pwhalen has quit [Read error: Operation timed out]
parabyte has quit [Ping timeout: 252 seconds]
Night-Shade has quit [Ping timeout: 272 seconds]
pwhalen has joined #linux-sunxi
hero100 has joined #linux-sunxi
Seppoz_work has joined #linux-sunxi
enrico__ has quit [Quit: Bye]
newleaves has joined #linux-sunxi
Seppoz_work has quit [Ping timeout: 252 seconds]
parabyte has joined #linux-sunxi
printallthething has quit [Read error: Operation timed out]
bsdfox has quit [Ping timeout: 252 seconds]
printallthething has joined #linux-sunxi
bsdfox has joined #linux-sunxi
parabyte has quit [Read error: Connection reset by peer]
parabyte has joined #linux-sunxi
printallthething has quit [Quit: Leaving]
parabyte has quit [Client Quit]
rings_IIV has quit []
rings_IIV has joined #linux-sunxi
Nyuutwo has quit [Ping timeout: 272 seconds]
egbert has quit [Disconnected by services]
egbert has joined #linux-sunxi
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
brain_ has joined #linux-sunxi
ggcock has joined #linux-sunxi
bfree_ has quit [Ping timeout: 260 seconds]
bfree__ has joined #linux-sunxi
hipboi has joined #linux-sunxi
<brain_>
anyone familiar with the Ft5x_ts.c source code? what im doing is trying to figure out were the interrupts that are being generated by my touchscreen are going, the code from github does not register any interrupt handler for some reason. i know the hardware works since its fine with the stock driver. i have a special case were i MUST be able to recreate the driver. i've been at it for 3 months now, reversed the stock driver with IDA,
<brain_>
made many changes to the driver code to try to get it to read the interrupts from the chip and nothing is working
<brain_>
it is however showing up when I do "getevents" under android so the device is registered, just wont send the touch data due to no interrupts :/
<brain_>
If i could somehow create a BARE minimum driver that does nothing but register a handler i could at least use that information to incorporate it into the focaltech code im working with.
bsdfox has quit [Ping timeout: 248 seconds]
<Turl>
brain_: tried cat /proc/interrupts to see where they're going?
bsdfox has joined #linux-sunxi
bsdfox has quit [Changing host]
bsdfox has joined #linux-sunxi
<Turl>
brain_: the driver seems to pull the interrupt pin from the pin ctp_int_port on [ctp_para], have you checked that on your script.bin?
bsdfox has quit [Ping timeout: 276 seconds]
<brain_>
they simply are not getting read by the driver, i do not see them by incremented in the /proc/interrupts, I checked the fex file, the reason i am writing a custom driver is number one I trashed my eeprom but got it to an inverted touch to the point were I can redo the code to fix this and 2....the OEM has EVERYTHING hardcoded inside the module itself. I can change just about every line aside from the sizes in the fex and nothing chan
<brain_>
ges. i can see its all hardcoded inside the OEM module using IDA PRO.
<Turl>
brain_: they are really external interrupts via PIO
<Turl>
you should have a pin defined on ctp_int_port with the mux set up to EINT
<brain_>
right, if i change the int port i have no effect as long as im using OEM module..what your saying is i need to BE SURE there is a EINT in there but If i do not have the exact value, would I have a way of tracing down the interrupt via some means? i've already read up on external interrupts.
<Turl>
brain_: try to look for related printk calls and look around
<brain_>
ok ill give it a shot thanks.
<Turl>
odd, the 'netbook' allwinner market seems to have shrunk considerably
bsdfox has quit [Ping timeout: 252 seconds]
JohnDoe_71Rus has joined #linux-sunxi
bsdfox has joined #linux-sunxi
Gerwin_J has joined #linux-sunxi
bsdfox has quit [Ping timeout: 252 seconds]
bsdfox has joined #linux-sunxi
<hipboi>
hmm...
<hipboi>
dram init failed
[7] has quit [Disconnected by services]
TheSeven has joined #linux-sunxi
phoenix_ has joined #linux-sunxi
rz2k has joined #linux-sunxi
brain_ has quit [Ping timeout: 276 seconds]
brain_ has joined #linux-sunxi
newleaves has quit [Quit: Leaving]
bsdfox has quit [Ping timeout: 276 seconds]
bsdfox has joined #linux-sunxi
<phoenix_>
anyone know anything about adding ethernet phy drivers?
bfree_ has joined #linux-sunxi
bfree__ has quit [Ping timeout: 276 seconds]
<phoenix_>
i'm trying to add a driver for the ksz8864 ethernet switch from micrel. i've got the source code but i'm not sure how to link it into the kernel
<phoenix_>
in the guide from micrel they say "In your MAC network driver “probe” function, to connect KSZ8895 PAL, call this function: phydev = ksz8895_mii_connect(pdev, netdev, phy_addr, phy_mode );"?
<libv>
heh, could this really be a case of my driver getting the size vastly overstated, and the kernel driver not checking the bounds of the allocation it is trying to map?
<libv>
that's pretty pathetic.
kivutar has joined #linux-sunxi
<itdaniher>
looks like BCM20710 datasheet @ download.csdn.net/download/bdwgbdwg/5325785, but it's some sort of chinese scribd
<itdaniher>
that's the BLE-capable chip in the ap6210 module
<itdaniher>
if anyone with better language skills than yours truly takes the time to grab a copy and mirror it on github or mega, I'll drop them a few bitcents
rellla has joined #linux-sunxi
maz_ has joined #linux-sunxi
rz2k has quit []
shineworld has left #linux-sunxi ["Leaving"]
bsdfox has quit [Ping timeout: 245 seconds]
<phoenix_>
i need help with phy drivers
kivutar has quit [Quit: Ex-Chat]
<libv>
phoenix_: what device do you need this for?
