issue_ has quit [Remote host closed the connection]
akaizen has joined #linux-sunxi
Andy-D has quit [Ping timeout: 260 seconds]
xinj has joined #linux-sunxi
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
linkmauve1 has quit [Ping timeout: 256 seconds]
bluse_ has joined #linux-sunxi
bluse has quit [Ping timeout: 256 seconds]
Skaag_ has quit [Ping timeout: 250 seconds]
dlan has quit [Ping timeout: 240 seconds]
dlan has joined #linux-sunxi
dlan has quit [Ping timeout: 264 seconds]
FreezingCold has quit [Ping timeout: 255 seconds]
dlan has joined #linux-sunxi
dlan has quit [Changing host]
dlan has joined #linux-sunxi
FreezingCold has joined #linux-sunxi
<wens>
oliv3r: benn is leaving as well?
torindel has quit [Ping timeout: 272 seconds]
mmarker has quit [Remote host closed the connection]
<libv>
heh, there's no such thing as a full clock reset on sunxi-3.4
<libv>
just the video0 pll multiplier is set badly
<libv>
which then of course throws the hdmi block off
<libv>
we will need some bits of display driver put into our platform code, otherwise things will fuck up.
FreezingCold has quit [Ping timeout: 250 seconds]
rz2k has quit [Read error: Connection reset by peer]
phh has quit [Ping timeout: 256 seconds]
naobsd has joined #linux-sunxi
montjoie[home] has joined #linux-sunxi
bhag has joined #linux-sunxi
physis has quit [Ping timeout: 260 seconds]
bhag has quit [Ping timeout: 272 seconds]
bhag has joined #linux-sunxi
Skaag has joined #linux-sunxi
physis has joined #linux-sunxi
GM_ has joined #linux-sunxi
akaizen has quit [Remote host closed the connection]
akaizen has joined #linux-sunxi
GM_ has left #linux-sunxi [#linux-sunxi]
akaizen has quit [Ping timeout: 240 seconds]
bhag has quit [Quit: Leaving]
bhag has joined #linux-sunxi
naobsd has quit [Quit: Page closed]
sehraf has joined #linux-sunxi
bhag has quit [Quit: Leaving]
y0g1 has joined #linux-sunxi
hansg has joined #linux-sunxi
Quarx has joined #linux-sunxi
hansg has quit [Quit: Leaving]
Black_Horseman has quit [Ping timeout: 250 seconds]
tomboy64 has joined #linux-sunxi
Skaag has quit [Ping timeout: 264 seconds]
Quarx has quit [Ping timeout: 250 seconds]
hramrach has joined #linux-sunxi
Quarx has joined #linux-sunxi
Andy-D has joined #linux-sunxi
akaizen has joined #linux-sunxi
yann_s|AFK has joined #linux-sunxi
yann_s|AFK is now known as yann_s
Quarx has quit [Ping timeout: 250 seconds]
akaizen has quit [Ping timeout: 240 seconds]
Quarx has joined #linux-sunxi
kivutar has joined #linux-sunxi
orly_owl_ is now known as orly_owl
orly_owl has quit [Changing host]
orly_owl has joined #linux-sunxi
bonbons has joined #linux-sunxi
dlan has quit [Ping timeout: 260 seconds]
Skaag has joined #linux-sunxi
dlan has joined #linux-sunxi
Andy-D has quit [Ping timeout: 240 seconds]
rz2k has joined #linux-sunxi
libcg has joined #linux-sunxi
Quarx|2 has joined #linux-sunxi
dlan has quit [Ping timeout: 245 seconds]
Quarx has quit [Ping timeout: 250 seconds]
Quarx|2 has quit [Client Quit]
dlan has joined #linux-sunxi
netlynx has joined #linux-sunxi
netlynx has quit [Changing host]
netlynx has joined #linux-sunxi
montjoie[home] has quit [Quit: leaving]
montjoie[home] has joined #linux-sunxi
paulk-collins has joined #linux-sunxi
BorgCuba has joined #linux-sunxi
libcg has quit [Ping timeout: 240 seconds]
xinj has quit [Read error: Connection reset by peer]
Quarx has joined #linux-sunxi
xinj has joined #linux-sunxi
physis has quit []
bgal has joined #linux-sunxi
Renard has joined #linux-sunxi
libcg has joined #linux-sunxi
Quarx has quit [Ping timeout: 250 seconds]
Quarx has joined #linux-sunxi
kivutar has quit [Quit: Ex-Chat]
Nazcafan has joined #linux-sunxi
FreezingCold has joined #linux-sunxi
