hno changed the topic of #linux-sunxi to: /Allwinner/sunxi development discussion - Don't ask to ask. Just ask and wait! - See | | Logs at
ganbold__ has quit [Remote host closed the connection]
naobsd has quit [Quit: Page closed]
rings_VII has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
rings_IIV has quit [Ping timeout: 264 seconds]
kaspter has quit [Ping timeout: 245 seconds]
ganbold has joined #linux-sunxi
rings_IIV has joined #linux-sunxi
rings_VII has quit [Ping timeout: 260 seconds]
drachensun has joined #linux-sunxi
rz2k has quit []
mcbrick has quit [Quit: Leaving]
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
vinifr_ has quit [Quit: Saindo]
rings_IIV has quit [Ping timeout: 260 seconds]
kaspter has joined #linux-sunxi
rings_IIV has joined #linux-sunxi
rings_VII has joined #linux-sunxi
rings_IIV has quit [Ping timeout: 264 seconds]
popolon has quit [Quit: Quitte]
hipboi has joined #linux-sunxi
[7] has quit [Disconnected by services]
TheSeven has joined #linux-sunxi
joshzhao has joined #linux-sunxi
rings_IIV has joined #linux-sunxi
rings_VII has quit [Ping timeout: 264 seconds]
woprr has quit [Ping timeout: 246 seconds]
woprr has joined #linux-sunxi
woprr has quit [Client Quit]
woprr has joined #linux-sunxi
dwilkins has quit [Ping timeout: 256 seconds]
akaizen has quit [Read error: Connection reset by peer]
akaizen has joined #linux-sunxi
dwilkins has joined #linux-sunxi
iamfrankenstein has quit [Ping timeout: 264 seconds]
bamvor has quit [Quit: Leaving]
JohnDoe_71Rus has joined #linux-sunxi
JohnDoe_71Rus has quit [Ping timeout: 246 seconds]
<setkeh> hey guys i just unplugged and cleaned my cubie cover and plugged it back in but now all i can see on hdmi is a black screen it doesent have any output at all there are no obvious errors from serial any idea's??
<setkeh> startx also complains no screens detected
deasy has joined #linux-sunxi
rellla has joined #linux-sunxi
<setkeh> [ 905.875] (EE) open /dev/fb0: No such file or directory
<setkeh> from the xorglog after it load the sunxifb
n01 has joined #linux-sunxi
wingrime has joined #linux-sunxi
geecko has joined #linux-sunxi
libv_ has joined #linux-sunxi
libv has quit [Ping timeout: 260 seconds]
libv has joined #linux-sunxi
libv_ has quit [Ping timeout: 260 seconds]
vicenteH has quit [Ping timeout: 246 seconds]
deasy has quit [Quit: Quitte]
<mnemoc> I'll shortly kill for moving it not another hw
<mnemoc> without btrfs!
kaspter has quit [Ping timeout: 264 seconds]
ykchavan has joined #linux-sunxi
mouchon has quit [Remote host closed the connection]
notmart has joined #linux-sunxi
vicenteH has joined #linux-sunxi
<oliv3r> yay!
<oliv3r> boo btrfs!
<oliv3r> (i really like btrfs's concepts, just needs to stablize first)
shineworld has joined #linux-sunxi
<rm> what in your experience is not stable enough about btrfs?
<ssvb> setkeh: do you actually have /dev/fb0 ?
<setkeh> no
popolon has joined #linux-sunxi
<ssvb> what kind of kernal are you using and with what config?
<setkeh> just stock kernel
<setkeh> everything is stock
deasy has joined #linux-sunxi
<ssvb> I don't know what is stock
<setkeh> it was working before i halted it the otherday
<setkeh> means i havent changed anything
<mnemoc> rm: Turl can tell you how many times btrfs has killed :(
<rm> how many?
<rm> you're doing something wrong, then
<rm> I am running btrfs on perhaps 8-10 systems with zero problems
<andoma> mnemoc: ENOSPC issues or other stuff or what?
<mnemoc> beside using the file system i don't know what can be done wrong there...
<mnemoc> [6190949.217036] btrfs-transacti D ffffffff81806240 0 4202 2 0x00000000
<mnemoc> [6190949.217042] ffff8801b7b5db80 0000000000000046 ffff88000a6802e8 0000000002f4fa34
<mnemoc> [6190949.217048] ffff8801b7b5dfd8 ffff8801b7b5dfd8 ffff8801b7b5dfd8 00000000000137c0
<mnemoc> [6190949.217053] ffff88032b611700 ffff8802da0e5c00 ffff8801b7b5db60 ffff88033fc34080
<mnemoc> [6190949.217058] Call Trace:
<mnemoc> [6190949.217067] [<ffffffff81118780>] ? __lock_page+0x70/0x70
<mnemoc> [6190949.217073] [<ffffffff8165cbbf>] schedule+0x3f/0x60
<mnemoc> [6190949.217076] [<ffffffff8165cc6f>] io_schedule+0x8f/0xd0
<wingrime> mnemoc: btrfs not looks production grade
<mnemoc> ugly kind of issues
<mnemoc> at least not in ubuntu's 12.04 kernel
<andoma> yeah that's a bit too ancient for btrfs i think
<wingrime> mnemoc: why not use debian ?
<mnemoc> because i already installed ubuntu ;-)
<oliv3r> i do hope btrfs stabalizes within a year or two
<oliv3r> it's just too cool
<rm> it is stable at the latest kernels
<andoma> i run it on 3.8 (without any fancy features turned on).. works like a clock
<rm> and the btrfs developers will be the first people to tell you you're insane using btrfs on 3.0 or 3.5
<wingrime> rm: what minimal partition size for btrfs
<rm> dunno
<rm> you don't want it for very small partitions anyway
<wingrime> rm: if I will make incremental backup infrastructure
<wingrime> rm: btrfs can be nice
akaizen has quit [Remote host closed the connection]
akaizen has joined #linux-sunxi
<Turl> so far it looks to me it's not too tolerant to failures or misconfiguration, and high maintenance
<Turl> I'll stick to ZoL for the fancy for the time being :)
<oliv3r> Turl: mornin'
<Turl> morning oliv3r
<Turl> mripard: :)
rellla2 has joined #linux-sunxi
ssvb has quit [Ping timeout: 256 seconds]
ssvb has joined #linux-sunxi
rellla2 has quit [Remote host closed the connection]
uwe__ is now known as uwe_
shineworld has quit [Quit: Leaving]
<ykchavan> is website down?
<n01> ykchavan: 10:24 < mnemoc> I'll shortly kill for moving it not another hw
<Turl> s/not/to/
<Turl> did you guys know olimex gave away a A13 olinuxino micro last friday?
<buZz> one?
