USB HOST MODE! (Coming soon)

[Frappacino]

they must be ex-blizzard devs (this is a compliment btw)

it is done when it is done

[sophocha]

oh, wow, this is great.Kudos to everybody for trying to work this out!

[chivar]

Originally Posted by Frappacino

they must be ex-blizzard devs (this is a compliment btw)

it is done when it is done

hahaha just like starcraft when it was anounced several years back

[joerg_rw]

Just to keep you updated: We're experimenting with PR1.3 kernel, 2 different versions. One with SarahN's and Mohammad's and my patches, the other one a complete "rewrite from scratch" based on stock kernel and doing "the right thing" by PaulFertser.
Results are quite promising. Paul got first connects and ENUM fails (whitelist etc foo, as usual) with his shiny new version.
Keep fingers crossed we'll not run in roadblockers again :-D

[edit]Mohammad has confirmed hostmode basically working in Paul's kernel. "Stop bme; Echo host >mode; start vboost; echo F >/proc/driver/musb_hdrc;". Even works for multiple connects without booting, when properly reverted. We'll sort out things like whitelist, force_session (F), FS/HS/LS auto-detect etc during the next days[/edit]
[edit2]untested first take, so you might get the picture more easily: http://pastebin.com/NgDufWej[/edit2]
[edit3]
[2010-10-31 11:09:05] <PaulFertser> It's Nokia who ****ed up the WHITELIST functionality!
[2010-10-31 11:11:51] <PaulFertser> It works as advertised in the mainline but nokia decided that's not good enough for them and screwed it up without changing the docs even.
[2010-10-31 11:28:59] <PaulFertser> [ 215.335906] generic-usb 0003:04D9:1702.0001: input: USB HID v1.10 Keyboard [ USB Keyboard] on usb-musb_hdrc-1/input0
[2010-10-31 11:29:34] <PaulFertser> (the latest commit to git makes the kernel without whitelist behave)
[2010-10-31 11:33:14] <PaulFertser> This keyboard is actually low-speed and autodetection doesn't work yet.
[2010-10-31 11:34:49] <PaulFertser> But i can type on it etc.
[2010-10-31 11:35:43] <SpeedEvil> you mean you force low-speed somehow?
[2010-10-31 11:37:47] <PaulFertser> SpeedEvil: yes, the only difference of my local tree comparing to h-e-naul-experimental is a 2-lines patch to force low-speed.
[2010-10-31 11:39:10] <PaulFertser> And after i did boostoff, hostoff, start bme, i could connect it to my laptop all right.
[2010-10-31 11:41:43] <PaulFertser> And to a charger as well (though it required another restart of bme, probably that's my "tweak" to twl4030, not sure).
[2010-10-31 11:44:28] <PaulFertser> So heh, i get first really working hostmode here (with my own patches)
[/edit3]

cheers
jOERG

[MohammadAG]

Slight correction to the above.

echo host > /sys/devices/platform/musb_hdrc/mode
echo F > /proc/driver/musb_hdrc
*connect*
VBUS

No timing issues involved, I accidentally disconnected the hub, and was able to re-enumerate easily with echo F..., compiled a kernel without WHITELIST and... let's just say that didn't go well

joerg_rw's point still stands, it's cleaner than before, and works well (with hubs at least)

[joerg_rw]

Originally Posted by joerg_rw

[2010-10-31 11:44:28] <PaulFertser> So heh, i get first really working hostmode here (with my own patches)

Congrats Paul :-D
Brilliant work!

yes, next thing pending is a Alpha release :-D


[2010-10-31 13:18:58] <PaulFertser> DocScrutinizer: "released" my scripts to the "Files" section.
...
[2010-10-31 13:29:00] <PaulFertser> And i need to actually change the patch to a saner way, the code should be the same for both CONNECT irq and manual connect request.
[2010-10-31 13:29:18] <PaulFertser> Also i need to sort it out about FS/HS mess. Currently only LS works reliably.
...
[2010-10-31 13:34:58] <PaulFertser> I see no sense in releasing something that doesn't work with mass storage.




