Turl changed the topic of #linux-sunxi to: Allwinner/sunxi development discussion - Don't ask to ask. Just ask and wait! - See http://linux-sunxi.org | https://github.com/linux-sunxi/ | Logs at http://irclog.whitequark.org/linux-sunxi | #blameOliver
<ssvb> Turl: and I guess dma-buf did not exist at the time when ump had been introduced, so something had to fill the void
<Turl> I believe the other 3 graphic stacks use ION
<ssvb> Turl: you mentioned "some sunxi devices use UMP, others ION, others some samsung thing", does it mean that not every android device vendor is using allwinner sdk as-is?
<Turl> ssvb: I then proceeded to s/sunxi/mali/ :p
<ssvb> Turl: oops, sorry, got it
<Turl> ssvb: sunxi vendors use aw's sdk as is
<Turl> at times I wonder if they even use it, maybe they just buy preflashed nands from wits :p
<libv> ok, package update of just dri-dev also worked
<libv> ssvb: do you still want the patches for 9.0*?
<ssvb> libv: yes, even if there is still no lima stuff there
<libv> okies, creating the tarball and then sending it.
<ssvb> libv: where does "apt-get source" usually find the sources?
<libv> ssvb: first off, there are source directives in /etc/apt/sources.list
<Turl> ssvb: on the deb-src sources.list entry
<libv> ssvb: then apt-get update tries to collect a Sources file, just like it gathers a Packages file
<libv> ssvb: sent
<ssvb> I mean for the sunxi mesa package, does apt-get source work right now? and if yes, does it show the full url of the downloaded orig and diff tarballs?
<libv> i haven't tested that, but i am reasonably confident :)
<libv> let's see
<Turl> g'night guys :)
<libv> yup
<libv> Turl: sleep well
popolon has quit [Quit: Quitte]
eebrah|noPC has quit [Ping timeout: 248 seconds]
<libv> even apt-get source mesa does the right thing with the preferences set correctly
<libv> ssvb: let me see whether i can easily debianize sunxifb :)
<libv> fbturbo?
<ssvb> odroid, rockchip and rpi are not sunxi devices
<ssvb> this was making it a bit confusing
<libv> well, this mesa isn't sunxi specific either :)
<libv> the naming is a bit bold though
<ssvb> right, it has to provide something to be a really "better fbdev"
<libv> :)
* ssvb still needs to add a generic neon xv implementation to rightfully claim this name
<naobsd> preparing fbturbo dpkg?
<libv> naobsd: yes
<naobsd> very nice :) I want to use it on my RK environment too :)
<naobsd> (NEON part for now)
<libv> ssvb: i will send you a patch for the dri driver naming
<libv> naobsd: i also am in the process of creating hacked mesa packages which are a precursor for the lima driver
<libv> naobsd: so you might want to use those too, as soon as things start trickling out
<naobsd> oh, lima, I want to try it :)
<libv> hold your horses.
<ssvb> naobsd: which mali version is used for RK?
<naobsd> unruly horse, it seems I need to do more work to stable it ;)
<ssvb> libv: ok
<naobsd> Mali400-MP4
<naobsd> on some RK30/31. some(RK3168) has SGX
<libv> ssvb: this patch allows mesa-dri to select the correct driver
<libv> the binary driver is unaffected by it, only when mesa-dri is fully installed does it do something
<libv> what other fixes i will need, i will find out in the next few
<libv> but this trivial one definitely is needed
<libv> and it is the only change a much earlier version of sunxifb needed
<ssvb> libv: ok, I can apply it now, the binary driver indeed ignores the name and should be fine (this could be actually added to the commit message)
<libv> feel free to amend the commit msg :)
<libv> that was a quick build
<libv> and now for fbturbo... *sigh*
soul is now known as Guest97485
<libv> ssvb: an extra check for /usr/include/ump/ump.h is needed
<libv> ssvb: it bombed on the odroid, as i do not have a proper sunxi-mali installation there, but the hardkernel provided mess
<ssvb> libv: this looks somewhat familiar - https://github.com/ssvb/xf86-video-fbturbo/issues/19
<libv> yup
<libv> AC_CHECK_HEADER([ump/ump.h],[],[AC_ERROR([...])])
