<oliv3r>
e.g. the uboot from boot0 -> boot1 -> boot.asf -> u-boot
<robb83>
ohh
<robb83>
thanks, thats new for me
<oliv3r>
we have practical zero u-boot support right now :) i started a few patches, but that's on very low prio rigt now :)
n01 has joined #linux-sunxi
<robb83>
want to deal with it? difficult to rewrite sd?
iamfrankenstein has joined #linux-sunxi
<robb83>
or modify my nand directly with usb firmware updater or android ?
<oliv3r>
it's a lot of work, and don't want to deal with it right now, also I lack a31 hardware, and with the powerVR in their, my motiviation is limited. I started it because it was interesting and exciting at start ;)
<oliv3r>
but as i write more,it becomes harder to test as i have no hardware :)
<oliv3r>
in a few weeks time i might work on the dramc a bit more
<robb83>
same are, no hardware :(
ykchavan has quit [Quit: Leaving]
sdk has joined #linux-sunxi
shineworld has quit [Quit: Leaving]
popolon has joined #linux-sunxi
<sdk>
hi, I'm having some difficulty getting my A10s (olimex) board to boot. The SPL loads but then the next stage of u-boot never starts. Any help would be appreciated! Thanks.
<oliv3r>
oh i'm not the only one who told puneet B to stop asking for lychee support; good
<oliv3r>
Turl: ping
<oliv3r>
Turl: ping re: ahci willis wang code
deasy has quit [Quit: Quitte]
fredy has quit [*.net *.split]
jlj has quit [*.net *.split]
FunkyPenguin has quit [*.net *.split]
MadSpark has quit [*.net *.split]
<robb83>
sdk: sorry, i lame, but how you can probe ?
MadSpark has joined #linux-sunxi
fredy has joined #linux-sunxi
fluo75|2 has joined #linux-sunxi
FunkyPenguin has joined #linux-sunxi
<sdk>
i'm using serial cable
fluo75 has quit [Read error: No buffer space available]
<oliv3r>
i guess using the size (in kb) from your current u-boot should be a good indication
hramrach has quit [Remote host closed the connection]
<sdk>
ok I'll give it a try.
<oliv3r>
226 it looks like
hramrach has joined #linux-sunxi
<sdk>
nope it makes no difference. it's strange
<oliv3r>
SD card broken?
<oliv3r>
dd if=/dev/mmcblk0 of=dump.img bs=64k
<oliv3r>
(dump entire SD card)
<oliv3r>
then, write it to your own card and see if it works, I suspect it also does not work
<sdk>
yesterday I wrote the olimex image to it, and it worked. so I don't think that it's broken
<oliv3r>
you can also first try to copy SPL (maybe SPL is wrong)
<sdk>
yes I'll try that
<oliv3r>
sd cards break without notice :)
<oliv3r>
there's no flash wear lvling layer, or not a good one
<sdk>
yeah you're right :)
<oliv3r>
mine actually blew up and cracked
<oliv3r>
it was only 128mb so i didn't care, i used it for mainline test kernels
<oliv3r>
but it baffled me
<oliv3r>
woprr: did you find if it all really overlapped
<sdk>
mmh. now SPL not booting as well. it's weird, first I dd 23 records to spl-orig, then dd'ing that file to the card it says 31 records out.
<oliv3r>
spl is 22 or 23 bytes, to that sounds sane
<oliv3r>
ok try downloading SPL and u-boot from olimex, I can't imagine them not hosting the binaries somewhere
<oliv3r>
they are very OSS friendly
<oliv3r>
get those and dd those (saner then extracting and writing)
<oliv3r>
SPL can be no larger then 24k
<oliv3r>
so you could try extracting 24k and writing 24k 'just in case'
<oliv3r>
the BROM will not ever read more then 24k (hardware limit) so that should be safe
<sdk>
ok, i'll look for the files on olimex site..
rz2k has joined #linux-sunxi
<oliv3r>
sdk: the SPL and u-boot from git _should_ work, we just have to find out why it's not working now
<sdk>
yes. well the only differences wrt olimex article I can see are that I'm using gcc4.7 and saying "make .. arm-linux-gnueabihf-" instead of "arm-linux-gnueabi-"
<oliv3r>
that string is just a name that you give to your toolchain, weren't there some issues with gcc 4.7? i don't know if those have been resolved and if those applied to u-boot at all
<oliv3r>
let me build a spl + u-boot for you
<oliv3r>
in 20 minutes
<oliv3r>
gotta test my own u-boot first
<oliv3r>
to see if the code still works ;)
<sdk>
ok tnx
shineworld has joined #linux-sunxi
akaizen has quit [Read error: Connection reset by peer]
akaizen has joined #linux-sunxi
<oliv3r>
bah mine's not working give me a bit more time ;)
<sdk>
ok, hope it will work..
<sdk>
i got it working now. using the olimex git snapshot git checkout v2013.07-rc1-sunxi and re-building then it works!
