15:26 <Wizzup> pali still around?

15:26 <Wizzup> ah, nvm, didn't read properly.

16:44 <Wizzup> Pali: quick question, does usb host mode work on the mainline kernel?

16:44 <Wizzup> I see there's a lot of awesome work done on mainline kernel for the n900

16:44 <Pali> Wizzup: I did not tested it

16:44 <Wizzup> I'm going to build it in a bit, but I am wondering if I can just try to boot maemo with it, or if that is a stupid idea

16:44 <Pali> maybe...

16:44 <Wizzup> loving usb host mode with charging btw

16:44 <Pali> and if it work then it have other API!

16:44 <Wizzup> yeah, I was afraid of that

16:45 <Pali> (incompatible with existing gui apps)

16:45 <Wizzup> basically I want to get usb displaylink working with the n900 usb host mode

16:45 <Wizzup> and that driver appeared somewhere in 2.6.38

16:45 <Wizzup> so I can either try to backport it, or just try a much newer kernel

16:45 <Pali> try to backport it

16:45 <Wizzup> [I have a few n900 devices I can toy with]

16:45 <Wizzup> ok

16:45 <DocScrutinizer05> N900 usb hostmode _never_ will work on any mainline kernel, since it needs nasty hacks

16:46 <Wizzup> DocScrutinizer05: mainline kernel + a few patches is also mainline for me ;)

16:46 <Wizzup> I just mean any _recent_ version

16:46 <DocScrutinizer05> the hacks we developed are independant of kernel version

16:47 <DocScrutinizer05> basically

16:47 <Pali> DocScrutinizer05: upstream kernel exports that debug registers to userspace

16:47 <DocScrutinizer05> ooooh :-)

16:47 <Pali> so you could be able to implement it in userspace...

16:47 <DocScrutinizer05> yes, indeed

16:47 <Pali> but I have no idea, if it is really works

16:47 <Wizzup> is there any documentation for running mainline on the n900, besides the repos + elinux guide on how to build it?

16:47 <Wizzup> I guess I should just try, like with most things ;)

16:47 <Pali> I saw registers exported via some debugfs entry

16:48 <Pali> no idea if it works or not

16:48 <DocScrutinizer05> ,ight actually work

16:48 <Wizzup> backporting is probably nice as well, since then other people could benefit

16:48 <DocScrutinizer05> I mean, it's no witchcraft

16:48 <Pali> musb is buggy...

16:48 <Wizzup> would need to get familiar with the sdk/stuff again, but doable probably

16:49 <Wizzup> still, in the long run, mainline kernels are cool :)

16:49 <DocScrutinizer05> there's one gotcha that you can't change the debug registers anymore when session established (aka VBUS applied)

16:49 <DocScrutinizer05> some idiocy in musb core

16:49 <DocScrutinizer05> the hardware

16:50 <DocScrutinizer05> the Mentorgraphics USB is really nasty weird critter

16:52 <DocScrutinizer05> instead of simply telling software about events and software statemachine would reconfigure the hardware, the musb core implements its own hw state engine that you can't control via software. nevertheless you need a second state machine in sw to stay in sync of what's going on in hardware statemachine

16:55 <DocScrutinizer05> IF you could control hw state machine from sw, you easily could force it into hostmode even without a "ID pin grounded" signal coming up the ULPI interface bus from PHY

16:57 <DocScrutinizer05> this been the whole crux why initially (without having the in depth knowledge yet) _I_ said "of course we make that critter do hostmode", and Nokia engineers (with a tad more of knowledge) told "it can't get done", then finally I found that with debug registers it still is sort of possible, alas in a very clumsy way

17:01 <DocScrutinizer05> the rest was "simply" tweaking the software so it wouldn't work as usual but rather do a special debug mode, plus kicking out of a few obstacles that are mandatory from USB specs, like whitelist/blacklist of devices and power negotiation in ENUM etc

17:02 <DocScrutinizer05> basically the latter was needed to turn OTG into pretty much standard USB host

17:03 <DocScrutinizer05> e.g. OTG doesn't allow hubs at all

23:14 <DocScrutinizer05> #7 \o/

23:14 <DocScrutinizer05> 7/20