gzamboni has quit [Read error: Operation timed out]
<hno>
lkcl, that uenv.txt change do not look right.
<hno>
lkcl, and the fex file is in sunxi-boards these days.
<hno>
no need to clone Allwinner-info.
<hno>
And... easier if you just drop the FAT partition completely. It is not needed. Use /boot in the rootfs instead.
Dessimat0r has quit [Quit: Computer has gone to sleep.]
steev has joined #arm-netbook
<vgrade>
hno: it is possible to use the old method of booting with things on sdcard?
ganbold__ has quit [Ping timeout: 252 seconds]
ganbold_ has joined #arm-netbook
ganbold__ has joined #arm-netbook
ganbold_ has quit [Ping timeout: 252 seconds]
eebrah_ has joined #arm-netbook
<lkcl>
hno: ahh ok
eebrah_ has quit [Read error: Connection reset by peer]
<lkcl>
ooooo, usb host networking is recognised off the eoma68 interface, ooooo. nearly got a keyboard and mouse and stuff
<lkcl>
wooow
eebrah_ has joined #arm-netbook
<penguin42>
lkcl: Find some that actually work and you could make a little router appliance with an EOMA slot
<lkcl>
penguin42: yeah :)
<hno>
vgrade, what do you mean?
<vgrade>
hno: with cubie, mele a1000 and others we prepare a sdcard with xxx-spl.bin and u-boot.bin before first part and then in first part have kernel etc with rootfs in second part
<vgrade>
can we still use this setup with eoma
wingrime has quit [Ping timeout: 264 seconds]
<lkcl>
vgrade: no because wits-tech fucked up the board. correction: the board is tiny, extremely well-made with 2 GND planes and only 6 layers.
<lkcl>
there's not enough room for the SD0 signals to cross over to get to the SDcard slot
<lkcl>
so instead you can use usb-fel
<hno>
vgrade, EOMA can't boot from sdcard. But other than that yes.
<lkcl>
or you can put spl.bin and u-boot.bin onto NAND
<lkcl>
and ask it to detect SD3
<hno>
yes, that works.
ganbold has quit [Ping timeout: 252 seconds]
<vgrade>
right understand now, thanks
<vgrade>
so fixed for the first batch
<lkcl>
when we've sold a few thousand cards, we'll have the proper finance to do a respin. use the 2 16-bit cheap-and-cheerful chinese DRAMs
<lkcl>
vgrade: no. the first lot you'll have to jump through a few hoops (including using the really really easy usb-fel in sunxi-tools)
<hno>
vgrade, there is no need for a separate boot partition on any A1x board.
<lkcl>
hno wrote fel-boot.bin
<vgrade>
so fel-boot enables spl.bin and u-boot.bin into nand
<lkcl>
no fel-boot.bin *replaces* spl.bin
<vgrade>
to be written
<lkcl>
and once you've loaded fel-boot.bin with fex you can then push u-boot.bin *directly* over into DDR memory
<lkcl>
here's an example:
<hno>
fel-boot.bin is a stripped down SPL. It's source is only some u-boot config settings and a linker script.
<lkcl>
+ fel write 0x2000 fel-boot.bin
<lkcl>
+ fel exe 0x2000
<lkcl>
+ fel write 0x4a000000 u-boot.bin
<lkcl>
+ fel exe 0x4a000000
<vgrade>
ok and that needs to be done one time only for each board
<lkcl>
so it replaces the fuck-awful proprietary phoenix tools and livesuit with a 7k bootloader and a 1000 line c program (fex.c)
<lkcl>
vgrade: no, you do that every time. it's incredibly handy and very quick.
<lkcl>
it *is* livesuite, except command-line, and no fucking about making an image.
<hno>
lkcl, once only when we finish the NAND parts..
<lkcl>
thanks to hno i even had it loading an initramfs directly into memory
<vgrade>
was wondering how you would make a tablet which had to be kickstarted from usb
<lkcl>
vgrade: you don't.
<lkcl>
why would anyone want to do that?
ganbold has joined #arm-netbook
<lkcl>
somebody else might, but i am not going to be doing any such project.
<lkcl>
vgrade: did you imagine that i would be involved with such a project?
<vgrade>
so we still need extra work to get eoma booting standalone?
<lkcl>
vgrade: eoma is a standard. eoma68 is also a standard. you must be referring to the eoma68-a10 cpu card, is that right?:
<vgrade>
yes
<hno>
vgrade, yes and no. You can drop in the eoma64 u-boot.bin in nanda and it will boot from sdcard.
<lkcl>
vgrade: what hno said :)
<lkcl>
so you boot using USB-fel
wm_ has joined #arm-netbook
<lkcl>
you get into an initramfs (for example)
<lkcl>
you put u-boot.bin into the initramfs
<lkcl>
a program in there writes that u-boot into nanda
<lkcl>
and you then power off.
<hno>
there remains some work to do this without having allwinner bootloeader installed in nand.
<lkcl>
forever thereafter, because that u-boot (and spl.bin) is now on nanda, that device will always boot from NAND
<lkcl>
*sigh* yehhh....
<lkcl>
yuk
abesis has joined #arm-netbook
<hno>
but it's well within grasp. Nothing magic needed.
<lkcl>
i think aseigo will amongst other people be looking forward to that.
<lkcl>
hno: is it something that perhaps one of aseigo's team could handle?
eFfeM has quit [Quit: Leaving.]
<hno>
there is two tasks needed
<hno>
a) Reshuffle the sunxi mtd code so it fits properly in the mtd framework
<hno>
b) Update u-boot mtd framework so it can handle the large page MLC flash chips seen on most Allwinner devices.
<lkcl>
ok
<hno>
I don't know what people is on aseigo's team, but they are very welcome to join the effort.
<lkcl>
hno: they've got some good people. worthwhile seeing if any of them have some spare time
<hno>
Oh, right.. one more task. Sort out the license of the u-boot DMA driver for Allwinner chips, or alternatively reimplement it. Current license header is not acceptable.
<lkcl>
:)
<lkcl>
hno: we have word from eva that all code is to be GPL'd. if you can take that as an official announcement, it means that we have explicit permission to add headers on their behalf.
<hno>
The word is that it's meant to be GPL, but was overlooked in the license cleanup earlier.
<lkcl>
acknowledging their copyright
<lkcl>
it's probably a good idea to tell them though :)
<vgrade>
hno: I'm confused, you say drop in u-boot.bin into nada and it will boot from sdcard. Then "there remains some work to do without having allwinner bootloader installed in nand".
<hno>
I do not like touching others copyright header when it says "All rights reserved"
<vgrade>
so are you saying you would have to have allwinner bootloader AND u-boot.bin in nand to boot from sdcard
<lkcl>
vgrade: as well as putting the allwinner bootloader in. or doing the work hno mentioned
<lkcl>
vgrade: yes.
<hno>
vgrade, you flash the board once with livesuid to get allwinner bootloader installed, then replace u-boot.bin with the one we want.
<lkcl>
vgrade: as you can imagine we're pretty pissed with wits-tech for not telling us that this was what was going to be done.
<hno>
well, we will sort out the NAND boot are part before too long.
<hno>
boot area part.
<lkcl>
yay
<vgrade>
hno: after loading the allwinner bootloader with livesuite how is the u-boot.bin written to nand?
<hno>
The nanda partition can be accessed from whatever system is installed in the NAND, or sending the character 2 on the console when boot1 starts.
<hno>
when doing the latter boot1 enters USB mass storage mode and exports the partition as an usb mass storage device.
<vgrade>
right ok
<hno>
but as we have no LRADC keys on the card there is no relable method to exit mass storage mode, so often requires a couple attempts before it gets written properly.
<hno>
err.. it's 1 for mass storage mode. 2 is FEL mode.
<vgrade>
so livesuite writes both boot0, boot1 and u-boot.bin and I assume a filesystem image to nand all in one go
<hno>
yes
<hno>
u-boot.bin is in one of the filesystem images it writes (nanda)