kivutar has joined #linux-sunxi
physis has joined #linux-sunxi
Nazcafan has quit [Quit: Quitte]
link0_ is now known as link0
afaerber_ has joined #linux-sunxi
libcg has quit [Ping timeout: 255 seconds]
kivutar has quit [Ping timeout: 240 seconds]
afaerber has quit [Ping timeout: 240 seconds]
ricardocrudo has joined #linux-sunxi
afaerber_ is now known as afaerber
kivutar has joined #linux-sunxi
kivutar has quit [Read error: Connection reset by peer]
Skaag has quit [Ping timeout: 240 seconds]
linkmauve1 has joined #linux-sunxi
<libv>
pfff, i have mode, i have text and a logo
<libv>
but i have lost all respect for u-boot and denx
xeros has quit [Ping timeout: 256 seconds]
<libv>
who was that working on coreboot?
<libv>
he should get cheered on.
<ssvb>
libv: the grass is always greener on the other side :)
<ssvb>
I'm sure you are going to find more than one reason to complain about coreboot or anything else
<libv>
i was the first to add native display support at coreboot
<libv>
i found loads of issues
<libv>
particularly their terrible use of the rtc
<libv>
which made it impossible for the display support to work out of the box
<libv>
but nothing as bad as this
<ssvb>
anyway, "i have mode, i have text and a logo" sounds like a great progress
<ssvb>
thanks for your hard work
<libv>
ssvb: there are two interfaces here apparently
<libv>
one is cfbconsole
<libv>
which allows you to return NULL if you found nothing
<libv>
which is what i am using now
<libv>
and there is lcd
<libv>
which has modern things like separate pitch definition
<libv>
but which does not allow you to state "nope, i found nothing, go on without me"
leviathanch2 has quit [Ping timeout: 250 seconds]
<libv>
it's really rather awkward, as they are pretty stuck in embedded thinking, and no-one has gone and rewritten the cfb drivers to work with the "newer" infrastructure
<libv>
then... absolutely nothing is documented
<libv>
it is not clear what is an interface and what isn't
<libv>
weak/strong symbols hold this thing together
<libv>
like a house of cards.
<libv>
it's amazing
<ssvb>
you can probably send a polite e-mail and ask the guys
<libv>
oh, it works, they'll just get a single patch, at best
<libv>
if they complain about my {} i complain about their memset_io stupidity
<libv>
{} is there so i can have an int i and run over the register space manually
<libv>
really though, coreboot is the way to go
<libv>
they have like driver objects
<libv>
and callbacks
<ssvb>
be sure to use the 'checkpatch.pl' script, the formal rules are nice because they can be automatically checked and enforced
<libv>
all that modern nineties stuff
<libv>
yeah, i know
<libv>
ssvb: this simplefb stuff is for the lcd driver only
<libv>
i'd probably have to go copy the code and hack it to make it work outside of the lcd "namespace"
<ssvb>
but what about the Raspberry Pi? I don't think that it is commonly used with LCD
<libv>
it only has an lcd driver.
<libv>
fits in with their edid code
<libv>
embedded ostrich thinking
Skaag has joined #linux-sunxi
abdel has joined #linux-sunxi
rz2k has quit []
<abdel>
hi , to everbody !! i am trying to root an a23 based tablet. The proccess that i am trying is reboot in recovery mode, upload busybox renamed as sh to /system/bin/ by using adb , installing bussybox to te recovery environment, the i mount /system , i copied de su binary of Superuser ap to /system/xbin/su i set correct owner 0:0 , and permisions 67455 then i unmount and reboot, but when i reboot and run su in terminal i get an error saying :"permision d
<abdel>
enial broadcast aks to run as user -1 but is calling from user 0 , this requiers android.permission.interact_accross_users", any idea?
