<libv>
tonikasch: read up on our wiki on what the cedrus guys are doing, and what media they are testing with
<Turl>
what you need first of all is a simpleish, working program to RE, so you may start from there
<libv>
right
<libv>
first goal is to capture and replay a simple test case
<libv>
and then you can gradually start prying the captured data apart
<tonikasch>
sorry, keyboard gone, brb
tonikasch has quit [Remote host closed the connection]
* Turl
wonders how did he manage to write that
tonikasch has joined #linux-sunxi
<tonikasch>
ok
<tonikasch>
a newbie question, i assume that has to be done in android
<libv>
tonikasch: anywhere where you have working binaries
<Turl>
tonikasch: preferably would want to do it outside of android
<Turl>
because it's easier to work on there
<libv>
tonikasch: android is a pain, but it's not impossible
<libv>
i spent half my lima time under android
<libv>
and the other half on sunxi :p
<Turl>
libv: have you had to suffer the pain of gdbservering over adb? :)
<libv>
yup
<tonikasch>
haha, well, we have no binaries in linux... well, we have some code of libvpu.so that performs some opening and closing of the vpu and perhaps some other thing
<libv>
Turl: it actually wasn't too bad
<libv>
Turl: the worst time was actually banging my head in the lack of LD_PRELOAD and trying to hack elf headers
<Turl>
libv: I suffered the most because I was trying to debug an android daemon that was autorun from init
<libv>
ouch
<Turl>
so each time it died I had to reattach the gdbserver, reconnect my target, ..
<libv>
i had to deal with init things when working for intel. really insane shit that you had to rebuild the iso image (android-x86) to be able to change the scripts
<libv>
my second mali target (mali-400) had very flaky usb
<libv>
transfer a big file (like an image from a test) and adb would die
<Turl>
libv: you worked on android x86? you've been everywhere :P
<libv>
for intel, for a bit yeah, implementing hw composer, for no reason whatsoever
<libv>
noticing, once again, just how shortsighted the ruling graphics driver mob is.
<tonikasch>
anyway, do you think libhybris is worth a try?
<libv>
tonikasch: you would have to ask the cedrus guys
<libv>
it might be more trouble than it is worth
<tonikasch>
ok, i'll ask them
<libv>
once you have your workflow going, it is not that impossible to work with android
<libv>
i had my Makefiles deal with most androidy things for me
<tonikasch>
:)
tonikasch has quit [Ping timeout: 240 seconds]
tonikasch has joined #linux-sunxi
t3st3r has quit [Ping timeout: 265 seconds]
geecko has quit [Remote host closed the connection]
t3st3r has joined #linux-sunxi
bertrik has quit [Remote host closed the connection]
<tonikasch>
btw, libhybris won't be enabled on libraries supplied already-build-by-the-maker, will it?
<tonikasch>
I mean, if the vpu libraries are shipped in android tree as .so files, libhybris is not going to work, i guess
<tonikasch>
(in android build tree...)
<Turl>
tonikasch: it's designed to work with blobs as far as I know
FreezingCold has joined #linux-sunxi
<tonikasch>
oh! ok, so I'll continue that way first and then start with android RE, thanks
devgiant has quit [Quit: Leaving]
indy has quit [Ping timeout: 240 seconds]
indy has joined #linux-sunxi
deffrag has quit [Ping timeout: 240 seconds]
deffrag has joined #linux-sunxi
tonikasch has quit [Excess Flood]
tonikasch has joined #linux-sunxi
tonikasch has quit [Remote host closed the connection]
buZz has quit [Ping timeout: 252 seconds]
buZz has joined #linux-sunxi
buZz is now known as Guest63663
Guest63663 is now known as buZz
techn_ has quit [Ping timeout: 240 seconds]
lkcl_ has joined #linux-sunxi
Net147 has joined #linux-sunxi
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
akaizen has quit [Remote host closed the connection]
t3st3r has quit [Ping timeout: 265 seconds]
FreezingCold has quit [Read error: Operation timed out]
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
FreezingCold has joined #linux-sunxi
TheSeven has quit [Disconnected by services]
[7] has joined #linux-sunxi
