hno changed the topic of #linux-sunxi to: Allwinner/sunxi development discussion - Don't ask to ask. Just ask! - See | | Logs at
BJfreeman has quit [Ping timeout: 251 seconds]
leowt has joined #linux-sunxi
leowt has quit [Quit: leowt]
\\Mr_C\\ has quit []
\\Mr_C\\ has joined #linux-sunxi
<drachensun> hramrach: thanks, minor patch, needs a , at the end of the dbg string
<drachensun> but it compiles
<drachensun> well wait, what is the supposed to do
<drachensun> now I have a nandf
<drachensun> ok, so everything is bumped back one, nandd is now now nande
<drachensun> and I guess nanda is really nand? the initrd to boot is a little delicate so it will take some work to test but it looks like it might work
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
piyushverma has quit [Quit: Konversation terminated!]
<drachensun> well no joy but I think I noticed what I was missing on the image creation side to make big partitions
<drachensun> you have to take their number and divide by 2 to see the real size, so I thought I had made the partition big enough but in fact, it was half what it should be
<drachensun> oh if I could have read the comments in Chinese
piyushverma has joined #linux-sunxi
TheSeven has quit [Disconnected by services]
[7] has joined #linux-sunxi
piyushverma has quit [Ping timeout: 255 seconds]
Soru has quit [Quit: No Ping reply in 180 seconds.]
Soru has joined #linux-sunxi
piyushverma has joined #linux-sunxi
cajg_ has joined #linux-sunxi
cajg has quit [Ping timeout: 264 seconds]
BJfreeman has quit [Quit: had a good time]
Quarx has joined #linux-sunxi
Soru has quit [Remote host closed the connection]
wingrime has joined #linux-sunxi
rz2k has joined #linux-sunxi
<hramrach_> drachensun: it's supposed to make the full nand partition as first
<hramrach_> but needs some tuning to make its name nand instead of shifting the names
<hramrach_> or just add it to end for testing
ganbold_ has joined #linux-sunxi
<hramrach_> drachensun: updated:
<hramrach_> should fix the naming as well
piyushverma has quit [Read error: Operation timed out]
piyushverma has joined #linux-sunxi
<rz2k> we have a20 nand sources or we launching sun[45]i nand on sun7i?
* rz2k bit out of the loop
<hramrach_> this is a patch that touches the nandlib glue
ganbold_ has quit [Remote host closed the connection]
Tsvetan has joined #linux-sunxi
<Tsvetan> just to let you know mnemoc is released from the hospital and geting better, got e-mail from him this morning
<hno> drachensun, right. It needs a small patch to expose the whole nand block device including partition table.
rellla has joined #linux-sunxi
arete74_ has quit [Ping timeout: 256 seconds]
<hramrach_> hno: it works in sun4i but not sun7i
<hramrach_> because mbr2disks is part of nandlib there
<hno> well,, nandlib is just an ar archive.
<hno> easy to substitute objects, and would expect things at this level to be the same. But check function signatures with gdb if unsure.
piyushverma has quit [Ping timeout: 256 seconds]
<naobsd> which wiki page is good for A20 sources
<naobsd> for pointing sources for A20 boards
<naobsd> such as hno's u-boot
<naobsd> for now I added link to Hans de Goede's repo on my site, but probably it's not good place ;)
<rz2k> naobsd: may be ?
_whitelogger has joined #linux-sunxi
cajg_ is now known as cajg
rz2k has quit [Read error: Connection reset by peer]
sanka has joined #linux-sunxi
piyushverma has joined #linux-sunxi
\\Mr_C\\ has quit []
<hramrach_> where is script.bin with boota?
<hramrach_> hmm, the kernel hangs at aw_clkevt_init: sun7i_timer0_clockevent mult 6442450, max_delta_ns 170000, min_delta_ns 1000, cpumask 0xc04607e0, irq 54
<wingrime> hramrach: can you make jpeg deocder using my description?
piyushverma has quit [Ping timeout: 256 seconds]
piyushverma has joined #linux-sunxi
<wingrime> nove: ping
<oliv3r> Tsvetan: thanks for sharing; very happy to hear
\\Mr_C\\ has joined #linux-sunxi
<oliv3r> says all there is to be said about uSD booting?
<oliv3r> hah my olimex SPL runs on cubie2 :D
<oliv3r> memory timings probably aren't that different
<ssvb> oliv3r: memory timings are quite funny, these look like the sets of magic numbers for all the same memory controller
<wingrime> ssvb: please check , can you consider this is enougt for jpeg PoC ?