<libv>
abdel: are you just here to root your tablet?
<libv>
abdel: or are you hoping to put a proper linux on it?
<libv>
abdel: in any case, have you see that automated notice that hit you when you joined the channel?
<abdel>
i need to root it to have a more confortable environment to work in
<libv>
so you are just here to root your android.
<libv>
abdel: use adb.
<abdel>
in fact i need to root it to be able to change the touch screen config
<abdel>
adb is not working in normal mode
<libv>
abdel: and don't try asking us again, we are not here to help people with their random android questions
<libv>
use slate-droid like normal android users
<abdel>
ok
<abdel>
sorry for asking in the wrong place,
<libv>
i think you have asked in many places before.
<abdel>
in fact is the first irc channel
<libv>
yes, android users tend not to do irc channels
<abdel>
i had been looking in lots of forums before and the web
<libv>
but that does comfirm that you have asked in many places before
woprr has quit [Ping timeout: 256 seconds]
<libv>
abdel: since you are too lazy to look at our wiki and identify your device, let me guess for you
<libv>
i haven't heard about shadow registers for ages, but you usually do not tend to use them from userspace even if they exist
<marcin_>
by shadow registers, I mean situation when you have pair of register, one for application and second for hardware. They can be flipped by e.g. vblank.
<marcin_>
OK
<ssvb>
marcin_: if I understand it correctly, the kernel driver postpones the layer changes until the vsync interrupt handler
<ssvb>
marcin_: so when you use the ioctls, the changes are not applied immediately
<ssvb>
marcin_: if you know that the framerate is low (25fps for video), then you don't need for an explicit FBIO_WAITFORVSYNC ioctl
<ssvb>
marcin_: if the framerate is higher than the monitor refresh rate and you need to throttle it, then FBIO_WAITFORVSYNC may be useful
<marcin_>
ssvb: Thank you for explanation.
abdel has left #linux-sunxi ["Saliendo"]
<libv>
marcin_: for me, shadow registers just are registers that re-appear at a different address
<libv>
but what you mean is somewhat true
<ssvb>
marcin_: are you developing new code? or just want to verify that the video playback is tear-free?
<marcin_>
Maybe I call the technique incorrectly, but CPU sees one register in one address.
ddc has joined #linux-sunxi
<ddc>
I would like to know if there is someone is working on mainline CSI driver
<libv>
the registers of the composer can be configured so they only apply when triggered manually
<libv>
no automatic vblank sync though
<libv>
which is what you usually also find
<libv>
but then, i haven't gone and tested that through completely
xinj has quit [Ping timeout: 264 seconds]
<ddc>
I've looked in the wiki and I don't see anyone working on it. I would like to start on it
ricardocrudo has quit [Ping timeout: 240 seconds]
<marcin_>
ssvb: I am trying to understand difference between mali and fbturbo code. You wrote post about reasons, why mali code is slow, but I want to go through the code.
<libv>
i think shadow registers might be the correct term for that, but then, little of what we set through registers are applied as directly as things used to be in the vga days
<libv>
debe is at 0x800 of its supposed range, and you have to manually null the range once you enable the engine
<libv>
so i think someone messed up one bit
<ssvb>
marcin_: are you primarily interested in 2D, 3D or maybe video playback?
<ssvb>
marcin_: because these are all substantially different use cases
ddc has quit [Ping timeout: 272 seconds]
<marcin_>
ssvb: I think that easiest part for me is video playback. I don't know X stack deeply - I have a problem with 2D part.
<libv>
what do you mean with a problem with 2d part?
<libv>
2d is 2d acceleration, which is what g2d is for
<ssvb>
marcin_: what exactly are you trying to achieve? maybe you are interested in supporting some other non-allwinner platform?
<libv>
drawing lines, boxes, moving bits around on the fb
<marcin_>
2D not from SOC perspective, but X itself. As I compare fbturbo with mali, I see that fbturbo hasn't exa part - fb fallback are used? Take for example window, when no a compositor is used.