<libv> or add an AC_DEFINE for that
<ssvb> thanks, that's a good point
<libv> ssvb: for some reason, turbofb is what my fingers like to type :)
<libv> great. black screen
<libv> and thanks to the big idiot at redhat, this is not indicative anymore.
<libv> we used to get the stipple at least, when nothing else was drawn to the fb
<ssvb> the /var/log/Xorg.0.log file may have something
<libv> ssvb: having been developing for X for the last decade, i would never have looked there :p
<libv> but there wasn't anything obvious there
<ssvb> modesetting might be causing troubles
<libv> heh, after a reboot, bringing up X crashes the device...
* libv is going to get serial out on this one
<ssvb> there is a somewhat known problem with xf86-video-modesetting, which can see the mali drm and try to use it
<ssvb> unsuccessfully, but still messing up things
<libv> this would not be a problem, as it was running sunxifb just fine before
<libv> ah, kernel oopsed
<libv> hrm.
<libv> disp_ump: disagrees about version of symbol disp_get_ump_secure_id
<libv> mymy
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
<libv> nah, that's not the issue
<ssvb> how old is your kernel?
<libv> quite old, but this exactly the sort of thing i want to avoid with what i do on mali.
<libv> are there known crashers there?
<libv> [ 386.070000] PC is at _ump_ukk_size_get+0x4c/0x124 [ump]
<ssvb> there were some bugs in ump buffers wrapping a very long time ago
<libv> right, frame #1 is ump_size_get_wrapper
<ssvb> like not allocating ump buffers at 4K boundary
<libv> this is a null dereference, so that's not it
<libv> anyway, checking changes to our kernel ump
<libv> there seem to have been no changes to sunxi-3.0 ump
<ssvb> is the kernel built with CONFIG_FB_SUNXI_UMP=y?
<libv> yes
rz2k has joined #linux-sunxi
<libv> i think this is about requesting ump size on an non-existent buffer
<ssvb> yes
<ssvb> libv: and I see that if GET_UMP_SECURE_ID_BUF is not implemented, the ioctl returns 0 (success) - https://github.com/linux-sunxi/linux-sunxi/blob/sunxi-3.0/drivers/video/sunxi/disp/dev_fb.c#L1410
<ssvb> libv: looks like I have a regression in the sanity checks here - https://github.com/ssvb/xf86-video-fbturbo/blob/master/src/sunxi_mali_ump_dri2.c#L980
<libv> this was not a problem before apparently
<libv> but then, you never requested the size before
<ssvb> ioctl never fails, and the mali->ump_fb_secure_id is not initialized to UMP_INVALID_SECURE_ID before ioctl
<libv> perhaps this is the reason: ump_id1 = ump_id2 = ump_id_fb = UMP_INVALID_SECURE_ID;
<libv> before the ioctl call
<libv> that was there before, and it could be that this does not happen in libump
<ssvb> yes, that's why it looks like a regression
<libv> ah, right, it is not set to that beforehand
<libv> not libump, but this code
<libv> kernel side is horrible, but yes, regression indeed
<ssvb> it's strange that the kernel does not provide valid ump buffers for the framebuffer, but I have not looked at the 3.0 kernel since a long time ago
<ssvb> nowadays I think everyone has moved to sunxi-3.4
<libv> sure, but the fix is now obvious
<libv> let me reboot and find out
<libv> yup, setting mali->ump_alternative_fb_secure_id to UMP_INVALID_SECURE_ID works
<libv> do you have this covered already, or do you want a patch for this?
<ssvb> it's fine either way
<libv> lima is not happy though
<libv> this is why i do not change a working setup in the middle of development ;)
<ssvb> I kind of warned about the major changes in DRI2 handling to make it behave more like the mali blob expects :) https://github.com/ssvb/xf86-video-fbturbo/wiki/Mali400-DRI2
<libv> right
<ssvb> are you requesting a single DRI2BufferBackLeft via DRI2GetBuffers or two buffers at once?