<juanfont>
ping rellla
<phoenix_>
libv: its a board I made, very similar to cubie. it has a ksz8864 ethernet switch
<wens>
has 2 interfaces which can operate in either phy mode or mac mode
atiti has joined #linux-sunxi
Night-Shade has joined #linux-sunxi
<oliv3r>
very interesting :)
<mripard>
ccaione: pong
<phoenix_>
oliv3r: any ideas?
atiti has quit [Remote host closed the connection]
<phoenix_>
oliv3r: i know basically nothing about this stuff. it looks to me like i'm missing a mac driver?
atiti has joined #linux-sunxi
focus has quit [Remote host closed the connection]
<itdaniher>
wens: where might I find the datasheet?
<oliv3r>
phoenix_: our mac driver is the gmac/stmmac; check the mailing list for wens's patches, should be helpfull enough
<JohnDoe_71Rus>
oliv3r: thanks. I do not know yet vid pid
<atiti>
hey dudez
<atiti>
just got my olimex a20 dev board
<wens>
itdaniher: can't find it anymore :(
<itdaniher>
shucks
<itdaniher>
if you dig it up, can you ping me?
<itdaniher>
I'd like to add BLE to the capabilities of my tiny computer collection
<itdaniher>
bonus points for adding it for 'free' aka however many hours of kmod dev
panda84kde has joined #linux-sunxi
<wens>
I found a leaked version from gempak a few months ago
<wens>
seems it disappeared
Nyuutwo has joined #linux-sunxi
pwhalen has quit [Read error: Operation timed out]
newleaves has quit [Quit: Leaving]
newleaves has joined #linux-sunxi
popolon has joined #linux-sunxi
popolon has quit [Changing host]
popolon has joined #linux-sunxi
bsdfox has quit [Ping timeout: 246 seconds]
bsdfox has joined #linux-sunxi
bfree__ has joined #linux-sunxi
bfree_ has quit [Ping timeout: 260 seconds]
ggcock has quit [Quit: Leaving]
notmart has joined #linux-sunxi
<bbrezillon>
Hi all, I'd like to reference all the people that previously work on the sunxi NAND in the next version of my sunxi nand driver (as a copyright line, or any way you'd like), because all I'm doing is cleaning up things in oder to mainline the NAND driver :)
jinzo has joined #linux-sunxi
jinzo has joined #linux-sunxi
focus has joined #linux-sunxi
<oliv3r>
bbrezillon: i know that hno, slapin, rz2k and yuq have worked on it. I don't know what code you actually ended up using :)
<bbrezillon>
oliv3r: a bit of yuq's code (but not much), but actually I got a lot of information from their reverse engineering work
<oliv3r>
fair nuff :)
<oliv3r>
slapin and hno mostly worked in the u-boot branch on hno's repo
<oliv3r>
rz2k i linked you last time his tree with bug fixes
<oliv3r>
and yuq, well you know that one ;)
enrico_ has joined #linux-sunxi
newleaves has quit [Quit: Leaving]
rz2k has joined #linux-sunxi
<hno>
hi bbrezillon
<oliv3r>
hey hno
<hno>
hi hi
bfree_ has joined #linux-sunxi
<bbrezillon>
hi
bfree__ has quit [Ping timeout: 276 seconds]
<slapin_>
bbrezillon: ohai!
<slapin_>
hno: hi!
<hno>
hi slapin_ whole nand crew here except for yuq :)
<oliv3r>
hehe, so many nand-folk
<oliv3r>
exciting times!
<bbrezillon>
great
<bbrezillon>
I was asking if you'd like me to include a copyright line for each of you guys in the next version of my NAND driver
<bbrezillon>
after all you did all the work ;)
<bbrezillon>
(almost all the work :P)
<oliv3r>
bbrezillon: still a basic driver with no ecc and no dma right? (which is fine, just curious)
<bbrezillon>
this WE I managed to get HW ECC and HW RND working
<bbrezillon>
still no DMA
<oliv3r>
bbrezillon: oh wow, big progress
<oliv3r>
DMA probably relies on DMA engine anyway
<bbrezillon>
I just need some time to clean it up
<bbrezillon>
yes, I heard DMA is not supoorted in mainline yet
<oliv3r>
so you ran into the issues described by others? OOB problems, bad block crazyness every 50 - 100 reboots etc
bfree__ has joined #linux-sunxi
bfree_ has quit [Ping timeout: 260 seconds]
<bbrezillon>
nope, not yet
<oliv3r>
what FS are you using? I assume ubifs?
<bbrezillon>
yes I tried UBIFS
CoSM0 has quit [Quit: ciao]
<bbrezillon>
but this should work with ext4 on top of mtdblock (I guess that's what is closer to the sunxi block driver, right ?)
CoSM0 has joined #linux-sunxi
<slapin_>
bbrezillon: NEVER use mtdblock for flash device access using real filesystem
<bbrezillon>
of course you won't get any Wear Leveling, and it might quickly introduce Bad blocks
<slapin_>
bbrezillon: and very fast
<Wizzup>
f2fs maybe?
<oliv3r>
bbrezillon: yeah but it's begging for problems; since mtdblock exposes the raw flash device in a way, so using ext4 ontop of that will cause bad blocks etc (mtdblock does not do wear leveling etc)
<bbrezillon>
What is used on the android image ?
<mripard>
Wizzup: nope, neither
<Wizzup>
oh
<mripard>
f2fs is a block filesystem too
<oliv3r>
Wizzup: nah, f2fs should be used on 'smart' fs, like ontop of libnand, on ssd's, on uSD or USB-flash
<mripard>
it doesn't handle the bad blocks at all
<slapin_>
bbrezillon: allwinner android uses special block device
<bbrezillon>
the best way is to use UBI, and then whatever you want
<oliv3r>
bbrezillon: android image uses ext4 ontop of libnand, and libnand does the wearlvling etc
<bbrezillon>
it does the WL work ?