<sdk>
so something must have changed along the way, breaking it
<oliv3r>
that's interesting
<oliv3r>
what hash tag is their snapshot that you are using?
<oliv3r>
for what it's worth, my SPL isn't booting on cubie2 either ;)
<mnemoc>
and all my test builds of the sun5i family failed :(
notmart has joined #linux-sunxi
notmart has joined #linux-sunxi
<sdk>
ok.. it must be something is broken in the u-boot. hope it will be fixed but for now i can go on and use the old version. I'm copying my own rootfs to the card now.. see how that goes.
<oliv3r>
hno: ping
<oliv3r>
sdk: well on the upside, it wasn't my patchseries that broke it :)
<oliv3r>
hno: 29689232f8a4d1640e551ef433176b81cecfa9c7 i just checked out and that still works, so gonna try to find which patch breaks it, but i have a very very slow testing method (build, upload, download, put on SD, put in cubie linaro, dd, reboot cubie
PiyushVerma has joined #linux-sunxi
jlj_ has quit [Read error: Connection reset by peer]
jlj has quit [Read error: Connection reset by peer]
<oliv3r>
hmm, i seem to miss-understand how git bisect works :S
<oliv3r>
29689232f8a4d1640e551ef433176b81cecfa9c7 is my last commit and i tested that, which works. so I'll use that as base reference (work on cb2)
<oliv3r>
so the breakage should be somewhere after that and before HEAD, so I use git bisect start HEAD 29689232f8a4d1640e551ef433176b81cecfa9c7 --
<oliv3r>
the first hash it checks out is: 934128bcd98106e764d2f492ac79c38ae53d] Merge branch 'master' of git://git.denx.de/u-boot-mpc85xx
<oliv3r>
so i assume it checks out 934128bcd98106e764d2f492ac79c38ae53d but when I do git log, my patches are all gone, boards.cfg lists no allwinner device?
<oliv3r>
so how do I use git bisect?
kaspter has joined #linux-sunxi
eagles0513875_ has joined #linux-sunxi
<eagles0513875_>
hey guys
<eagles0513875_>
how is everyone this morning
vinifr has joined #linux-sunxi
<shineworld>
hi eagles0513875_
<eagles0513875_>
how are you shineworld
<shineworld>
fine, thank you .:)
<eagles0513875_>
shineworld: i think im almost there with build root
<shineworld>
wow :) do you publish something somewhere ?
<shineworld>
you know there are noobs like me in the world :)
<eagles0513875_>
shineworld: i will post a blog post on my blog which needs a serious update :D
<eagles0513875_>
its a blog about everythign music to open source etc
<eagles0513875_>
shineworld: are you having issues working wiht buildroot and ur cubieboard2
<shineworld>
eagles0513875_, I never touch buildroot chain for cb2, just made my jobs manually... but I'm far to have what I need
<shineworld>
so I'm happy if someone share its jobs
<eagles0513875_>
shineworld: once i get build root a working setup i will post an article to my blog
<eagles0513875_>
in a nut shell shineworld im using the cubieboard scripts with some modifications for the CB2
<atsampson>
oliv3r: I'd normally do it as separate commands, although I don't know if it actually makes a difference
<atsampson>
oliv3r: i.e. checkout the latest version, "git bisect start", "git bisect bad" (to say this version's bad), then "git bisect good <the hash of HEAD>"
<atsampson>
then it should grab a version in the middle for you, and you can do "git bisect good/bad" as appropriate
hipboi has quit [Quit: Leaving]
<oliv3r>
atsampson: ok i try
massi has joined #linux-sunxi
sdk has quit [Quit: Leaving]
shineworld has quit [Quit: Leaving]
kaspter has quit [Ping timeout: 245 seconds]
massi has quit [Read error: Connection reset by peer]
sdk has joined #linux-sunxi
sdk has quit [Read error: Connection reset by peer]
<rellla>
oliv3r: 35 days later, 2nd cubie2 arrived. some discussions with customs, but tax-free at the end ;p
<oliv3r>
2nd?!
<oliv3r>
lucky oyu!
<oliv3r>
my cb1 is ethernetl ess, and my cb2 is having mmc issues it seems
<rellla>
i don't know what to do with all my devices :\
<n01>
eagles0513875_: asap you have something working submit the patches to buildroot ml ... sometimes the approval process is fairly long
<oliv3r>
rellla: so why do you have 2?
benoitm974 has joined #linux-sunxi
<rellla>
oliv3r: i begged tom for one *dev* board, when he announced it. but heard nothing, so a month later i ordered one the official way including two cases...