<wingrime> oliv3r: pcb related
<wingrime> oliv3r: timings may be recived using some check tool, that find maximum workable freqency
_BJFreeman has joined #linux-sunxi
<ssvb> oliv3r: the cubieboard2 memory timings use cas 9, the original cubieboard memory timings use cas 6
_BJFreeman is now known as BJfreeman
<ssvb> oliv3r: the settings from cubieboard1 also seem to work with cubieboard2, and reduce memory latency as expected from lower cas
<oliv3r> hno: more spam (resend of a previous on a20 tree; you can ignore the previous one, this one does it on sc)
<oliv3r> wingrime: livesuit may do this yea, but we still don't know how timings are derived :)
<oliv3r> i thought we had jpeg PoC :D
<wingrime> ?
<oliv3r> didn't nov submit one
<oliv3r> pretty sure he did
<wingrime> who
<oliv3r> nove*
<oliv3r> ssvb: it would be awesome if we could find out more about the DRAM controller; how to set it up, how to over/under 'clock' it with the cas settings etc
<oliv3r> but there's much happening that's currently unknown
<ssvb> oliv3r: maybe Tom needs to ping the right people, so that they could provide the right settings
<ssvb> oliv3r: CAS6 is probably related to DDR3-800, CAS9 is probably related to DDR3-1333
<ssvb> oliv3r: but what we probably really want are the timings for DDR3-1066, because it is closer to 480MHz
<oliv3r> well in theory there's really just 2 factors
<oliv3r> DRAM chips (which have certain timing requirements
<oliv3r> board layout which reduces those timings
<oliv3r> what would want to know for starters, is if the dram controller is Allwinner IP; or bought IP
<oliv3r> and i think it's not AW IP
<oliv3r> if it is, we really would want docs :)
<mripard_> $
<mripard_> hmmm, wrong window :)
<oliv3r> mripard_: hah! and there you go never answering my after your accusations! shame Shame I say!
<ssvb> mripard_: coincidentally, $ could probably solve this problem too :)
<mripard_> oliv3r: ?
<mripard_> ssvb: and it happens to be the "sync" shortcut in mutt as well :)
<mripard_> oh yeah, I forgot about this one
<mripard_> nah, I asked about that because marvell has a mbus driver currently being posted to the mailing lists
<mripard_> so I thought you just came across this patch, that's all
<wingrime> oliv3r: oliv3r: usb can be used from mdev with small patch currently?
<mripard_> I'm not accusing you of anything :)
<oliv3r> mripard_: yes, yes you are! you said i was looking at the marvel i2c
<mripard_> no
<oliv3r> and then ... just dissapear
<mripard_> I asked if you were looking at "marvell patches"
<oliv3r> lol
<oliv3r> see!
<oliv3r> acusation!
<oliv3r> anyway
<mripard_> not "*my* marvell patches" :)
<oliv3r> :p
<oliv3r> what is an mbus controller?
<oliv3r> a10 manual had it all cut out
<oliv3r> a13 manual left it in tact
<oliv3r> and more interestingly, dram controller sets up the clock for (gps?!) and mbus
<oliv3r> so was curious what this mbus is and why the dram controller could possibly ever need it
<oliv3r> wingrime: i have no clue :(
<mripard_> in marvell's case, it's an IP that configures the physical memory layout
<oliv3r> Memory Bus
<oliv3r> i doubt it's that
<oliv3r> well the thing is
<mripard_> so that you can decide that say the uart0 is at a given offset instead of another one
<mripard_> in allwinner case, I have no idea
<oliv3r> in the register list, the mbus is past everything, even after the mali GPU
<oliv3r> but i don't see anything being changed in the mbus registers, only the clock being setup
<oliv3r> but its interesting still, it could be a memory bus controller of some sorts
<oliv3r> so it might actually be important
<oliv3r> but GPS, really?
<ssvb> wingrime: thanks, that's a nice update to the cedarx wiki
<ssvb> wingrime: but first I want to find out what is the current status of cedarx support for a20 (with the binary blobs)
<oliv3r> wingrime: i'll make tables monday
<mripard_> oliv3r: I don't know really
<ssvb> g2d hardware in a20 seems to be the same as in a10, which means no premultiplied alpha support :(
<wingrime> ssvb: I still not have any a20 hw
paulk-desktop has joined #linux-sunxi
<ssvb> wingrime: have you filed a web form request for cubieboard2?