<buZz> ZOMG
<buZz> i guess the A13 users list went up 500% now
<Turl> buZz: lol
<Turl> buZz: they give away a board each friday :)
<buZz> O-M-G
<Turl> (or device, often it's motors, lcds, batteries or other components)
iamfrankenstein has joined #linux-sunxi
<wingrime> Turl: I have saw a Loooot china tablets with a13
<wingrime> Turl: this is realy looow end
<wingrime> 50-60$ per device
<Turl> wingrime: china tablets not OSHW with lots of pins :P
<deasy> i can be interested in good china products, as my zte blade
<wingrime> Turl: a13 is eTQFP )
<wingrime> Turl: just solder wire to SoC 0
<hno> olimex friday quiz is quite nice, very good marketing.
<hno> and Olimex boards are very much worth their price if you need a dev board or embedded board. But clearly not if what you want is a tablet.
<oliv3r> Turl: that's pretty cool; but i wanna focus on a20 mostly now
<oliv3r> but yeah it's fully Opensource, which is win to me
<Turl> oliv3r: I'll get an SD hopefully today and give you a hand with that :)
<oliv3r> i'll send you 4 512MB Sd cards
<Turl> they should've put the fel thing on a jumper or a button far from microUSB
<oliv3r> Turl: is it still breaking or is it now working
<Turl> it works, but fel is a pain to boot to :)
<oliv3r> howcome?
<Turl> the button is right under the microUSB
<Turl> a pain to hold while powering on
<oliv3r> oh yeah
<oliv3r> i never use FEL mode
<oliv3r> but i have SD cards out the wazoo :p
<oliv3r> which you refuse to take
<Turl> they'll arrive by the time we get cubietrucks :p
<oliv3r> good; so you are in despare need for more
<oliv3r> i'll just send you a few postcards, and use some tape to stick it on there :)
<oliv3r> 1 Sd per postcard
<Turl> lol
<oliv3r> what, far more reliable
<oliv3r> it's called 'post-raid0'
<oliv3r> raid1 even
<Turl> lol
<Turl> bbl
<oliv3r> go learn
hipboi has quit [Quit: Leaving]
<hno> Turl, just use an SD card thta jumps to FEL. There is an image in suxi-tools.
<oliv3r> he doesn't have enough SD cards :p
<hno> you only need one...
<hno> the rest is done over FEL.
<hno> and the same FEL boot card works in all a10/a13/a10s/a20 devices, and likely a31 as well (but there you still need to hold button...)
<oliv3r> yeah but sometimes you want fel, sometimes you want to boot something else )
<ganbold> is down or the problem is in my side?
<oliv3r> it's down
<oliv3r> we're migrating from an old host to a new host
<oliv3r> mnemoc should be able to give us a transistion ETA
<mnemoc> oliv3r: data is moved. configuration comes now
<oliv3r> ah, still moving it all
<oliv3r> thought you put the site in read-only mode; did all the transfering etc to the new host, then switch over DNS
<oliv3r> but all's well
<mnemoc> oliv3r: at $work$ :( and today is my last day to cancel the old server
<mnemoc> so i can't be .... sweet
<oliv3r> och
<oliv3r> ouch*
<oliv3r> okay :)
FergusL has joined #linux-sunxi
<FergusL> website down ?
<FergusL> nvm, no it's not down
<mnemoc> oliv3r: up
<mnemoc> FergusL: it was down
<mnemoc> now it's running on another hw
<FergusL> it was down just for a few moments ?
<mnemoc> no.... since 10:30 to 13:45 CET
deasy has quit [Quit: Quitte]
<oliv3r> testing
<mnemoc> much longer than expected, but $work$ was consuming precious time...
Black_Horseman has quit [Remote host closed the connection]
<mnemoc> is the new IP in case you have resolving problems
<oliv3r> seems to work
<mnemoc> hopefully better
<mnemoc> btw, is everyone happy with the current stage/sunxi-3.4 ?
<mnemoc> would like to merge it....
<mnemoc> and jump forward to the next 3.4.z
<wingrime> Rockchip rk2918 DMC
<wingrime> hno: ^ Willswang writed to wiki about DRAMC
<oliv3r> mnemoc: can't say, but you have my blessing :p
<oliv3r> wingrime: i thought that was you :p
<oliv3r> wingrime: it's probably atleast partially a port from HNO's rhombustech wiki page
<mnemoc> oliv3r: :)
<oliv3r> if he was real smart, he took my commits from u-boot and added those ;) I was lazy in adding it to the wiki
<oliv3r> mnemoc: i really have to start upgrading my tablet to 3.4, still runs GPL violating stock kernel, but i'd be touchscreen less; which I have a fix for, but just requires tons of time
<oliv3r> and since nobody is working on an android build, i haven't put time into it either
<oliv3r> so much to do, so little time
<mnemoc> oliv3r: consider it educational time
<mnemoc> and writting a TS driver also fits in your CV ;-)
<oliv3r> mnemoc: heh, i posted it ages ago; and i don't mind doing it, but the 4 lettered daemon from hell haunts me
<oliv3r> Time
<oliv3r> shit i forgot to start a build on my new htpc
<oliv3r> after trying fedora and debian, i'm falling back to gentoo for now
<oliv3r> should have tried openbricks I suppose
<hno> oliv3r, don't know if it's a copy of mine.. much less info than I have.
<wingrime> oliv3r: please move it to our wiki
<oliv3r> hno: i've noticed, i've started porting some info over
<oliv3r> wingrime: pff, too much work, to little time!
<oliv3r> i just noticed there's hardly anything moved :p
<oliv3r> so i'll see when ihave time
<oliv3r> 've moved the CCR register over :)
<oliv3r> maybe i find a few minutes to do the rest, there's not much yet out
deasy has joined #linux-sunxi
<hno> oliv3r, if you are copying then remember to continue copuing. I am not moving to mediawiki.
shineworld has joined #linux-sunxi
<oliv3r> i'll try to keep it sync
<rellla> has sun7i cedar ( already found its way into any 3.4-branch?
<deasy> hno, had another person than you actively make code for sunxi on uboot?
<deasy> (it's for me know if your branch is the reference if i want to build something on sdcard)
<oliv3r> rellla: if not, we should unify it rather then copying it
<oliv3r> cedarX engine is the same
<oliv3r> so kernel driver should be near the same
<oliv3r> someone on ML said that cedarX on A20 is using same binaries
<hno> deasy, yes there is several that have contributed to sunxi u-boot. The reference branchs is
<oliv3r> deasy: use
<oliv3r> i'm too slow :(
<deasy> :p
<hno> oliv3r is one of them.
<deasy> thank you both, ok i use git sunxi for reference as it seems logic
<rellla> oliv3r: sun7i should be added to wingrime's unifying patch and then pushed to stage/sunxi-3.4
<oliv3r> ideally yeah
<oliv3r> if wingrime cedarX merge hasn't been merged
<rellla> must the regs be hardcoded in sun4i_cedar.h?