<rellla>
now i need another case ;)
<oliv3r>
ahh
<oliv3r>
okay
<oliv3r>
a bought one ;)
fredy has quit [Excess Flood]
<rellla>
so i will kick out my good old dockstar. squeezeboxserver is running much faster on cubieboard...
fredy has joined #linux-sunxi
<oliv3r>
rellla: kool radio, i think my mom has something like that in the addict, i should claim it and make a networked player out of it aswell, probably not with TS though ;)
shineworld has joined #linux-sunxi
sdk has joined #linux-sunxi
wingrime has joined #linux-sunxi
<oliv3r>
atsampson: same result
<oliv3r>
i must be doing something wrong
<oliv3r>
maybe it's because of the two branches that got merged?
<oliv3r>
my git foo is too weak :(
benoitm974 has quit [Quit: benoitm974]
Super-noob has quit [Ping timeout: 268 seconds]
boycottg00gle has joined #linux-sunxi
<boycottg00gle>
see an uboot regression switching from sunxi-current (c0f3b94d7cdcd83042b35f02552520d107fd7f35) to sunxi 48845ec7) boot stopped working on olimex a20
<oliv3r>
heh, read the log :p
<oliv3r>
boycottg00gle: i noticed this aswell, but haven't figured out the why yet
<oliv3r>
and need hno/git bisect guru to test more
<boycottg00gle>
last line from serial reads: SUNXI SD/MMC: 0
<boycottg00gle>
a single "G" follows - then nothing. restoring older version now
<oliv3r>
grepping for my name in the log found a working version ;p
<boycottg00gle>
ah - reading the log
fredy has quit [Excess Flood]
benoitm974 has joined #linux-sunxi
massi has joined #linux-sunxi
fredy has joined #linux-sunxi
<oliv3r>
hmm, i may have found my issue, which may be unrelated ;)
<boycottg00gle>
c4163747 is newer than 29689232
<boycottg00gle>
but am i really sure i used c4163747 to build the working rev?
<oliv3r>
29689232 is quite old
rellla has quit [Ping timeout: 260 seconds]
<boycottg00gle>
restored uboot works
benoitm974 has quit [Ping timeout: 245 seconds]
<oliv3r>
i'm trying 218f643881c0dabd7e40cdb21a757416fa80afb2 right now, the one before the merge commit, and it doesn't work
<oliv3r>
i'm not getting your 'G'
<oliv3r>
i just get the SPL and after that nothing
<oliv3r>
sunxi sd/mmc: 0 is my last line
<oliv3r>
but need hno to talk about it more now ;)
<oliv3r>
hno: ^
benoitm974 has joined #linux-sunxi
<boycottg00gle>
working version first line: U-Boot SPL 2013.07-07794-gc0f3b94 (Aug 16 2013 - 09:05:21)
<oliv3r>
i'm not quite sure how that ID is made up tbh
<boycottg00gle>
me neither
<oliv3r>
i first thought it was the git hash tag, but that's not the case (the gc0f part)
<oliv3r>
it appears spl might work, but u-boot is broken
<boycottg00gle>
ah my local sunxi-current branch is on: c0f3b94d
<boycottg00gle>
this also matches the last chars in above line
<boycottg00gle>
regression somewhere between c0f3b94d and 218f6438 ?
<oliv3r>
ah so g does stand fr git; good to know
<oliv3r>
if i use the head sunxi-spl, in combination with an old (working u-boot) i get ### ERROR ### Please RESET the board ###
<oliv3r>
if i use sunxi-spl + u-boot both from head, i get no output at all
<mnemoc>
oliv3r: what hexa value was it supposed to be? :p
<oliv3r>
if i use old sunxi-spl + new u-boot, i get ...
<oliv3r>
mnemoc: huh?
<mnemoc>
oliv3r: it's a hash, 0-9a-f ... g is out of range unless the hash is in base 17
<mnemoc>
ignore me
<oliv3r>
mnemoc: g is the prefix for git :p
<oliv3r>
i think :)
<oliv3r>
ok and i lied, head u-boot with head SPL gives the ERROR please reset too
<mnemoc>
yes, like r in the svn world
<oliv3r>
:p
<oliv3r>
anyway, it appears u-boot is broken (atleast for a20)
<mnemoc>
and hansg's for-amery is broken for the sun5i family
<oliv3r>
everything appears to be broken!
<mnemoc>
my budget too
<oliv3r>
heh, duh who's isnt
deasy has joined #linux-sunxi
<mnemoc>
:)
<oliv3r>
good samaritans don't have money :p
<mnemoc>
true
tinti_ has joined #linux-sunxi
benoitm974 has quit [Ping timeout: 245 seconds]
<oliv3r>
maybe i should change fields
<oliv3r>
I love blobs!
<oliv3r>
propriatary code all the way!
* mnemoc
jus read "I love boobs!" :|
<oliv3r>
also that
<oliv3r>
who in their right mind wouldn't
<mnemoc>
ack, but boobs are kind of OT
Super-noob has joined #linux-sunxi
Small-noob has joined #linux-sunxi
<oliv3r>
never.