jemk has joined #linux-sunxi
<wingrime> ssvb: yes
<wingrime> ssvb: but russian mail is slowpocke
<oliv3r> ssvb: he lives in russia, while he is the closest to tom physically; russian mail takes the longest it is said :)
<wingrime> thay have insane logistics
<oliv3r> packages goes to moscow first
<oliv3r> where KGB investigate
<oliv3r> then its sent to USA
<oliv3r> where CIA investigates
<oliv3r> then it's sent to china, to make sure the russiand and the americans haven't messed with it
<oliv3r> and then, only then it's deliverd to wingrime
<ssvb> unless stolen by a postman I guess
<oliv3r> well mine arrived yesterday only
<oliv3r> and i was 2nd or so on the list! :p
<wingrime> ssvb: I add WMV9 reg short namse
<oliv3r> oh, have to look at for a20
<wingrime> ssvb: and can do same trick wirh VP8 but I still have no idea witch engine handle VP8
<wingrime> ssvb: how much VP8 close with H264 ?
<oliv3r> hmm, the chip ID field is really useless and bogus; since A20 it is reset to 'chip id A' again
<wingrime> olvi3r: I think we need CRC32 for BROM
<oliv3r> how will that help identifying the chip revision
<oliv3r> sun4i has version A, B and C
<oliv3r> i think there was a bug in the memory controller, some lines 'crossed'
<oliv3r> they fixed it in code
<naobsd> mmm, make cubieboard_fel_config and make cubieboard2_fel_config generate 10K+ bytes of u-boot-spl.bin :(
<wingrime> * VP8 Engine
<naobsd> if my understanding is correct, it must be smaller than 8K for fel write 0x2000 u-boot-spl.bin
<ssvb> about the Cortex-A7 clock frequency in Allwinner A20, my cubieboard2 can only be clocked up to 1.1GHz (1.2GHz is unstable)
<ssvb> and at 1GHz it seems to only need 1.3V (lower than the default 1.4V settings for the default 912MHz)
<jemk> wingrime: VP8 is decoded by h264 engine
<wingrime> jemk: cool, than we will have reg names
<wingrime> jemk: I tested you PoC
<wingrime> jemk: it worked , but not always
<wingrime> 1/3 cases for me
<wingrime> jemK: see updated page
<jemk> wingrime: did it show any errors or just didnt work?
<wingrime> jemk: no error messages, output files half broken
<wingrime> only first block lines deceded right way
<jemk> wingrime: thanks, i'll check that
BJfreeman has quit [Ping timeout: 252 seconds]
<wingrime> I think this is - reset IQ table register
hglm has joined #linux-sunxi
ganbold_ has joined #linux-sunxi
Soru has joined #linux-sunxi
jemk has quit [Ping timeout: 246 seconds]
<naobsd> hno: how to make u-boot-spl.bin smaller than 8k?
<naobsd> for fel
ganbold_ has quit [Remote host closed the connection]
ganbold_ has joined #linux-sunxi
<hglm> I noticed there is room for improvement in the various memcpy implementations for linux-sunxi -- both in userland and in the kernel, although I guess userland (glibc) is distro-dependent.
jemk has joined #linux-sunxi
<jemk> wingrime: you were faster, wiki edit collision xD
<wingrime> d))
<wingrime> jemk: I have added some info according vp8
<jemk> i see, but i only work on mpeg engine
<wingrime> jemk: I know
<wingrime> jemk: I now add data that I can easy get from blob
<jemk> poc is fixed btw
<wingrime> jemk: how we will do mpeg now ?
<jemk> im working on mpeg1+2, but first i need a way to display without this ppm crap
<wingrime> jemk: disp can do it directly
<jemk> i know, but how?
<wingrime> jemk: ssvb know disp better than me
<wingrime> jemk: I think how make jpeg decoding in-place
<wingrime> jemk: have you tryed this?
<jemk> what do you mean with in-place?
<wingrime> without memcopy to reserved mem
<jemk> one could read the file directly to reserved mem instead of mmaping it.
<jemk> but i dont think this is worth the effort, it wont gain much performance.
<jemk> compressed data is small compared to resulting picture
<wingrime> jemk: 2/3
<wingrime> jemk: one file still broken
<jemk> uhm, where do you get your files, mine worked xD
leowt has joined #linux-sunxi
wingrime-a13 has joined #linux-sunxi
<wingrime> jemk: try this
<wingrime> also do normal reset and print cedar version
<wingrime> ioctl(fd, IOCTL_SET_VE_FREQ, 160); <- 160 - are feqency in MHz
<wingrime> but I have 180 for me
<wingrime> in some trace
<hramrach_> naobsd: thanks. will try the script.bin hack
Soru has quit [Read error: Connection reset by peer]
Soru has joined #linux-sunxi
eebrah is now known as eebrah|away
_whitelogger has joined #linux-sunxi
leowt has quit [Quit: leowt]
Soru has quit [Quit: No Ping reply in 180 seconds.]
soul has joined #linux-sunxi
<jemk> wingrime: fixed, decoding works correctly, but my crap output function was wrong
<jemk> wingrime: haven't looked at freq yet, but kernel source says something about 160, 240 and 320
<wingrime> it in Mhz
<wingrime> jemk: ther any thing where I can help you with regs
<wingrime> ?
lioka has quit [Quit: leaving]
eebrah|away is now known as eebrah
cajg has quit [Ping timeout: 268 seconds]
cajg has joined #linux-sunxi
wingrime-a13 has quit [Quit: Lost terminal]
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
rellla has quit [Read error: Connection timed out]
rellla has joined #linux-sunxi
Dreadlish has quit [Changing host]
Dreadlish has joined #linux-sunxi
wingrime-a13 has joined #linux-sunxi
<wingrime-a13> ssvb: have problem with your driver on current sid
<wingrime-a13> ssvb: Couid not open module "sunxifb" in Xorg.log
Undertasker has joined #linux-sunxi
<hglm> wingrime: do you rerun autoreconf -vi and ./configure --prefix=/usr?
<wingrime-a13> very likly
<wingrime-a13> now try again
<wingrime-a13> but withot /usr
<wingrime-a13> *without user
<wingrime-a13> ok just --prefix was realy required
<hglm> OK. Did you notice faster window dragging with new driver in LXDE?
wingrime-a13 has quit [Quit: Lost terminal]
<wingrime> hglm: I use icewm
<wingrime> hglm: and can't find fast browser
<hramrach_> chrome? midori?
<hglm> Midori by default uses a diskcache in the home directory which is very slow on SDCARD.
<hglm> But you can move it to a ramdisk which makes Midori a lot faster.
<wingrime> hramrach: we now have jpeg poc
<wingrime> what we should do next
<hramrach_> drachensun: I built the patched kernel and it appears the code with nand_blk_register is not used
<hramrach_> also the kernel fails to boot because it cannot find root device
<wingrime> hraamrach_: ?
<drachensun> yeah, I had the boot issue as well
<drachensun> so when I reversed the order, the init code was happy again
<drachensun> but still the nand partitioner threw the same error on the whole nand device
<drachensun> however in looking at the fdisk output
<drachensun> I finally had by eureka moment about the partition size problem with dragon packing
<drachensun> and so now I can create the partition of the size I need
<drachensun> it should be good to fix in the future, but its not crucial, I can now get a whole X windows system on at flash
<hramrach_> is it possible that the nand driver is a module?
<hramrach_> will need to make a ramdisk then
<wingrime> jemk: still there?
<hramrach_> wingrime: as for jpg I won't have time for looking into that anytime soon
<wingrime> hraamrach_: it work
<hramrach_> good
<wingrime> hraamrach_: but what next
ganbold__ has joined #linux-sunxi
<hramrach_> drachensun: did you manage to build the nand module?
<hramrach_> looks liek the makefile is broken
ganbold_ has quit [Ping timeout: 256 seconds]
<drachensun> back
<drachensun> yeah, the nand driver is a module
<drachensun> I built it with the changes you mentioned, but like you said I couldn't add hno's suggestion as the mbr parts are in the library
<drachensun> and yeah, it uses a ramdisk to boot
<drachensun> which I think is actually going to be convienent since doesn't want to boot from an SD card
<wingrime> jemk: ping
<hramrach_> drachensun: this one might actually work
<hramrach_> but did not manage to boot custom kernel yet
<hramrach_> will need to build a SD card I guess because once I touch the initranfs in nand it might be hard to undo
<drachensun> I'm not sure, whats the goal of this patch? it touchs a bunch of mali pieces, the A31 doesn't have mali
<hramrach_> just ignore that part
<hramrach_> the makefiles are bogus there so had to patch them to try and build the module
<drachensun> ok, so thats only on your setup
<jemk> wingrime: pong
Undertasker has quit [Read error: Connection reset by peer]
Undertasker has joined #linux-sunxi
<wingrime> jemk: I just add manual for mpeg
<wingrime> jemk: but it in very early
<wingrime> can you comarea with your traces
rz2k has joined #linux-sunxi
<jemk> wingrime: i only did mpeg1+2 but it looks similar
jemk has quit [Remote host closed the connection]
Undertasker has quit [Read error: Connection reset by peer]
Undertasker has joined #linux-sunxi
Quarx has quit []
foubarre has joined #linux-sunxi
<foubarre> hello. Are there working images for cubieboard A20 ? (mainly interested in I2C being active in this configururation)
<ssvb> foubarre: right now it's more like "build it yourself" situation, some instructions are available here -
<foubarre> aw. Ok, thank you
eebrah is now known as eebrah|away
<rm> just get an A10
<foubarre> rm: yes, however i am interested in testing specificities of the A20 and buying both cards is not something i can afford..