<oliv3r> what do you mean
<oliv3r> yeah for now that's how it is
<hno> gah.. usual strategy of banging every possibly releated module directly instead of using suitable driver interfaces?
vinifr has joined #linux-sunxi
hipboi has joined #linux-sunxi
<oliv3r> hno: yeah it's really sad
<oliv3r> hno: but it'll be changed soon enough ;)
<hno> and no actual cedar register info at all..
jemk has joined #linux-sunxi
vicenteH has quit [Ping timeout: 246 seconds]
<oliv3r> no, jemk and wingrime have been very busy to get that sorted
<oliv3r> hno: actually, the driver is SO bad, it just opens a hole to let a userspace driver manipulate the registers directly via an ioctl
rz2k has joined #linux-sunxi
<oliv3r> 'write reg'ioctl and read reg ioctl
<oliv3r> very well done :)
<setkeh> hey guys im still having trouble starting X the error is /dev/fb0 file not found it was working 2 days ago and i havent changed anything since any suggestions ??
<mnemoc> oliv3r: "bad" is relative.... that allows them keep more of the IP hidden in a closed blob without breaking the GPL
<rellla> bah. too long urls ;)
<setkeh> << My xorg logfile
<oliv3r> rellla: i think so, from what i've heard/read; cedarX may have seen a small upgrade, but the blob is still the same, so the kernel headers should work from sun4i, but sun7i might be' cleaned'
<wingrime> oliv3r: actualy cedarx blob have branches dependng on SoC version
<wingrime> oliv3r: for example I find some reigster that a13 have but a10 not have
<hno> kernel cedar driver don't know any details. It's all in the userspace blobs.
<oliv3r> wingrime: so for unified sun457i cedarX that's ok (kernel side)
<oliv3r> for your RE work that's much harder :(
ganbold_ has joined #linux-sunxi
<wingrime> oliv3r: also, but it seems for a13 cedarx down-clocked
<wingrime> 150 Mhz vs 180Mhz
<wingrime> (nearly)
<wingrime> thats only one difference I have noticed with diff
<wingrime> some parts of driver not ever used
<hno> setkeh, double check that you haven't changed script.bin..
<hno> setkeh, and what does dmesg say?
<hno> error says you have no framebuffer.
<oliv3r> wingrime: well a13 is a castrated something, so it's not that strange that they did that
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<oliv3r> wingrime: could be that those parts are from sun3i
<setkeh> hno: i havent changed anything the board has been off for 3 days and checking dmesg now
<setkeh> doesent look to be anything amiss
<hno> setkeh, I don't see any messages at all from the framebuffer driver.
<setkeh> i have fbdev and xunxifb installed
<setkeh> s/xunxifb/sunxifb
<hno> that's xorg drivers. You also need the kernel driver loaded.
<setkeh> i followed this when i installed it
ganbold__ has joined #linux-sunxi
<hno> setkeh, try "modprobe hdmi"
<setkeh> kk
ganbold_ has quit [Remote host closed the connection]
<setkeh> hno: ok x still wont start but my display isent saying no input anymore
<hno> setkeh, from the archlinkx link you provided it seems you are missing step 2.
<setkeh> hno: warning: cubieboard-modules-load-1-1 is up to date -- reinstalling
<setkeh> warning: cubieboard-modules-load is flagged out of date
<hno> there a nice hint on where the problem is :)
<setkeh> well considering i havent updated in that time either it shouldent have affected the system ??
<hno> setkeh, try "modprobe -a lcd; disp fbcon"
<hno> err. no ;
<hno> setkeh, try "modprobe -a lcd disp fbcon"
<setkeh> kk
<FergusL> is there anything to do to enable wifi dongles ? (cubie2 with cubian)
<hno> setkeh, the updated question is more what updates may you or any automatic update have installed since you last successfully booted with X.Org working.
<FergusL> just tried two of them, the second has a message in dmesg that it loads but then "ERR: script_parser_fetch usb_wifi_usbc_num failed"
<setkeh> hno: the is no automatic update in arch
<hno> FergusL, what wifi dongle are you using?
<setkeh> and no change with new modprobes
<hno> setkeh, dmesg time agin.
<FergusL> RT5370 Wireless Adapter ID 148f:5370
<FergusL> hno ^
<setkeh> dmesg
<hno> setkeh, no new messages at all? Did any of the modules load?
<FergusL> installing the firmware, as per google search
<setkeh> [root@SETKEH-CUBIE ~]# lsmod
<setkeh> Module Size Used by
<setkeh> [root@SETKEH-CUBIE ~]#
<setkeh> aparently not
ykchavan has quit [Quit: Leaving]
vinifr has quit [Quit: Saindo]
<FergusL> anybody using cubian here ?
<oliv3r> not me, sorry
<setkeh> im going to try it now lol alarm is annoying me now spent all day trying to fix this stupid thing
<FergusL> try what ?
<wingrime> jemk: what mean slice in h264 context?
ykchavan has joined #linux-sunxi
<jemk> wingrime: see specification, pictures are divided into one or more slices
<wingrime> jemk: you mean, thats not full size frame?
<jemk> wingrime: it can be full frame, or only parts of it
<wingrime> jemk: so if you poke h264 trigger it still not give you full size frame
<wingrime> ?
<jemk> wingrime: no, you have to do this again for all slices in picture
deasy has joined #linux-sunxi
<WarheadsSE> setkeh: probably because it didn't load the module?
<WarheadsSE> Helps if your uname matches your linux-sun7i pkgver
<hno> oliv3r, seems gcc-4.8 produces slightly larger binaries making SPL >24K..
<WarheadsSE> cute
hipboi has quit [Read error: Connection reset by peer]
<oliv3r> hno: ouch :( i haven't updated my gcc beyond 4.6.3 yet
<FergusL> "could not insert 'rt5370sta': Cannot allocate memory" what could cause this ?
<hno> oliv3r, I am on 4.7 something normally, but started to get some complaints from users after Falcon mode support was enabled.
<hno> lets see how large it really is...
<hno> Hm... odd. Still complains about size.
<hno> 4.8: 0x6084, 4.7:0x5dfc. That's a fairly significant -Os size regression in 4.8 compared to 4.7.
naobsd has joined #linux-sunxi
<oliv3r> wow taht's quite a lot
<hno> 2.6% increase in size.
<deasy> new optimizations for arm?
<deasy> maybe use -O0 ?
rellla has quit [Quit: Nettalk6 -]
<setkeh> WarheadsSE: upgrading linux-sun7i hasent helped either
panda84kde has joined #linux-sunxi
<hno> deasy, -Os is optimize for minimal size.
<deasy> it as O2 with "normally" optimization who don't make the binary swells
<deasy> is*
<hno> -Os makes considerably smaller binaries than -O2
<deasy> but i have never compare them(O0 and Os)
<mnemoc> speed optimizations usually rely in larger asm
<deasy> that's why i say it
<hno> -O0 is much larger than -O2 and -Os
<deasy> (in the past i was think than Os especially for size but it seems not true)
<hno> mnemoc, same optimization levels in both..