<marcin_>
I assume that window has direct representation on framebuffer. But in mali in exa part I have only function like modifypixmapheader in which I have width and height but not x, y?
<marcin_>
etc.
<marcin_>
A lot of things to check.
rds has joined #linux-sunxi
<rds>
good to see that LIBV is happily coding now....
<libv>
grmbl.
<rds>
ddc: great!, you go it. Folks are waiting for patches!
<libv>
rds: you are so sickeningly self-serving
<libv>
yet you contribute nothing
<rds>
why? I am encouraging people with more skills to get the job done! what is wrong with that ?
woprr has joined #linux-sunxi
rds has quit [Quit: Page closed]
<marcin_>
ssvb: I would like to speed up "graphics" in one of custom board with A20. It uses Mali DDX and X with a compositor. I wonder in case of use COW only, if will be a difference between fbturbo and mali?
<marcin_>
BTW. Is it list of things/ideas to implement in e.g. fbturbo? I hope that maybe I can help in some areas?
kivutar has joined #linux-sunxi
physis has quit []
ricardocrudo has joined #linux-sunxi
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
tomboy64 has quit [Write error: Connection reset by peer]
Skaag has quit [Read error: Connection reset by peer]
<ssvb>
I initially thought that it had been caused by the UMP buffers allocation overhead, but later confirmed that EXA internal bookkeeping significantly contributes to the performance reduction
popolon_ has quit [Quit: Quitte]
<marcin_>
I understand. So fbturbo and mali should have similar performance if only dri2copyarea/dri2swapbuffer functions will be used (artifical case when every X client, will use GLES for example)?
<ssvb>
not sure if I understand the question
Skaag has joined #linux-sunxi
<ssvb>
do you mean GLES only performance (no X11 drawing operations of any kind), and xf86-video-mali patched to use sunxi disp layers?
<ssvb>
please note that the DRI2 implementation in xf86-video-mali relies on EXA, and removing the EXA dependency takes a bit of effort
<marcin_>
Assume that every X client application use GLES to "window drawing" and a compositor also uses GLES for composition, I think that no EXA/COMP/RENDER etc is used. To swapping data DRI2 will be used. In this case performance of fbturbo and mali should be equal due to use only DRI2?
<marcin_>
Ok.
<marcin_>
I have to check codes one more time, with your information in mind.
<ssvb>
if you have a custom setup where X11 drawing operations can be avoided, then probably using just the mali framebufer driver variant would be the best choice
<ssvb>
the Qt5 people prefer to use this approach
TheSeven has joined #linux-sunxi
ricardocrudo has quit [Ping timeout: 240 seconds]
<linkmauve1>
ssvb, related to that, what is missing to get a working Wayland compositor? For example unaccelerated and running on top of X11.
<marcin_>
Wayland compositors typically expect drm not fb... Am I right?
<ssvb>
linkmauve1: if you don't care about X11 compatibility, you can use libhybris to get wayland (in a roundabout way and with the use of android mali blobs)
<ssvb>
linkmauve1: if you are asking about a compatibility layer for running wayland applications in the X11 desktop (kind of backwards of what XWayland does), then I'm not sure if anybody has investigated this
<linkmauve1>
No, what I meant was like Weston does on x86, being an X11 window in which one can launch Wayland clients.
<linkmauve1>
marcin_, Weston doesn’t depend on DRM AFAIK, it can run plainly unaccelerated on top of fbdev.
<ssvb>
linkmauve1: yes, afaik you can run wayland compositor on top of X11 (not tested this on sunxi hardware), and wayland clients on top of this wayland compositor
<ssvb>
which already seems a bit too heavyweight to me, because this involves too many layers on top of each other
<linkmauve1>
Of course.
<linkmauve1>
Once I get my A10 working again I’ll try Weston’s fbdev backend.
<ssvb>
this should totally work
<ssvb>
but what kind of applications do you want to run in this setup?
<ssvb>
if these are just the legacy X11 applications, then Wayland/Weston/XWayland are not really any better than native X11
deasy has joined #linux-sunxi
<ssvb>
most likely (significantly?) slower than native X11, but I need to run benchmarks to see what kind of performance we can expect
<linkmauve1>
Most of the software I use are either console-based or already running natively on Wayland.