<oliv3r>
i dunno if you can just stick whatever you want ontop of ubi
<oliv3r>
i think eventually that's the plan
<bbrezillon>
yes you can
<bbrezillon>
it will expose an mtd dev (which can in turn expose an mtdblock dev using the mtdblock layer)
<bbrezillon>
you end up with a block device on top of NAND with a good WL support :)
FreezingCold has quit [Read error: Connection reset by peer]
FreezingCold has joined #linux-sunxi
<bbrezillon>
(I forget to tell you need gluebi to get mtd dev from ubi volumes)
<oliv3r>
bbrezillon: interesting; anyway, i did read that ubi(fs) hasn't solved all problems with nand flash yet, it's still very much WiP
<oliv3r>
bbrezillon: you should try jffs, as that should expose a fair share of problems (oobo etc)
<bbrezillon>
yes but then you're stucked with the JFFS2 file system
<bbrezillon>
with UBI (not UBIFS) you can put whatever you want on top
<oliv3r>
what do you mean, stuck?
<oliv3r>
well yeah
<oliv3r>
but you want to test your mtd driver :p
<oliv3r>
so if you use jffs2, you may encounter all the problems found by the others ;)
<bbrezillon>
(yes these were generic statements regarding NAND WL, of cource I can test my driver with JFFS2 if you wish)
_massi_ has joined #linux-sunxi
<oliv3r>
bbrezillon: i just recall people running into problems when using jffs2 as it uses oob's
<oliv3r>
bbrezillon: btw, I guess it won't be horrible if the first version of the driver only supports ubi(fs)
<bbrezillon>
it will support JFFS2, because it does not depends on the NAND driver
<bbrezillon>
you just have to take care when defining the OOB layout
<bbrezillon>
but I others encounters theses OOB issues, I will too :)
<oliv3r>
bbrezillon: well yuq, slapin/rz2k's work all didn't support jffs2 :p
<bbrezillon>
but I -> but if
<oliv3r>
bbrezillon: hehe, just pointing you that way, as it's a known problem
<oliv3r>
but hey, we can always say 'nfc doesn't support oob's
<bbrezillon>
it does (at least from what I've experienced) :)
<oliv3r>
but ubi doesn't use oob's at all though?
<oliv3r>
(I really have to read up on what oob is and why a FS would need it ;)
<oliv3r>
i only repeat what i know and heard
<bbrezillon>
No UBI does not use aby bytes in the OOB area
<bbrezillon>
but relies on the NAND ECC calculation
hipboi has quit [Quit: Leaving]
bsdfox has quit [Ping timeout: 252 seconds]
FreezingCold has quit [Ping timeout: 248 seconds]
<oliv3r>
the way I understand it, is the ECC controller needs those ECC's?
<bbrezillon>
as far as SOFT ECC is working (and it is: I tested it), I'd say the OOB area actually store the right informations
<bbrezillon>
here's the sunxi HW ECC layout : (4 bytes of available OOB + N bytes for HW ECC) * X
<oliv3r>
and i think jffs2 needs more then just 4 bytes right?
<bbrezillon>
N depends on the chosen strength (e.g. 40 bits per 1024 block => 70 ECC bytes)
<oliv3r>
so i'll be lazy and just pick your brain :p (i'll google it later)
<oliv3r>
what exactly IS oob? just some additional storage in the nand chip for 'other' uses?
<bbrezillon>
X => Number of sectos = (page_size / ecc_blk_size)
<oliv3r>
e.g. for ECC and some spare for the FS to use?
<bbrezillon>
yes that's it
<bbrezillon>
it is mainly used to store ECC data
<bbrezillon>
but JFFS2 use it to store FS informations
<oliv3r>
block devices don't have this do they?
<bbrezillon>
no they don't
<bbrezillon>
(or at least AFAIK)
bsdfox has joined #linux-sunxi
<oliv3r>
so if you use software ECC, the oob becomes much bigger and thus jffs2 has room?
<oliv3r>
or is it not freely asignable
<bbrezillon>
BTW if you define the OOB layout correctly JFFS2 will know where it can put it's FS data
<bbrezillon>
yes with SOFT BCH ECC, the ECC bytes are stored at the end of the OOB area
<bbrezillon>
this leaves space at the beginning of the OOB area till the start of the ECC byte free
<oliv3r>
i don't understand, if you use software ECC, I would have thought the oob wouldn't be used at all for ecc
<oliv3r>
so with HW ECC you use the start of the OOB and with soft ECC it usees the end?
<bbrezillon>
even with SOFT ECC, you need to store DATA to be able to correct inconsistencies in the NAND page
<oliv3r>
i would have expected that softecc stores it at the end of the page; so your page would actually be smaller :p
<oliv3r>
so why the inconcistency between hard and soft ECC?
<bbrezillon>
with HW ECC there are holes of 4 bytes every N bytes and it starts at the beginning of the OOB area
<bbrezillon>
ah, that's the question, because HW vendors decide how this is easily done in HW, whereas softaware guys think about how this can be easily handle in software :D
<arete74>
else
<oliv3r>
bbrezillon: so there is a difference, ok :p
<oliv3r>
so the hardware guys just assume the entire oob 'block' is for ecc, and the fs can use whatever is leftover
<oliv3r>
so the nfc was likly designed with no spare oob's
<bbrezillon>
yes it is => 4 bytes every X bytes of ECC + the end of the OOB area
<oliv3r>
so there's holes of 4 bytes every ECC and at the end
<bbrezillon>
but it's not helping to store contiguous large data
<oliv3r>
whatever is left over; which isn't enough for jffs2, or jffs2 doesn't know how to map this
<oliv3r>
'large' being very relative here :p
<bbrezillon>
JFFS2 should know according to the ECC layout defined by the NAND driver
<oliv3r>
so ideally; jffs2 would need an extension to work with holed oob's, e.g. an oob LUT
<oliv3r>
bbrezillon: ah so jffs2 knows how to deal with these holes allready?