<deasy> Os just keep away some of O2
<hno> '-Os'
<hno> Optimize for size. '-Os' enables all '-O2' optimizations that do
<hno> not typically increase code size. It also performs further
<hno> optimizations designed to reduce code size.
<deasy> i have this page under eyes ;p
<hno> 4.7 documentation is identical from what I can tell.
<deasy> yes yes it's, i have just never compare these option in size
<ssvb> WarheadsSE: why sun7i when setkeh has "chip-id: A10 (AW1623 revision C)"?
<WarheadsSE> ssvb: He did his own
<WarheadsSE> that's not my kernel
<WarheadsSE> setkeh: CB2 or CB1/
geecko has quit [Quit: Quitte]
<setkeh> CB1
<setkeh> srry i forgot to mention that
<ssvb> setkeh: can you check /lib/modules directory and verify that you have something like 3.0.57-5-ARCH subdirectory there?
<setkeh> sure do
<ssvb> if you have /proc/config.gz file, then you can "zcat /proc/config.gz | grep FB" to check whether the framebuffer modules are compiled at all
<ssvb> but who has built this "stock kernel"?
\\Mr_C\\ has quit []
<ssvb> setkeh: framebuffer is compiled as a module in your kernel, you need to load the relevant modules in the right order -
<ssvb> setkeh: if you can't 'modprobe lcd' without getting it really loaded, then your /lib/modules/3.0.57-5-ARCH is likely fubared
<ssvb> setkeh: btw, the framebuffer driver is statically compiled in the kernel using the standard defconfigs to make it much easier for the users
<wingrime> jemk: have you ever tryed H263 video ?
<wingrime> jemk: 0x204 may be header for it
<setkeh> ssvb: hmm problem is no matter what module i load lsmos stays empty
<setkeh> lsmod*
<ssvb> setkeh: ... then your /lib/modules/3.0.57-5-ARCH is likely fubared
<wingrime> jemk: also a20 manual claim that we may have in new hardware SorenSpark and VP6
<jemk> wingrime: no, i havn't, and no, 0x204 is header for h264, i only havn't found out how it is used
<setkeh> ok ill swap to the default sun7i one
<buZz> is cedarx working on A20 ?
iamfrankenstein has quit [Ping timeout: 260 seconds]
atiti has joined #linux-sunxi
<wingrime> jemk: 0x210 is used ?
<jemk> wingrime: yes, persumably for weighted prediction
<wingrime> jemk: thats bad,
<wingrime> jemk: It means same registers have multiple uses
<wingrime> MACC_H264_MB_NB1 0x0264 is used ?
<jemk> wingrime: haven't seen 0x264 yet, but where else is 0x210 used?
<wingrime> jemk: have you new blob for a20
<wingrime> ?
<jemk> wingrime: i don't have a20, so no
<wingrime> jemk: If a20 manual correct we have Soren Spark and VP6
<wingrime> jemk: at least in a20
<wingrime> oh, no VP6 we must have in mpeg engine
shineworld has quit [Ping timeout: 260 seconds]
shineworld has joined #linux-sunxi
<wingrime> jemk: are you tryed MACC_H264_SDROT_CTRL
<wingrime> ?
<wingrime> jemk: it must work like MPEG one
tinti has joined #linux-sunxi
rellla has joined #linux-sunxi
\\Mr_C\\ has joined #linux-sunxi
<wingrime> jemk: also MACC_H264_TRIG should have input color space like mpeg?
<jemk> wingrime: i don't know yet, but i think it can only decode yuv 4:2:0, as other colorspace is only allowed in higher h264 profiles
<merbanan> SorenSpark is h263 with other headers
<setkeh> ssvb: youstill about bud ??
shineworld has quit [Quit: Leaving]
<setkeh> or any one able to tell me how to change the timeout on uboot so i can make it wait longer for keypress to drop to sunxi shell ??
massi has joined #linux-sunxi
<deasy> big diff between O2 and Os :p
mcbrick has joined #linux-sunxi
<WarheadsSE> setkeh is covered, btw
ganbold__ has quit [Remote host closed the connection]
<setkeh> yes ty :)
vinifr has joined #linux-sunxi
<vinifr> wigyori, ops, here ping
ykchavan has quit [Quit: Leaving]
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
vinifr has quit [Quit: Saindo]
<oliv3r> hno: hmm, with -Os and 4.6.3 I get File size: 0x5ef4
<oliv3r> IF that's the SPL's size
<oliv3r> it's what make outputs at the end
<oliv3r> Data Size: 228120 Bytes = 222.77 kB = 0.22 MB
<oliv3r> that's for u-boot.img i belive
n01 has quit [Ping timeout: 245 seconds]
massi has quit [Remote host closed the connection]
BJfreeman has quit [Read error: Connection reset by peer]
leowt has joined #linux-sunxi
robb83 has joined #linux-sunxi
merbanan has quit [Ping timeout: 246 seconds]
panda84kde has quit [Quit: Konversation terminated!]
jemk has quit [Ping timeout: 264 seconds]
akaizen_ has joined #linux-sunxi
<FergusL> what is the right way to setup console resolution ?
akaizen has quit [Ping timeout: 240 seconds]
<robb83> anyone can use ttyUSB0 as console at boot time ? :)
<oliv3r> robb83: that's a very confusing question
<oliv3r> i have a usb -> uart converter connected to by PC, that's ttyUSB1 (could be 0), that in turn is connected to UART0 on my board. UART0 on the board however is ttyS0
naobsd has quit [Quit: Page closed]
<oliv3r> and i have that connected often when booting to check debug output
<robb83> and inversily, so my "board" no have UART port in case, but usb
<oliv3r> you can't send debug output via usb -> uart
<robb83> ok, thx :(
<oliv3r> i dont' think our EHCI driver does debugging and if it does, that won't help you with grub
<oliv3r> and every board has UART output via uSD -> UART adapter
<robb83> yes yes, but still under construction :)
<oliv3r> if you have a tablet, that you can't dissasemble or can't find any pads to solder the uart too; then that's your only option
<robb83> hmm, thx!
<oliv3r> wingrime: ping
<bfree> grub? what did I miss ;) (I've failed so far to get grub arm-uboot to actually boot a kernel)
jemk has joined #linux-sunxi
<wingrime> oliv3r: pong
<oliv3r> you mentioned the pengpod a few days ago?
<oliv3r> the guy who did it is in the channel here :)
<wingrime> ))
<oliv3r> he's actually doing developer discounts
<oliv3r> but shipping is a killer
<wingrime> oliv3r: I have a10 and a13 tablets
<oliv3r> a10 only here
<oliv3r> since when do you have a10 tab?