Small-noob has quit [Client Quit]
<oliv3r>
boycottg00gle: it's only the SPL that appears to be broken, i can use an old SPL 2013.07-07792-g2968923 (is what I use) with head u-boot U-Boot 2013.07-08154-g48845ec
<oliv3r>
and look how beautifully those lines aline (well on my 80 char terminal they do)
<oliv3r>
and now things are really strange
<oliv3r>
i load the kernel from the sd card, the dtb from the sd card, and it boots the kernel from nand
benoitm974 has joined #linux-sunxi
ganbold_ has joined #linux-sunxi
<oliv3r>
oh maybe i lie
<oliv3r>
mripard: sid works on A20 aswell :D
<oliv3r>
will spend the next hour if i have time left to clean up my tree, then hopefully sned it off tonight
<oliv3r>
tomorrow probably u-boot fixing time :(
<mripard>
oliv3r: cool
<mripard>
what's the status of all your patches?
<mripard>
did gkh picked some?
benoitm974_ has joined #linux-sunxi
benoitm974 has quit [Ping timeout: 245 seconds]
<oliv3r>
mripard: the sysfs ones, yeah
<oliv3r>
mripard: the sid ones, none I don't think, but that's ok
<oliv3r>
i'll try to write the fdt documentation as it should too, and add A20 support in comments, so far, i don't know of sid in a31, but on the wits board it appears to be zero too i think and only the 'short' version
benoitm974_ has quit [Ping timeout: 245 seconds]
<mripard>
oliv3r: as a general principle, only care about the hardware you have.
<mripard>
it's always a good starting point :)
<oliv3r>
well you know i boycot the a31 :p
<oliv3r>
but i couldn't find anything sensible in a31
<Turl>
my cubie got an extra entry on the tracking woot
<oliv3r>
mripard: anyhow, i don't see why the driver shouldn't support it ;)
<Turl>
it's still
<oliv3r>
Turl: poor turl
<Turl>
on the post office* :P
hramrach has quit [Ping timeout: 240 seconds]
<oliv3r>
atleast they are performing labor, so progress
<Turl>
oliv3r: I need to move to russia, maybe I'll get my stuff faster
<oliv3r>
Turl: rofl
<Turl>
oliv3r: besides it's cold :D
<oliv3r>
Turl: did you read my ping above?
<Turl>
oliv3r: probably not
<Turl>
oliv3r: please resend :P
<oliv3r>
i forgot :(
<FR^2>
*lol*
<oliv3r>
Turl: mailling list, you responded to someone
<jinzo>
regarding china post, if anyone doesen't know - Singapore Post is wonderfull for me. Still cheap, and <14 days for everything I order
<Turl>
oliv3r: hm?
benoitm974 has joined #linux-sunxi
<oliv3r>
i haven't read all of the mails yet; but the ahci patch he sent, is it working?
<Turl>
he sent an ahci patch?
benoitm974 has quit [Client Quit]
<oliv3r>
i think he added sunxi support to the stadard ahci stuff
<eagles0513875_>
n01: ok :)
<oliv3r>
Re: [linux-sunxi] Re: [PATCH v0 1/1] [stage/sunxi-3.] use ahci platform driver for sunxi
<oliv3r>
it didn't seem like a huge patch
ykchavan has joined #linux-sunxi
shineworld has quit [Quit: Leaving]
<eagles0513875_>
n01: hopefully this time is the charm
<oliv3r>
so getting it to mailine status shouldn't be too much work; e.g. mmc u-boot + kernel; then sata!
<oliv3r>
Turl: but my real question, I most likly forgot :p
vinifr has quit [Ping timeout: 246 seconds]
<eagles0513875_>
hey Turl :)
<rz2k>
oliv3r: iirc it has TonOfCamelCase and whitespace damage
<rz2k>
so until Wills will run checkpatch and etc. there's no mainline for sata
<rz2k>
:p
<oliv3r>
rz2k: of course, but that's all easy
atiti has quit [Ping timeout: 264 seconds]
<oliv3r>
rz2k: i was thinking more along th elines 'i'll take his patch, clean it up and fix it up and submit it'
rellla has joined #linux-sunxi
<eagles0513875_>
holy crap batman :) n01 its working :)
<n01>
eagles0513875_: good
<n01>
eagles0513875_: do you reach the login?
<eagles0513875_>
n01: it drops me down here ** No partition table - mmc 0 ** sun7i# seems like it wants me to type some commands or something
<n01>
eagles0513875_: before submitting the patches make sure to get al least to the login
<n01>
paste the log
<Turl>
mripard: someone on the list is running their cb with mainline "24/24" :o
<mripard>
Turl: yeah, he told me quite some time ago
<mripard>
I'm impressed he only had 6 warnings in a 2 month time :)
<oliv3r>
wow, not bad at all
<bfree>
mine is usually on debian's ~mainline armmp kernel w/rootfs on nbd unless I boot it into a 3.4 to access the sd card to put on a new kernel or uboot or something. no complaints/issues that I've noticed
<mripard>
bfree: which board ? cubie as well ?
