USB hostmode status: beta.

Thanks to everyone part of the h-e-n team, and most importantly, thanks to Joerg Reisenweber for starting the project and leading it, and thanks to Paul Fertser for the new kernel patches, written from scratch.

This release has been tested by a few of us, but it has worked perfectly on all devices.

You will need to a female to female USB adapter, and the stock USB cable.
(You can also use a Nokia CA-101D, a shorter version of the stock cable, a classmate gave it to me, and I found it perfect for hostmode, pic below)

I highly recommend you DO NOT! use direct adapters (microUSB to full size in one adapter || directly plugged into port), using those adapter will put a lot of stress on the USB port and may cause to be torn out! I suggest using at least a small cable like the Nokia CA-101D below with a female to female adapter, or a cable that is microUSB to full, but it should not be a direct adapter!



Paul has found out that it is impossible (at least at the time of writing) to detect speed changes while VBUS is running, so for now, you have to select speeds manually.
Use common sense in doing so, high speed is for devices that require a lot of writing and are fairly new (flash drives/HDDs/CD-drives, new hubs).
Full speed devices are devices that do not require a lot of USB activity, such as BT/2.4GHz mouse transceivers.
Low speed devices are normal mice and keyboards.

A help window is available and will guide you through the steps needed.
The mount button is currently hardcoded to mount /dev/sd*, default for most USB flash drives and HDDs.
For NTFS support, do modprobe ntfs as root
For ISO (CDs/DVDs - most of them), do modprobe isofs (also as root).
CDs will have the dev node /dev/sr0, so you'll need to do mount /dev/sr0 /mnt etc...
Disabling tracker by adding /media/flash to tracker-cfg's non watch configuration is recomended, use tracker-cfg UI.






(note, full speed devices are 1.1, the screenshot is old, fixed in release)

Demo of DVD player on N900:
http://www.youtube.com/watch?v=zDzUG-bno6Q

To install:
Install hostmode-gui from -devel, the package depends on kernel-feature-usbhost, so it should be compatible with any kernel that provides this (in debian/control), kernel-power provides this, and therefore, if no other kernel that does provide it is supported, it will be installed.

The GUI now checks for API version instead of kernel version, with the latest API being 20101110, also in kernel-power.



Known problems/bugs:
lsusb sometimes needs to be executed to return statemachine to original state.
Battery meter is not updated, thanks for the closed bme Nokia
Bugs not related to host mode, but kernel-power (see relevant threads, we don't have to do with these issues):
fcam-drivers doesn't work on certain devices for some reason, please see kernel-power's thread, quick fix: uninstall fcam-drivers.

Support thread: http://talk.maemo.org/showthread.php?t=65570


Please note that donations go to the team, and not one specific person. Details here

Hostmode development team:
Project Admins:
Joerg Reisenweber
Paul Fertser

Developers:
Chitrang Srivastava
Dave na
Hagen Meyer
Ian Stirling
Jacek Milewicz
Klaus Kurzmann
Mohammad Abu-Garbeyyeh
Mugur Enache
Sarah Newman
Sonny Jim
Tom Tanner

Awesome work. I'm sure a LOT of people have desired this since the beginning.

Wow good work, wiill the phone auto detect devices like USB storage devices or other things?

thank u guys for all your hard work

One question: is h-e-n kernel based on power-kernel? I would love to have it installed but cannot live without overclocking

Sorry for anther post, after reading, the kernel will be replaced, does this mean no overclocking?

Looks to me like we need a version of kernel_power+pr1.3 patches+usb-hostmode, but I'm not entirely sure how to make my own kernel version.

Unfortunately yes

HS=USB2.0, FS=USB1.1, LS=USB1.0
USB2.0/HS-peripherals will also work in USB1.1/FS mode, though much slower.

