So you have to make sure that you are never disconnected due to noise. But you didnt get the dabble interrupt, right? If not, then there are two problems. BR Ulf
I have two thumb drives same manufacture, different models. One drive I can insert and it mounts just fine with a udev rule. I can take that drive in and out as many times as I like and it always mounts and unmounts. The other drive will lock up the Host, I can reset it by reading \dev\bus\usb\001\001 file. Curious why one drive works all the time and the other fails, I put an Oscilloscope on the USB host Power P3-p1, noticed a small voltage 250mv sag when inserting the drive that always works, the drive that always fails has a dip of 1V. I placed a 10V 150uF Tantalum capacitor between U8 Pins 6,7 & 8 and ground to stiffen up the rail. now both drives work without fail. Hope this helps.
troy
I set up a crowd-funding project on
http://igg.me/at/eMagii/x/5581172
for people that wants to have USB host hot-plugging on the Beaglebone Black
I’ve ponied up for the $125 donation. Gotta have this running right.
I’ve also put in $125 and must say that I haven’t had any troubles since using Ulf’s kernel.
Thanks Ulf
Using a usb extention cable also fixes my issue with thumbdrive.
Where do I go to get Ulf’s kernel? I would like to test but I don’t know where to look. Is there a patch somewhere that I missed?
Where do I go to get Ulf's kernel? I would like to test but I don't know where to look. Is there a patch somewhere that I missed?
Check The http://igg.me link in your post.
Ok, got it. I need to contribute to try it out. Here is my problem, and I would like to hear if this solves it. I have a hub that uses the SMSC 2512 hub chip (USB2512QFN36). Mass storage devices work as long as it’s inserted at power up. Any hot plugging after boot will not work. Does your kernel fix this problem?
Also, I have some other items in the kernel that I need enabled. If I contribute $50 or $125 I only get whatever you built into your kernel… right? Could I give you my .config and you build it?
Ok, got it. I need to contribute to try it out. Here is my problem, and I would like to hear if this solves it. I have a hub that uses the SMSC 2512 hub chip (USB2512QFN36). Mass storage devices work as long as it's inserted at power up. Any hot plugging after boot will not work. Does your kernel fix this problem?
What I find with the standard kernel, is that if You have a hub inserted at boot time, you can hot-plug stuff using the hub. If you remove the hub then you have to reset the device.
If you have a device inserted at boot, it is detected and is functional.
If you remove it after the boot and reinsert it, it is no longer detected.
For some devices, removal will crash the kernel within a minute.
With my patched kernel, I can remove hub and / or devices and have them be detected and functional after boot, and can do this multiple times.
Tested with a mouse, keyboard, USB Serial and Hub.
Also, I have some other items in the kernel that I need enabled. If I contribute $50 or $125 I only get whatever you built into your kernel... right? Could I give you my .config and you build it?
My idea was $50 will get you a root file system without source.
$100 will get you the rootfs, and then you get an Angstrom layer which builds
the kernel with the patches after the end of the campaign.
You can start with the binary, and if you want source, chip in another $100 later,
but this will be more expensive than getting the $100 perk.
Obviously, once someone get the source, they have every right to publish it
according to the GPL, including the right not to ship it.
I will proably do that myself at some stage, so the perks essentially buys you time.
The $125 is my hourly rate for any extra work, so you need to select an option
above, and then an additional $125.
1 extra hour should be sufficient for rebuilding the kernel.
Ok, got it. I need to contribute to try it out. Here is my problem, and I would like to hear if this solves it. I have a hub that uses the SMSC 2512 hub chip (USB2512QFN36). Mass storage devices work as long as it’s inserted at power up. Any hot plugging after boot will not work. Does your kernel fix this problem?
What I find with the standard kernel, is that if You have a hub inserted at boot time, you can hot-plug stuff using the hub. If you remove the hub then you have to reset the device.
If you have a device inserted at boot, it is detected and is functional.
If you remove it after the boot and reinsert it, it is no longer detected.
For some devices, removal will crash the kernel within a minute.With my patched kernel, I can remove hub and / or devices and have them be detected and functional after boot, and can do this multiple times.
Tested with a mouse, keyboard, USB Serial and Hub.Also, I have some other items in the kernel that I need enabled. If I contribute $50 or $125 I only get whatever you built into your kernel… right? Could I give you my .config and you build it?
My idea was $50 will get you a root file system without source.
‘Without source’ go read the GPL again…
>
> > Ok, got it. I need to contribute to try it out. Here is my problem, and I would like to hear if this solves it. I have a hub that uses the SMSC 2512 hub chip (USB2512QFN36). Mass storage devices work as long as it's inserted at power up. Any hot plugging after boot will not work. Does your kernel fix this problem?
>
> What I find with the standard kernel, is that if You have a hub inserted at boot time, you can hot-plug stuff using the hub. If you remove the hub then you have to reset the device.
> If you have a device inserted at boot, it is detected and is functional.
> If you remove it after the boot and reinsert it, it is no longer detected.
> For some devices, removal will crash the kernel within a minute.
>
> With my patched kernel, I can remove hub and / or devices and have them be detected and functional after boot, and can do this multiple times.
> Tested with a mouse, keyboard, USB Serial and Hub.
>
> >
> >
> >
> > Also, I have some other items in the kernel that I need enabled. If I contribute $50 or $125 I only get whatever you built into your kernel... right? Could I give you my .config and you build it?
>
> My idea was $50 will get you a root file system without source.
'Without source' go read the GPL again...