<bfree>
yes, cubie1
<Turl>
bfree: :)
<eagles0513875_>
ok guys i am making some progress with booting now the next issue i am having is no login screen. here is my current output to the serial port http://pastebin.com/Nff8PBrN
<eagles0513875_>
any ideas would be greatly appreciated
<n01>
eagles0513875_: it is not that you don't have login screen. Your kernel is not booting at all
<eagles0513875_>
Turl: any ideas as to the above as I am using the kernel you had suggested yesterday the latest revision on that branch
<boycottg00gle>
after updating uboot from 645e9015 to a3cca421 boot still works but hdmi is off again (seems like hdmi working / not working is quite random)
<mripard>
Turl: ah, it's not really that important then
<mripard>
but yeah, we should fix it
<Turl>
mripard: I can add a patch to my series fixing all of them after the fact
<Turl>
mripard: or we can respin all the not yet merged series to get it right there and then fix the remaining cases
<mripard>
I'll need to do some tweaks on those init sections as well for other drivers, so whatever you prefer
<mripard>
seriously, for that kind of stuff, I'd really like not to respin a new serie and get to the end of the to-merge list once again.
<Turl>
mripard: ok then, let's fix all of it after the fact
<Turl>
mripard: I'll handle clocks and you handle the other drivers
<mripard>
especially since the ARM merge window is almost closed by now, and neither the pinctrl or clocks patches are merged for no reason (yet).
<eagles0513875_>
not sure where those odd characters are coming from
<boycottg00gle>
oliv3r: for me 218f6438 works
<mripard>
Turl: for sid, we had some reasons
<mripard>
for watchdog, the maintainer seems to be MIA
<mripard>
if we miss 3.12, I'll carry it without his Acked-By in my branch
<boycottg00gle>
regression somewhere between 218f6438 or c4163747 and 48845ec7 ?
<n01>
eagles0513875_: I think uboot is not reading from sd
<eagles0513875_>
n01: where is it trying to read from?
<n01>
how do you create that card
<eagles0513875_>
i let the script here sudo ./mkcubiecard.sh ../../../output/images/ /dev/sdb format and put the stuff it needs on the card
<oliv3r>
boycottg00gle: i'm just confused on how to check what changed inbetween, but that's my lack of git-foo
<eagles0513875_>
n01: will paste you whats in the script
<n01>
yep that is my script and it works for cubie1
<eagles0513875_>
ok
<eagles0513875_>
whats next in the recipe change
<oliv3r>
boycottg00gle: is that working on a20?
<n01>
eagles0513875_: stop into uboot and give the commands one by one
<boycottg00gle>
oliv3r: using emacs/magit myself most of the time (ymmv)
<oliv3r>
<- vim!
<boycottg00gle>
oliv3r: working on olimex a20
<eagles0513875_>
n01: you mean dont let it autoboot
<oliv3r>
boycottg00gle: hmm, strange, because taht's before the merge
<n01>
yes
<oliv3r>
boycottg00gle: so it breaks in the last 2 patches?
<boycottg00gle>
oliv3r: no for me the old/sunxi-current works
FR^2 has quit [Quit: Connection reset by peer]
<oliv3r>
boycottg00gle: i'll test it tonight :)
<oliv3r>
218f6438 looks hopefull
<eagles0513875_>
n01: if i stop the autoboot it drops me down to what there was pasted earlier
<n01>
ok, then execute the commands
<boycottg00gle>
oliv3r: maybe i have some time to manually bisect further i will just paste here in irc
<oliv3r>
i'll keep an eye out ;)
<oliv3r>
but if ti's working, it means it the breakage is only int he last 2 patches
<eagles0513875_>
n01: sadly seeing as im building all this in a vm im having some issues where i cant type anything in the serial terminal i have running :(
<n01>
it's time to install linux w/o crap vm
<eagles0513875_>
i wonder n01 can i build with buidl root on mac osx?
<n01>
c'mon ... just install linux on mac
<eagles0513875_>
ok ok ok ill do that but that will have to wait until tomorrow
<n01>
take macports and install minicom
<eagles0513875_>
ok sadly i have to run now :( will continue working on this tomorrow
<eagles0513875_>
n01: could i request a branch be created where i can push and pull my changes from
<n01>
to whom?
jemk has joined #linux-sunxi
* n01
wonders if "whom" is correct *_*
panda84kde has joined #linux-sunxi
<eagles0513875_>
dunno for build root n01 so i can push all my changes there and then it can be merged to master when i have something which works
<n01>
eagles0513875_: just use a local repo on your machine and export changes using patches
<eagles0513875_>
ok
_BJFreeman has joined #linux-sunxi
_BJFreeman is now known as BJfreeman
eagles0513875_ has quit [Ping timeout: 250 seconds]
<Turl>
don't worry, typing on the wrong window has happened to all of us
<n01>
boycottg00gle: your mother is on IRC? cool
<boycottg00gle>
n01: ?