Switching speeds "on the fly" is not possible for now (to elaborate on Mohammad's introduction: We need VBUS to detect the speed of device attached, but we can't switch speed accordingly as long as VBUS applied - musb-core doesn't allow that). You have to stop hostmode and start over from beginning. Also the much requested hostmode while charging (N900 works as USB host but does not supply VBUS 5V - this will be done by a modified hub or a Y-cable: http://wiki.openmoko.org/wiki/Specia...aka_Y-Cable.29) is not yet supported (however see http://talk.maemo.org/showthread.php...203#post921203 ). Well folks, this is a BETA :-), though Mohammad once again managed to pack a shiny package that looks better than some products tagged "final gold", thanks for that Mohammad :-)

Special thanks to Paul who did a brilliant job, like always. Also a special thanks to Sarah Newman without her patches we probably wouldn't be anywhere today. And a special thanks to all the people who shared their knowledge with me in endless private chats but preferred to stay in the background for different reasons.

Enjoy hostmode Beta, cheers
jOERG


ps.: As it seems it always somehow vanishes: N900 can deliver 200mA VBUS power (actually current, that's a hardware limitation of bq24150 chip), as compared to a standard 500mA of a PC host, this means power greedy peripherals like e.g HDD, UMTS sticks, mug warmers, etc, will cause hostmode to fail. An easy tested way to deal with this problem is to use a powered hub.
If you want to try with externally powered things, then you should plug in the peripheral same moment you push the VBUS button in hostmode GUI

ps2: Until Mohammad updates the info in #1: We are not shipping a dedicated hostmode kernel anymore with h-e-n GUI. You are supposed to install kernel-power v45 (or newer), which has hostmode patches integrated. Installing h-e-n won't do that for you. Please refer to kernel-power howto and readme and FAQ, for details on how to install and how to remove this non-stock kernel. There are known issues in multiboot with kernel-power, as well as in fcam drivers, which don't work for v45 kernel-power at date of this writing (unrelated to h-e-n!). Kernel-power v45 comes with uBoot, which should replace multiboot. It also should come with a fix (actually a revert of a patch) for the locked@500 bug Mohammad mentions in next post.
The 'mount' button and function of h-e-n is a beta-test convenience thing, and is known to not work under all arbitrary circumstances. It will vanish in final version of h-e-n anyway, and any problems are WONTFIX

As a rule of thumb, this thread is about problems (and unexpected success) to provide 5V VBUS and to enable hostmode. As soon as your device is showing up in dmesg as "usb 1-1: New USB device found, idVendor=xxxx, idProduct=yyyy...", h-e-n considers this as mission accomplished.
Everything beyond are derived issues not directly related to h-e-n, so please report them either in the problems-thread, or on the forum of the involved subsystems which actually throw the error. Esp mounting, though still supported by a button in h-e-n, is NO topic for this thread


recommended links: http://talk.maemo.org/showthread.php...esg#post875006
http://talk.maemo.org/showthread.php...032#post895032
http://www.youtube.com/watch?v=fkCDyUO0sKQ&NR=1

LATEST NEWS:
there are missing modules in powerkernel46. See http://talk.maemo.org/showthread.php...219#post971219 - we're aware now and working on it. Shall be fixed in recent PK48 (which is said to work http://talk.maemo.org/showpost.php?p...&postcount=861 ). For PK47 see http://talk.maemo.org/showthread.php...91#post1116491 - thanks blue_led :-)

see http://talk.maemo.org/showpost.php?p...&postcount=857 for latest devel-version of parts of h-e-n, with augmented error-reporting (notifies on overload [huge HDD], runs a notification LED pattern to let you know your battery is draining) [edit 2012-02-27: fixed link]

cheers
jOERG

Well, most of you don't know this, but, PR1.3 introduced two things:
Backports of new kernel patches to musb, which really helped with hostmode.
And.. a new bug, if you connect the wall charger to the device, 100% of the time it will stay on 500 (sometimes 550?) MHz, we do not know the behavior of an overclocked device when this bug is triggered on it, if it stays clocked @ 900 (for example), then we're going to see some dead devices pretty soon.
Anyways, hostmode's patches require 1.3's patches to musb, so until titan updates his kernel, it's a no go, feel free to contact his email address (see http://maemo.org/packages/view/kernel-power-flasher and click his name) and request an update.