smartboyhw | xnox, you here? I want some bit of help... | 06:37 |
---|---|---|
smartboyhw | xnox, forget about the help now | 07:49 |
=== ogra-cb__ is now known as ogra-cb | ||
smartboyhw | cjwatson, er can you approve the Ubuntu Studio -> Wubi code merge? | 14:24 |
cjwatson | smartboyhw: done, and bumped everything to 13.04 since I don't think it makes sense to build more versions for quantal at this point | 14:31 |
smartboyhw | cjwatson, thank you:D | 14:31 |
cjwatson | ev: could you build wubi r275 for raring, please? | 14:31 |
ev | cjwatson: will do now | 14:49 |
cjwatson | ta | 14:49 |
ev | cjwatson: done | 15:08 |
* ogra_ still waits for the first wishlist bug for wubi on win RT :) | 15:09 | |
cjwatson | ogra_: which reminds me, I was wondering how totally insane it would be to use something wubi-esque as a way to dual-boot android and ubuntu on the nexus7 | 15:12 |
cjwatson | conditional on having the grub port working of course | 15:12 |
cjwatson | it almost makes more sense on a constrained device than on a full desktop/laptop system, in some ways - fewer worries about upgrade / disk growth / etc. | 15:13 |
ogra_ | heh, well, might be possible though there are easier ways for dual boot (just flash the bootimg to the recovery partition) that simply the userspace (flash-kernel) doesnt cover | 15:13 |
ogra_ | doesnt lupin use a readonly squashfs ? | 15:13 |
* ogra_ has never looked deeply into wubi | 15:14 | |
cjwatson | no, it's an ordinary filesystem | 15:14 |
cjwatson | the weirdness is that it's all loop-mounted on a single file in the containing fs | 15:14 |
cjwatson | and the boot loader and initramfs conspire to make that as invisible as possible | 15:14 |
ogra_ | needing overlayfs ? | 15:15 |
cjwatson | no | 15:15 |
cjwatson | it's just ext4 | 15:15 |
cjwatson | the inner filesystem doesn't actually matter | 15:15 |
ogra_ | ah, great, since thats usually missing in the kernels on these devices | 15:15 |
cjwatson | it does need a smart boot loader, and it needs the loop module, but aside from that the required kernel support is not desperately advanced | 15:15 |
* ogra_ will take a look, is grub an actual requirement ? | 15:16 | |
cjwatson | realistically yes | 15:16 |
ogra_ | hmm | 15:16 |
cjwatson | but the grub arm port was coming along quite well last I looked | 15:16 |
cjwatson | (need to find time to go back and supply more review feedback) | 15:16 |
ogra_ | well, you will likely need to chainload it somehow, else it would need very close HW support | 15:16 |
cjwatson | it's grub on top of uboot | 15:17 |
cjwatson | the linaro guys are doing it | 15:17 |
ogra_ | hmm, so the question is if it also works on fastboot | 15:17 |
cjwatson | so it only needs a relatively small amount of hardware-specific gubbins | 15:17 |
cjwatson | that I don't know | 15:17 |
ogra_ | i would guess so, but needs to be verified | 15:17 |
cjwatson | you can sort of do it without grub if you're prepared to tolerate copying the kernel and initrd out of /boot to somewhere the loader can read it - but for dual-boot support you still want a menu anyway | 15:17 |
ogra_ | which then gets hairy on something like the nexus7 | 15:18 |
cjwatson | and the userspace support for wubi without grub2 has almost certainly atrophied | 15:18 |
cjwatson | (grub legacy wasn't smart enough to do the loop-mount dance) | 15:18 |
ogra_ | we would need sme HW key support for the vol up/down keys | 15:18 |
cjwatson | you'd need a touch driver or hw key support, yes | 15:18 |
cjwatson | how do they present? | 15:18 |
ogra_ | and some trick to generate an "enter" as well | 15:18 |
ogra_ | just /dev/input | 15:19 |
cjwatson | I mean to the kernel | 15:19 |
cjwatson | just gpio pins or something? | 15:19 |
ogra_ | hmm, i would have to check, but likely | 15:19 |
cjwatson | probably wouldn't be that hard to write a really stupid grub input driver for them if it came to it | 15:19 |
ogra_ | yeah, just saying, menus arent easy on tablets in console mode :) | 15:20 |
ogra_ | the annoying thing about switching image models is sadly that flash-kernel is still living off a hardcoded DB without even the capability to override, so it currently can only support one type of booting per device | 15:21 |
ogra_ | though i guess when we use grub we can completely skip flash-kernel and rely on having the right stuff in /boot | 15:24 |
cjwatson | sadly it just moves the goalposts | 15:27 |
cjwatson | yes, you don't need to tell uboot where to find the kernel/initramfs | 15:27 |
cjwatson | but you do still need to tell it where to find grub | 15:27 |
cjwatson | which is an equivalent problem and you'd almost certainly end up using nearly all of the same code | 15:27 |
ogra_ | well, i wont change the fastboot/u-boot config after install and grub is unlikely to move around, so that part is fine | 15:30 |
ogra_ | the only thing a user might touch is the cmdline and thats living in grub-land | 15:31 |
cjwatson | it won't move around, but the installer still needs to know how to put it in place | 15:34 |
cjwatson | and afaics that's still going to involve something that looks startlingly like flash-kernel | 15:34 |
ogra_ | heh, yes, well, it could just use flash-kernel (or some of its functions) | 15:41 |
ogra_ | that missing override functionality is definitely a bug i want to fix before release, but its more whishlist than task | 15:42 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!