<boycottg00gle>
ah :)
<boycottg00gle>
moment
<n01>
:)
notmart has quit [Quit: notmart terminated!]
hramrach has quit [Ping timeout: 240 seconds]
hramrach has joined #linux-sunxi
iamfrankenstein has quit [Quit: Nettalk6 - www.ntalk.de]
<boycottg00gle>
hmm git bisect doesn't work out in that case
BJfreeman has quit [Read error: Connection reset by peer]
rellla has quit [Ping timeout: 264 seconds]
ZetaNeta has joined #linux-sunxi
notmart has joined #linux-sunxi
notmart has quit [Changing host]
notmart has joined #linux-sunxi
hramrach has quit [Ping timeout: 240 seconds]
boycottg00gle has quit [Remote host closed the connection]
massi has quit [Quit: Sto andando via]
<bfree>
anyone familiar with the kernel/initrd/dtb addresses care to look at http://paste.debian.net/28006/ and see if they can figure out why it would hang there? grub was ext4loaded to 0x4600000 and set to put itself at 0x4800000 and works fine until it tries to boot at which point grubs debug=all leaves hanging at that point in the paste (I'm compiling an earlyprintk kernel atm to see if that shines any light)
hramrach has joined #linux-sunxi
<steev>
grub?
<steev>
uboot is the bootloader, why are you adding in grub?
<bfree>
cause it's nicer (a usable menu, updatable config file, doesn't need kernels+initrds processed by mkimage, etc etc), cause I can (try), cause it's new(ish) and "just for fun". it sits on top of uboot
hramrach has quit [Ping timeout: 240 seconds]
n01 has quit [Ping timeout: 264 seconds]
hramrach has joined #linux-sunxi
hramrach has quit [Ping timeout: 240 seconds]
hramrach has joined #linux-sunxi
iamfrankenstein has joined #linux-sunxi
<oliv3r>
boybut that's a really really small patch with nothing related really
panda84kde has quit [Quit: Konversation terminated!]
steev has quit [Ping timeout: 246 seconds]
<oliv3r>
2013.07-07802-g218f643 works, confirmed
<bfree>
earlyprintk just gets me "Uncompressing Linux... done, booting the kernel." :-/
<bfree>
$grubdev guess is: "possibly your platform needs exporting of info about e.g. memory map (most likely) ... GRUB already has memory map. You need to pass it. It may be a variable in device-tree to set or it may be an additional command line argument." ... anyone have any ideas what that might be/mean/need?
jemk has quit [Ping timeout: 264 seconds]
<mripard>
bfree: how did you enable earlyprintk?
<mripard>
(plus, you should take a look at barebox if you want something nicer than u-boot :))
<bfree>
I stole the config from prior debian kernel, make oldconfig, make menuconfig, enabled LL and earlyprintk and set it to sunxi uart0
<mripard>
that should be right
<mripard>
you have to pass earlyprintk in the bootargs as well though
<bfree>
yep I did (otherwise I wouldn't have seen the "Uncompressing Linux... done, booting the kernel." unless I'm missing something, I wasn't seeing it before)
<oliv3r>
earlyprintk is set in bootargs i think by default (via u-boot that is)
<mripard>
he is not using u-boot
<bfree>
well grub on top of u-boot ... grubs arm-uboot platform
<oliv3r>
well then yer own your own bud!
<bfree>
I've had $grub_dev and mripard chip in, so not 100% on my own ;) but yes, I know I'm wandering off into the wild wild west
<mripard>
oliv3r: competition is always good, you should favor it :)
steev has joined #linux-sunxi
hramrach has quit [Ping timeout: 240 seconds]
<bfree>
as it leverages u-boot it's not really competition ;)
<mripard>
bfree: grub has no first-stage mechanism on ARM?
<bfree>
mripard: corrent, it uses u-boot API
<mripard>
so, for now, you have U-Boot SPL -> U-Boot -> Grub -> Linux, right?
<bfree>
mripard: correct
* mripard
wonders if the new U-boot Falcon Mode could be used
<mripard>
that would be cool :)
<bfree>
it's only recently merged to grub trunk having been around a while as a branch. I don't think they'd any interest in re-inventing all the board support from scratch ;)
<oliv3r>
falcon mode?
<oliv3r>
mripard: oh i'm not against it at all :p
<mripard>
oliv3r: only the SPL starting the kernel
<oliv3r>
dunno if possible
<mripard>
if it is possible to do what ?
hramrach has joined #linux-sunxi
<oliv3r>
spl is really small, 24k only
<oliv3r>
and i think it needs to do more then just load the kernel?
<oliv3r>
well, it inits dram; it loads 'a file to dram' and executes it
<oliv3r>
so maybe it could be possible
<oliv3r>
hno: ^
boycottg00gle has joined #linux-sunxi
<mripard>
anyway, can't you guys start working on barebox instead ? :)
<oliv3r>
*facepalm*
<oliv3r>
spreading thin resources is a bad idea however :p
<mripard>
yeah, well
<oliv3r>
how can i apply have a commit?