<bbrezillon>
I guess so
<oliv3r>
(i'm just trying to learn and understand here ;)
<oliv3r>
then i wonder what slapin and rz2k ran into with regards to oob
torindel has quit [Read error: Connection reset by peer]
<oliv3r>
there's oobsize and oobavail
<bbrezillon>
oobavail is the number of bytes available for non ECC use
<bbrezillon>
and then you define the oobavailable map using the oobfree table
<bbrezillon>
each entry is a range of available data
torindel has joined #linux-sunxi
<bbrezillon>
(offset: start of the range, size: size of the range)
<bbrezillon>
eccbytes store the number of bytes used for ECC
<rz2k>
I've ran into not understanding the oob_layout and how it needs to be set up regarding the HW ECC activity :( but it was working for some time 10-50 reboots.
<bbrezillon>
and eventually eccpos encore the position of the ECC bytes within the OOB area
<bbrezillon>
here's is the code I've developped to define the oob layout:
<rz2k>
bbrezillon: thanks for figuring this all out!
<oliv3r>
rz2k: so he 'fixed' the issues we had then? :D
<rz2k>
sort of more clearly declared what is happenning
<rz2k>
:p
<oliv3r>
bbrezillon: you extracted that from the original libnand?
<oliv3r>
i'm quite excited about seeing this work getting merged
<oliv3r>
and even more importandly, backported
<oliv3r>
i guess 3.10 will be the only 'old' one to see this work; backporting to 3.4 is probably to much effort to be worth it
<oliv3r>
anybody tested the vdpau stuff on a20?
<rellla>
oliv3r: some vdr users are reporting success with vdr and vdpau. basically.
rz2k has quit [Ping timeout: 276 seconds]
<oliv3r>
ok so it 'works' normally
<rellla>
yes.
brain_ has quit [Quit: Leaving]
kivutar has joined #linux-sunxi
rz2k has joined #linux-sunxi
FreezingCold has joined #linux-sunxi
rz2k has quit [Ping timeout: 245 seconds]
<oliv3r>
rellla: do you happen to have vdpau_sunxi.so compiled for arm for me?
<oliv3r>
preverably a recent one?
<oliv3r>
jemk shared one iwth me, but he allready deleted it from his dropbox again
fredy has quit [Excess Flood]
<rellla>
oliv3r: not now. maybe on late evening.
fredy has joined #linux-sunxi
<oliv3r>
i really should just install a native compiler etc on my demo box
HeHoPMaJIeH has quit [Remote host closed the connection]
JohnDoe_71Rus has quit [Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org]
<oliv3r>
biting the bullet, was hoping to do it all in packages/precompiled; but this will have to do
ssvb has quit [Quit: Leaving]
kivutar has quit [Ping timeout: 246 seconds]
<hno>
bbrezillon, you are welcome to add a copyright line with my name if you think it's warranted.
kivutar has joined #linux-sunxi
<hno>
and as slapin said, mdtblock is not at all what it's name says. Never use mtdblock as a block device, it's not what it is made for.
<hno>
you will only cause problems if you try to.
rz2k has joined #linux-sunxi
<oliv3r>
unless you wanna 'test' something i suppose ;)
<slapin_>
bbrezillon: and mine, and mine... :)
<hno>
oliv3r, mtdblock is not even usitable for testing something.
<hno>
it's main use is to provide a block handle for jffs etc, which for some reason needs a block device handle even if only doing mtd transactions.
ganbold_ has joined #linux-sunxi
<hno>
the other puspose of mtdblock is to providee read-only support for simple block devices on primarily NOR flash. I.e. read-only access to a FAT filesystem stored on NOR flash.
<hno>
it can also do read-write but it's not meant to be used.
kivutar has quit [Read error: Operation timed out]
phoenix_ has quit [Ping timeout: 272 seconds]
fredy has quit [Excess Flood]
eebrah|mobile has joined #linux-sunxi
fredy has joined #linux-sunxi
hipboi has joined #linux-sunxi
<binaryferret>
Hi guys, I bought a bunch of Allwinner A13 tablets but got tablets that have Rockchip RK2928 socs. Are these just allwinners with a different name?
<oliv3r>
hno: you mean an older version of jffs, or even jffs2?
nove has joined #linux-sunxi
<hno>
binaryferret, that's just some seller who don't know a shit of what they are selling, copy-pasting specifications from something similar.
<binaryferret>
Cheers hno. Bummer. I was ready to start hacking.
<hno>
oliv3r, it's for mount to find what device to mount. Not sure which versions require it.
<balage>
arokux, it's about the old Hackberry A10 usb power issue
<arokux>
balage: yes, I've seen it. I was busy last time and on the weekend. I was going to take a look at it during this week. it was found what code causes the problem I believe, but there was no consensus as how to solve it as I remember.
<balage>
arokux, it's okay, I just wasn't sure you got it
<arokux>
balage: no problem, thanks for remembering.
<balage>
arokux, if there is anything I can do for you to help just send me an email
<arokux>
balage: ok
Black_Horseman has quit [Ping timeout: 248 seconds]
eebrah|mobile has quit [Read error: Connection reset by peer]
balage has quit [Quit: leaving]
eebrah|mobile has joined #linux-sunxi
bsdfox has quit [Ping timeout: 260 seconds]
hipboi_ has joined #linux-sunxi
hipboi has quit [Ping timeout: 245 seconds]
bsdfox has joined #linux-sunxi
paulk-collins has joined #linux-sunxi
<jinzo>
Hm, I have not been following closely lately - but how's the A23 performance?
<libv>
jinzo: i am not sure anyone tried it yet
<libv>
jinzo: supposed to clock a lot better, but i haven't bothered myself
<jinzo>
oh - I just encountered a quite-cheap A23 tablet and it looks interesting
<oliv3r>
jinzo: i think libv's tablet runs at a default clock of 1.5 Ghz
<libv>
jinzo: yes, they sell in the space where the a13 tablets sold
<oliv3r>
which should be atleast 50% faster then a20
<jinzo>
that's quite considerable then
<jinzo>
yeah, 41€ for a tablet isn't that bad actually :P
<libv>
oliv3r: that's the extreme speed, i think the normal cpufreq limit will be a bit lower, perhaps comparable to a10 again
<oliv3r>
i wonder how stable it runs at 'extreme'
<oliv3r>
if it can run full time at extreme, then fine; but i have my doubts
<oliv3r>
maybe a few seconds max
<jinzo>
also there's not much info about cedarx in A23
<oliv3r>
i would expect it to be very similar
<jinzo>
but I see it's working quite nice in A10
<jinzo>
even VP8? That's, nice.
eebrah|mobile has quit [Ping timeout: 252 seconds]
<jinzo>
and VDPAU work, I don't have enough time for this stuff :/
<oliv3r>
do we have vp8 yet?