<oliv3r> i thoguht you only had a13
<wingrime> oliv3r: I not see any difference between pengpod and any comon a10 tablet
<oliv3r> it's supported by drachensun :)
<wingrime> oliv3r: one month ago
<oliv3r> what made you get it?
<oliv3r> does it have more then 512 mb ram?
<wingrime> oliv3r: 1 gb
<oliv3r> good
<oliv3r> mine has 512 mb and is ram starved
<wingrime> oliv3r: realy chap
<wingrime> oliv3r: but it have led for charge
<wingrime> oliv3r: and full size usb port
<oliv3r> reall ynice
<wingrime> I also still not figured why battery indicators in linux lie
merbanan has joined #linux-sunxi
<wingrime> oliv3r: I hope pengipod move to a20 soon
<drachensun> hello
<oliv3r> i think there's a31 pengpods too
<oliv3r> hi drachensun, mr. pengpod himself ;)
<drachensun> let me catch up on the conversation here
<oliv3r> not much really
<drachensun> oh yeah, so oliv3r, turns out the mele tv box doesn't seem to take a flash by any convention means we can figure out
<oliv3r> drachensun: mele m5?
<drachensun> I would still like to get you an A31 for your U-boot work but we need a better candidate
<drachensun> I've got the 1000G I think?
<oliv3r> id only get frustrated by having yet a nother device requireing a blob ;)
<wingrime> I have no a31 hw (
<oliv3r> but swamped in $work this week
<oliv3r> drachensun: i have no mele, I want to get an m5
<oliv3r> i have a10 tab; a10 cb1, a20 cb2, a20 olimex
<oliv3r> wingrime: you don't want a31 :p powervr! evil! :p
<drachensun> wingrime: if I had known it was for you we could have applied a discount
<drachensun> but the A10 models are going away, we can't get them anymore
<drachensun> something is coming up but we aren't quite ready for the official announcement
<oliv3r> a20? :p
<wingrime> drachensun: actualy looking at photos on site I figured that actualy rebrended china)
<drachensun> oliv3r: I dont get how powervr is any worse than Mali, its just that there is already some work to reverse engineer Mali and no one has taken on the powervr from what I can tell
<drachensun> I do have a working libhybris for it though
<oliv3r> or do you have actually a40 news!!
<oliv3r> drachensun: ask libv :p he worked with powervr, he says it's a huge pile of crap and steered people clear of even trying to RE it
<oliv3r> freedreno based part would be cool too
<drachensun> wingrime: yeah, thats exactly what it is, and as oliv3r said we provide support, images and a preflash for those that don't know how
<oliv3r> rockchip are trying, but I'd hope not, splitting work away from A10 to RK is not good for either i'd think
<drachensun> oliv3r: I see, well good to know even if things are going really well its not worthing trying to RE
leowt has quit [Quit: leowt]
<oliv3r> drachensun: Mele 1000G is a31 device?
<drachensun> yup
<oliv3r> drachensun: i think libv worked for the company doing the initial powervr drivers at some point, or something along those lines
<drachensun> no, I have no special A40 news either
<oliv3r> drachensun: how is flash not working?
<oliv3r> drachensun: bummer, big.LITTLE a40 would be cool if you had the scoop
<wingrime> drachensun: but , it you have direct contact to china factory you can try request some nice features like 9000 mAh accamulator
<drachensun> it doesn't have a traditional button, its got that special button to force the A31 to boot from an SD
<drachensun> I think
<wingrime> oliv3r: I think you can use a40 like pan for breakfast
<oliv3r> ohhh
<drachensun> but there isn't a good way to make bootable SD cards that I know of, mripard apparently had to do a lot of different hijinks to get it to work on his
<oliv3r> yeah a31 requires a 'boot selector' there's more then 1 pin that influences boot order
<wingrime> oliv3r: (like some new nvidia boards
<drachensun> but so for now, I see no way to get it into fel mode, short of wipe the boot sector of the flash
<oliv3r> even if I had a31 hardware (which I really odn't wanna burn energy on) a31 u-boot requries a LOT of work still, it's a completly new memory controller, i've done some work on the dram controller allready, but i've got maybe 15% done of that
<oliv3r> and spotted errors in my pushed code :p
<wingrime> drachensun: you are saw batt indicator bug?
<drachensun> I've got a sales contact with mele, hopefully its possible and its just a matter of translating his answers properly
<oliv3r> drachensun: do you have the possibility to take high-res images inside? top/bottom PCB ideally
<oliv3r> only thing I know is that there's now 2 pins that have 4 combinations to short
<wingrime> oliv3r: as I find yesterday fex dram config not so magic like looks
<oliv3r> it it depends entirly what they decided to map out at all
<drachensun> oliv3r: realistically probably not, I've got no time right now, there is info on CNX though
<oliv3r> wingrime: check my last patches that where pushed to sunxi/u-boot i tried to clarify a lot of settings best I could
<oliv3r> drachensun: ill double check that
<wingrime> drachensun: have you request tablets directly form factory?
<drachensun> wingrime: Which on exactly, I never figured out a Linux graphical monitor that would work, various users submitted text based ones that would read the AXP output
<oliv3r> not working for anybody else?
<oliv3r> it's rerouting to some google url
<oliv3r> nvm, google is being ghey
akaizen_ has quit [Remote host closed the connection]
<drachensun> wingrime: pretty much, but there are a lot of limits on what I can get
<oliv3r> drachensun: anyway, only thing I wanted tos ay is; you probably know this:
<oliv3r> which you probably are totally aware of
<drachensun> we are trying to get the volume up so we can buy from a larger, more honest supplier
<drachensun> oliv3r: yup
<oliv3r> and I"m sure you've tried press fel key without any SD card
<wingrime> drachensun: that bug shown in KDE for example
<drachensun> oliv3r: Yeah, it doesn't go to fel mode, so probably that button isn't quite what I think it is
<wingrime> drachensun: I have tablet with nice full sized usb
<oliv3r> drachensun: do you know what it does do?
<oliv3r> yeah full USB + extra otg port is luxurary :D but awesome for tablet
<oliv3r> drachensun: the only thing I can imagine, that that key is mapping check_boot_sel bit 0
<wingrime> oliv3r: but building quality tootaly crap
<drachensun> oliv3r: they claim you hold down the button and plug in the power to flash it (like normal) but two of us tried on two different devices today, no joy
<oliv3r> so it toggles between 00 (nand, normal) and 01 (SD2)
<oliv3r> and the only way to make it fail to fel mode, is if the SD2 is corrupt
<wingrime> oliv3r: I afrand disasemble tablet agian
<oliv3r> (from that graphic)
geecko has joined #linux-sunxi
<oliv3r> wingrime: lol what?
<oliv3r> drachensun: they claim it's normal FEL button, but I doubt it
<wingrime> oliv3r: afraid, plastic locks too weak
<drachensun> wingrime: I'm not sure what you mean, you mean the KDE battery widget seems to work but isn't accurate?