<libv> sunxifb is just as happy as before
<libv> i renender to just gl_fb->Attachment[BUFFER_BACK_LEFT].Renderbuffe
<ssvb> it may make sense to look at the x11trace log to see what mesa is trying to do
<ssvb> libv: also perhaps try to set the 'need_window_resize_bug_workaround' variable to 0 here - https://github.com/ssvb/xf86-video-fbturbo/blob/master/src/sunxi_mali_ump_dri2.c#L379
<libv> i seem to get a perfectly valid physical address for both buffers
<libv> and both buffers seem to be enabled...
<ssvb> and it fails how?
<libv> the messaging is not too clear on that, but iirc, this is a reported segfault
<ssvb> ok, I'll go to sleep now
<libv> different ump ids than on the old driver, but the same address...
<libv> ok, good night, and thanks
<ssvb> just looks like trying to be more compatible with mali r3p0 blobs bugs had its price :-(
<ssvb> but hopefully this is fixable
<libv> if i switch to front_left, things are fine
<libv> es2gears just works
<ssvb> just to see that no frames are dropped and they are presented in the right order
<ssvb> it's quite likely that something is messed up, but can't be easily observed with es2gears :)
<ssvb> the window title should match the actual background color
<ssvb> ok, now I'm really going to sleep, see you later
<libv> sigsegv ;)
<libv> interesting
<libv> but that seems to be in the lima driver...
aep has quit [Ping timeout: 248 seconds]
aep has joined #linux-sunxi
ykchavan has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
Gerwin_J has quit [Quit: Gerwin_J]
TheSeven has quit [Disconnected by services]
[7] has joined #linux-sunxi
eebrah has joined #linux-sunxi
eebrah is now known as Guest14740
Guest14740 is now known as eebrah_
ykchavan has quit [Ping timeout: 240 seconds]
eebrah_ has quit [Read error: Connection reset by peer]
acr has joined #linux-sunxi
<acr> hi
<acr> libavcodec on cedarx howto?
<rz2k> acr: we now have only proof of concept version of vdpau https://github.com/linux-sunxi/libvdpau-sunxi
<rz2k> and XBMC and VLC ports
<acr> agrrr
<acr> there no correlation VDPAU and libavcodec
hipboi_ has quit [Ping timeout: 265 seconds]
hipboi_ has joined #linux-sunxi
deasy has joined #linux-sunxi
fredy has quit [Excess Flood]
benn````` has joined #linux-sunxi
fredy has joined #linux-sunxi
benn```` has quit [Ping timeout: 245 seconds]
<rz2k> acr: if you feel like you want to have libavcodec - go write your own implementation
benn````` has quit [Ping timeout: 248 seconds]
Black_Horseman has quit [Remote host closed the connection]
hipboi_ has quit [Ping timeout: 240 seconds]
fredy has quit [Excess Flood]
fredy has joined #linux-sunxi
allen has joined #linux-sunxi
hipboi_ has joined #linux-sunxi
rellla has joined #linux-sunxi
shineworld has joined #linux-sunxi
shineworld has quit [Client Quit]
shineworld has joined #linux-sunxi
shineworld has left #linux-sunxi [#linux-sunxi]
<acr> i try to compile vlc
<acr> configure: error: "Could not find Cedar development headers"
<rz2k> last time I checked http://linux-sunxi.org/VLC it did work
allen has quit [Ping timeout: 245 seconds]
<acr> u croscompile vlc?
eebrah has joined #linux-sunxi
eebrah is now known as Guest19122
allen has joined #linux-sunxi
Guest97485 has quit [Remote host closed the connection]
<rz2k> then edit it if it is wrong
Soru has joined #linux-sunxi
<rz2k> also half of the answers in that topic are from me.