<oliv3r>
cool
<oliv3r>
but i think they are working on encoders now
<oliv3r>
they masterd jpeg, mpeg and h264, so now focusing on those encoders as there's some expertise there now
<jinzo>
but I still need to get a phone before I can start fiddling with sunxi stuff. The day needs more hours...
<mnemoc>
it seems the trick is to not sleep
<mnemoc>
but the body tends to dislike that on the long run
<jinzo>
mnemoc, do the jiyagus (or however are they spelled) still work ok?
<mnemoc>
yes
<mnemoc>
both jiayu g3, and the jiayu g4 :p
<jinzo>
there're a bit more models on the market at the moment
<jinzo>
but I'm looking for something along the lines
<mnemoc>
the g3s is verly much like the g3
<mnemoc>
but slimmer and lighter
<mnemoc>
with better internals
<jinzo>
there's a g5 now too - but didn't have time to check it out properly
<jinzo>
there's also g3t if I'm not mistaken
<mnemoc>
nice
<jinzo>
And I'm always looking at the Lenovos (got one for my friend a while back, and it works quite good)
<mnemoc>
I'm going to replace my x121e (with dead SSD) with an e145
<mnemoc>
a4-5000/12GB/240GB Intel ssd
<oliv3r>
sounds nice
<oliv3r>
and expensive!
<mnemoc>
~700E
<mnemoc>
including the ssd
<oliv3r>
ouch
<jinzo>
a4 ? What do I know - I'm very disappointed with the A8 my gf has
<oliv3r>
can I change the cpufreq via the kernel commandline?
<oliv3r>
cpufreq.governor=performance makes uboot unhappy for some reason
<jinzo>
but that's the first generation of that A* if I'm not mistaken
<mnemoc>
oliv3r: imo expensive for a $work$ laptop starts at 1k
wingrime has joined #linux-sunxi
<jinzo>
on the other hand I presume you're running linux anyway (the windows drivers are unimaginable shit that I have no idea how they passed on to sell - but that was probably 2 years ago)
<jinzo>
also mnemoc now that you're in DE - you know of http://geizhals.de/ right?
_newleaves_ has joined #linux-sunxi
<oliv3r>
mnemoc: if your $work pays for it; it doesn't matter does it :p
<oliv3r>
i think my T42 had a new value of 2200 Euro
<oliv3r>
in 2004 obviously
pwhalen has joined #linux-sunxi
<oliv3r>
it's more or less like pricewatch.com
<oliv3r>
or tweakers.net/pricewatch :)
<mnemoc>
oliv3r: no, this is a personal personal
<mnemoc>
oliv3r: at work I got a thinkpad L530
<mnemoc>
oliv3r: but can't take it home :p
<oliv3r>
then why are they hadning out laptops?
<mnemoc>
instead of desktop, and to carry to meetings
<oliv3r>
i'll try to look at 3.10-script tonight; want to get libvdpau going on the CT today
<oliv3r>
meetings, *sigh* meetings are overrated
<mnemoc>
:)
<oliv3r>
people spend far to much time 'meeting' then actual doing 'work' (well around here anyway)
<jinzo>
One of the previous "startup" gigs I interviewed at had an interesting take on meetings
<mnemoc>
i'll be offline until friday, then... hopefully with the new laptop, resume the rebasing of 3.4 on top of 3.10
<jinzo>
mandatory every day meetings and you had to pay something if you were late/didn't show up
<jinzo>
ofcourse the "startup" was not paid/only equity.
<jinzo>
so I guess some people do love/do take seriously meetings
<mnemoc>
:)
<oliv3r>
i understand that ou need to meet to talk things over; but some people do more meetings in a day, then actual work
<ccaione>
meetings are for bosses who know nothing about actual work
<oliv3r>
thank you, that.
<oliv3r>
:p
<oliv3r>
ccaione: have you compiled libvdpau yet?
rz2k has quit []
<oliv3r>
oh crap; appearantly i have commit access to libvpau-sunxi and if you edit it via the webinterface, it doesn't generate a pull request by default
kivutar has joined #linux-sunxi
<arete74>
s
gzamboni_ has quit [Read error: Operation timed out]
eebrah|mobile has joined #linux-sunxi
gzamboni_ has joined #linux-sunxi
FreezingCold has quit [Ping timeout: 260 seconds]
FreezingCold has joined #linux-sunxi
_newleaves_ has quit [Ping timeout: 245 seconds]
eebrah|mobile has quit [Read error: Connection reset by peer]
jemk has joined #linux-sunxi
<oliv3r>
jemk: vdpau doesn't work with mpv does it?
<libv>
aha...
<libv>
so it prepares space in the page directory for the size it actually wants to allocate, and then goes off and tries to map the actual size of the buffer instead of the size it tries to allocate.
<oliv3r>
libv: the mali 2GB thing?
<libv>
the mali crashing thing :)
<jemk>
oliv3r: here it works with mpv better than with mplayer
<oliv3r>
jemk: good; the commandline the same?
<libv>
which is irrespective of the 2GB thing :)
<oliv3r>
jemk: because I get error opening/initializing the the selected video_out (-vo_ device :(
<jemk>
oliv3r: mpv --hwdec=vdpau --vo=vdpau or something like this
<jemk>
oliv3r: VDPAU_TRACE=1
nove has quit [Ping timeout: 252 seconds]
<oliv3r>
ah different option; ok
<oliv3r>
i copy/pasted the mplayer options from the wiki
<oliv3r>
er README
<oliv3r>
jemk: also i wanted to do a pull request via the github interface; but i messed up; so double check those commits if you will :(
<oliv3r>
sorry!
