
<adj_>
now that i found a card that worked, any new test should be faster (i hope)
<ssvb>
adj_: I just made a silly typo in /boot/setup/boot-a10s.cmd (and /boot/setup/boot-a10s.scr which is generated from it)
<ssvb>
adj_: it needs to refer to /boot/setup/sun5i-a10s-bootsetup.dtb instead of /boot/setup/sun4i-a10s-bootsetup.dtb
<ssvb>
adj_: I can re-generate the image or you can alternatively just fix the /boot/setup/boot-a10s.cmd and run bootscr.sh if you have u-boot-tools installed on your linux computer
wenbin has joined #linux-sunxi
<ssvb>
adj_: bootscr.sh is in the same /boot/setup directory
<linkmauve1>
And at only 50 fps maximum, but that’s expected.
<linkmauve1>
By “it” I mean my game engine. :)
<ssvb>
linkmauve1: you are welcome :)
<linkmauve1>
I had to disable framebuffer scaling in order to get it to render.
mmarker has joined #linux-sunxi
<linkmauve1>
And there is tearing, despite it rendering to 50 fps (there is a builtin watchdog setting up a delay between each frame if vsync isn’t enabled).
<User1>
this is what I get /lib/modules/3.4.29+/linux-sunxi/scripts/gen_initramfs_list.sh: Cannot open '../../linux-sunxi/rootfs/rootfs.cpio.xz' .... make[3]: *** [usr/initramfs_data.cpio.xz] Error 1 and make[2]: *** [usr] Error 2 ... make[1]: *** [sub-make] Error 2 ... make: *** [all] Error 2
<ssvb>
linkmauve1: this is all strange, but this day seems to be full of surprises
<ssvb>
User1: why don't you ask that blog author for support?
<linkmauve1>
Anyway, it’s working, I’ll get the ALARM people to fix the build options of sdl2 to make it work without recompilation, but tomorrow.
adj__ has joined #linux-sunxi
<User1>
I tried. Seems kind of dead
<linkmauve1>
And after I clean up the various hacks I introduced to PyTouhou. ^^'
<adj__>
ssvb, sorry for the delay, i'm fixing another computer
adj_ has quit [Ping timeout: 255 seconds]
<adj__>
y fixed the image and now boots well
<adj__>
do you want me to install, login as root?
<ssvb>
adj_: no problems, thanks a lot for your help
<ssvb>
adj__: does the input work?
<adj__>
at least console, hdmi and usb host for the keyboard works
<adj__>
yes
<adj__>
with a powered hub
<adj__>
and shell on the initrdimage works
<ssvb>
btw, does this board have FEL button? it is sometimes labeled as UBOOT
<adj__>
it has a button "BOOT/REC"
<ssvb>
adj__: most likely that's it, one more possible alternative to a USB keyboard :)
<linkmauve1>
Alright, the actual patch amounts to just removing the two-passes rendering, and forcing the output to be 640×480.
<linkmauve1>
I’m pretty happy this is the only thing not working on GLES. ^^
<linkmauve1>
Especially since, for three years, we haven’t developed with GLES in mind at all.
<linkmauve1>
Anyway, good night, and thank you again. \o_
<adj__>
anything more to try?
<adj__>
ssvb, i go to bed soon
<ssvb>
adj__: maybe just check that it can identify your board correctly (this is printed in the title of the window), and also suggest the correct u-boot build for your board
<adj__>
the title of the window where i run screen?
<ssvb>
adj__: the title of the menu window on your hdmi monitor
<ssvb>
adj__: yes, and the same menu should be displayed on the serial console
<ssvb>
adj__: this SD card image does not do much, but the whole point is that it can boot on a large set of Allwinner devices and serve as an installer for some real linux distribution :)
<adj__>
i'm planning to port plan9 to the olinuxino lime2 (A20)
<adj__>
i will need much help from you
<adj__>
my console is mainly garbage, is it because the encoding?
<ssvb>
adj__: regarding console, maybe just try to close it, open a new one and start screen again
<ssvb>
adj__: your current console has probably misinterpreted some garbage during earlier tests as command sequences, which reconfigure console output
<adj__>
no, it happens if i run it again
<ssvb>
adj__: was there the same problem with the sd card image from olimex?