skaczor has quit [Quit: leaving]
pseudomind has joined #linux-sunxi
Seppoz has quit [Remote host closed the connection]
<rellla>
jemk: thanks, i'll try that later. i must enjoy this sunny weather :p bbl
bertrik has quit [Remote host closed the connection]
nove has joined #linux-sunxi
theskilledworker has joined #linux-sunxi
smotocel69 has joined #linux-sunxi
leviathanch2 has quit [Ping timeout: 240 seconds]
geecko has joined #linux-sunxi
<nove>
there is no escape from android, new binaries will be always be there first
leviathanch2 has joined #linux-sunxi
<nove>
if we don't what to be late, we must get openmax support in libhydris
vincenzo has joined #linux-sunxi
vincenzo has quit [Client Quit]
Gerwin_J has joined #linux-sunxi
<nove>
and finally, people are speaking of RE other vpu
techn_ has quit [Ping timeout: 265 seconds]
<nove>
and in this case a on2 vpu, is a bit odd, but it looks like they have source code
jemk has quit [Ping timeout: 246 seconds]
leviathanch2 has quit [Ping timeout: 240 seconds]
<nove>
very strange
<oliv3r>
on sun5i; what else is going frmo PLL6 (which should be sata)
<nove>
let's worship might google, so that we get another tada source
<nove>
complain is easy, but nobody wan't to do the dirty laundry
<wens>
oliv3r: wild guess, usb phy clocks maybe
<wens>
docs don't really say where they come from
<oliv3r>
i think i did it on the wiki
<oliv3r>
APB1 and {NAND_CLK, <unknown>_CLK, SD[0123]_CLK, TS_CLK, SS_CLK, SPI[0123]_CLK, IR[01]_CLK}
<oliv3r>
all can run of pll6
<oliv3r>
(with the {} bit optionally from pll5)
kuldeepdhaka has joined #linux-sunxi
rz2k has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
<mnemoc>
,pom
<mnemoc>
err
<mnemoc>
moin
<wens>
haven't had your coffee? :p
Black_Horseman has quit [Quit: Zwi se logou mou!!!]
MarkusR has joined #linux-sunxi
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
hawi_ has joined #linux-sunxi
hero100 has quit [Quit: Leaving]
techn_ has joined #linux-sunxi
ssvb has quit [*.net *.split]
FunkyPenguin has quit [*.net *.split]
steev has quit [*.net *.split]
yang has quit [*.net *.split]
uwe_ has quit [*.net *.split]
bamvor has quit [*.net *.split]
sulky has quit [*.net *.split]
ganbold has quit [*.net *.split]
MSameer has quit [*.net *.split]
setkeh1 has quit [*.net *.split]
andoma has quit [*.net *.split]
hno has quit [*.net *.split]
bjoernb has quit [*.net *.split]
petr has quit [*.net *.split]
swabbles has quit [*.net *.split]
cazzacarna has quit [*.net *.split]
kenny has quit [*.net *.split]
swabbles has joined #linux-sunxi
bjoernb has joined #linux-sunxi
bamvor has joined #linux-sunxi
hno has joined #linux-sunxi
kenny has joined #linux-sunxi
cazzacarna has joined #linux-sunxi
ssvb has joined #linux-sunxi
andoma has joined #linux-sunxi
petr has joined #linux-sunxi
yang has joined #linux-sunxi
setkeh1 has joined #linux-sunxi
ganbold has joined #linux-sunxi
sulky has joined #linux-sunxi
FunkyPenguin has joined #linux-sunxi
Guest47151 has joined #linux-sunxi
MSameer has joined #linux-sunxi
uwe_ has joined #linux-sunxi
woprr has quit [Excess Flood]
woprr has joined #linux-sunxi
t3st3r has joined #linux-sunxi
<oliv3r>
ijc: what clock is pll6 being setup for sata support? i'm going over the a20 usermanual and it states, that if pll6 is used for sata support, pll6 must be 100 MHz
<mnemoc>
wens: indeed :\
<oliv3r>
ijc: ah wait, we init pll6 to a safe 600 mhz; the moment sata out is enabled, pll6 gets reconfigured to /6 so 100 MHz
<oliv3r>
ijc: but def. something take into account
<oliv3r>
i documented it all on the wiki for now
MarkusR has quit [Quit: Verlassend]
leviathanch2 has joined #linux-sunxi
binaryferret has left #linux-sunxi [#linux-sunxi]
leviathanch2 has quit [Ping timeout: 240 seconds]
leviathanch2 has joined #linux-sunxi
geecko has quit [Remote host closed the connection]
geecko has joined #linux-sunxi
Gerwin_J has quit [Read error: Connection reset by peer]
<ijc>
hno: Any chance you could reply to the mainlining thread offering your S-o-b for the contributions which you made to u-boot-sinxi.git which ended up incorporate there?