<mnemoc> moin
<acr> this build instruction only for native build not for crosscompilation
<mnemoc> hopefully we can get pre-compiled vlc packages in the apt and yum and... repos too
<rz2k> acr: you can setup crosscompile for vlc, but if I remember correctly, it had problems with autoconf, so I ended up building it natively, takes about 5 hours
Guest19122 has quit [Ping timeout: 265 seconds]
Soru has quit [Read error: No route to host]
Soru_ has joined #linux-sunxi
ykchavan has joined #linux-sunxi
\\Mr_C\\ has joined #linux-sunxi
rellla has quit [Quit: Nettalk6 - www.ntalk.de]
\\Mr_C\\ has quit []
allen has quit [Remote host closed the connection]
Sonicadvance1_ has joined #linux-sunxi
Sonicadvance1 has quit [Ping timeout: 240 seconds]
<gzamboni> moin, iupi, my C book just arrived :)
hramrach has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
notmart has joined #linux-sunxi
Sonicadvance1_ is now known as Sonicadvance1
Black_Horseman has quit [Read error: Operation timed out]
Black_Horseman has joined #linux-sunxi
rz2k has quit []
\\Mr_C\\ has joined #linux-sunxi
Gerwin_J has joined #linux-sunxi
Black_Horseman has quit [Quit: Αποχώρησε]
ssvb has quit [Read error: Connection timed out]
ssvb has joined #linux-sunxi
ykchavan has quit [Ping timeout: 240 seconds]
sun4ihdmi has joined #linux-sunxi
sun4ihdmi has quit [Quit: Page closed]
Gerwin_J has quit [Quit: Gerwin_J]
FDCX has joined #linux-sunxi
ykchavan has joined #linux-sunxi
paulk-collins has joined #linux-sunxi
Gerwin_J has joined #linux-sunxi
<libv> heh, the crasher is due to a bad egl config not being handled correctly (because of "old" mesa cruft)
<libv> ah, EGL_SAMPLES, 4
popolon has joined #linux-sunxi
Soru_ has quit [Read error: Connection reset by peer]
Soru has joined #linux-sunxi
popolon has quit [Quit: Quitte]
paulk-collins has quit [Quit: Ex-Chat]
paulk-collins has joined #linux-sunxi
rellla has joined #linux-sunxi
rellla has quit [Remote host closed the connection]
ykchavan has quit [Read error: Operation timed out]
Gerwin_J has quit [Quit: Gerwin_J]
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
notmart has quit [Quit: notmart terminated!]
fredy has quit [Excess Flood]
fredy has joined #linux-sunxi
iamfrankenstein has quit [Ping timeout: 240 seconds]
drachenphone has joined #linux-sunxi
focus_it has joined #linux-sunxi
<focus_it> hi - spend all day upgrade my cubieboard2 distro linaro ubuntu desktop 13.04 to 13.10 - it kinda worked - the xubuntu, lxde works, and it seems quicker, but gnome fallback has been taken off and instead you gnome flashback - that and gnome flashback (no effects) and ubuntu sessions all give blank screens. Anyone any ideas how to get gnome flashback working or gnome fallback put back in?
eebrah has joined #linux-sunxi
eebrah is now known as Guest51518
Guest51518 is now known as eebrah_
andreaswatch has joined #linux-sunxi
andreaswatch has quit [Remote host closed the connection]
popolon has joined #linux-sunxi
eebrah_ has quit [Read error: Connection reset by peer]
deasy has joined #linux-sunxi
drachenphone has quit [Remote host closed the connection]
raiden334341 has joined #linux-sunxi
<raiden334341> hi friends
<raiden334341> has anybody been able to boot the Q88 tablet with linux yet?
<libv> i need to go create a libump package now, as without that, this fbturbo package is actually useless.
<libv> so -sunxi1 should happen soon
<WarheadsSE> libv: I am looking forward to packaging that for arch..
<ssvb> libv: do you mean useless for mali support and gles acceleration?
<libv> ssvb: at this point, that package is useless, unless one runs our installation at sunxi-mali manually
<libv> i also do not want to go and package the mali libs
<libv> but we do need libump packaged
<libv> i will make limare dlopen libmali instead of linking directly to it. that way there could be a helpful warning
<raiden334341> do you guys know about a13 tablets?