<adj__>
i don't remember, and i lost the image
<adj__>
i go to bed, so i will investigate tomorrow
<ssvb>
ok
<ssvb>
good night
<User1>
# if only one file is specified and it is .cpio file then use it direct as fs .. What does fs mean?
<adj__>
the only microsd that has worked is the one that had the olimex image and now yours
<adj__>
good night
<NiteHawk>
User1, it's likely "filesystem" - the .cpio archive will usually represent an initramfs
<User1>
ok
F1skr has quit [Quit: WeeChat 1.0.1]
cnxsoft has joined #linux-sunxi
mmarker has quit [Read error: Connection reset by peer]
mmarker has joined #linux-sunxi
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
User1 has quit [Quit: Page closed]
wingrime has quit [Quit: Lost terminal]
popolon has quit [Quit: WeeChat 1.0.1]
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
xenoxaos has quit [Ping timeout: 258 seconds]
xenoxaos has joined #linux-sunxi
xenoxaos has quit [Ping timeout: 258 seconds]
xenoxaos has joined #linux-sunxi
viccuad has quit [Read error: Connection reset by peer]
cnxsoft has quit [Quit: cnxsoft]
akaizen has joined #linux-sunxi
TheSeven has quit [Ping timeout: 244 seconds]
TheSeven has joined #linux-sunxi
Andy-D has joined #linux-sunxi
cnxsoft has joined #linux-sunxi
JohnDoe_71Rus has joined #linux-sunxi
konradoo77 has joined #linux-sunxi
konradoo77 has quit [Ping timeout: 258 seconds]
konradoo77 has joined #linux-sunxi
konradoo77 has quit [Ping timeout: 240 seconds]
FreezingCold has joined #linux-sunxi
cubear has joined #linux-sunxi
Andy-D has quit [Ping timeout: 250 seconds]
wenbin_ has joined #linux-sunxi
wenbin has quit [Ping timeout: 255 seconds]
wenbin__ has joined #linux-sunxi
wenbin_ has quit [Ping timeout: 250 seconds]
domidumont has joined #linux-sunxi
TheSeven has quit [Ping timeout: 244 seconds]
Andy-D has joined #linux-sunxi
wenbin_ has joined #linux-sunxi
wenbin__ has quit [Ping timeout: 250 seconds]
wenbin__ has joined #linux-sunxi
wenbin_ has quit [Ping timeout: 264 seconds]
domidumont has quit [Ping timeout: 264 seconds]
Andy-D has quit [Quit: Alive/Dead]
LinAdmin has joined #linux-sunxi
domidumont has joined #linux-sunxi
HeHoPMaJIeH has joined #linux-sunxi
alexxy has quit [Ping timeout: 264 seconds]
kurain has joined #linux-sunxi
kurain1 has quit [Ping timeout: 256 seconds]
naobsd has quit [Quit: Page closed]
_massi has joined #linux-sunxi
shineworld has joined #linux-sunxi
shineworld has left #linux-sunxi [#linux-sunxi]
jemk has quit [Remote host closed the connection]
naobsd has joined #linux-sunxi
notmart has joined #linux-sunxi
FR^2 has joined #linux-sunxi
<wens>
mripard: do you think we should put the cpufreq opps in the DT or in mach-sunxi/sunxi.c?
<mripard>
wens: in the DT
<mripard>
the A31 rev D opps will probably end up in mach-sunxi though
<mripard>
but it should be an exception
naobsd has quit [Ping timeout: 246 seconds]
<wens>
for a31 we'd need to use new bindings, like rev-abc-operating-points / rev-d-operating-points
<mripard>
I don't think that would work
<mripard>
what I had in mind was more put the default (ie A/B/C) in the DT
<mripard>
and have a quirk in mach-sunxi that would remove this OPP, and feed new ones instead
<mripard>
but I don't have any rev D hardware :/
<mripard>
do you?
<wens>
i don't know if you could tweak the opp table after cpufreq has loaded
<wens>
no :(
<mripard>
well, you can tweak it before ;)
<mripard>
cpufreq-dt is a module_platform_driver
<mripard>
so it's initialized pretty late
<wens>
cpufreq-dt will still load the default one from the DT
<EL3>
And if I dont get this to work, can anyone tell me what board/s could act like a usb client?
