<theOzzieRat>
can someone tell me where changes to sunxi-bsp get submitted?
egbert has quit [Ping timeout: 260 seconds]
egbert has joined #linux-sunxi
rellla has joined #linux-sunxi
jpsminix has joined #linux-sunxi
n01 has quit [Disconnected by services]
n01|afk is now known as n01
botton has joined #linux-sunxi
botton has left #linux-sunxi [#linux-sunxi]
shineworld has joined #linux-sunxi
fredy has quit [Ping timeout: 256 seconds]
fredy has joined #linux-sunxi
<oliv3r>
lkcl: why is there going to be an A31 based card? I understand you want the quad for its awesomeness, but the powerVR will put it into an iffy sittuation
torqu3e has quit [Read error: Connection reset by peer]
torqu3e has joined #linux-sunxi
<ssvb>
libv: sunxi disp properly supports rgb565 since ages
<ssvb>
do you have any problem with it?
<libv>
ssvb: i haven't tried yet :)
<libv>
ssvb: but it's ok, apart from X11 buggering up, i now have rgba8888 working on the odroid
<ssvb>
though there is 'scaler_mode', which is 32bpp only
<libv>
poking at the x driver now to see what is causing that
<libv>
sunxi is really already lightyears ahead compared to exynos
<ssvb>
and also xf86-video-mali seems to have problems with 16bpp
<libv>
no matter how insane disp is, it is an actual fbdev driver
<libv>
xf86-video-mali on exynos has problems with 32bpp :)
<libv>
currently collecting all releases from ARM
<ssvb>
well, I guess something might be hardcoded for 32bpp in r3p0 and for 16bpp in r3p2 version of xf86-video-mali :)
<ssvb>
xf86-video-mali also seems to have troubles with dual monitor setup
<libv>
wait and see what i find, i hope mdrjr does get the hardkernel code posted, i bet there is a difference between that code and the arm code, even if it is only small
<ssvb>
yeah, good luck
jpsminix has quit [Remote host closed the connection]
<ssvb>
exynos display driver only needs some support for hardware cursor and overlays (or at least fbiopan for page flipping), then it's basically happy sailing
<libv>
well...
<libv>
you've read the crap i had to go through yesterday to be able to map FB to the mali?
<libv>
with the whole v4l2 based mess?
<ssvb>
yes, of course
<ssvb>
I had a look at this crap myself a few months ago
<libv>
maybe the main display driver is better
<libv>
but for that you need an LCD
<ssvb>
yes, and I don't have LCD for exynos
<ssvb>
though I also have arm chromebook with exynos5 and it should be more or less the same
<libv>
hehe, the arm xf86-video-mali driver cannot decide whether it is MIT licensed or all rights reserved
<libv>
one release it's this, next release it's that
<ssvb>
btw, you did a good job getting 32bpp and the physical framebuffer address for exynos4 hdmi
<ssvb>
thanks
<libv>
both are ugly hacks
<libv>
but vb2-fb deserves nothing less
<jelly>
libv: all rights reversed... by MIT.
<jelly>
unintentional typo makes things even better
<libv>
jelly: no, MIT replaced with ARM confidential, all rights reserved, blahblah
<libv>
it's quite amazing given the tiny amount of churn this xf86 driver gets
<libv>
scons? for an X driver? really?
<libv>
*sigh*
<libv>
ARM has issues
<libv>
news at 12 :)
rz2k has joined #linux-sunxi
Kadosch has joined #linux-sunxi
JCQ has quit [Ping timeout: 260 seconds]
JCQ has joined #linux-sunxi
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
BJfreeman has quit [Ping timeout: 246 seconds]
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
BJfreeman has quit [Quit: had a good time]
ganbold_ has joined #linux-sunxi
xlab_ has quit [Remote host closed the connection]
Xlab has joined #linux-sunxi
<rm>
so does CedarX currently work?
<hramrach>
kind of
<hramrach>
reportedly some media files play back incorrectly
<hramrach>
and the hacked vlc crashes a lot
<rm>
so perhaps I should enable it in my .config even at the cost of some MB of static RAM reserve
<hramrach>
you could use it for media server, yes
<hramrach>
if you had a software that can make use it and not crash
<oliv3r>
any word on NEW cedar libraries?