A lot of products ship equipment with binaries only, but they must make an offer to ship the source to anyone receiving a binary as well, which I do.
It is possible to charge for both binaries and sources according to the GPL.
The amount is up to the vendor.
There is actually a possible violation of the GPL since you should not charge more for the source code, than for the binary. Only discovered tt fter the campaign was set up and if anyone complains, I will settle that directly in a GPL compliant way.
Cannot change the deal right now, due to the rules of IndieGoGo.
Anyway, if I release the source after the campaign, then that settles everything.
'Without source' go read the GPL again...
A lot of products ship equipment with binaries only, but they must make an offer to ship the source to anyone receiving a binary as well, which I do.
It is possible to charge for both binaries and sources according to the GPL.
I'm not arguing that point...
The amount is up to the vendor.
There is actually a possible violation of the GPL since you should not charge more for the source code, than for the binary. Only discovered tt fter the campaign was set up and if anyone complains, I will settle that directly in a GPL compliant way.
Read again what you wrote:
My idea was $50 will get you a root file system without source.
That's a pure violation...
Regards,
I don’t have any problem with your hourly rate, but I’m not confident your patched kernel will fix my problem. Here’s why… I bought the SMSC evaluation kit for the 2512 chip that supports two downstream ports. I plug it into the BBB and it recognizes it. Any mass storage device that I have plugged in at boot works just fine. If I unplug the device and plug it back in to the hub while running, I get the Babble Interrupt error.
I’ve read that devices plugged into hubs work just fine, but it’s definitely not the case for me and the SMSC 2512 chip. Where are you located at? I’m tempted to send you one of my SMSC evaluation kits and have you try it on your end.
>> 'Without source' go read the GPL again...
>>
> A lot of products ship equipment with binaries only, but they must make an offer to ship the source to anyone receiving a binary as well, which I do.
> It is possible to charge for both binaries and sources according to the GPL.
I'm not arguing that point...
> The amount is up to the vendor.
> There is actually a possible violation of the GPL since you should not charge more for the source code, than for the binary. Only discovered tt fter the campaign was set up and if anyone complains, I will settle that directly in a GPL compliant way.
Read again what you wrote:
> My idea was $50 will get you a root file system without source.
That's a pure violation...
Kernel is released under GPLv2 and this allows me to ship object code with an offer to ship source code for the cost of distribution. I believe that this is what I an doing.
Apparently there are several variants og GPLv2, but I am using the contents of the COPYING
file in the kernel source which claims that this is *the* version to use if in doubt.
I have seen other versions, which seems to support your interpretation.
I do want to comply, so if you find that I am violating anything in the COPYING file,
then please point out the paragraph.
/ulf
I don't have any problem with your hourly rate, but I'm not confident your patched kernel will fix my problem. Here's why... I bought the SMSC evaluation kit for the 2512 chip that supports two downstream ports. I plug it into the BBB and it recognizes it. Any mass storage device that I have plugged in at boot works just fine. If I unplug the device and plug it back in to the hub while running, I get the Babble Interrupt error.
The babble Interrupt may still occur, but so far the devices still seems to be detected on a patched kernel.
Before sending anything, I should probably test locally.
If it doesnt work, then there may be other things that need to change in the driver.
If it does work, I would be happy to test your devkit.
Live in Stockholm, Sweden.
The following codes can workaround that…
#!/bin/sh
echo “on” > /sys/bus/usb/devices/usb1/power/control
while [ 1 ]; do
cat /dev/bus/usb/001/001 > /dev/null
sleep 5
cat /dev/bus/usb/001/002 > /dev/null
sleep 5
done
I am getting this same issue on my new BBB running ubuntu saucy 13.10. My FTDI peripheral loads up and works fine for ahile, but then the babble interrupt inevitably comes, and nothing I do can allow the system to recover. Is there anything I can change to be able to recover from this without hard rebooting the system (Only thing that fixes it right now)?
I have also tried these commands where usb 1-1 is the port in question on my system:
echo ‘1-1’ |sudo tee /sys/bus/usb/drivers/usb/unbind
echo ‘1-1’ |sudo tee /sys/bus/usb/drivers/usb/bind
and it seems to be trying to reset, but it fails with error -110. Any ideas what that might be?
Transcript of dmesg when I try to debug it and reset the usb port after Babble Interrupt:
[12624.364551] musb_host_rx 1717: Rx interrupt with no errors or packet!
[12633.120879] CAUTION: musb: Babble Interrupt Occurred
[42284.311147] usb 1-1: unregistering interface 1-1:1.0
[42284.318828] ftdi_sio ttyUSB0: FTDI USB Serial Device converter now disconnected from ttyUSB0
[42284.319153] ftdi_sio 1-1:1.0: device disconnected
[42284.319505] usb 1-1: usb_disable_device nuking non-ep0 URBs
[42289.318728] usb 1-1: tee timed out on ep0out len=0/0
[42294.243524] usb 1-1: usb_probe_device
[42294.243594] usb 1-1: configuration #1 chosen from 1 choice
[42299.242612] usb 1-1: tee timed out on ep0out len=0/0
[42299.242711] usb 1-1: can’t set config #1, error -110
Ulf, does your kernel patch interrupt data streams? In other words, for example, if I’m running a webcam and I hotplug another device, will data transfers from the webcam be affected?
I do no think the problem exists if you have a hub connected and try to hotplug using the hub.
The patch is then not needed.
Have you seen this happening?
Best Regards
Ulf Samuelsson
25 jan 2014 kl. 02:23 skrev Dan Lipsitt <dan@typeamachines.com>: