hno changed the topic of #linux-sunxi to: /Allwinner/sunxi development discussion - Don't ask to ask. Just ask and wait! - See | | Logs at
fluo75 has quit [Read error: Connection reset by peer]
fluo75 has joined #linux-sunxi
naobsd has joined #linux-sunxi
fluo75 has quit [Read error: Connection reset by peer]
fluo75 has joined #linux-sunxi
Offshore has quit [Ping timeout: 240 seconds]
Black_Horseman has quit [Quit: Zwi se logou mou!!!]
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
\\Mr_C\\ has quit []
egbert has joined #linux-sunxi
egbert_ has quit [Ping timeout: 240 seconds]
techn_ has joined #linux-sunxi
\\Mr_C\\ has joined #linux-sunxi
wingrime_ has quit [Ping timeout: 264 seconds]
wingrime has joined #linux-sunxi
BJfreeman has quit [Quit: had a good time]
bsdfox has quit [Ping timeout: 256 seconds]
Black_Horseman has quit [Ping timeout: 276 seconds]
[7] has quit [Disconnected by services]
TheSeven has joined #linux-sunxi
<wingrime> mnemoc: ping
rz2k has quit []
bsdfox has joined #linux-sunxi
bsdfox_ has joined #linux-sunxi
ganbold_ has quit [Remote host closed the connection]
eebrah|away is now known as eebrah
bsdfox has quit [Ping timeout: 264 seconds]
rachie_53uo has joined #linux-sunxi
rachie_53uo has quit [Client Quit]
Tartarus has quit [Ping timeout: 240 seconds]
Tartarus has joined #linux-sunxi
<wingrime> mnemoc: ping
eebrah is now known as eebrah|away
ganbold_ has joined #linux-sunxi
deasy has quit [Ping timeout: 276 seconds]
rings_VII has joined #linux-sunxi
deasy has joined #linux-sunxi
rings_IIV has quit [Ping timeout: 240 seconds]
Black_Horseman has joined #linux-sunxi
<wingrime> hno: ping
<wingrime> fel write 0x44000000 ../linux/arch/arm/boot/uImage
<wingrime> libusb usb_bulk_send error -1
Black_Horseman has quit [Remote host closed the connection]
ganbold_ has quit [Ping timeout: 246 seconds]
bsdfox_ has quit [Ping timeout: 248 seconds]
atiti has joined #linux-sunxi
\\Mr_C\\ has quit []
eebrah|away is now known as eebrah
<hno> wingrime, you lost contact with the device.
\\Mr_C\\ has joined #linux-sunxi
<wingrime> hno: it happends always when I try send 3rd image
<wingrime> hno: If I send spl + uboot it works normaly
<wingrime> hno: if I try send kenel too, it happends
<oliv3r> who all have mele m5
<oliv3r> (a20 mele)
<oliv3r> i'm thinking of getting one to hack on
<oliv3r> well use it as xbmc box in the (further future)
<oliv3r> but there's cheaper 'clones'
<oliv3r> jesurun that looks quite sleek
<oliv3r> then of course there's the A20 'hdmi sticks'
<oliv3r> anybody have an inside view of the m5? wanna know if we can tab more of the analog pins to get shvs or scart(rgb) out of it
<oliv3r> if you have an m5, would you mind posting a foto of the PCB on the wiki?
<hno> wingrime, what device?
<hno> wingrime, and is there any difference if you send the images in different order?
<oliv3r> libv: others still agree with you :)
shineworld has joined #linux-sunxi
deasy has quit [Remote host closed the connection]
eebrah is now known as eebrah|away
<hno> oliv3r, that is still AtomBIOS... "only" another glue around the open sourced AtomBIOS runtime parser.
<oliv3r> i haven't read the inside, but from the comments/story, it seemed like the intention was to fully replace the bios?
<oliv3r> loadable firmware*
<hno> oliv3r, only the x86 parts. It relies on the original vbios as source for the atombios "tables".
<hno> atombios do not qualify as "firmware" in my view of the world.
<hno> it's a program running in a vm.
<oliv3r> yeah but wasn't it the intention to replace it entirly?
<hno> There is very little stated about intention that I can find. Only "open source video bios for Radeon".
<oliv3r> lets see how it evolves
<hno> I don't expect it to replace AtomBIOS. But still, having a open sourced VBIOS using AtomBIOS is nice.
<hramrach> well, atombios is used for modesetting in radeon
<hramrach> it's buggy and/or undocumented and as a result modesetting works really poorly
<hramrach> there is no intention to replace atombios by the ati payed developers. They would have to release more information about the cards. and the driver might even work then
<hramrach> I don't think radeon driver as is qualifies as free software
<hramrach> it's just blobware with smaller blobs
<hno> hramrach, agreed, but we are not talking about ATI here or even the radeon driver.
<hno> but yes. the open sourced VBIOS would not qualify as free software either.
<hno> the source does, but not the resulting VBIOS.
<hramrach> it uses the blob as well?
<hno> hramrach, yes, it uses AtomBIOS tables from the proprietary VBIOS.
shineworld has quit [Ping timeout: 264 seconds]
<hramrach> you probably want to read the connector layout and stuff from the tables
<hramrach> but then some cards might not have them at all
ganbold_ has joined #linux-sunxi
<hno> hramrach, Unlikely to see a VBIOS not using and providing AtomBIOS.
<hramrach> it has happened for nouveau
mouchon has joined #linux-sunxi
mouchon has quit [Quit: Page closed]
<hno> hramrach, there is no nouveau VBIOS.
<hramrach> but nouveau does not use the bios, does it?
<hno> hramrach, as far as I know it does not use the bios.
<hramrach> so you don't need one then :p
<hno> Linux in general do not need VBIOS.
<hramrach> radeon driver does :/
<hno> radeon driver "needs" AtomBIOS, which is part of the ATI VBIOS.
<hno> "needs" because AMD do not document registers an no one has the patience to reverse the register details any more.
<hno> partly due to AtomBIOS..
Black_Horseman has joined #linux-sunxi
Black_Horseman has quit [Changing host]
Black_Horseman has joined #linux-sunxi
Offshore has joined #linux-sunxi
pacopad has joined #linux-sunxi
Black_Horseman has quit [Remote host closed the connection]
rellla has joined #linux-sunxi
<mnemoc> moin
<wingrime> mnemoc: why you picked only one patch?
<wingrime> mnemoc: other fix build issues too
<mnemoc> wingrime: i want to fix non-sun7i first
<wingrime> mnemoc: ok, whats now
<wingrime> mnemoc: but gadget still broken
<mnemoc> yes, and that makes all defconfigs except sun[45]i fail
<wingrime> mnemoc: better reverse that commit
<mnemoc> the merge or the defconfig?
<mnemoc> techn_: !!
<mnemoc> fixing small mistakes sounds better than reverting needed large changes as a whole
<wingrime> mnemoc: I change config for enable gadget and get this
<mnemoc> i was going to try hansg's sunxi-3.4
<mnemoc> i'm only testing vanilla defconfigs, automatically
<wingrime> mnemoc: techn patch broke gadget support again
<mnemoc> :'(
<wingrime> mnemoc: but If I remove #ifdef default sun4i builds good
<mnemoc> yes
<mnemoc> but the other 6 sunxi defconfigs still fail
<wingrime> give me log, I take a look
<mnemoc> sun4i_crane
<mnemoc> i'll give a try to hansg's sunxi-3.4 now (instead of -for-amery)
<wingrime> b20c046c7b844392ea86fab370e7660e06614720
<wingrime> this patch add new problems with android.c
<mnemoc> hansg's sunxi-3.4 is building fine....
<libv> oliv3r: it's doomed to failure.
<mnemoc> but haven't finited yet
<mnemoc> libv: android?
<libv> oliv3r: nobody is talking about unpacking and packing existing ati bioses or uploading them to the board roms
<libv> mnemoc: openradeonbios
<mnemoc> don't know the details, but sounds pretty greedy
<libv> oliv3r: that is the first step, anything beyond that is just whiny stupidity
<mnemoc> even coreboot is pretty difficult to spread
<wingrime> libv: sounds interesting , becose videocards can have vendor made limits, locked cores , freqs etc
<libv> wingrime: did you ever hear about the radeonhd project?
<wingrime> libv: yes, long ago , but I never had radionhd supported video card
<wingrime> libv: I still have radeon 9225 somewhere
<libv> anyway, we made mshopfs atombios disassembler public after radeonhd had died (as i stopped it from going out, knowing full well that the airlieds of this world would otherwise use this directly and thrash us for it too - they found another successful route in the end though)
<libv> this would be the starting point of any such project
deasy has joined #linux-sunxi
<wingrime> libv: did you find something intersting in disassembled listing ?
<mnemoc> wingrime: hno: bad news, only sun7i_defconfig builds with hansg's sunxi-3.4 :(
<wingrime> mnemoc: I have solution
<wingrime> mnemoc: I just checked git log
<mnemoc> reverting sunxi-usb unification oesn't sounds like a solution
<wingrime> mnemoc: 11-12 patches
<wingrime> mnemoc: or you can cherry pick that you prefer
<wingrime> mnemoc: this patches must be on separated branch untill it mature
<mnemoc> are they only on stage/sunxi-3.4?
<mnemoc> or hansg's?
<wingrime> hansg's
<libv> wingrime: endlessly so
<wingrime> mnemoc: broke usb support even worser
<mnemoc> can you make a branch removing those you think need to be kept aside until they mature enough?