<hno> naobsd, see boards.cfg
Superpelican has joined #linux-sunxi
<Superpelican> Hello
<Superpelican> May I ask if there is already support for the sun4i_crane (inet crane) in the sunxi 3.0 source tree?
<hno> Superpelican, I don't know sun4i_crane as sush, but we have full ethernet support in all sunxi kernels.
<rm> what is "crane" or "inet crane" anyways
<Superpelican> well it seems to be the board that is in my tablet
<rm> ...wrong answer :p
<hno> I think crane is the name used for one of the dev boards.
<Superpelican> it's also the name of a TI OMAP based dev board
<Superpelican> my tablet is a rebranded version of the POV Protab 2 XXL
<Superpelican> The Android System Info app said my board was a inet crane
<rm> I am not sure what this means, but afaik it's certainly not a board name from some particular tablet
<rm> more like codename of "sunxi-with-android"
<Superpelican> well the page I linked to says build product: crane-evb
<Superpelican> I just booted up my tablet
<Superpelican> will check
<Superpelican> The "Android System Info" app lists the information under "BuildInfos"
<Superpelican> "Board: crane"
<Superpelican> "Brand: iNet"
<Superpelican> "Device: crane-inet"
<Superpelican> "Model: PC1007"
<Superpelican> "Product: crane_inet"
soul has quit [Read error: Connection reset by peer]
<Superpelican> rm, hno:But don't I need to use a sys_config for my specific board/device?
<Superpelican> rm, hno:I do have a script.bin of my tablet, which I extracted using the adb shell
<Superpelican> rm, hno:So if I can just use that with the linux-sunxi kernel
Superpelican has quit [Quit: Konversation terminated!]
Soru has joined #linux-sunxi
<hglm> Superpelican: Yes, using the script.bin from Android will help make the LCD screen work, apart from HDMI
ganbold__ has quit [Remote host closed the connection]
ganbold_ has joined #linux-sunxi
eebrah|away is now known as eebrah
Soru has quit [Read error: Connection reset by peer]
Soru has joined #linux-sunxi
<hno> Superpelican left...
<hno> but product info from most Allwinner customers is very often unchanged from the evaulation board sources they started from.
<hglm> yeah most 1GB devices have the same DRAM parameters, but on a tablet the LCD screen will be different so you need the correct script.bin to get the LCD working.
leowt has joined #linux-sunxi
BJfreeman has quit [Ping timeout: 252 seconds]
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
wingrime has quit [Read error: Connection reset by peer]
wingrime has joined #linux-sunxi
rellla has quit [Remote host closed the connection]
leowt has quit [Quit: leowt]
_BJFreeman has joined #linux-sunxi
BJfreeman is now known as Guest7102
_BJFreeman is now known as BJfreeman
Guest7102 has quit [Ping timeout: 252 seconds]
servili007 has joined #linux-sunxi
eebrah is now known as eebrah|away
rz2k has quit []
wingrime has quit [Read error: Connection reset by peer]
wingrime has joined #linux-sunxi
hglm has quit [Quit: leaving]
Undertasker has quit [Quit: Leaving.]
gzamboni has quit [Remote host closed the connection]
gzamboni has joined #linux-sunxi
heffer has quit [Ping timeout: 256 seconds]
<hramrach_> and most out of factory script.bin are bogus :/
leowt has joined #linux-sunxi
leowt has left #linux-sunxi [#linux-sunxi]
<ssvb> does anyone know if A20/A31 devices have any temperature sensor for throttling?
paulk-desktop has quit [Quit: Ex-Chat]
BJfreeman has quit [Ping timeout: 252 seconds]
wingrime has quit [Ping timeout: 248 seconds]
leowt has joined #linux-sunxi
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
_BJFreeman has joined #linux-sunxi
BJfreeman is now known as Guest6510
_BJFreeman is now known as BJfreeman
Guest6510 has quit [Ping timeout: 252 seconds]
Soru has quit [Read error: Operation timed out]
derethor has quit [Remote host closed the connection]
Soru has joined #linux-sunxi
sanka has quit [Remote host closed the connection]