<oliv3r>
I may send an email to eva, asking her what the status is, as in some video she claimed that 'all sources are with RH'
<oliv3r>
(except the GPU of course)
<rm>
why is it still CONFIG_VIDEO_SUN4I_CEDAR in 3.4, is there no unification possible to SUNXI?
rs_ has joined #linux-sunxi
tinti has joined #linux-sunxi
rs_ has left #linux-sunxi [#linux-sunxi]
<oliv3r>
rm: cedar has been unified
<oliv3r>
rm: wingrime posted patches on the ML
Kadosch has quit [Quit: Page closed]
rz2k has quit [Ping timeout: 264 seconds]
<ssvb>
oliv3r: have you or wingrime verified that cedar still works after these patches?
<ssvb>
hramrach: are you using CedarX yourself?
<hramrach>
ssvb: I tried playuing some video with vlc
<hramrach>
worked ok for very small test video sometimes but always crashes on 720p videos
<hramrach>
not sure if it's teh higher res or different file format
<hramrach>
did not do tests with many videos
<hramrach>
after it crashes the video layer stays and vlc no longer works at all
<hramrach>
until reboot
<ssvb>
yeah, video layers can be leaked on process termination, that's a bug in disp
<hramrach>
it's possibly not CedarX fault at all except the wrapper does not tear down the layer and fails to re-initialize it when already there
<ssvb>
it's possible to forcefully disable the layers without rebooting
<hramrach>
the video output driver should do just that but it fails instead
<ssvb>
well, the video output driver can request a layer, but this request will fail if there are no more unallocated layers available
<hramrach>
well, it can forcefully disable layers when there are none
<hramrach>
except if there was some tracking whic layers are in use they could be disabled automagically and not leaked
<ssvb>
yes, that's why it is a bug in sunxi disp :)
<ssvb>
hramrach: about cedarx, was your test run with the latest sunxi kernel?
<ssvb>
I mean, there don't seem to be many cedarx users, the changes in the kernel can very easily introduce regressions
<hramrach>
semi-recent
<hramrach>
it was vlc that crashed, though
<hramrach>
so it's obviously bugged regardless of what state is otehr stuff
<hramrach>
others complain that vlc crashes, too
<ssvb>
if it ever worked better than this, then somebody might want to run a bisect investigation
<hramrach>
I just picked random file with actual movie and it crashed. not sure if it ever did any better
<hramrach>
and both vlc and the kernel evolve so you whould need a working kernel *and* vlc hash to bisect
<ssvb>
hramrach: btw, https://github.com/willswang does not seem to have any open issues in vlc and libcedarx repositories (some old issues are closed)
<ssvb>
maybe some feedback from the users would be welcome there? :)
<oliv3r>
ssvb: never used cedarX :p only saw the patches; only got a cubieboard yesterday! :D
<oliv3r>
ssvb: anybody doing any disp work that you know of? libv is really working on mali; anybody besides you?
<ssvb>
oliv3r: I believe that would be techn_ and hansg
* ssvb
has not been doing anything sunxi related for weeks
<hramrach>
ssvb: it's kinda pointless. it might be upstream vlc bug. and to do any decent debugging you would need the layer not leaking so that you don't have to reboot every time after playing a video
<hramrach>
how can it leak, anyway?
<hramrach>
it's configured in ioctl, right?
<ssvb>
hramrach: yes, you get a magic cookie via DISP_CMD_LAYER_REQUEST ioctl and need to release it via DISP_CMD_LAYER_RELEASE ioctl after you are done
<ssvb>
if some process does not release a layer, then it is leaked
<hramrach>
can't it tear down the layer on fd close?
<ssvb>
it should, but it doesn't
<ssvb>
as a workaround it is possible to just do DISP_CMD_LAYER_RELEASE for a range of possible layer id values, and they are always the same
<rellla>
ssvb: which patches do you mean? maybe i can do some test with the xbmc-port to look, if there are any changes in cedarx behaviour.
<rellla>
ssvb: if you are weak enough, try stage/Frodo or stage/master! i didn't it yet
<ssvb>
rellla: I can use any rootfs, it's not a problem
<rellla>
ssvb: i heard of problems with linaro 12.04. due to some python-version conflicts. maybe there have to be made some changes in the makefile first. i use debian/sid without problems.
<hramrach>
I don't know if it is possible to identify uniquely a fd and craft a dummy fd that can never be allocated for internal layer allocation so I store the allocated layers in fd data
<ssvb>
hramrach: the patch does not look too ugly to me
<hramrach>
not among the disp code I guess :s
<hramrach>
any set functions in the kernel?