<jemk>
:D nice commits
<oliv3r>
so --vc is not needed right?
<oliv3r>
hehe, sorry! it was unintended
<jemk>
no. thats --hwdec
<oliv3r>
ah ok
wingrime has quit [Ping timeout: 252 seconds]
<jemk>
should i keep them or do a force push?
<oliv3r>
jemk: error opening the vo device :(
<oliv3r>
jemk: i haven't done them locally
<oliv3r>
well i edited the files etc; but if you like them, keept hem as it; it's ok
<oliv3r>
if i export the VDPAU_TRACE where does it end up?
<oliv3r>
and modprobe sun4i_cedar_mod is enough, right?
<jemk>
if they don't break things its ok, but it looks funny, three commits for two small things
<oliv3r>
it wasn't inteded to be 3 commits :(
<jemk>
you need r/w access to /dev/{cedar,disp}
<oliv3r>
you can squash it and reposih it; that was my intention
<oliv3r>
ah of course!
<jemk>
and g2d for the osd branch
<oliv3r>
i'll do it without osd for now;
<oliv3r>
jemk you are awesome
<oliv3r>
jemk mind if i add this info to the readme?
<oliv3r>
mpeg2 via vdpau of big buck bunny, 1080p 6600K short takes up 135 CPU though;
<jemk>
np, but wait till i squashed your commits ;)
<oliv3r>
i will
<oliv3r>
it IS hardware encoded though; as software encode runs at 105% cpu and is like 1 fps; wheras with vpau it's fluid; dunno the FPS though; haven't found the keybinding to print fps
<oliv3r>
hmm, strange though it runs smooth, but says 'mpeg_decode_postinit() failure, falling back to software decoding
<oliv3r>
h264 works though; great; i'll only do that anywa y:p
<jemk>
oliv3r: add --hwdec-codecs=all for mpeg
<oliv3r>
how confusing! --hwdec=vdpau for h264; --hwdec-codecs=all for mpeg
<jemk>
i hope nobody minds the forced push...
<oliv3r>
what if i do both options for either file? :)
<oliv3r>
nobody iwll notice!
<jemk>
both are ok
<jemk>
mpv changed the whole commandline and made a little mess in some points it seemd
<oliv3r>
i'll try the g2d 1 week before fosdem
<oliv3r>
but works; still 135% cpu usage on the _short mpeg2
<oliv3r>
but h264 runs great; 13.5 cpu usage of X; 9% for mpv; 8% for the term it runs in
<oliv3r>
and it might not even be on 'performance' governor
<jemk>
huh, h264 low cpu and mpeg high?
gzamboni_ has quit [Ping timeout: 265 seconds]
brain_ has joined #linux-sunxi
<jemk>
maybe the audio, if mpeg has 6ch ac3 it could be cpu intensive
<oliv3r>
mpeg started low, but quickly jumped
<oliv3r>
i use -ao null
<oliv3r>
jemk: anyway, thanks you helped me out a lot; this shoulud be amazing for next month
<jemk>
and a lot of cpu goes into the XClearWindow call i do each frame... But i did not find a way to avoid this
<oliv3r>
for mpeg2?
<jemk>
always
<oliv3r>
well mpeg2 runs extremly slow with software dec, but with hardware dec it runs smooth
<oliv3r>
so it must be enabled
<oliv3r>
just cpu load is high (on a20)
<oliv3r>
what's the longest you let it loop for?
<oliv3r>
besides BBB at the highest encoding, anything else worth while to show? sintel i keep hearing you about, is that more intens (and do people know this?)
<jemk>
i watched whole films with it already
<oliv3r>
after fosdem (or if i ffind meanwhile but I doubt that) i'll see if i can see what's the mpeg2 issue
<brain_>
after more than 3 months...closer to 5 really. i FINALLY got my own custom built Focaltech driver working! screw the damn OEM driver with aLL the hardcoded settings which made the tablet useless after a currupted eeprom. I am now working to correct the axis and get everything calibrated. WHAT A DAMN NIGHTMARE this has been but god have i picked up some serious driver coding techniques along the way.
<oliv3r>
the most important mpeg2 data it has to play back well though; imo is sdtv; e.g. dvb-t and dvb-s2 (and dvb-c) broadcasts
<oliv3r>
h264 appearnalty allready works
<brain_>
THANK you to all those who have helped me over the past few months, I know i only pop in once in a while when i run out of ideas.
<jemk>
sintel uses more features the (older?) armhf blobs were not able to decode
<oliv3r>
brain_: we eagerly await a beatiful focaltech driver
<oliv3r>
jemk: the older allwinner supplied drivers you mean; what about cedrus? Can it handle it?
<jemk>
oliv3r: vdpau can decode it
<oliv3r>
smoothly?
<brain_>
I plan to release everything on my github and i'll come here and let you guy decide were to put it, I also wrote and EEPROM flasher, Reader, and Blanker utility which is standalone and accepts .bin or .intel hex files
<jemk>
yes
<brain_>
I need some rest, gonna do a serious code cleanup and get this all going, feel great
<oliv3r>
brain_: to upload firmwares into the mCU?
<oliv3r>
brain_: amazing; awesome dude :)
<brain_>
yes, and also read them out
<oliv3r>
brain_: firmware could be handled via the default firmware stuff int he kernel though? via /lib/fimrware loading?
<oliv3r>
brain_: this will be a hugely annoying venture, getting firmwares from touchscreens, hosting them, finding out which is good, which is bad; should get RE-ed eventually
<oliv3r>
brain_: but rest now
<brain_>
well yes it could, but like i said my system is kinda oddball so I had to make a utility to quickly flash things
<oliv3r>
jemk: so cedrus is allready better then cedarX; amazing; will use that
<jemk>
oliv3r: the android blobs from allwinner are even better, but don't tell them ;)
<oliv3r>
then ours?