<mnemoc> then i test it, on success you comment on the ML, and I push the stuff?
<wingrime> mnemoc: check hansg's "git log" only 11-12 patches related usb
<wingrime> mnemoc: only 11-12 git reverse in proper order
<wingrime> mnemoc: when it becomes mature enought , we can cherry-pick or use ML patchset
mcbrick has quit [Quit: Leaving]
mcbrick has joined #linux-sunxi
<wingrime> all usb patches after 7d474df5280aa16263cd288ea712c468297ffc87
<wingrime> must be reserved
naobsd has quit [Quit: Page closed]
deasy has quit [Quit: Quitte]
<wingrime> a20 have same IP USB so , normaly it must work with sun4i or sun5i
<wingrime> techn patch mix unification and Kconfig changes
rz2k has joined #linux-sunxi
Soru has joined #linux-sunxi
<hno> mnemoc, not so bad news. hansg is aware there is issues and plan to resume getthing things in shape when back.
<hno> mnemoc, are you using the current tree from earlier today?
<mnemoc> hno: tried hansg's sunxi-3.4 from some hours ago
<hno> mnemoc, ok.
<mnemoc> removing hramrach's usb patches seems to help
mcbrick has quit [Ping timeout: 264 seconds]
<wingrime> mnemoc: hramrach's ?
<mnemoc> the one changing a tristate to boolean
<hramrach> without that one I could not build the sun4i config
<mnemoc> build test still running, can't tell yet
<hramrach> but I have probably a different config
<wingrime> hno: about usb-boot script , normaly if I use SPL + UBOOT only - all works normaly , otherwice if I try send using your script SPL+UBOOT+UIMAGE
<wingrime> hno: this happends
<mnemoc> hramrach: I'm testing only defconfigs
<mnemoc> all sunxi ones
<hramrach> I am testing only useful configs
ykchavan has joined #linux-sunxi
<mnemoc> well.... all failed anyway
<mnemoc> will try stock stage/sunxi-3.4 .... this is all too fishy
<hramrach> I specifically made those config changes when the usb part did not build
<hramrach> but there might be some configurations which are still possible but don't build :s
* mnemoc building defconfigs over current linux-sunxi's stage/sunxi-3.4
<wingrime> hramrach: What do you think about leave usb-patches in separated branch untill it mature (unification)
<wingrime> ?
<mnemoc> if origin/stage/sunxi-3.4 builds fine, I'll go for old-school cherrypicking instead of pull
rings_VII has quit []
<hramrach> wingrime: I did not test the otg after the unification. with the extra patches that prevent selecting the OTG as module the thing builds for sun4i but that's all I can say
<hramrach> also the andriod.c does not build iirc
<hramrach> android.c
<mnemoc> lovely origin/stage/sunxi-3.4 is broken too
<mnemoc> except for sun4i and sun5i
<hramrach> the andriod configs don't build?
<mnemoc> 1m
<mnemoc> /bin/sh: 1: cd: can't cd to drivers/gpu/mali/mali
<mnemoc> wtf
<hramrach> yes, there is some unreliability wiht parallel builds
<hramrach> sometimes something fails randomly
<mnemoc> mach/sys_config.h .... that's easy to fix
<hramrach> that only gives more errors for me
<mnemoc> and here we go again
* mnemoc cries
<mnemoc> wtf this I pull this stuff to stage without testing it myself?!
<mnemoc> s/this/did/
<mnemoc> trust-no-one (tm)
<hramrach> maybe making branches for these major changes is a good idea - the usb thing and the a20 merge
<mnemoc> +1
<hramrach> because there are numerous small fixes in stage or floating around
<mnemoc> so, stage-3.4 is broken
<wingrime> mnemoc: witch ramdisk are you using (too lazy for build it myself) ?
<wingrime> soorry
<wingrime> mripard_: witch ramdisk are you using (too lazy for build it myself) ?
<mnemoc> none
<wingrime> mnemoc: mainline have no mmc support yet
jemk has joined #linux-sunxi
<hramrach> see you
deasy has joined #linux-sunxi
deasy has quit [Client Quit]
<wingrime> oliv3r: ping
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
<wingrime> mripard_: ping
ganbold_ has quit [Remote host closed the connection]
<bfree> wingrime: if your kernel has emac (easy enough to patch in if not) then you can use a network rootfs. nbd (or nfs I'm sure but I've not tried it) with an initrd built by debian's initramfs-tools (you'll have to mkimage it to put it in u-boot format)
* mnemoc in shock
<mnemoc> meh. read emacs
<wingrime> bfree: probem realy with uboot and dts
<wingrime> ERROR: Did not find a cmdline Flattened Device Tree
bamvor has quit [Ping timeout: 260 seconds]
jemk has quit [Ping timeout: 248 seconds]
<Turl> mnemoc: emacs is an OS, it just lacks a good text editor imo
<buZz> you can run nano inside emacs
<Turl> ooo well that fixes it :)
<mnemoc> Turl: :)
bamvor has joined #linux-sunxi
bsdfox has joined #linux-sunxi
<hno> wingrime, did you give the fdt address in bootm?
<wingrime> hno: yes
<wingrime> bootm 0x40008000 - 0x4100000
<hno> wingrime, one 0 missing there.
<hno> bootm 0x40008000 - 0x41000000
<wingrime> hno: I will fix wiki
<wingrime> hno: thanks , but now Starting kernel ...
<wingrime> hno: and silince
<Turl> wingrime: enable DEBUG_LL, choose sunxi uart0 on config and then enable EARLY_PRINTK
<Turl> and pass earlyprintk on cmdline
atiti has quit [Read error: Connection reset by peer]
mcbrick has joined #linux-sunxi
<wingrime> Turl: are you about mainline ?
<Turl> yes
<Turl> yes
<Turl> err, not my terminal :)
<wingrime> Turl: wtf, it must work without
<wingrime> such things
<Turl> wingrime: yea, but with such things you can know why it isn't working instead of guessing :)
<wingrime> Turl: jtag more better than it
<Turl> jtag is too low level
<Turl> it's probably something simple that you're missing
<wingrime> Turl: jtag can get access to ramconsole
<Turl> you need ramconsole working for that
<Turl> I'm betting it doesn't even go as far as choosing the right machine
<wingrime> Turl: also, are you know witch jtag suit out mmc-jtag connector?
<Turl> no idea, never used jtag myself
<Turl> hno probably knows
<mnemoc> the mapping is documented in the wiki
<mnemoc> but it's not an standard jtag connector
ykchavan has quit [Ping timeout: 264 seconds]
<wingrime> Turl: nice, looks like mripard not added sun7i machine id
ykchavan has joined #linux-sunxi
<wingrime> wtf
atiti has joined #linux-sunxi
<wingrime> what a hell
<hno> wingrime, what do you mean?
<wingrime> hno: take a look at paste
<hno> wingrime, are you using an initrd?
<hno> is this with earlyprintk?
<wingrime> hno: earlyprintk used
<wingrime> hno: no still not using any initrd
<wingrime> hno: Ok I hope it clean issue after git pull
<hno> wingrime, hmm.. numbers don't add up for me.. bootm 0x40008000, Legacy Image at 40008000, Load Address: 40008000, Entry Point: 40008000, XIP Kernel Image ... OK.
<hno> That can't be XIP.
<hno> but maybe I am just confused.. don't know.
<wingrime> hno: it maybe my ability find probelms ....
<wingrime> just do make again
<hno> wingrime, how did you make the uImage?
<wingrime> make uImage + toolchan + load address + dtbs
<hno> wingrime, if it is a XIP uImage then it needs to be loaded to entry_point - 64, i.e. 0x40007fc0.
<hno> you can't execute the u-boot image header.
<hno> don't know why u-boot accepts that as valid. Does not make sense.
fredy has quit [Excess Flood]
fredy has joined #linux-sunxi
jemk has joined #linux-sunxi
<wingrime> hno: that worked
<wingrime> see
<wingrime> normaly I want see panic "can't sync" but silince again...
<hno> wingrime, looks quite good. Now the problem is your ttyS0 settings making you blinded...
<hno> missing console=ttyS0,115200
jemk has quit [Ping timeout: 268 seconds]
<hno> most likely switches to 9600 or whatever is the default if no baud rate is given.
jemk has joined #linux-sunxi
<wingrime> extraargs=console=ttyS0,115200
<wingrime> I hope uboot parse it before I press key for stop automatic boot
<hno> no, that is parsed by the default boot script.
<hno> your commandline was very short.. [ 0.000000] Kernel command line: earlyprintk
<wingrime> hno: how compile/decompile script?
<hno> wingrime, printenv
<hno> but you are not using any if you are doing bootm manually.
<hno> your responsibility to set bootargs accordingly then.
<hno> wingrime, yes that looks familiar.
<wingrime> hno: so whats a problem
<hno> i<hno> but you are not using any if you are doing bootm manually.
<hno> <hno> your responsibility to set bootargs accordingly then.
<wingrime> hno: ok, how compile/decompile boot.scr?
<hno> If you do not interrupt u-boot then it runs the bootcmd macro, which does a little dance and eventually ends up in running the setargs macro to compose the command line.
<hno> wingrime, are you using a boot.scr?
<hno> a boot.scr is just an u-image of an u-boot shell script.
<wingrime> yes, it present on mmc
<hno> wingrime, are you using it? It's only used if you do not interrupt the boot process.
<hno> to "decompile" dd if=boot.scr of=boot.uboot-sh bs=64 skip=1
<wingrime> hno: strange solution where possible use simple plain text instead
<wingrime> hno: what proper parameters for mkimage ?
<wingrime> hno: I want compile sctipt
<hno> wingrime, -T script
<wingrime> thanks
* hno thinks this is mentioned in the sunxi u-boot wiki...
<hno> Yes it is.
utzig has joined #linux-sunxi
<hno> wingrime, I mostly use uenv.txt these days. simpler.
<hno> but for mainline booting a boot.scr is needed. the built-in boot script is tailored for allwinner style booting.
<hno> well, possible to override from uenv.txt as well, but messier.
<Turl> hno: I boot mainline with uEnv.txt on the cubieboard
<Turl> I don't have the script handy now, but I think it was just a matter of overriding the default command that was run
<hno> Turl, yes there is several ways of doing it with uenv.txt. Either override variables & macros as needed, or set uenvcmd to your boot script.
<Turl> on the olinuxino I was lazier so I just set the env to my liking on the shell and then ran saveenv
<hno> that works too, but not so easy to automate image creation or control from Linux which kernel boots next reboot.
<Turl> hno: about the card from the other day, it seems to not work correctly on other devices either
<hno> Turl, memory short... failing SD card?
<Turl> yeah, I don't know what is it
<Turl> it's sad, because it was a 16G card
<hno> what symptoms?
<Turl> hno: it goes straight to nand, then works if I reset from there
<Turl> on my phone it's not detected, errors -110 and -123 on the mmc code
<hno> that is seen if the card takes too long to become ready.
<Turl> if I fill the code with sleeps it works.. sometimes
<hno> You can try erasing the card.
<hno> what code?
<Turl> dd with 0s? or what kind of erasing?
<Turl> hno: mmc linux code, around sd.c or such
<hno> no, the kind of erasing that trim does.
<Turl> hno: what tool can I use to do that?
* hno tries to remember how to do that at device level..
<hno> Turl, hdparm has options for raw trim at device level.
<Turl> wingrime: yeah, I don't think mainline has SMP support just yet
<Turl> mripard_: ^
<Turl> hno: I'll try, thanks
<wingrime> mripard
<wingrime> mripard_: ping
<hno> Turl, an as usual the standard comment "EXCEPTIONALLY DANGEROUS. DO NOT USE THIS OPTION!!" in the manual...
<Turl> hno: it has a "" script
<hno> Turl, there is also special programs for formatting sdcards.
<hno> that's a monster script. But looks nice.
<Turl> /dev/sdc: DSM/TRIM command not supported (continuing with dry-run).
<wingrime> Turl: witch initrd are you used?
<wingrime> Turl: I done it years before manualy with ulibc + busybox , but have you ready to go?
<hno> Turl, run it with a real mmc controller. I.e. on the Allwinner device.
<hno> usb readers is a bit limiting..
<wingrime> hno: have you any prebuilded initrd
<wingrime> hno: busybox+ulibc
<wingrime> ?
<Turl> wingrime: buildroot initramfs
<Turl> wingrime: I think this is an old copy of it, should be ok to test if it boots
<wingrime> Turl: thanks
<hno> wingrime, I have my miniroot..
pacopad has quit [Quit: pacopad]
<wingrime> Turl: are you load it with fatload ot it must be build in kernel?
<hno> haven't pushed a prebuilt one, only takes a minute to build from source.
<Turl> wingrime: I embed it in the kernel
<Turl> but you should be able to get it to load separately
<wingrime> Turl I will try
<hno> Turl, how large is your initramfs?
<wingrime> 6.8 mb
<hno> compressed or uncompressed?
<bfree> wingrime: if not building it in see (e.g. note the "setenv initrd_high 0xffffffff")
<wingrime> bfree: thanks
<hno> -rw-rw-r--. 1 henrik henrik 2940928 20 maj 00.40 initramfs-3.4.43+.cpio
<hno> with 1.8M modules in it..
<hno> -rw-rw-r--. 1 henrik henrik 1463531 20 maj 00.40 initramfs-3.4.43+.img compressed u-boot image.
<wingrime> hno: Turl: bfree: thanks , I finaly get woring dev env
<Turl> hno: it's an uncompressed cpio
<Turl> wingrime: yw :)
bsdfox has quit [Ping timeout: 276 seconds]
<wingrime> Turl: are you currently devel something for mainline ?
<wingrime> jemk: ping
rellla has quit [Remote host closed the connection]
<oliv3r> wingrime: pong
<oliv3r> libv: sorry for the noise then, it _sounded_ interesting ;)
<wingrime> oliv3r: how you mainline?
<wingrime> oliv3r: I finaly get mainline working on a20
<libv> oliv3r: nah, it was no noise, i even let myself be dragged into the shitthrowing :)
<wingrime> oliv3r: but with some strange uboot settings
<wingrime> libv: have you ever sended patches to mainline ?
<libv> wingrime: nope
<hno> wingrime, what strange u-boot settings you need?
<wingrime> firstly stange image base
<hno> the xip problem?
<wingrime> yes
<wingrime> also initrd_high 0xffffffff
<hno> wingrime, u-boot should have rejected that as a non-XIP load address and relocated the image as needed. Don't know why there is a special case to allow that broken usage..
<Turl> wingrime: clocks stuff, and PM
<hno> initrd_high is a bit strange indeed. Should not be needed, but might be related to fdt_high which is also needed if you use a sufficeintly large zimage.
<wingrime> hno: also your usb script still not works when you ask it push uImage too
<hno> odd.
<hno> wingrime, what device do you have?
<hno> cubieboard2?
<wingrime> hno: yes
<wingrime> hno: but I have also two a10 tablets and one a13
<wingrime> hno: also I saw some sun3i hw in shop , but price unreasonable for such crap hw
<hno> our sun3i support is likely borked. No one using that ever.
<wingrime> hno: but not always I can get information about CPU
<wingrime> hno: that was tft ebook reader.... so crap
<wingrime> hno: if it was elink....
<hno> wingrime, cb2 booted fine for me.
<wingrime> hno: but It still mystery for me why a10 so slow in comparea with other hardware
<wingrime> hno: I think there deep problems with memory ctrl,and more
<hno> well, so far as u-boot finding the uimage and initramfs. Don't have a working cb2 kernel yet..
<hno> wingrime, unfortunately docs on the memory controller is a bit limited, but we are doing our best at guessing what the AW code means.
vinifr has joined #linux-sunxi
<wingrime> hno: tpr stuff?
<oliv3r> hramrach: how does radeon driver not qualify as free software? Politics aside, afaik it's as free as linux is, linux sometimes talks to the PC bios aswell (not very often, but i think there's an option to ask the bios about certain things like disks)
<wingrime> libv: how do you think can a10 becomes most open-sourced arm device without any prop blobs?
<libv> wingrime: it is better supported than any other mali based device out there
<hno> wingrime, there is a lot more than the tpr to worry about. And for tpr we have some nice looking magic numbers.
<libv> and it was the first device to allow me to work on a proper gnu/linux with working binaries, which is a massive simplification for RE work
<wingrime> hno: do you tryed overclock feq and increase cas for it ?
<wingrime> libv: next rms notebook candidate
<libv> wingrime: i hope that none of us here have that as our goal
<hno> oliv3r, it's a definition matter. It links in the AtomBIOS code at runtime. Parsed, but still code.
<libv> even radeonhd did that
<libv> this to gather board layout information
<libv> and to call ASICInit, which then ended up replacing a full x86 bios int10
<libv> funny thing... we had been bashed by ATI for not wanting to use atombios much for a month or so already
<oliv3r> wingrime: turl and I use turl's ramdisk for mainline, it's simple but works:
<hno> wingrime, I don't overclock any of my equipment. But I do fix obvious errors in timings etc.
<libv> and we wanted to use ASICInit as it was only a few hundred lines of disassembled microcode compared to a full 64kB bios running int10
<libv> so bridgman organized a call with some fglrx driver developers and us
<libv> and then we shot off the question and ask, so, how are you guys using ASICInit, can we get any info or description on it
<wingrime> libv: also I thinking that a10 can be considered mostly free from hw backdors unlike modern intel cpu, that now have microcontroller in southbridge that can full access to main memory and listen tcp prot...
<libv> the answer was: "I don't know, we just use int10 on linux"
<libv> that was the only time we got to speak to any engineer of ATI
<libv> wingrime: yes, a10 seems to be one of the free-est possible SoCs out there
<hno> well, the A10 do have some ARM Trustzone support, but from specifications it looks like key management is fundamentally flawed.
<wingrime> libv: thanks jemk we now have jpeg and mpeg12 PoC ,
<libv> cool
<wingrime> libv: take a looks my and jemk discovers
<hno> yes, soon only CAN, Memorystick and something else I don't remember is dark spots on the A10.
<wingrime> hno: pata
<wingrime> hno: It seems to be AW boght IP and place it in ASIC in hope that it can be used sometime
<oliv3r> Turl: hno mmc doesn't have TRIM though does it? They don't even do wear leveling or only some very basic? I don't know if you can with hdparm and mmc, but with IDE/SATA you do a 'security erease' to completly reset a drive
<wingrime> oliv3r: mmc ctrl can send commands to mmc directly, usb mmc readers use mass storage specification that limited on that
<oliv3r> wingrime: mripard's a20 mainline code? i'm still doing a10 :p
<wingrime> oliv3r: Some years ago I used attiny gpio for read mmc directly (SPI) but attiny have 128 byte memory so, I can't read sector fully
<wingrime> oliv3r: yes a20 branch
<oliv3r> wingrime: well i always load my kernels to 0x48000000 and the dtb to 0x43000000, maybe that's causing you heartache?