<oliv3r> drachensun: you describe it as bot_sel button, but who knows, it's just some brainfart
<wingrime> drachensun: it show no precentage when you dissconnect tablet
<drachensun> well typically boot_sel is hard wired with resistors to be the nand flash
<drachensun> I guess you could set it on a button, I see what you mean, but there is no second SD card on this device
<oliv3r> unless the main SD card is SD2 :)
<wingrime> drachensun: it feel only cable exsitness
<wingrime> *exsist
<oliv3r> only way to find out is trace sd card to about where it enters the soc, and how far it is from SD0
<drachensun> wingrime: I see, yeah, most of the monitors don't work at all with the AXP driver. I think there was some discussion about that on the mailing list
<oliv3r> script.bin might tell you of course
<wingrime> drachensun: thats problem in kpower logic
<hno> oliv3r, so the linaro 4.7 toolchain is a winner, and 4.8 no go..
<wingrime> drachensun: if you read sysfs by hand , all reads good
<drachensun> wingrime: yup, thats what the guys in my forums were doing
<oliv3r> hno: let me build 4.6.3 with -O2 again see hwat that reports
<drachensun> they wrote some script so it would always show on a semi transparent desktop background
<drachensun> I thought it was pretty slick
<oliv3r> heh
<oliv3r> conky? wasn't something like that
<drachensun> wingrime: well I was going to take that on this week for the A31 and see what is missing to make the standard linux monitoring apps understand the driver
<drachensun> oliv3r: sounds right
<oliv3r> hno: 4.6.3 File size: 0x5ef4
<oliv3r> with -O2, so -Os actually makes it bigger
<oliv3r> no i lie, it's nearly the same size
robb83 has quit [Quit: - A hand crafted IRC client]
<wingrime> drachensun: some sysfs file must be present or some not , kpower have strange logic for it
<wingrime> drachensun: witch ts are you using in pengipod ?
<wingrime> *TS IC
<hno> oliv3r, are you really really sure that's an -O2 build?
<oliv3r> well i edited
<oliv3r> did a rm -r on build/cubie2*
<oliv3r> and i did a random pause somewhere to check if my -Os was being used
<oliv3r> but no, i have no clue where to actually change it
<hno> 4.8 -O2 is 0x6d3c
<hno> make ... OPTFLAGS="-g -O2"
<oliv3r> HOSTCFLAGS>-----= -Wall -Wstrict-prototypes -O2 -fomit-frame-pointer \
<oliv3r> that's the line I changed
<hno> That's for compiling the host tools (mkimage etc).
<hno> 4.7 -O2 is 0x6714
<oliv3r> ok OPTFLAGS=; let me try that
<oliv3r> but howcome there's size differences though
<oliv3r> nvm there weren't i'm confused
<oliv3r> arm-pc-linux-gnueabi-gcc -g -g -Os -ffunction-sections -fdata-sections -fno-common -ffixed-r8 -msoft-float -D__KERNEL__ -ffunction-sections -fdata-sections -DCONFIG_SYS_TEXT_BASE=0x4a000000 -DCONFIG_SPL_TEXT_BASE=0x20 -DCONFIG_SPL_PAD_TO=24576 -DCONFIG_SPL_BUILD -I/silo/build/sunxi-bsp/build/cubieboard2-u-boot/include2 -I/silo/build/sunxi-bsp/build/cubieboard2-u-boot/include -I/silo/build/sunxi-bsp/u-boot-sunxi/include -fno-builtin -ffreestanding
<hno> there is a size diff of +/-0
<oliv3r> so I see -Os there
<oliv3r> File size: 0x5ef4
<oliv3r> same size
<hno> did you specify -O2?
<drachensun> wingrime: The typical ft5x_ts
<hno> because that output suggests you specified -g -Os
<drachensun> wingrime: though the latest devices have some weird ft5x06 driver
<hno> oliv3r, 4.6 -O2 gives 0x67b4 for me. 4.6.2 20120613 (release) [ARM/embedded-4_6-branch revision 188521]
<oliv3r> hmm
<drachensun> and an inet_ctp driver that depends on it (or vice versa, can't remember)
<drachensun> I've gotten some source for the ft5x06 but not the inet_ctp
<drachensun> so I haven't been able to get the touch screen working
<oliv3r> -O2 breaks the build
<oliv3r> arm-pc-linux-gnueabi-ld.bfd: arch/arm/cpu/armv7/sunxi/libsunxi.o: relocation R_ARM_MOVW_ABS_NC against `a local symbol' can not be used when making a shared object; recompile with -fPIC
<oliv3r> arch/arm/cpu/armv7/sunxi/libsunxi.o: could not read symbols: Bad value
<drachensun> the A20 tablets are coming with the same driver, its getting to be a problem
<hno> oliv3r, odd
<oliv3r> drachensun: eeck, so we should sort proper mailined ts drivers :)
<oliv3r> hno: is OPTFLAGS used for anything else?
<oliv3r> adding -fPIC to my OPTFLAGS it doesn't help; strange
<oliv3r> also, i'm not running a hardenend kernel on my buildbox
<oliv3r> if i leave out -O2 it's fine (which builds the default)
<drachensun> wingrime: About the battery, yes, I think I'm just going to have to dig into the kpower program and see what is missing. I'll share what I find so we can catch up the axp209 as well
<hno> kernel do not matter
<hno> -fPIC is not about hardening.
<oliv3r> position independant code afaik
<hno> yes
<oliv3r> anyhow, default IS -O2
<hno> you changed
<oliv3r> so i don't understand why OPTFLAGS="-Os" works, but "-O2" does not
<drachensun> oliv3r: ugh, its been a constant thing, they have a new nandflash chip I can't find the settings for as well. That is why I have found a new supplier who ACTUALLY RELEASES THEIR SOURCE, hard to believe right?
<oliv3r> nope
<oliv3r> drachensun: impossible! :p
<hno> default is -Os
<oliv3r> oh reall
<drachensun> oliv3r: only problem is they have a high MOQ so if I can't bring in enough interest its not going to be possible to buy from them
<oliv3r> hno: of course you are right, the -O2 was for 'tools'
<oliv3r> drachensun: *sigh*
<oliv3r> hno: what happens if you build with OPTFLAGS="-O2"?
robb83 has joined #linux-sunxi
<hno> oliv3r, file gets too big.
<oliv3r> oh, so it does some chcks while building and just errors out halfway?
<oliv3r> well with -O3 we get: arm-pc-linux-gnueabi-ld.bfd: address 0x838c of u-boot-spl section `.text' is not within region `.sram'
<hno> Well, I have increased the limit so build succeeds.
<oliv3r> so that's allready way to big :)
<oliv3r> no point for this little test, if -Os is the most efficient size wise
<oliv3r> then that's what it has to stay anyhow
<oliv3r> since we want the smallest possible binary
<hno> that's CONFIG_SPL_MAX_SIZE in include/configs/sunxi-common.