<mripard>
we're all hobbyist, you can't blame anyone for doing the stuff they find fun, and not the boring stuff that needs to be done :)
<oliv3r>
iw ant to test only parts of a patch
<oliv3r>
commit*
<mripard>
apply it, and play with stash/checkout -p ?
<oliv3r>
checkout -p? is that like add -p?
<mripard>
yes
<mripard>
but the other way around
<mripard>
you discard bits of a patch, instead of adding them
<oliv3r>
so i checkout badhash-1; then checkout -p badhash
<robb83>
anyone know why work qemu with "u-boot" elf binary and why not "u-boot.bin" ? (but tutorials use u-boot.bin)
sud0x3 has joined #linux-sunxi
rellla has joined #linux-sunxi
sud0x3 has left #linux-sunxi ["Leaving"]
<WarheadsSE>
robb83: considered `file u-boot.bin` ?
<robb83>
hmm, u-boot.bin: ERROR: vasprintf failed (Invalid or incomplete multibyte or wide character)
<mripard>
bfree: I have no idea :(
xtofury has joined #linux-sunxi
<mripard>
I don't see it patching the memory node
<mripard>
though
sdk has quit [Quit: Leaving]
<xtofury>
hey turl, if you're around, thanks for that tip about make mrproper....
<xtofury>
it's weird I had the dependencies installed, but when I went to deal with make menuconfig, getting the 64 bit libraries for libncurses5 worked, but I noticed it removed libncurses5-dev:i386
<xtofury>
so I tried to compile last night clean, and noticed it failed on the packer, missing u-boot.bin, and another bug with boot.fex --> I believe this is because of the removed dependency i386 version. so I put it back on, is there a way to get through this without another 12 hours of compiling to find out?
<bfree>
"patching the memory node is probably what's needed to be done"
<bfree>
bah phonecall just came to drag me out to meet someone ... will give me head some peace anyway for a little while I suppose :-p
<xtofury>
hmmmm u-boot is broken, maybe I'll be holding off on that a20 purchase for a month or two, lol....
<oliv3r>
2013.07-07802-g218f643i run now appearantly ok
mcbrick has joined #linux-sunxi
<hno>
oliv3r, is there any error message?
<oliv3r>
sort of
<oliv3r>
sec
<oliv3r>
please reset board :p
<hno>
very nice & informative.
<oliv3r>
yeah so useless
<oliv3r>
i can't spot any code changes that seem to be related
<hno>
Ok, I see what it is. The common spl expects u-boot.bin as an uimage, not a raw image.
<hno>
e4c444b34b6ff6b347e552805ee89c48f85785bb
wingrime has quit [Read error: Connection reset by peer]
<hno>
and load address needs to be adjusted -0x40 for that to work.
<hno>
(uimage header is 0x40 bytes)
<oliv3r>
ah ok
<oliv3r>
well atleast you spotted it really fast :D
<hno>
oliv3r, try "make u-boot.img" and use that instead.
<hno>
load address is already adjusted it seems.
<hno>
hm.. maybe it's not really..
<oliv3r>
i should start looking at cleaning up my old branches now too, i think my wip/a20 isn't needed anymore as that's merged, cleanup patches are merged too i think, good ogod
mcbrick has quit [Quit: Leaving]
mcbrick has joined #linux-sunxi
<boycottg00gle>
anything to test?
Tartarus_ has joined #linux-sunxi
Tartarus_ has quit [Remote host closed the connection]
<hno>
boycottg00gle, in u-boot yes.
<hno>
see above.
<hno>
<hno> oliv3r, try "make u-boot.img" and use that instead.
<hno>
boycottg00gle ^
ykchavan has quit [Read error: No route to host]
ykchavan has joined #linux-sunxi
<boycottg00gle>
ok will test
rz2k has quit []
<hno>
works for me on Cubieboard2
<boycottg00gle>
looks good
<boycottg00gle>
system up
<boycottg00gle>
now if only hdmi would work as well
xtofury has quit [Ping timeout: 268 seconds]
notmart_ has joined #linux-sunxi
notmart has quit [Read error: Operation timed out]
hramrach has quit [Ping timeout: 240 seconds]
n01 has joined #linux-sunxi
n01 has quit [Client Quit]
n01 has joined #linux-sunxi
<oliv3r>
so we have a fix? anything to pull ;)
<boycottg00gle>
<hno> <hno> oliv3r, try "make u-boot.img" and use that instead.
<oliv3r>
boycottg00gle: lol i saw that :p but i ment, is there something in git I can test over the manual fix :)
<oliv3r>
did the defaults for printing console stuff change in the last few weeks in u-boot?