<oliv3r>
crap :p
<oliv3r>
well; give it time!
<oliv3r>
i assigned 64mb to cedarx; will that be enough?
<jemk>
the most obvious thing we miss is decoding of interlaced h264, the other parts should be pretty complete
gzamboni_ has joined #linux-sunxi
<oliv3r>
ugh; i hate interlaced content, but for broadcasting stuff; most is interlaced
<oliv3r>
so what happens if you feed it interlaced content?
<jemk>
for normal hd it should be enough
<jemk>
in the meantime you get an error, in the beginning it overwrote your kernel xD
<oliv3r>
O.o
<jemk>
in memory of course
<libv>
aha
<libv>
they are shipping a23 tablets with hdmi
<oliv3r>
i'll offer it a default of 128mb then just to be safe
<oliv3r>
libv: how is that done? i thought a13 didn't have hdmi; if a23 DOES have hdmi; that makes it a quite interesting chip
<jemk>
80mb is the allwinner default
<oliv3r>
i wonder if it'll have emac
<libv>
so i got twice unlucky with my supposed Q88 tablet
<oliv3r>
if it has hdmi + gmac; then it is a nice chip for stb's too
<libv>
yeah
<libv>
but i guess all the capacity is currently geared at q88 replacements
<oliv3r>
all assumptions are out the door for now though; we thought it'd be a dual a13
<oliv3r>
but a13 didn't have hdmi, so that won't fly
<oliv3r>
it's not a gul-wing chip anyway
<oliv3r>
so there are potentially enough pins
<libv>
oliv3r: but why take the bga form factor then?
<libv>
if it were just a dual core version of a13
<oliv3r>
more pins
<libv>
if it really were a dual core a13, then that wouldn't be needed
<oliv3r>
well most their production lines might be on bga, so it could be 'easier'
<libv>
anyway, perhaps tsvetan will make us all happy
<oliv3r>
we do have the sdk; but doubt we can find anything usefull
<oliv3r>
since drivers != hardware :)
<libv>
right
<oliv3r>
espcially with their newly found idea of unifying it all
<oliv3r>
but it's very interestignt o say the least
<oliv3r>
shit hometime, and sintel needs 4 more minutes to download :S
<oliv3r>
anyway i got cedar running with zero blobs :p
<libv>
cool :)
<libv>
i am doing yet another sun7i_defconfig build...
<libv>
hence me having a dig around aliexpress
<atiti>
anyone has a good vdpau example? (other than the mplayer source code)
e-ndy has quit [Remote host closed the connection]
<ccaione>
oliv3r: never compiled
nikrou has joined #linux-sunxi
shineworld has joined #linux-sunxi
<nikrou>
I try to register on wiki but my ip seems to be registered as an open proxy !
nikrou has quit [Client Quit]
t3st3r has quit [Remote host closed the connection]
nieuwbie has joined #linux-sunxi
<nieuwbie>
hey my uart stopped to work
t3st3r has joined #linux-sunxi
<nieuwbie>
everything seams to be alright with the chip (I tested it by touching rx with ground and got an output - correct me if Im wrong here)
<nieuwbie>
by stopped to work I mean it gives no output.
fredy has quit [Excess Flood]
fredy has joined #linux-sunxi
deasy has joined #linux-sunxi
rz2k has joined #linux-sunxi
atiti has quit [Ping timeout: 260 seconds]
hipboi_ has quit [Remote host closed the connection]
rz2k has quit []
<libv>
hrm... fixed that bug
<libv>
and kind of figured out why it was triggered.
<libv>
for some reason, i am being told to render to the actual FB instead of to a buffer
Fusing has joined #linux-sunxi
<hno>
oliv3r, jfffs uses mtdblock for mount to find what device to mount. Not sure which versions require it.
<brain_>
Thank you Turl, only thing left to do is get some sleep and tidy up the code, oh and make MANY backups after all this time!
Fusing has quit [Ping timeout: 276 seconds]
Fusing has joined #linux-sunxi
deasy has quit [Ping timeout: 272 seconds]
FreezingCold has quit [Ping timeout: 248 seconds]
everseened has joined #linux-sunxi
e-ndy has joined #linux-sunxi
Quarx has quit []
deasy has joined #linux-sunxi
sspiff has quit [Quit: Konversation terminated!]
deasy has quit [Quit: Nom d'un quark, c'est Edmonton !]
Fusing has quit [Ping timeout: 252 seconds]
agath_at_work has quit [Remote host closed the connection]
agath_at_work has joined #linux-sunxi
Seppoz_work has joined #linux-sunxi
enrico_ has quit [Quit: Bye]
FreezingCold has joined #linux-sunxi
FreezingCold has quit [Ping timeout: 265 seconds]
jemk has quit [Ping timeout: 245 seconds]
Fusing has joined #linux-sunxi
popolon has quit [Quit: Quitte]
adb_ has joined #linux-sunxi
_massi_ has quit [Quit: Leaving]
FreezingCold has joined #linux-sunxi
<nieuwbie>
so nobody had similiar problem with UART?
bbrezillon has quit [Quit: Ex-Chat]
<brain_>
Is anyone aware of the Focaltech source code thats floating around that has a "Scaling hack" added to the driver? I was using it a while back and seem to have lost it over the past few months, tried searching my history but nothing is coming up. I remember it had many options for adjusting the demensions in the code. I'd like to put those changes into the driver im working with.
popolon has joined #linux-sunxi
<brain_>
I wish I could remember who added the hack :( it was in the header I believe.
<brain_>
only time my uart on the tablet gives me issues is if I pull on the port too much and break a lead, usually the ground in my case.
panda84kde has quit [Quit: Konversation terminated!]
<nieuwbie>
brain_: well I use jump wires and double side tape so maybe thats the issue.