<hno> mksunxiboot is slightly broken as well, silentrly truncating at 24K-20
notmart has quit [Quit: notmart terminated!]
<hno> pushed sunxi-spl-max-size to my repo with limits lifted to 32KB.
<hno> need to check actual size limit on the dierent boards.
akaizen has joined #linux-sunxi
<hno> limit on a20 seems to be ~0x7700
<oliv3r> which is 32k?
<hno> ox8000
<hno> same on a10
<oliv3r> so almost 32k; 28k i'd guess?
<hno> ~30K
<oliv3r> 30464 bytes
<oliv3r> 29.75k
<hno> yes. 7800 is 30K.
<oliv3r> i cheated :p
<oliv3r> i hAd no clue:)
<hno> 800 is 1K.
<hno> no, that's 2K.
<oliv3r> i should know this stuff
* hno too
<oliv3r> by heart i mean
<oliv3r> but i don't at all :(
<hno> aha, A10s OLinuXino MICRO have another AXP... more to fix up for Falcon :)
<hno> did a little powerbutton magic for forced u-boot fallback in Falcon SPL mode.
<hno> A10s have the same limit.
<oliv3r> oh really? i tried to make it run, didn't get very far; gave up way to soon
<hno> Haven't tried to load that large SPL yet, only checked what memory is being used.
<hno> right.. need to round down to sector size. Soe 0x7600-20
<oliv3r> sounds like very exact science :p
atiti has quit [Ping timeout: 276 seconds]
<wingrime> oliv3r: may be write ps/2 driver for a10 ?
<hno> an SPL of 0x7600 works fine.
heffer_ is now known as heffer
<oliv3r> i was thinking of that too wingrime for the far future
<oliv3r> but cubie doesn't bring out pins
<oliv3r> nice, that gives us alot of extra space
<wingrime> oliv3r: olimex do, but you need only 4 wire for ps/2
tlhingan_ has joined #linux-sunxi
<wingrime> gnd , +5 V , sync, data
<jemk> do we have a cedar player that doesn't drop frames (when being traced)?
<tlhingan_> Does anybody here have any experience with the lima driver?
<jemk> vlc allways drops exactly these frames i need
<wingrime> jemk: disable memcopy dump feature
<wingrime> jemk: our valgrind plugin have option to disable memcopy log
<oliv3r> wingrime: yeah, clock + data
<oliv3r> so 2 really; if you ignore power
<hno> oliv3r, you need a shared GND between the devices if using external power. But yes.
<oliv3r> well i ment 'soc pins'
<hno> USB is also only two pins.
<jemk> wingrime: you mean --show-suppressions=no ? that i already have
<wingrime> jemk: may be not
<hno> Wonder where the limit is for NAND.
<oliv3r> i guess you can best compare it to I2C
<oliv3r> hno: what limit?
<hno> SPL size.
<oliv3r> ohh, you mean bigger SPL size when boot0 resides in nand?
<oliv3r> I expect it to be the same tbh
<hno> seems it's even higher for NAND, 7d60, rounded down in KB, minus 0x20 for header, gives 7be0 (31712)
<oliv3r> but why make it 2k bigger?
<wingrime> rellla: ping
<oliv3r> i can understand they only want to reserve some max size for it (Like sram-A) but why the 24k I don't understa,d if they read the size from the header anyway
<rellla> wingrime:pong
<wingrime> oliv3r: stack and vars
<wingrime> rellla: do you know how drop memcopy form traces
<wingrime> jemk: --no-drop-late-frames
<wingrime> jemk: ^ -- do you tryed ?
<wingrime> jemk: --no-skip-frames
<rellla> wingrime: no clue at all
<wingrime> jemk: ?
<hno> oliv3r, what?
<jemk> wingrime: im trying, needs time
<oliv3r> hno: why would the SPL size differ in size?
<oliv3r> why hardlock it to 30k, and not to whatever is in the headersize (with a max of SRAM-A size)
<hno> because it has different drivers.
<hno> it's not a hardlock. Only trying to find where the limits are.
<hno> so make fails rather than boot crashing.
<jemk> wingrime: works, thx
<oliv3r> hno: well why differentiate between spl on either sizes i mean
<oliv3r> i don't understand why make the defines even diff.
<jemk> wingrime: but now im getting mad. the traced regs match exactly what im doing, but my output is full of artifacts. wtf
<hno> oliv3r, I don't know yet what size the NAND SPL driver will end up at.
<hno> but likely far below limits.
<wingrime> jemk: diff 100% same ?
<hno> just pushed a change that moves u-boot.img to 40K offset, and allow SPL to grow up to 0x75E0.
<wingrime> jemk: what kind of artifacts ?
<oliv3r> hno: what's the status of the MTD work? i've totall ymissed any news on it
<oliv3r> hno: will that leave some space between SPL and u-boot now?
<jemk> wingrime: not 100%, but all regs match (well, except memory addresses and such things cause i use other memory allocation)
<hno> oliv3r, yuq have done some progress, but apparently still have problems understanding the randomizer for normal NAND accesses, and also problems with the OOB data.
<wingrime> jemk: maybe 1k allign?
<jemk> wingrime: all aligned
rellla has quit [Remote host closed the connection]
<oliv3r> hno: ah so nothing new then
<wingrime> jemk: ve-sram writes same ?
<jemk> wingrime: i hope xD it's hard to check for sure, i would need some sort of emulator
<wingrime> jemk: we have replay tool
<hno> oliv3r, there is a driver for the NAND boot area now, which is important.
<wingrime> jemk: nove not here ((
<jemk> wingrime: blob overwrites old data all the time, so there is no real dump in the trace. i would need to reorder all writes to get linear dump
<oliv3r> anybody else working on it except yuq? i know rz2k was, but he had to pause? and slapin was being excited, but gone again
<hno> now just need to relearn all users to write u-boot.img to the new offset...
<oliv3r> yeah
<wingrime> jemk: for every frame?
<jemk> wingrime: yes, dump for every frame
<jemk> wingrime: but i don't think this would show anything new, i double checked everything
pacopad has quit [Quit: pacopad]
boycottg00gle has joined #linux-sunxi
<wingrime> jemk: may be something with your display code
<geecko> any clue on what could be wrong here?
<wingrime> jemk: you a get it only with some h264
<wingrime> ?
<jemk> wingrime: no, display is all right
<oliv3r> ssvb: ^ that's probably up your ally :)
<wingrime> jemk: witch artifacts?
<wingrime> jemk: what kind of artifacts?
<jemk> wingrime: typical block artifacts
<wingrime> jemk: like very low compression?
<jemk> wingrime: hm, more like what you see after frames were dropped, but they aren't dropped, i singlestep
<wingrime> jemk: like that?