[2010-11-03 23:30:10] <PaulFertser> DocScrutinizer: funny thing, i've somewhat reproduced your result of connect after writing TESTMODE=0
[2010-11-03 23:34:36] <PaulFertser> DocScrutinizer: this trick (disable FORCE_HOST) doesn't work with my FR but i wonder if it enables to get true HS.
[2010-11-03 23:35:07] <DocScrutinizer> possibly
[2010-11-03 23:37:02] <DocScrutinizer> I guess my log of successfully connecting to memstick should be 2.0 though, as they won't sell a 1.1 2GB stick, will they?
[2010-11-03 23:41:32] <PaulFertser> DocScrutinizer: probably so, i'm testing myself atm.
[2010-11-03 23:44:31] <PaulFertser> Nov 4 01:44:07 192.168.1.132 kernel: [ 458.688720] usb 1-1: new high speed USB device using musb_hdrc and address 25
[2010-11-03 23:44:39] <PaulFertser> Heh so it works
[2010-11-03 23:46:07] <DocScrutinizer> on N900? \o/
[2010-11-03 23:46:14] <PaulFertser> With my latest kernel and b51fe29c61a2fa167513b2ecdb0fffcb30eab6d6 reverted (just reenable the HS) and this sequence: hoston (TESTMODE<-FORCE_HOST, disable IRQs), booston, echo A > /proc/drivers/musb_hdrc (TESTMODE<-0), that's ****ing all needed it seems.
[2010-11-03 23:46:36] <DocScrutinizer> lemme post that :-D

[2010-11-03 23:46:56] <PaulFertser> DocScrutinizer: and LS is still supported with "hoston; booston; mconnect"

[2010-11-03 23:47:18] <PaulFertser> (where mconnect being echo F > ... and it does the same that CONNECT irq does)
[2010-11-03 23:48:57] <PaulFertser> Apparently some minor delays needed between booston and echo A.
[2010-11-03 23:49:26] <PaulFertser> Yep, hoston; booston; sleep 1; echo A > /proc/driver/musb_hdrc
[2010-11-03 23:57:52] <DocScrutinizer> only question now: will it blend, err properly disable hostmode and revert to normal operation?
[2010-11-04 00:00:34] <PaulFertser> Ok, pushed.
[2010-11-04 00:01:27] <PaulFertser> DocScrutinizer: yes, works fine after "boostoff; hostoff; start bme".


We'll get this into kernel soonish :-D. No need to patch hostmode.py


[Do 4. November 2010] |[10:12:48] <DocScrutinizer>| ok, til tonight I might find the time to finalize jrbme script with hostmode and vbus control (plus maybe charging, bq27200 reports, logging, notifications, etc)
[Do 4. November 2010] |[10:13:06] <DocScrutinizer>| so we should see an Alpha this week
[Do 4. November 2010] |[10:13:29] <PaulFertser>| Sounds great
[Do 4. November 2010] |[10:14:16] <DocScrutinizer>| I thought about red flashing indicator LED for notifying VBUS boost is eating battery (in hostmode)

alas I have to cope with this thread here

then
[Do 4. November 2010] |[21:26:18] <MohammadAG_>| DocScrutinizer, don't use the python version, please
[Do 4. November 2010] |[21:26:24] <MohammadAG_>| And don't use noob packaging :P
...
[Do 4. November 2010] |[21:37:27] <MohammadAG>| sigh, why do I even discuss this **** with you
[Do 4. November 2010] |[21:37:32] <MohammadAG>| you always argue with invalid facts
[Do 4. November 2010] |[21:37:44] <MohammadAG>| I'm getting bored of all of this
[Do 4. November 2010] |[21:37:55] <DocScrutinizer51>| sorry then

So I decided not to bother anymore about this thread and other annoying debates, and rather take some days off

cya all in a week or sth


if you want to express your excitement, please don't spam this thread! there's a thanks button right down there ->

[Larswad]

Let me say joerg_rw and MohammadAG that we're all very glad to have you doing this work, it is really promising!
At the same time let me say in the most kind way: I am waiting impatiently like hell for OTG on the N900.
I hope some day it ends up in extras.

The N900 won't be able to power a 2.5" harddrive right? Has it been tested by someone here yet?
And USB gamepad of some sort would be interesting for some games and emulators.

[!!Nokia N900!!]

when it is expected to be in Extras-Testing ??

[jacktanner]

Originally Posted by !!Nokia N900!!

when it is expected to be in Extras-Testing ??

When it's ready.

[!!Nokia N900!!]

Originally Posted by jacktanner

When it's ready.

What an answer. Of-course when it is ready. but i mean how fast??