Gerwin_J has joined #linux-sunxi
<ijc>
AFAICT most of the other people who didn't give an S-o-b at the time were contributing either to a file with a clear GPL header or it was code which hasn't been included in the upstreaming series.
Seppoz has quit [Ping timeout: 252 seconds]
<ijc>
It looks like I'm going to need to do a deeper dive but having your S-o-b would be very helpful since it would move a big bunch of code into the "unambiguously OK" pile.
<ijc>
hipboi_: ^ If you would also be happy to also provide an S-o-b then that would be super awesome.
<ijc>
and saying "All of my contributions to u-boot-sunxi.git are Signed-of-by: Your Name <you@example.com>" would be perfectly adequate.
<ijc>
TBH I don't think this is strictly speaking needed, due to DCO clause (b), but it would certinaly short circuit a long and tedious argument with Wolfgang on the topic
pwhalen has quit [Remote host closed the connection]
rigo88 has joined #linux-sunxi
deasy has joined #linux-sunxi
F1skr has joined #linux-sunxi
xinj has joined #linux-sunxi
leviathanch2 has quit [Ping timeout: 245 seconds]
rigo88 has quit [Quit: Page closed]
leviathanch2 has joined #linux-sunxi
akaizen has joined #linux-sunxi
xinj has quit [Ping timeout: 265 seconds]
hawi_ has quit [Ping timeout: 240 seconds]
<Turl>
mripard: ping ping? :)
<mrnuke>
ijc: yeah. Wolfgang's nitpicking on the SOB made no sense. Though I think he was more concerned about saying that "j" and some_weird_nick wrote something, instead of using the real name
<mrnuke>
UARGH! -ELONGPHRASE, -ENOSENSE
Wizzup has quit [Ping timeout: 245 seconds]
<mrnuke>
ijc: I think Wolfgang didn't like the use of nicks as opposed to real names, in the commit message
<Turl>
mrnuke: you sending email as "mrnuke" probably doesn't help much :p
<mrnuke>
Turl: it all depends on which machine I'm sending it from
<mrnuke>
Turl: and yeah, mrnuke helps. That Wolfgang guy should know whom he's dealing with.
<Turl>
mrnuke: you mean the guy who created uboot? :)
leviathanch2 has quit [Remote host closed the connection]
<mrnuke>
yeah. him. he created uboot. I wrote code for coreboot
<mrnuke>
mripard: should I resend v3 of the sun4i-spi long transfers patch?
<mripard>
didn't you send it already ?
<mripard>
I definitely have a v3 in my inbox.
<mrnuke>
OK. I thought it might have gotten lost. All is good then
hawi_ has joined #linux-sunxi
Wizzup_ has joined #linux-sunxi
<hno>
ijc, sure. I did have s-o-b lines in sunxi-patchqueue.
<mripard>
mrnuke: I have a few more comments though, don't send the v4 right now
<mripard>
I'm replying to it just now.
<mrnuke>
mripard: no v4. Waiting for your comments first
<mripard>
(it's really minor things though)
<ccaione>
uhm, since when is legit to cast void * to int without getting a warning?
<mrnuke>
ccaione: I did that? where?
<ccaione>
mrnuke: you? O_O see broonie last email
<mrnuke>
-ENOPARSE
<ijc>
hno: There's still quite a few un-signed off patches even in the pq, although I've not looked to see if they are ones which matter for the initial upstreaming series so an explicit mail would still be very much appreciated.