<jemk> wingrime: exactly
<jemk> wingrime: only near moving parts
<hno> oliv3r, can you check that I have updated what is needed in
<oliv3r> i'll follow the wiki tomorrow to see if it still works
<wingrime> jemk: thats artifacts present when you run mplayer without trace tool?
<rz2k> oliv3r:
<rz2k> oliv3r: everything that needs to be done was done
<rz2k> i've uploaded my stuff to my tree
<rz2k> only needs to be ported to u-boot
<oliv3r> rz2k: all done? inc oob?
<wingrime> jemk: I have seen such , whan mplayer work in HARD framedrop mode
<jemk> wingrime: sure, even when i single step, every frame is decoded, no drops
<rz2k> everything works, except random and ubifs fails like hell with hardware ECC
<rz2k> so you need to use fs without oob usage :( or fix the driver, yuq recently brought that up on mailing list
<rz2k> read his and my replies
<rz2k> maybe you can figure out ubifs settings somehow so it wont fail after 2-150 reboots/remounts
<oliv3r> yeah i read all that
<oliv3r> was hoping things had improved since
<wingrime> jemk: and do you checked log with " grep <reg> | wc -l " thats you realy do same amount of work like orignial blob
<rz2k> its usable :)
<jemk> wingrime: no, i dont do same amount, cause i write ve-sram without that many double writes
Black_Horseman has joined #linux-sunxi
Black_Horseman has quit [Changing host]
Black_Horseman has joined #linux-sunxi
<wingrime> jemk: can you check , (count how much access to some reg , thats must be accessed only once per slice)
<wingrime> with grep and wc , so , we will be sure that you realy do all slices like blob not lost something
<jemk> wingrime: the normal regs are all same, only 0x2e[04] are different
<wingrime> paste me both traces, so I can play with it
<wingrime> jemk: wait, it do double wrties, what it means?
<wingrime> jemk: ptr is autoincremented
<jemk> wingrime: and they set it to single addresses often and do only one write there
<wingrime> jemk: are you thinked that cedar have internal state
<wingrime> jemk: so cedar can write to ve-ram too
<wingrime> jemk: also, may be 0x02e0 used by cedar not only for write , may be it need be set to specific offset before write
<wingrime> jemk: and more , are you sure all ve-ram rewrited for every slice? sram can (and I think must) save internal state between slices , (like some prediction info)
<jemk> wingrime: they rewrite everything, so that shouldnt be a problem
<wingrime> jemk: witch buffers are you using?
<jemk> wingrime: i have a pretty good idea what's wrong, but i can't find how this would be done right
<wingrime> jemk: ?
<jemk> wingrime: there is some buffer, if i initialize it to something != 0 the artifacts get less and in simpler videos vanish completely
<jemk> wingrime: problem is, blob doesn't initialize this buffer
<jemk> wingrime: i tried different alignments, huge oversize to prevent overflow and so on. nothing helped, only initializing
<wingrime> jemk: witch buffer?
<wingrime> address please
<jemk> wingrime: don't know what it is for. every frame has chroma and luma address and some extra buffer i think for some motion info
<jemk> wingrime: this extra buffer is the problem
<wingrime> jemk: witch register for it?
<jemk> wingrime: no register, its written into the framelist in ve-sram
<jemk> wingrime: as all frame related things
<wingrime> ....
<wingrime> jemk: can you document this
<wingrime> jemk: framelist to wiki
<jemk> wingrime: simple list of buffer addresses for output and reference frames
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
<jemk> wingrime: as h264 can have many reference frames registers won't work
<wingrime> jemk: thats ONLY writes to ve-ram, nothig more?
<jemk> wingrime: no, also weighted prediction tables and reference order list
<jemk> wingrime: last is referencing the addresses from the list
<jemk> wingrime: i'll try to document this somehow, but it is much
<wingrime> jemk: it hard to tell for me, I simple unsure witch logic it use
<wingrime> jemk: framelist count is limited , or , fixed , or can varry?
<wingrime> jemk: I need simply know , ve-sram internal offsets are fixed or behaves like stack
<jemk> wingrime: blob writes enough 0x0 to store 18 frames there, don't know if all can be used
<jemk> wingrime: it realy writes to fixed offsets, i've checked that with uboot yesterday
<wingrime> jemk: 0x2e0 autoincremet after any write to 0x2e4 ?
<jemk> wingrime: yes
<wingrime> jemk: do you tested videos without weighted prediction ?
<jemk> wingrime: yes, same problem
<jemk> wingrime: but there they completely vanish if i initialize those buffers (which can't be the solution)
<jemk> wingrime: except trace would miss something, that would be bad
<wingrime> jemk: 'those buffers' I think maybe cedar must write to it
geecko has quit [Quit: Quitte]
<wingrime> jemk: some predictions vector
<jemk> wingrime: at least it should, but something goes wrong here
<wingrime> jemk: this buffers, blob write 0 to it?
<wingrime> jemk: or not even touch
robb83 has quit [Quit: - A hand crafted IRC client]
<jemk> wingrime: doesn't touch, at least not traced.
boycottg00gle has quit [Remote host closed the connection]
<wingrime> jemk: I think thats related internal cedarx state ( variables) that must change after every step
tlhingan_ has quit [Quit: Page closed]
<wingrime> jemk: so you can try add /dev/kmem code to dump it after you step
<wingrime> jemk: also , we can fix ammt valgrind tool , and make it do same after TRIG
<wingrime> jemk: I hope this is possible
<wingrime> jemk: we need nove
<wingrime> that writed ammt
<wingrime> jemk: or send me your both traces, and I try parse it with python and get <known> mem state tomorrow
<wingrime> jemk: firstly we need know , is cedar touch sram and use it for internal needs
<wingrime> jemk: you have other ideas?
<jemk> wingrime: i think i have no more ideas today, im completely drained, maybe tomorrow
<wingrime> jemk: then bzip logs (yours and vlc's) and drop somewhere
<wingrime> jemk: If I get it today, tomorow, I try parse it
<Seppoz> whats changed in uboot?
<hno> Seppoz, SD carf offset where u-boot.img needs to be written. And the fact that u-boot.img needs to be used and not u-boot.bin.
wingrime has quit [Ping timeout: 245 seconds]
jemk has quit [Remote host closed the connection]
<Turl> back
<Turl> did anyone ping me?
merbanan has quit [Ping timeout: 256 seconds]
<hno> Turl, not really. Last message to you was me suggesting to use an SD card to force FEL mode.
<hno> just after you left.
<Turl> hno: ah
<Turl> hno: well, the point is that I don't have spare SDs :)
<Turl> otherwise I'd be using uboot on mmc
rz2k has quit []
tinti has quit [Remote host closed the connection]
dwilkins has quit [Ping timeout: 240 seconds]
dwilkins has joined #linux-sunxi
rings_IIV has quit []
hramrach has quit [Ping timeout: 240 seconds]
hramrach has joined #linux-sunxi