<oliv3r>
anybody familiar with the sunxi-mmc around? I'm curious how well the sd drive can handle errors, especially when there are long wires involved
<oliv3r>
EL3: did you install the g_midi module?
<EL3>
Yea I chose midi support in the kernel setup
<ssvb>
there is a table with glmark2-es2 benchmark results
cosm has joined #linux-sunxi
<ssvb>
linkmauve1: iirc, you mentioned that enabling the 'scaler mode' was not working correctly
<ssvb>
linkmauve1: but it is improving memory bandwidth utilization, especially with FullHD displays
<linkmauve1>
libMali and Xorg don’t seem to take much time, most of the time my game is spending is in libpython.
<linkmauve1>
Ah btw, I set my display to 720p at 60Hz, instead of 50Hz.
<linkmauve1>
My conclusion is that my game is still CPU-bound, even on ARM. :/
<ssvb>
which cpufreq governor are you using?
<linkmauve1>
According to cpufreq-info, performance.
<linkmauve1>
“current CPU frequency is 1.01 GHz (asserted by call to hardware).”
<ssvb>
ok, this is good
quitte has joined #linux-sunxi
<ssvb>
oliv3r: congratulations!
<ssvb>
oliv3r: how much of this book's content is still up to date nowadays? ;-) the software is a fast moving target
<wens>
iirc it's geared towards 3.4 for desktop usage
quitte has quit [Read error: Connection reset by peer]
<linkmauve1>
The Xorg process doesn’t seem to call memcpy a lot, and it’s using only ~7% of the CPU.
<linkmauve1>
It’s currently running at about 80~100 fps, which I find quite good actually.
<linkmauve1>
The only strange performance issue I get is when rendering the 640×480 game area to a texture first, and then scaling it with another draw call.
<linkmauve1>
It’s like it halves the framerate.
<linkmauve1>
Down to 30~40 fps when doing that.
<linkmauve1>
“1.55% libc-2.20.so [.] _int_malloc” is much higher than before.
<linkmauve1>
Err, confused malloc with memcpy. >_<
<linkmauve1>
“0.61% libc-2.20.so [.] memcpy”
<linkmauve1>
“0.58% [.] _gles_gb_buffer_object_data_range_cache_dirty” now is on top of the libMali dso.
<linkmauve1>
My guess would be that I’m breaking some asynchronousness in the driver or GPU, by using a single texture and updating it every frame before rendering from it.
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<linkmauve1>
Oh! The game is now using only ~35% of the CPU instead of ~70 without that rendering method.
<linkmauve1>
Xorg 4% instead of 7%.
<linkmauve1>
I would guess all that unused CPU is spent waiting for the GPU.
<linkmauve1>
But no idea how to confirm that. :/
<linkmauve1>
In a few days I’m going to a friend’s place, and he has an A20 Lime, I could compare that way, maybe.
<linkmauve1>
Since my engine is entirely single-threaded.
deasy has joined #linux-sunxi
cubear has joined #linux-sunxi
notmart has quit [Quit: notmart terminated!]
konradoo77 has joined #linux-sunxi
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<Turl>
oliv3r: congrats on publishing! :)
viccuad has joined #linux-sunxi
quitte has joined #linux-sunxi
mawe242 has joined #linux-sunxi
<mawe242>
i've noticed my a20-olinuxino-micro has the "usb-hardware-sc" problem, where that process is in constant "D" state. the only solution i found on the net so far is to disable otg. does anybody know a better solution? (on 3.4 kernel)
<linkmauve1>
It’s not a process AFAIK, but a kernel thread.
<linkmauve1>
It’s also in D state here, but I haven’t encountered any issue due to that.
<mawe242>
i've got a constant load of 1 due to that
konradoo77 has quit [Ping timeout: 245 seconds]
konradoo77 has joined #linux-sunxi
paulk-collins has quit [Quit: Quitte]
paulk-collins has joined #linux-sunxi
blueLumocolor has quit [Ping timeout: 272 seconds]
<Atlantic777>
libv: hey, I tested that touchscreen driver. It works on A20, too, without any changes.
