Same problem here, description:

http://lists.maemo.org/pipermail/maemo-developers/2008-March/017217.html

I noted a few OTG discussions on LKML, might be relevant:

http://lkml.org/lkml/2008/2/12/112
http://lkml.org/lkml/2008/1/20/112

This is a regression from N800 where host does work.

CC to Quim.

I confirm Charles patch allows me to use USB key and USB keyboard in host mode.

Note: I was using kernel from:

~ $ cat /etc/osso_software_version
RX-34+RX-44_2008SE_2.2007.50-2_PR_MR0

Is there an USB change for the latest?
RX-34+RX-44_2008SE_2.2007.51-3_PR_MR0~/QR

(In reply to comment #4)
> Note: I was using kernel from:
> 
> ~ $ cat /etc/osso_software_version
> RX-34+RX-44_2008SE_2.2007.50-2_PR_MR0
> 
> Is there an USB change for the latest?
> RX-34+RX-44_2008SE_2.2007.51-3_PR_MR0~/QR
> 

I'm on 51-3 myself. I think they both use the same kernel(rx-34-2.6.21.0) but
I'm not 100% sure on that...

Charles Werbick

The more I think about it, the patch should probably have-

+       if (!last_state) last_state = 1;

instead of-

+       if (!last_state) last_state = new_state;

Otherwise the first state change via sysfs is ignored...

cheers,
Charles Werbick

Created an attachment (id=744) [details]
patch in musb_core.c to prevent hang at b_idle on n810

This patch calls usb_stop() then usb_start() on mode change, basically forcing
cleanup so that the USB mode can be changed on the n810 via sysfs reliably.

Comment 8 Andre Klapper 2008-06-06 17:17:20 UTC

reassigning MDK's bugs to Rodrigo.

Comment 9 Andre Klapper 2008-06-09 12:36:59 UTC

OK, this bug bites a lot of people and according to emails and comments the
patch works fine.
High priority, let me push this for Diablo (or its updates).

Comment 10 Andre Klapper 2008-06-09 12:38:28 UTC

another one, copying from today's #maemo irc channel:

<zap> Anybody had ever problems with enabling USB host on a N810? I have two
things here, one works fine and other doesn't see anything I connect to it
(same cable, same USB pen drive)
<zap> I have two N810s
<zap> flashed them both with the same firmware
<zap> I use the usbcontrol python script
<zap> when I enable Host mode, the one that works switches from b_idle to
a_wait_vrise
<zap> the other stays in b_idle
<zap> peripherial mode on both works fine
<zap> cat /sys/whatever/state shows b_idle as well
<andre___> sounds a bit like https://bugs.maemo.org/show_bug.cgi?id=3026
<zap> yes, that's it, thank you very much

Yes, my situation is quite similar to what's described in the first report,
with a few differences:

- If I plug the N810 into the PC USB port, then unmount the USB drives and plug
it off, it won't allow to switch controller mode anyways.

- If I plug the N810 into the PC USB port, then unmount the USB drives and then
immediately switch the mode to 'host' (without taking the plug out of the PC),
it switches to a_wait_vrse, but then if I plug it out of the PC and try to plug
into N810 an USB pen drive, it doesn't work. The USB pen drive seems not
powered as the LED doesn't light up, and N810 doesn't notice something is
inserted.

I have access to another N810 and it works just fine with the same cable and
the same USB A-A gender converter. I have no problems switching to host mode
and connecting the same USB pen drive, keyboard etc.

On both units "cat /proc/cpuinfo" says they're a "Nokia RX-44 rev 24202524".

(In reply to comment #11)
> Yes, my situation is quite similar to what's described in the first report,
> with a few differences:
>

Andrew,

Sounds very similar. Have you tried the patched kernel? It's available here-
http://rapidshare.com/files/105578563/zImage.html

(It's a stock kernel. The only modification is the usb-host patch. So if you're
running any other kernel mods (i.e.- FB rotate support) you'll need to download
the patch and compile yourself...)

Cheers,
Charles Werbick

Comment 13 Andre Klapper 2008-06-16 16:30:57 UTC

Forwarding some questions from the internal bugticket, would be great if
someone could answer this:


Does it work with a correct micro-A cable?
The sysfs interface is only there for hacking/testing/development purposes. The
correct way to use usb otg is with proper certified cables.

Please provide more detailed output from dmesg if you have one of the messy
devices:

# echo D5 > /proc/drivers/musb_hdrc

Will increase the loglevel of the driver giving me more valuable information.

A few printk in drivers/usb/musb/tusb6010.c:musb_platform_set_mode() also
helps. 

Try also grounding the id pin in the micro-B cable and see if it automatically
switches to host mode.