<libv> raiden334341: have you tried looking at linux-sunxi.org ?
<raiden334341> yes, i couldnt find a way to boot this tablet
<raiden334341> i already tried many images
<libv> raiden334341: have you tried reading up on things at linux-sunxi.org?
<raiden334341> of course
<libv> raiden334341: then why are your questions that hazy?
<raiden334341> its not hazy. its if anyone could boot this q88 a13 tablet yet
<libv> i am pretty certain that someone has succeeded in the past
<libv> raiden334341: how many sunxi devices do you think there are?
<libv> raiden334341: what are the chances of the owner of such a device being awake at this very moment?
<raiden334341> look, if you dont know how to help , or dont want to, just say it, ok
<libv> raiden334341: try creating an sd card from scratch.
<libv> instead of randomly throwing "images" at it
<libv> if you had read through linux-sunxi.org, you would've tried this
<raiden334341> there are ready to use images for many devices
<libv> raiden334341: i know how to help, i just also can spot it when people ask useless questions
<libv> and i know that attempting to answer those usually are a waste of perfectly good time
<raiden334341> useless for you maybe
<raiden334341> anyway, i hope u get soon a girlfriend, that way u could eventually feel calmer
<libv> raiden334341: did you find an image that was specifically meant for your q88?
<libv> raiden334341: yeah, i really should
<libv> raiden334341: i tell my other half that all the time
<libv> raiden334341: so, did you find an image that was specifically meant for your device?
<libv> raiden334341: given that you asked "has anyone ever ran a linux on the q88", i am pretty certain that the answer to that question is a big fat "No."
<raiden334341> libv, i only found a13 images, but nobody specified q88 in them
<raiden334341> since its an a13 tablet, i imagine its possible
<libv> 21:13 < libv> raiden334341: try creating an sd card from scratch.
<libv> 21:14 < libv> instead of randomly throwing "images" at it
stekern has quit [Ping timeout: 245 seconds]
<libv> ssvb: i am sending you the xorg.conf.d patch as well, that should bring fbturbo in line with the orig.tar.gz on packages
fredy has quit [Excess Flood]
n01 has joined #linux-sunxi
vinifr has joined #linux-sunxi
<steev> wtf
<steev> raiden334341: you really should not insult people who are trying to help you
fredy has joined #linux-sunxi
<steev> libv: fbturbo in line with what orig.tar.gz?
<libv> to be absolutely honest, i wasn't really trying to help raiden334341, as his questions were too hazy for even trying to help him
<libv> waste of time imho
<libv> steev: it's code
<steev> i realize that
<libv> steev: and you haven't done the secret handshake yet :p
<steev> oh, right
<steev> if the secret handshake involves ubuntu, i will never do it
<libv> *sigh*
<steev> but that makes sense
<steev> libv: you're right, i shouldn't hate ubuntu, it's mark and canonical that i hate
<steev> ogra and all the other fine devs in #ubuntu-arm are awesome
<libv> i hate the blind canonical bashers more
<steev> oem services though can go suck it
<libv> as i once was on the receiving side of their idiocy too
<steev> libv: oh i have very good reasons to hate canonical and mark
<steev> including but not limited to, going to companies that we were attempting to work with and telling them that since ubuntu ran on our hardware, they needed to pay canonical for support
<libv> well, i am with mark on the mir thing.
<libv> or better said, mark is with me.
<steev> idgaf about mir
<steev> software side is meh, i don't use it, it doesn't affect me
<libv> i had the same wankers throw endless mud at radeonhd
<libv> and they then reinvented it, and did a much less free or advanced version of it
<libv> and that was then claimed to be the non-evil solution
<libv> this mud throwing usually boils down to people from a given corner of the linux world
<libv> and their target shifts depending on marketshare of the targets
<libv> anyway, i have ubuntu on my sunxi systems atm, so i package stuff for it now
<libv> since this part of packages.linux-sunxi.org was up for grabs anyway, this allowed me to go and experiment at will
raiden334341 has left #linux-sunxi [#linux-sunxi]
pirea has joined #linux-sunxi
pirea has quit [Client Quit]
pirea has joined #linux-sunxi
<oliv3r> i find ubuntu to be a perfectly acceptable target for now, it's 'popular' so it's good
<oliv3r> bt, back to work ;)
<oliv3r> but
cesman has quit [Quit: Konversation terminated!]