<oliv3r>
before i never had to do anything, just load up spl + u-boot and load a earlyprintk uart0 enabled kernel
<oliv3r>
mripard: is it correct that A20 only shows 1 cpu in /proc/cpuinfo and that hardware reads as: Allwinner A1X (Device Tree) (instead of A2x)
fredy has joined #linux-sunxi
<mripard>
oliv3r: yes
<oliv3r>
mripard: why does it say A1X instead of A2x? (and why not A1x :p)
<mripard>
is it important?
<oliv3r>
purely cosmetic :p
<oliv3r>
mripard: oh and hno said falcon mode should work (in theory) and we do init pretty much everything in SPL
akaizen has quit [Ping timeout: 276 seconds]
<hno>
oliv3r, and then I said that there is some more CONFIG_... flags needed that we do not set.
<oliv3r>
yeah, which is why I said (in theory) :)
<oliv3r>
rellla: reading an email from 19th of august on ML; Patrick Wood says A20 cedarX doesn't really work right now, benn huang sent 2 A20 cubies to Allwinner devs and see if they care to look into it
akaizen has joined #linux-sunxi
shineworld has quit [Quit: Leaving]
ZetaNeta has quit [Ping timeout: 245 seconds]
iamfrankenstein has quit [Read error: Connection reset by peer]
iamfrankenstein has joined #linux-sunxi
fredy has quit [Ping timeout: 245 seconds]
fredy has joined #linux-sunxi
fredy has quit [Ping timeout: 245 seconds]
ZetaNeta has joined #linux-sunxi
deasy has quit [Quit: Quitte]
<[7]>
hno: sorry for contacting you directly with this, but I think you're involved with sun4i MTD and boot0/1 replacement work, right?
<[7]>
if so, can you give me a quick overview about the status of that work? is it already possible to completely get rid of allwinner's FTL and write a (working) uboot SPL to the raw NAND instead, and use yaffs2 or something from there on?
<[7]>
if that's already possible I'd love to check it out and experiment with it :)
<hno>
[7] I don't think yaffs2 works yet. oob data is not supported yet, and I do not know if the oob data made available by the Nand ECC controller is sufficient for yaffs2.
<[7]>
well, I'd prefer yaffs2 if possible, but would be fine with any flash file system that's faster than this crappy FTL for now
<[7]>
that might be one of the areas for me to experiment with :)
<hno>
I don't think the speed issues is due to the FTL really..
<[7]>
interesting... I'd want to experiment with it regardless. this proprietary allwinner stuff is just annoying me
<[7]>
(and I've just experienced filesystem corruption while using it)
fredy has joined #linux-sunxi
<[7]>
so can this be done already? if so, which git branches (or patches?) would I need to use?
<[7]>
(also what's the matter with OOB data? which data does the nand ECC controller "make available"? do you mean that the ECC controller is putting the checksums at a specific position within the OOB bytes where it might conflict with other stuff?)
<hno>
The Nand controlle rhave hardware ECC + Randomizer support, and as such manages the OOB data area as well.
<hno>
There is support for "user data" in the ECC area, but I am not sure how much is made available. But I am sure it is no more than 32 bit per 1K data.
<[7]>
ouch, that's small
<[7]>
probably too small for yaffs2 indeed. do other flash filesystems require less space there?
<[7]>
it's also kinda odd because it seems to be incompatible with the way how flash memory manufacturers typically mark factory-known bad blocks?
<hno>
UBI don't use OOB at all.
fredy has quit [Ping timeout: 245 seconds]
n01 has quit [Quit: leaving]
benoitm974 has joined #linux-sunxi
<hno>
bad block management is a chapter on it's own. OOB bad block markers is only meant to be used on initial (format time) scan of the NAND. After that the driver is supposed to maintain it's own tables.
<[7]>
...which would screw up that information in case it is ever reformatted?
<[7]>
both yaffs2 and the Whimory FTL that I've worked with kept that byte as is, assuming anything other than 0xff means bad, and writing an implementation-specific value to own discovered bad blocks
<hno>
To access (read) raw OOB data the ECC + randomizer is disabled. This is also needed to detect blank pages. But as above, once formatted the driver really should not need to do this.
benoitm974 has quit [Ping timeout: 245 seconds]
<hno>
Doing a complete reformad of a Nand is generally a bad idea. You want to preserve erase counters and some other data.
<[7]>
for file systems which even have those (yaffs2 doesn't). and I fully agree for FTLs, as long as nothing got corrupted.
<[7]>
ok, so assuming I go with UBI... what is necessary to get that set up?
benoitm974 has joined #linux-sunxi
benoitm974_ has joined #linux-sunxi
benoitm974 has quit [Ping timeout: 245 seconds]
benoitm974_ is now known as benoitm974
fredy has joined #linux-sunxi
<hno>
[7] you need the mtb branch of kernel & u-boot.
<[7]>
of which repositories?
<[7]>
doesn't seem to exist in linux-sunxi/linux-sunxi, so I guess I need a special one?