Check if the same bug appears in current linux-omap tree, if it doesn't then
it's a bug in our tree if it does then it might be hw issue, although I need
more information to state that more certainly.

(In reply to comment #13)
> Forwarding some questions from the internal bugticket, would be great if
> someone could answer this:
> 
> 
> Does it work with a correct micro-A cable?
> The sysfs interface is only there for hacking/testing/development purposes. The
> correct way to use usb otg is with proper certified cables.
> 

Yes, a micro-A host cable still works with the broken kernel. The problem is
with the software ID. And while I'm all about using the proper cables, Nokia
does not sell one. (Nor have I found one online.) All of the cables I've seen
for sale are hacked micro-B motorola phone adapters. Because of this, 95% of
n810 users are using sysfs to change modes. 

Nokia offering host cables for the n810 would be an acceptable fix for this bug
though. 


> Please provide more detailed output from dmesg if you have one of the messy
> devices:
> 
> # echo D5 > /proc/drivers/musb_hdrc
> 
> Will increase the loglevel of the driver giving me more valuable information.
> 
> A few printk in drivers/usb/musb/tusb6010.c:musb_platform_set_mode() also
> helps. 
> 
> Try also grounding the id pin in the micro-B cable and see if it automatically
> switches to host mode.
> 

I'm running the patched kernel. So not much help there. However I *did*
increase the logging level while looking into the problem. It didn't seem to
give me any extra(useful) info. If I have time, I'll flash the stock kernel and
post some output... As for trying the generic omap kernel, I'm not quite that
brave. (Perhaps if I had more than one n810 ;)

cheers,
Charles Werbick

According to your description , I try to reproduce the bug,and found the actual
outcome is all right;it is changed into "b_wait_bcon" mode waiting for a slave
device to plug in, this mode is belong to host mode but just for waiting. when
I plug the memory to the N810 and check the mode ,it becomes "b_host" mode.

Internet Tablet OS: maemo linux based OS2008
Version :2.2007.51-3

Created an attachment (id=822) [details]
Demonstrating the mode can be changed

I didn't modify any source code before reproducing it.So I think this bug has
been fixed in my OS version.

(In reply to comment #17)
> I didn't modify any source code before reproducing it.So I think this bug has
> been fixed in my OS version.

Did you ever experience this bug on your device? AFAICT, it is a hardware bug
that only affects perhaps one in 50 n810's... If you can't reproduce it, your
n810 is likely not one that is affected.

Cheers,
Charles Werbick

Comment 19 Andre Klapper 2008-07-14 18:54:29 UTC

removing invalid target milestone.

Comment 20 Andre Klapper 2008-08-04 14:26:26 UTC

Internal ticket was closed as WORKSFORME, see quote below.
So maybe it's really a hardware issue as mentioned in comment 18?
If we don't get some clarification I'm also going to close this report as
WORKSFORME.


--quote:--

Can't reproduce it:
/ # echo host > /sys/devices/platform/musb_hdrc/mode 
/ # [   48.992187] usb 1-1: new high speed USB device using musb_hdrc and
address 2
[   49.132812] usb 1-1: Product: Cruzer Mini
[   49.140625] usb 1-1: Manufacturer: SanDisk Corporation
[   49.148437] usb 1-1: SerialNumber: SNDK9A73141999307702
[   49.148437] usb 1-1: device v0781 p5150 is not supported
[   49.171875] usb 1-1: configuration #1 chosen from 1 choice
[   49.203125] scsi0 : SCSI emulation for USB Mass Storage devices
[   54.226562] scsi 0:0:0:0: Direct-Access     SanDisk  Cruzer Mini      0.2 
PQ: 0 ANSI: 2
[   54.273437] SCSI device sda: 2001888 512-byte hdwr sectors (1025 MB)
[   54.281250] sda: Write Protect is off
[   54.289062] sda: assuming drive cache: write through
[   54.304687] SCSI device sda: 2001888 512-byte hdwr sectors (1025 MB)
[   54.312500] sda: Write Protect is off
[   54.312500] sda: assuming drive cache: write through
[   54.320312]  sda:<7>musb_start_urb 214: qh c7d9ad60 urb c7da3540 dev2
ep1out-bulk, hw_ep 1
, ffcfe000/31
[   54.335937]  sda1
[   54.367187] sd 0:0:0:0: Attached scsi removable disk sda
[   62.156250] usb 1-1: USB disconnect, address 2
/ # echo otg > /sys/devices/platform/musb_hdrc/mode 
/ # [   78.187500] usb0: high speed config #1: 8 mA, Ethernet Gadget, using CDC
Ethernet
It works with the N810 hardware, marking WORKSFORME.

Comment 21 Andre Klapper 2009-01-05 13:17:50 UTC

Closing as per last comment.