rellla has joined #linux-sunxi
<oliv3r> rellla: heya
stekern has joined #linux-sunxi
pirea has quit [Quit: Leaving]
raiden334341 has joined #linux-sunxi
<steev> libv: i tend to not throw mud, i just don't use it. i prefer to build myself, but i know that isn't for everyone
Black_Horseman has joined #linux-sunxi
<libv> steev: i did not state that you were throwing mud at canonical over this mir idiocy...
<libv> all this negativity currently being thrown at ubuntu does have a source though
<steev> libv: i'm just being captain obvious
<steev> it has its uses, just none for me, and any of my own personal needs
<libv> and it is the same source that threw endless crap at SuSE linux because novell had a bigger patent portfolio than microsoft and microsoft officially stated that it would not sue novell (and in the second order, would make it impossible to sue other linux players)...
<libv> i still fail to see what was so wrong about that, ms actually handed x0 million usd to SuSE for the next 5 or 6 years due to this industry standard patent licensing deal
<libv> but the amount of shit was endless.
<libv> and the exact same game is happening now
rellla has quit [Quit: Nettalk6 - www.ntalk.de]
<libv> canonical is perceived as too big a threat, and so every little thing is blown out of proportion, and everyone mindless joins the shitthrowing contest
<libv> and because of that, i now get to hear "urgh, ubuntu" when i show some packages that i made, which seem to be the first packages ever created for anything sunxi
<libv> i would not have heard such negativity a year ago
<steev> i will always say urgh, ubuntu :P
<steev> and i would have said it a year ago
<libv> sure, but i heard it a few times before now
<steev> *shrug*
<steev> every time i mention gentoo i get to hear lolricer, funrollloops or whatever
<steev> for me, it's about not installing things that i'll never use
<libv> i stopped caring about distribution X or Y, i am happy that i do not have to use android for my lima stuff
n01 has quit [Read error: Connection reset by peer]
<steev> why you no android?
<steev> i WOULD like to play with ubuntu touch
n01 has joined #linux-sunxi
<steev> that stuff looks interesting
<libv> linaro was easily available for sunxi type hw, so it got installed, and i ended up using that on all my lima development devices, so as to not increase the nightmare even further
<libv> libhybris :(
<steev> oh right
<steev> linaro... heh
<steev> i've no idea about libhybris - and i can't complain too much about the binary drivers as i don't have the technical know-how or ability to write my own graphics drivers, and the one that i would like to see advanced, i can't assist on at all
<steev> would like to see freedreno extended to work on imx5
<steev> but i've seen the gpu driver sources so no assistance from me :(
diego71 has quit [Ping timeout: 252 seconds]
diego71 has joined #linux-sunxi
FDCX_ has joined #linux-sunxi
FDCX has quit [Ping timeout: 272 seconds]
raiden334341 has left #linux-sunxi [#linux-sunxi]
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
ZetaNeta has quit [Ping timeout: 248 seconds]
n01 has quit [Ping timeout: 248 seconds]
paulk-collins has quit [Quit: Ex-Chat]
ZetaNeta has joined #linux-sunxi
ZetaNeta has joined #linux-sunxi
ZetaNeta has quit [Changing host]
Black_Horseman has quit [Quit: Αποχώρησε]
vinifr has quit [Remote host closed the connection]
Black_Horseman has joined #linux-sunxi
rz2k has joined #linux-sunxi
<Turl> libv: what DE do you use on your ubuntu on sunxi?
<libv> lxde apparently
<libv> not that i care much, anything that doesn't get in the way of testing things :)
<Turl> :)
<Turl> hm, torvalds pushed rc6 and mturquette is still awol :(
popolon has quit [Quit: Quitte]