<hramrach>
binaryferret: #linux-rockchip for rk2928
VargaD_ is now known as VargaD
<hramrach>
they thend to be eaven cheaper than a13
FreezingCold has quit [Ping timeout: 248 seconds]
<hramrach>
you can boot from uSD but only if you like short some pins on the nand so the SoC is unable to read the bootloader from there
<hramrach>
you should probably try to look over there for somebody who hacked on an actual rk2928
<nieuwbie>
.
<nieuwbie>
hack+0101
<nieuwbie>
are you there?
jemk has joined #linux-sunxi
<nieuwbie>
seriously I have no idea why this uart doesnt work.
everseened has quit [Quit: Page closed]
Fusing has quit [Ping timeout: 246 seconds]
<ccaione>
how do you know that is the uart that is not working and not the entire device
nikrou has joined #linux-sunxi
nikrou1 has joined #linux-sunxi
fredy has quit [Excess Flood]
nikrou has quit [Ping timeout: 276 seconds]
fredy has joined #linux-sunxi
<nieuwbie>
ccaione: well first of all I got an output when I touched rx with grn
<nieuwbie>
gnd
<nieuwbie>
second of all its working
<nieuwbie>
I connected rx - rx and tx - tx and gnd - gnd
<nieuwbie>
and should be rx tx tx rx gnd gnd
<nieuwbie>
now another problem I have Polaroid tablet MIDC
<nieuwbie>
and trying to fix bootloop.so I berrybooted and it doesnt power on at al
<nieuwbie>
*all
<nieuwbie>
and I get nothing from my uart at all
<nieuwbie>
while when sd is out I get everything.
FR^2 has joined #linux-sunxi
ssvb has joined #linux-sunxi
<nikrou1>
Hi all
<nikrou1>
I open my device and I want to take usefull picture
<ssvb>
libv: is the mali issue fully resolved now?
bsdfox has quit [Ping timeout: 252 seconds]
netlynx has joined #linux-sunxi
netlynx has joined #linux-sunxi
netlynx has quit [Changing host]
adb_ has quit [Remote host closed the connection]
bsdfox has joined #linux-sunxi
shineworld has left #linux-sunxi ["Leaving"]
pfdm has joined #linux-sunxi
Seppoz_work has quit []
deasy has joined #linux-sunxi
eebrah|mobile has joined #linux-sunxi
<nieuwbie>
damn why is so difficult to boot os from sdcard just to fix bootloop. :/
nikrou1 is now known as nikrou
shineworld has joined #linux-sunxi
eebrah|mobile has quit [Read error: Connection reset by peer]
enrico_ has joined #linux-sunxi
notmart has quit [Quit: notmart terminated!]
_newleaves_ has joined #linux-sunxi
_newleaves_ has quit [Ping timeout: 252 seconds]
foubarre has joined #linux-sunxi
kivutar has quit [Ping timeout: 245 seconds]
netlynx has quit [Remote host closed the connection]
netlynx has joined #linux-sunxi
netlynx has quit [Changing host]
netlynx has joined #linux-sunxi
jemk has quit [Remote host closed the connection]
shineworld has quit [Quit: Leaving]
<Turl>
nieuwbie: did you get the right berryboot?
nikrou has quit [Ping timeout: 248 seconds]
netlynx has quit [Quit: Leaving]
foubarre has quit []
<nieuwbie>
Turl: No idea. Was able to write the 'wheezy' on sdand booted it via uart.
<nieuwbie>
my finger hurts from pushing 'y' in fsck. :/
<Nyuutwo>
how it is suspend supported? When I use halt command, it hangs after killing CPU and I have to press long power button to power off
<oliv3r>
Nyuutwo: that requires axp improvements
<oliv3r>
Nyuutwo: the SoC can't actually power off
<oliv3r>
android actually reboots the soc
<oliv3r>
sets a bit in the env partition; which tells boot.axp to start the 'charge app'
<Nyuutwo>
and this charging app really can power off
<oliv3r>
the axp IS capable of shuting down everything; but i thinkt he driver misses stuff (i do recall it working on an old fedora image on my tablet)
<oliv3r>
i think after a while it does yea; when battery is full
<Nyuutwo>
so we need some kind of axp support in u-boot?
<oliv3r>
not u-boot; the kernel itself
<oliv3r>
after it halts the soc; the axp needs to shut down the power
<ccaione>
oliv3r: I've started working on regulators today
* ssvb
got A10-OLinuXino-LIME today
<Nyuutwo>
I have strange feeling that connecting wires to i2c will help quicker rev engi than reading aw-boot conde
<Nyuutwo>
does anybody knows speed of i2c on axp209?
Black_Horseman has joined #linux-sunxi
<oliv3r>
ssvb: you pre-ordered one?
<oliv3r>
ssvb: how's the memory performance?
<ssvb>
oliv3r: Tsvetan actually sent a free one to me, also with a case included :)
<oliv3r>
pff
<oliv3r>
*shakes fist*
<oliv3r>
wasn't aware he was doing that with the lime :(
<oliv3r>
no LIME for demo then!
<oliv3r>
heh he's not even here
<ssvb>
oliv3r: you will probably have a chance to rob him at FOSDEM :)
<oliv3r>
:p
<ssvb>
but LIME has no VGA, would it be still useful for demo?
<oliv3r>
no :p
<oliv3r>
no VGA on the pin headers either?
<ssvb>
hmm, that could possibly work
<oliv3r>
*shakes fist*!
<oliv3r>
i'm quite satisfied with my demo setup for now
<oliv3r>
sintel + BBB run smooth with 35% cpu load
<oliv3r>
give or take; boots from SD, but runs from ssd nicely
<ssvb>
we still have a chance to fix this
<oliv3r>
oh! i tried to use the performance gov by default
<oliv3r>
but cpufreq.governor=performance gives u-boot errors; shit about use bla bla to set the env
<ssvb>
hmm, how so?
<oliv3r>
don't understand why, as we pass the resolution the same way, disp.screen_0=1024x786
<ssvb>
did you use quotes?
<oliv3r>
so why shouldn't I be able to add the governor, also u-boot doesn't know if it's a valid comandline arugment or not