<ijc>
mrnuke: Yeah, I tihnk you are right re the nicks. I've investigated the especially weird ones and I think they can be discounted since they were only adding support for boards which are not included in the series
<mrnuke>
ijc: I'd say just don't mention 'other' contributors. They should have (C) lines in the actual source, in which case SoBs certify section (b) of the certificate
rigo has joined #linux-sunxi
<rigo>
hi
<rellla>
jemk: the patch partially solves the problem, but i have some problems after 3rd, 4th, ... restart video layer disapperars this floods the logs. blit is called twice. once with 0:0 - 0:0, the other time with correct values. osd is shown. http://paste.debian.net/89101/
<rigo>
can you please help me what packages are needed to install the s2-liplianin-v39 driver package?
<rigo>
the first em comes when i try to build with make all (because of a ln problem. i solved it by re-creating the link) now i get this: http://paste.debian.net/89104/
<jelly-home>
rigo: do the s2-liplianin have a tree for linux 3.4 kernels?
<rigo>
im trying something now. maybe ive missed something before.
<rellla>
jemk: fine now. one more thing is, that the g2d_fill in rgba_clear after each video frame decreases performance rapidly. this has to be fixed to be called only once after the last g2d_blit
<rellla>
jemk: we move towards a useable vdr :p
netlynx has quit [Quit: Leaving]
hawi_ has joined #linux-sunxi
<captainigloo>
hi there, i have a question about sunxi and SPI bus
hawi_ has quit [Remote host closed the connection]
<captainigloo>
it's possible to have SPI transfer with 9-bits ?
Dodger78 has joined #linux-sunxi
jemk has joined #linux-sunxi
<jemk>
rellla: so the patch solves the problem? shall we patch the kernel or work around at vdpau side?
<jemk>
rellla: and rgba_clear should only call fill when the surface is dirty
<Turl>
captainigloo: on what SoC?
Wizzup_ is now known as Wizzup
<rellla>
jemk: i think we should do both. at minimum we need the vdpau patch, but it's better to solve the kernel bug, too.
<rellla>
jemk: G2D_DMA0_SIZE_REG and G2D_OUTPUT_SIZE_REG cannot handle -1 that good ...
<rellla>
jemk: fill should be called at surface close and change. but not after each video frame. iirc we talked about that already
<rellla>
jemk: you have a clue, why we there is a double g2d_blit without the zero-check in vdpau?
<jemk>
rellla: well, it can handle it, it interprets it as max size (8192px i guess). But that is not what one expects
<jemk>
rellla: i think these calls come from softhddevice, i don't know what and why they do this, but i don't think its good behaviour
<jemk>
the fill only happens if something was on the surface before, if we don't clear that it will be visible forever
<captainigloo>
Turl: a13 or a10
<jemk>
we can't really see if there is a change (maybe one could track all draw operations, but...) so we have to clear every frame, that's how vdpau works
<jemk>
normally the video gets rendered to the surface and overwrites the old osd, but as we show the video on an extra layer we have to clear the surface explicitly
<Turl>
captainigloo: the FIFOs there are 64 x 8 bits
<rellla>
jemk: ok. so no chance so far with g2d to fix that. nvidia, intel etc. are using 1 layer
<captainigloo>
Turl: cause i have an LCD screen wich i would like to connect to my a10 or a13 board, but it needs 9bit width spi transfer
<Turl>
captainigloo: well, I don't know much about the spi protocol
<Turl>
captainigloo: maybe you can send one 8 bit message with just 1 bit and then the 8 bits on the next one?
<jemk>
rellla: I don't know how they do it exactly, but the specification says that videomixer renders the video to the output_surface, overwriting old content
<captainigloo>
Turl: hum i don't think it's gona work
<captainigloo>
Turl: first bit is to say the 8 following bits are a comand or a data
<captainigloo>
Turl: i think it has to be 9 consecutives bit, at least it's how i understand the datasheet
<captainigloo>
i tried to send a 2 bytes btw, and it doens't work
<rellla>
jemk: ehm, why do we need XClearWindow? it runs ok without it ...