<oliv3r> while it's busy anyway, i got a few replies on ML with regards to sun3i; so i'll ask here again just in case, any reason to keep sun3i? I say lets just drop whatever bits we have to get rid of some cruft
<wingrime> oliv3r: I have problem with "offset - 64"
<wingrime> oliv3r: we can defently drop it from 3.4
<oliv3r> oh yay i've backread :p
<oliv3r> hno: do you think that 'trustzone' can be used for anything usefull?
<oliv3r> wingrime: maybe that's an a20 thing? I'll test tomorrow, i've only used a10 really
<wingrime> oliv3r: trustzone It seems for DRM
<wingrime> oliv3r: but I have not seen any soft that use it
<oliv3r> well could it be used for anything else
<wingrime> oliv3r: we have aw's cripto controller that much usefull
<wingrime> oliv3r: if It can give realy random unmbers
<hno> wingrime, yes PATA also,, completely forgot about that one. Seems they never bothered to port PATA driver to Linux at all.
<wingrime> hno: seems thay only bought IP and add it only
<oliv3r> they may not even have tested it :p
<oliv3r> 'but we do have pata!'
<oliv3r> i didn't know 'can' was so dark
<oliv3r> and we have GPS too :)
<hno> wingrime, it's defenitely used in their earlier sun2i & sun3i media player devices
ssvb has quit [Remote host closed the connection]
<oliv3r> ah of course
<oliv3r> sun4i is the 4th generation, so they may have simply not removed it
<hno> oliv3r, right. GPS is a quite blank area, but we do have a gps.ko with full debug info.
<oliv3r> if anybody ever would need it
<hno> but no known hardware using it.
<oliv3r> then there's the secret 'transportstream' though we atleast now have datasheet
<wingrime> hno: pata also can be used for Compact Flash
<wingrime> if I remeber right
<hno> wingrime, yes.
<wingrime> hno: also I have no idea what GPS means in A10 case
Black_Horseman has joined #linux-sunxi
Black_Horseman has joined #linux-sunxi
<hno> wingrime, GPS baseband controller, doing all the math.
<wingrime> hno: but seriosly a10 can contine product live as car pc , so maybe we see CAN
<oliv3r> as opposed to a serial/USB controller? dont' they do allt he math aswell?
<oliv3r> i guess its oposition would be a gpio GPS?
atiti has quit [Ping timeout: 268 seconds]
<oliv3r> wingrime: actually i've thought of this aswell
jemk has quit [Read error: Operation timed out]
<oliv3r> I don't know if the eoma-68 connects the CAN bus
<hno> oliv3r, A full GPS controller is antenna + radio + baseband.
<oliv3r> that's using A10
<wingrime> hno: you mean GPS backend ?
<oliv3r> so if you could use emoa-68 for that, you can upgrade it!
<wingrime> oliv3r: aw nowdays in not good shape
<hno> wingrime, no, baseband is the term used for the digial part.
<wingrime> hno: It requires good ADC for GPS
<wingrime> oliv3r: also do you know that LG produced low-end phones with MTK SoC
<wingrime> oliv3r: it a first time I see that hi-level company use china SoCs
<hno> wingrime, no, the interface between the radio and baseband in a GPS is all digital timing based.
<hno> and fairly low frequency.
<wingrime> hno: you means you need only downmix it?
<hno> what do you mean by downmix?
<wingrime> downconverter
<oliv3r> well i'm sure there's some ADC somewhere in the chain :)
<oliv3r> i mean, the signal you receive, is usually fed to an ADC rather then toggling a pin
<oliv3r> I mean, hooking up the antenna directly to a GPIO would be kinda insane :)
<oliv3r> (with voltage conversion)
deasy has joined #linux-sunxi
<wingrime> firstly must move HF signal to some low-freq
<wingrime> oliv3r: antenna can be connected to GPIO!
ssvb has joined #linux-sunxi
<wingrime> oliv3r: for transmit
<wingrime> oliv3r: but filter recomended
<oliv3r> well for transmit yeah, you can 'fake' a DAC that way
<oliv3r> and sometimes it doesn't even have to be that accurate
<wingrime> oliv3r: you must know about nyqvist frq
<oliv3r> but i don't think you can do that for receive
<wingrime> oliv3r: it prefectly accurate if you deal with nyqvist frequency, but without filter you will recive transmitted signal not only f but 2*f ,3*f 4*f
<wingrime> oliv3r: it can be used for reciving if you have second gpio! (or comparator)
<wingrime> oliv3r: or you have downconverter
<hno> wingrime, the RF part handles the demodulation. Data bit rate in GPS is 50 bps, from multiple concurrent sources on the same fequency.