<hramrach>
anyway, I try if the thing works later tonight hopefully
<ssvb>
the use of a generic set imho would be an overkill, we have very few layers and just a linear search works best in this case
<hramrach>
but by using a library function you get to blame somebody else when it explodes ;-)
ganbold_ has quit [Remote host closed the connection]
shineworld has left #linux-sunxi ["Leaving"]
focus has quit [Remote host closed the connection]
focus has joined #linux-sunxi
Kadosch has joined #linux-sunxi
torqu3e has quit [Quit: torqu3e]
<techn_>
hramrach: ssvb: I think we need to add disp session support
<techn_>
hramrach: your patch removes all layers when any disp handle closes. It should remove all layers for that handle
<techn_>
nevermind.. looks good :)
n01 has quit [Ping timeout: 245 seconds]
Xlab has quit [Remote host closed the connection]
hipboi has quit [Quit: Leaving]
paulk-desktop has joined #linux-sunxi
Kadosch has quit [Quit: Page closed]
torqu3e has joined #linux-sunxi
rellla has quit [Ping timeout: 264 seconds]
<paulk-desktop>
hi
<hramrach>
ssvb: hmm, the kernel hack does not work
<paulk-desktop>
I've just sent my EKTF2K touschsreen driver to the mailing list, could it please be reviewed and included quickly, I'm sure it would be very useful
eebrah has joined #linux-sunxi
<ssvb>
hramrach: yeah, looks like you have a bug there
<ssvb>
hramrach: BSP_disp_layer_release(j,i)
sankasan has joined #linux-sunxi
Wetomelo has quit [Ping timeout: 256 seconds]
<ssvb>
hramrach: just replace it with "BSP_disp_layer_release(j,data->layers[j].layer[i])" and it starts working fine
test3 has joined #linux-sunxi
test3 has left #linux-sunxi [#linux-sunxi]
<hramrach>
ssvb: fixed that one when copying the patch to the board
<hramrach>
but does not work anyway
<ssvb>
hmm, but it seems to work for me
<ssvb>
how do you test it?
<hramrach>
the userspace hack works
<hramrach>
I run vlc till it crashes nad the layer stays
<ssvb>
add some printk to trace what is happening with the layers?
<hramrach>
yes, would be good idea
<mnemoc>
re...
<mnemoc>
anything urgent in the ML I should take care of right now?
rellla has joined #linux-sunxi
<hramrach>
hello
<hramrach>
don't think there is anything that cannot wait a few more days ;-)
<mnemoc>
neat :)
RoadriverRail has quit [Ping timeout: 245 seconds]
<hramrach>
what you can do today postpone to the day after tomorrow and get a day free ;-)
Dave77 has joined #linux-sunxi
rz2k has joined #linux-sunxi
<mnemoc>
I just arrived from vacations today. after solving the urgents I'll finally be able to take better care of the postponed stuff on our ML
<hramrach>
ssvb: ugh, when you do make the kernel is not built so maybe I don't have the new one :s
<Dave77>
is there somewhere I can download the u-boot for NAND? (re: I don't know how 2 build)
<hramrach>
yeah, works now
_BJFreeman has joined #linux-sunxi
tinti has quit [Ping timeout: 272 seconds]
<hramrach>
posted to ml
<hramrach>
thanks for testing
Yaku321 has quit []
BJfreeman has quit [Ping timeout: 246 seconds]
rellla has quit [Remote host closed the connection]
tinti has joined #linux-sunxi
shineworld has joined #linux-sunxi
Dave77 has quit []
<techn_>
mnemoc: halt on those usb related stuff.. or inform what are going in
<techn_>
I'm doing unification / import from allwinner sdk
<techn_>
couple things are going to be fixed at same
<mnemoc>
techn_: sdk stuff not in the -dev ?
<techn_>
from a10 and a1x sdk
<mnemoc>
ok
paulk-desktop has quit [Quit: Ex-Chat]
n01 has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
shineworld has quit [Quit: Leaving]
mdfe has joined #linux-sunxi
n01 has quit [Ping timeout: 256 seconds]
gzamboni has quit [Remote host closed the connection]
mdfe has quit [Remote host closed the connection]
<oliv3r>
anybody know anything about ext4 on disk versioning?
tinti_ has joined #linux-sunxi
sankasan has quit [Remote host closed the connection]