<jemk>
rellla: if you don't mind a random background color you can remove the XClearWindow (with active osd, in non-osd mode the bg-color has to match for colorkeying) and see how much faster it gets
<jemk>
too slow
<captainigloo>
with support for a lot of lcd
<Turl>
captainigloo: have you asked in the list? someone there may know better
<captainigloo>
Turl: nope, i will do, thanks
<jemk>
rellla: XClearWindow each frame is wrong either way, but I'm still working on finding a better way
akaizen has joined #linux-sunxi
Dodger78 has left #linux-sunxi [#linux-sunxi]
orly_owl has quit [Ping timeout: 252 seconds]
orly_owl has joined #linux-sunxi
FreezingCold has quit [Quit: Out]
FreezingCold has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
<hramrach>
clear or not to clear that's a question
<hramrach>
I know some mplayer outputs would not clear parts of window not overwritten by video rendering and you would get OSD test piling one on top of another
<hramrach>
text
<ZetaNeta>
Hai guys. I just got a ACER Z130
<ZetaNeta>
And gonna put linux on it
<ZetaNeta>
Anyone can redirect me to the right channel for MT6572?
<ZetaNeta>
s/Anyone can/Can anyone/
<ZetaNeta>
s/Can anyone/Can someone/
<jemk>
rellla: you might want to test the latest git ;) if I didn't sleep already it should avoid unnecessary clears
<Turl>
ZetaNeta: maybe somebody on #omnirom-mt6589 can point you to the right direction
<hramrach>
I can do that on a PC as well but quite busy lately :/
<jemk>
rellla: um, I guess that's because of the invalid pic_order_cnt (0x7fffffff), that looks bad
souther has quit [Ping timeout: 246 seconds]
<jemk>
where should these reference frames come from, it's the first picture being decoded
souther has joined #linux-sunxi
Jachy has joined #linux-sunxi
akaizen has quit [Remote host closed the connection]
Fusing has joined #linux-sunxi
<hramrach>
maybe a damaged stream or stream that does not start with key frame (which can happen when you tune into a broadcast or such)
akaizen has joined #linux-sunxi
<rellla>
hramrach: thats what i'm doing
<jemk>
i think you are right, it's probably not a keyframe.
<jemk>
But with my current mess of h264 framelist code I don't think I can fix this easily
<jemk>
I have to redo this for interlaced support anyway, but always busy with fixing output things, i haven't had any time to work on the real things ;)
<hramrach>
can't you skip frames until the stream makes sense?
<rellla>
jemk: so now you can do real things again, as output is fixed basically :p
<jemk>
skipping would be a way, but also needs some code reordering to know the frame type earlier
<hramrach>
or fill the buffer list with black frames initially
rigo has quit [Quit: Page closed]
* nove
likes vdr osd and wanted to use again, but doesn't know what to watch
<jemk>
h264 code really needs a cleanup...
akaizen has quit [Read error: Connection reset by peer]
akaizen has joined #linux-sunxi
<jemk>
but not now, night all
<rellla>
cu
dlan^ has joined #linux-sunxi
dlan^ has joined #linux-sunxi
akaizen has quit [Remote host closed the connection]
jemk has quit [Ping timeout: 246 seconds]
dlan has quit [Ping timeout: 240 seconds]
nove has quit [Quit: nove]
Night-Shade has joined #linux-sunxi
akaizen has joined #linux-sunxi
Valduare has joined #linux-sunxi
<Valduare>
hi
<Valduare>
I have an mk802 with a10s cozyswan crapola device :P
<Valduare>
wanting to find someone else that has one to see if this is normal behavior for it or not. it most of the time wont boot when power is applied to it
<libv>
mk802 with a10s?
<Valduare>
aye cozyswan clone
<libv>
well, start off by working through the new device howto in our wiki
<Valduare>
it was my first mk device, amazon purchase.
<Valduare>
I was able to get it booting linux
<libv>
i myself have an a10s stick in transit
<Valduare>
but the screen did this fantastical red circular distortions then severe screen distortions then white screen
<libv>
does it boot android afterwards?
<Valduare>
just hangs
<Valduare>
this happend after i was on linux desktop
<libv>
and what if you let it cool down for a bit?
<Valduare>
just enough time to click on terminal and run ifconfig
<Valduare>
its not warm,
<Valduare>
fresh boot etc.
<libv>
what are you using to power it?
<Valduare>
it has a barrel jack power supply that came with it
<Valduare>
2 amp if I remember
<Valduare>
let me double check on the amperage
<Valduare>
yes 2
<libv>
i had massive issues with my mele a1000 in oktober 2012, due to too little power
<libv>
was it running android np on the same power supply?
<Valduare>
aye I found on amazon a seller of apple ipad power supplies real nice for 7 bucks a piece bought a handfull of them