Upstream kernel has already debugfs for musb. I do not know if it is enough for forced host mode or not, but due to bug changes in upstream musb driver and usb code, we cannot backport upstream musb to 2.6.28.

Another problem with enumerating is somewhere in kernel modules. Sometimes it is needed to unload and load g_file_storage driver, otherwise enumeration failing and flooding dmesg with lot of errors...

I really do not know why reloading g_file_storage is needed (maybe bug in ulpi/usb/gadget code?), but it fixed more errors when I used *my* usb devices. Last time when I looked into H-E-N code it does *not* unload & load gagdet driver and my usbmode.sh script doing it. So this is reason why I suggest also try to use usbmode.sh and not only H-E-N. I do not know if somebody else has similar problem with g_file_storage or not.

And if I want to start usb host mode with charging I must do: disconnect charger, switch to peripheral mode, then switch to usb host mode with charging and then after enumeration connect charger and change boost mode back to charger. I'm not able to enable host mode (+enumeration) with attached charger.

I do not have time, so I cannot look at problems and debug it...

All tests I've described were repeated (painfully) with BOTH usbmode.sh and H-E-N - of course, one after another, not at the same time So, the "main" problem that's hitting everyone doesn't seem to be related.

Correct, I wasn't implying that one can switch from peripheral to hostmode with charger attached (BTw, I think you mean "(...)then switch to usb host mode with boost(...)" after peripheral).

What I meant, is that you can get through all steps to get working device in hostmode with charging, THEN switch to Peripheral, THEN disconnect charger, THEN (optionally) physically de-attach USB client device (or not, doesn't matter), THEN enable hostmode with boost again, and get working device without problems.

But, if you try the same after hoostmode with boost only (without going to "hostmode with charging" step), after switching to peripheral, and trying "hostmode with boost" again, you won't - most likely - get enumeration, successfully. You will need to connect charger for a second - or do mumbo-zumbo rituals to "fake" connecting to charger, probably, from terminal - to be even able to enumerate device again.

For me, it happens 100% of time, if I try to re-enable hostmode after short time (<minute to few minutes) after last using of hostmode. I haven't checked if it "tickles" when attempting it after longer time.

Anyway, the point is (was), that if charging is present as last step oh hostmode before going back to peripheral - or, after it, but before another hostmode attempt - it works like charm. Otherwise, it tends to fail utterly.

/Estel

Maybe you can ask musb devs... I think I cannot help with this problem.

I'm working on a daemon script to connect a rtl2832 dvb-t key to the N900 to grant a good percentage of successful connections (while monitoring the battery status and current absorption) and i can confirm that very often reload the g_file_storage module unlocks the musb.

The typical dmesg message that identifies a no-go situation is:
"khubd timed out on ep0in"

So, at this moment i'm using a try-and-check approach, enabling hostmode, monitoring syslog messages and restarting hostmode and enumerate when needed.
Initially i was worried about the current absorption, but now that i'm monitoring it, i must say that the problem is absolutely not there but it's probably due to a poor bus initialization that changes completely his behavior at different speeds.

Actually, the musb driver that we are using into kernel power seems to be aligned with the TI sdk:
OMAP35x-psp-sdk-02.01.03.11.tgz

Is it possible to use the drivers from the OMAP35x-PSP-SDK-03.00.01.06.tgz ?
ref:
http://processors.wiki.ti.com/index....beta_3_release

or:
http://processors.wiki.ti.com/index....elease_Package

Hi guys!
I was wondering if someone could post most up-to-date version?
Thanks!

Of what?ten chars

I would like to use N900 with an external hdd and charge it at the same time. Please point me where to read how.

http://talk.maemo.org/showthread.php...203#post921203

Do I get it right that I should get:
1. External HDD
2. Y-Cable for external HDD
3. USB charger with at least 1A output
4. USB AF-AF adapter
5. Any USB cable for n900
6. (Optional) USB extender for charger so that I don't have to place hdd near the power adapter

And then do:
1. Connect everything together, do not power up charger
2. Set High-Speed hostmode, vbus boost, enumerate and mount
3. Plug in / power on charger
4. Enjoy?

After enumerating and pluging in a charger (mount is neutral here, not required), you need to disable vbus boost, too - not from H-E-N (as it will disable whole hostmode), but from terminal (boostoff).

If you want to do the same, but via much more convenient way, just use Pali's USB mode applet (of course alongside hostmode enabled kernel, like kernel-power). There, you just need to:

1. Click "USB Host mode (with boost)" button (it will select speed, booston, enumerate etc all for you, "automagically").

2. After step 1 ends its job, plug-in charger.

3. Click "USB Host mode (with charger)" and immediately enjoy.

If you pair the above with Pali's BME replacement and CSSU'ish battery applet, you will even have working battery level indication during hostmode or hostmode with charging.

Obviously, all hardware parts (female-to-female adapter and y-cable if you want charging) are still required, no magic here.

/Estel