<Atlantic777>
I just edited fex config and loaded driver. The problem is it's called sun4i_ts. I propose to rename it to sunxi_ts.
<Atlantic777>
Do you know maybe who to contact for that change in linux-sunxi kernel tree?
<libv>
Atlantic777: there is this guy on #linux-sunxi
<libv>
Atlantic777: called Atlantic777
<Atlantic777>
:)
<libv>
i am sure that he can provide such a trivial kernel patch and mail it into the ml
<Atlantic777>
so fork + pull request?
<Atlantic777>
Oh, ok, so patch is the way.
<libv>
just a sec, digging out the page
<Atlantic777>
Sorry, this would be my first patch commited. :)
wickwire has quit [Read error: Connection reset by peer]
konradoo87 has joined #linux-sunxi
<libv>
Atlantic777: np, people with good knowledge of the wiki can find the relevant pages rather quickly and paste the links
leviathanch_ has joined #linux-sunxi
nemunaire_ has joined #linux-sunxi
sherpa_ has joined #linux-sunxi
<Atlantic777>
And... I'm updating the wiki page about touchscreens as I promised yesterday.
<libv>
cool :)
FDCX_ has joined #linux-sunxi
konradoo77 has quit [Ping timeout: 244 seconds]
LinAdmin has quit [Ping timeout: 244 seconds]
mawe242 has quit [Ping timeout: 244 seconds]
viccuad has quit [Ping timeout: 244 seconds]
popolon has quit [Ping timeout: 244 seconds]
jukivili has quit [Ping timeout: 244 seconds]
VargaD has quit [Ping timeout: 244 seconds]
sfritsch has quit [Ping timeout: 244 seconds]
mawe242 has joined #linux-sunxi
Zboonet has quit [Ping timeout: 244 seconds]
leviathanch has quit [Ping timeout: 244 seconds]
FDCX has quit [Ping timeout: 244 seconds]
nemunaire has quit [Ping timeout: 244 seconds]
jukivili has joined #linux-sunxi
viccuad has joined #linux-sunxi
nemunaire_ is now known as nemunaire
<Atlantic777>
oh, allmighty netsplits...
<Atlantic777>
And... I've got a stupid question, too. I tried to compile kernel modules. So, make modules, but there wasn't any .ko files produced. Because of that, make modules_install also failed. Any ideas what could possibly go wrong?
<Atlantic777>
It's not like I'm building kernel and modules for the first time, I just didn't do it for some time and I really don't see why .ko weren't produced.
<libv>
did you follow our kernel build howto?
VargaD has joined #linux-sunxi
<libv>
cross compilation and stuff
* libv
really has no idea where Atlantic777 is coming from
<Atlantic777>
yup, I followed it
<Atlantic777>
And there wasn't any build errors.
<libv>
make ARCH=arm CROSSCOMPILE=... modules
<Atlantic777>
So, cross toolchain works.
popolon has joined #linux-sunxi
<Atlantic777>
I didn't even get any .ko files. It's not like they are compiler for wrong arch.
<Atlantic777>
Where do I come from? Is it because of my bad/strange English? Or you are wondering how much/little experience I have? :)
<libv>
the latter
konradoo87 has quit [Ping timeout: 264 seconds]
<Atlantic777>
Well, I'm a student, studying CS, low level stuff, but I was raised on linux and open source. Was using Gentoo during high school so... I have some experience. :D
<libv>
ok
<libv>
i still find "no modules build" a weird statement
<libv>
./drivers/input/touchscreen/sun4i-ts.ko
<libv>
does that not exist?
<Atlantic777>
find . | grep -i "\.ko$" didn't get any results
<Atlantic777>
Is it more clear that way? :D
<libv>
find . -iname "\*.ko" would also work :p
<Atlantic777>
I know, I know... I'm just used to the other way with grep. :D
<libv>
ok
<libv>
but you do have a uImage?
<Atlantic777>
nope
popolon has quit [Ping timeout: 265 seconds]
<Atlantic777>
I just wanted to build the modules. Is it necessary?
<libv>
i am not sure, i have never done this differently