<linkmauve1>
So I guess it can be better than X11, due to the lower complexity, but I’ll have to test.
<ssvb>
oh, that's great, do you have a list of these applications?
<ssvb>
I also want to see what can be already done in wayland
<linkmauve1>
Not really, but anything gtk3-, qt5-, efl- or sdl2-based will run fine.
<ssvb>
can gtk3, qt5, sdl2 run without GLES in wayland?
<linkmauve1>
Lately I tried the gtk3 port of Firefox (on x86), it crashed after a while but was working mostly fine.
<linkmauve1>
gtk3 doesn’t use gles at all, qt5 only for some things (qtquick IIRC), and sdl2 has many backends including a pure-software one.
<ssvb>
well, regarding the gtk3 port of firefox, in practice everything can work, but has a lot of bugs to iron out
<linkmauve1>
Only plugins don’t work yet, but I already don’t use any on X11. ^^
<linkmauve1>
In four days it’ll be my sixth anniversary of the removal of flash in my browser. :D
<ssvb>
I'm still mostly using gtk2 and qt4 applications on my desktop myself, and this is not because my PC or the graphics card is too weak for the newer stuff
<ssvb>
just the old software works fine, and the new software is too buggy or introduces controversial features (I can't stand gnome3)
<linkmauve1>
Heh, I do almost everything in a terminal, except browsing, gaming, and a few other things.
<ssvb>
linkmauve1: looks like you are one of the persons who have fun with wayland and enjoying it, good for you
<linkmauve1>
Yep. ^^
<ssvb>
you can perhaps have a look at the Raspberry Pi compositor and port it from DispManX to sunxi disp layers
<linkmauve1>
I bought a few A10 computers from Olimex, in the hope to help with Lima, but I still haven’t started anything. :/
<linkmauve1>
Where can I find that sunxi disp documentation?
ricardocrudo has joined #linux-sunxi
<linkmauve1>
Ok, found it.
<linkmauve1>
“The allwinner provided sunxi disp driver has many severe issues and needs to be significantly redesigned.” :)
<linkmauve1>
It does require proprietary data though. :/
<linkmauve1>
But people can use it to make an entirely free game based on it.
Skaag has joined #linux-sunxi
<ssvb>
is there an example of at least one entirely free game?
<linkmauve1>
Not yet, currently some sample data I wrote with only one enemy.
<linkmauve1>
I am not a game designer, so even that was hard for me. ^^'
<linkmauve1>
And it still requires a few images from the proprietary game, which I am progressively replacing with my own.
<libv>
gles is painful that way
<libv>
no applications whatsoever
<ssvb>
linkmauve1: the biggest problem with GLES is not bad drivers, but almost total absence of any usable non-proprietary linux applications
<linkmauve1>
Really?
<libv>
glmark2-es is a horrible mess that is counterproductive for developing graphics drivers
<ssvb>
linkmauve1: I tried to find the applications, but did not succeed much
<libv>
compared to glmark2, ioquake3 is a bed of roses
<linkmauve1>
Well, I will finish the removal of those dependencies on proprietary data quickly then.
<ssvb>
linkmauve1: we can perhaps collect a list of usable GLES applications on some page in the linux-sunxi wiki
<linkmauve1>
I already got it running on top of libMali, so I know it works.
<libv>
half the planet seems to be using my lima tests these days it seems
<linkmauve1>
The performances were abysimal though, I haven’t searched why yet.
<linkmauve1>
ssvb, it’s too bad the Mali 400 can’t do GLES 3.0, there is Dolphin which can emulate tons of GameCube and Wii games which is written against GLES 3.0.
<ssvb>
linkmauve1: I think that the best contribution to the lima driver from your side would be to continue developing your game engine and ensure the availability of at least one non-proprietary game :)
<linkmauve1>
Great, I will!
<linkmauve1>
I’m going to Germany tomorrow (if I can find a car going there…) to visit a friend who happens to want to make a game.