<wingrime> hno: do you know witch modulation used? QAM? PSK?
<hno> the baseband decodes the demodulated signal, separates the differerent signals and correlate timinig between them to find your position.
<hno> by knowing the position of each sattelite, and timing differences in receiving signal from each.
ykchavan has quit [Quit: Leaving]
<wingrime> oliv3r: 1000mA tablet (
<Turl> wingrime: that's the sun3i guy :)
<hno> wingrime, "The RF signal transmitted by the satellites is a right hand circularly polarized Bi-phase modulated CDMA code with data impressed on it."
<wingrime> hno: than it PSK , with manchester code
<wingrime> hno: very simple
<utente> someone can use SPI interface? there are case of success to control exterlam chip0 as eeprom?
<wingrime> Turl: we need he onboard
<Turl> wingrime:
<utente> i talk about cubieboard
<oliv3r> as long as we don't have hardware, we can't write it anyhow
<oliv3r> so does anybody have a mele m5?
<hno> wingrime, yes, the signal as such is very simple.
<wingrime> by bin on 2 juin 2012
<wingrime> year ago
<oliv3r> Turl: whats with sysfslab?
<oliv3r> oh F20, i read A20
\\Mr_C\\ has quit []
<hno> wingrime, tries to give a brief intro.
<wingrime> hno: ok, coding not difficult
pitillo has quit [Ping timeout: 240 seconds]
<wingrime> a31s
<wingrime> still no modem
pitillo has joined #linux-sunxi
vinifr has quit [Quit: Saindo]
<oliv3r> usb modem
<oliv3r> i need to see the PCB of an M5
<hno> wingrime, the signal given to the digial baseband is 2 digial bit (SIGN, MAG), clock, and an SPI channel for controlling the radio chip. Same interface seen in many GPS chipsets.
<hno> digial baseband in A10.
wickwire has joined #linux-sunxi
<wickwire> Hi guys, does anyone know how to get Qt5 to successfully build with the sunxi mali libraries, for the Mali400 GPU?
<wickwire> I'm trying to build and use Qt5 with Cubieboard
<wickwire> Allwinner A10, Mali 400 GPU
<wickwire> I know there's libhybris, and Mer,
<wickwire> but I was trying to build Qt5 from source
<wickwire> rather than use the Mer pre-compiled binaries
<wickwire> and those seem to use libyhbris, a different approach...
_whitelogger has joined #linux-sunxi
<wingrime> wickwire: sorry , we mostly handle kernel stuff
<wickwire> oh ok
<wingrime> wickwire: #cubieboard for it better
<wickwire> ok I'll try there, thanks!
<ssvb> wickwire: and then download & configure & make Qt5
<wickwire> ssvb: I'm asking at #cubieboard, so far I've actually managed to build Qt5 with the mali binaries for framebuffer, but when I run an app, I get EGL Error : Could not create the egl surface: error = 0x300b error
<wickwire> thank you for the help anyways, I'll keep at it!
<ssvb> wickwire: does the simple gles test application run correctly?
<wickwire> I'm able to successfully run a coloured triangle on framebuffer
<wickwire> I used make config VERSION=r3p0 ABI=armhf EGL_TYPE=framebuffer
<wickwire> with the mali libs
<wickwire> but I'm still struggling with having a Qt app run properly
<wickwire> although Qt5 builds...
<ssvb> well, maybe Qt needs a few tweaks to work with the framebuffer mali drivers
\\Mr_C\\ has joined #linux-sunxi
<ssvb> just check where this EGL error comes from and try to fix this part of code (using the coloured triangle app sources as a reference)
<ssvb> Qt5 works with x11 mali drivers out of the box and needs no special tweaks
<ssvb> as for the framebuffer mali drivers and Qt5, probably you are the first to try this :)
Soru has quit [Remote host closed the connection]
<ssvb> wingrime: if this is for kernel only, maybe we need a userland-sunxi irc channel for these things?
<wingrime> ssvb: we have not so much people here
<wingrime> ssvb: I say "mostly"
<ssvb> ok
<wingrime> but exists some big projects that have #dev #user #etc channel
<wingrime> but there not so much people
dl9pf has quit [Ping timeout: 240 seconds]
<libv> separate irc channels or mailing lists are only useful for a limited time
<libv> after a projects peak they are just counterproductive
jlj has quit [Ping timeout: 264 seconds]
jlj has joined #linux-sunxi
<hno> ssvb, this channel is certainly NOT kernel only.
Offshore has quit [Ping timeout: 240 seconds]
jlj has quit [Ping timeout: 240 seconds]
jlj has joined #linux-sunxi
wickwire has quit [Remote host closed the connection]
<rz2k> damn he left
<rz2k> actually answer for him is to use qpa-xcb and x11 libs
<rz2k> it should work out of the box
<rz2k> iirc, i built it really long ago