USB - the Elephant in our Room


 
803 posts   Page 11 of 33   1 ... 8, 9, 10, 11, 12, 13, 14 ... 33
by dom » Thu Aug 16, 2012 3:46 pm
lintweaker wrote:Congrats, the new USB OTG driver greatly improves USB2 audio performance/usability. Even 192K FLACs now play almost flawless (still regular crackles).

Keep it going!


Thanks for the report. Glad it seems improved. We will get there eventually.
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 5076
Joined: Wed Aug 17, 2011 7:41 pm
Location: Cambridge
by PaulCheffus » Thu Aug 16, 2012 3:59 pm
dom wrote:We updated the synopsys driver to a newer version (2.94) last night.
Thanks to ddv2005 for tracking down a bug that had been breaking the newer version of the driver.


Hi

Please could you let me know how I get hold of the new driver, apt-get or rpi-update.

Cheers

Paul
Procrastination - The Thief of Time.
Posts: 381
Joined: Tue May 29, 2012 2:51 pm
Location: Lincolnshire UK
by dom » Thu Aug 16, 2012 4:03 pm
PaulCheffus wrote:Please could you let me know how I get hold of the new driver, apt-get or rpi-update.


rpi-update for now.
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 5076
Joined: Wed Aug 17, 2011 7:41 pm
Location: Cambridge
by kermitas » Thu Aug 16, 2012 4:18 pm
My small report after 12h of webcam tests in configuration:
- Fedora 17 remix test image 17 fully updated
- rootfs on pendrive
- sdhci-bcm2708.missing_status=0 sdhci-bcm2708.sync_after_dma=0 in /boot/cmdline.txt
- in one time only one webcam (of two: Microsoft LifeCam Cinema and Creative Socialize HD 1080) is plugged via powered usb cable (to eliminate powered usb hub because it is next usb device that can produce troubles)

Horizontal lines that I described here http://www.raspberrypi.org/phpBB3/viewtopic.php?f=63&t=12304 (Microsot LifeCam Cinema webcam) now disappeared :D :D.

It looks like "port X disabled by hub (EMI?), re-enabling..." http://www.raspberrypi.org/phpBB3/viewtopic.php?f=63&t=12247 also disappeared.

Looks like all things related with USB works better in this configuration and with latest firmware :D :D .

Webcams works stable for about 4-5 hours and then stop to send video. They are still present in lsusb, I can stil confugre them using v4l2-ctl but I have to do restart to get them work. But even with this problem situation is much better than before :).

Thank you 'dom' and other supporters! I think many people are waiting for possibility to eliminate working from pendrive (less devices connected to usb = less troubles and higher bandwidth for other things).

Arthur.
Posts: 108
Joined: Thu Jan 26, 2012 11:49 am
by Dexterp37 » Thu Aug 16, 2012 6:05 pm
My 2 cents :) With the update driver, the situation is definitely better! My Logitech Quickcam E 3500, directly hooked to the RPi, doesn't fail anymore with "select timeout" when trying to read images through v4l2. At least, it doesn't fail soon. It first grabs some frames (a random number > 50) and then fails. Definitely a progress!

If I put a powered hub between the camera and the RPi, the acquisition fails with a different error ("VIDIOC_S_FMT error 5, Input/output error", before it was just a "select timeout").

Keep up the good work guys!
Posts: 21
Joined: Mon Jul 30, 2012 7:51 pm
Location: Italy
by Aydan » Thu Aug 16, 2012 6:08 pm
Hi,

I have a few findings concerning staggered hubs.
Have a look here http://www.raspberrypi.org/phpBB3/viewtopic.php?f=28&t=10355&p=151571#p151571

Regards
Aydan
Posts: 552
Joined: Fri Apr 13, 2012 11:48 am
Location: Germany, near Lake Constance
by blavery » Thu Aug 16, 2012 10:07 pm
Well, 4 minutes of testing: (it's 8AM here)
1. No external hub.
2. Wireless keyboard/mousepad in one slot.
3. Wifi 5370 dongle in other slot. (Woops - lsusb -v reports 450mA max !!!)
4. Nothing on lan port.
Previously an impossible combo. Should be an illegal (450mA) arrangement also.
NOW IT WORKS.

Much more to test, but clearly something is performing better. Thanks.
Posts: 95
Joined: Sun Jul 01, 2012 2:57 am
Location: QLD. Australia
by zog » Thu Aug 16, 2012 10:09 pm
I agree USB stability is a major problem. I have debian wheezy running on an old i386 laptop and on the raspberrypi. I bought numerous bluetooth devices to try to get blue tooth running on the pi.
All the blue tooth devices worked on the old laptop.

I tested all my devices using a Belkin powered hub on the raspi to over come any power issues.

The only device I could get to work on the pi was a device using a cambridge silicon chip set, and then I have to be very carefull what else I plugged into the pi.

I solved my power supply issues by using a 2A powered usb and powering the pi from this device using a htc usb phone cable.

I got a usb Samba3G modem with a telit chip set up and running on the i386 in a matter of minutes. The pi recognises the device but won't talk to it at all. I was able to use putty to query the modem on the laptop, but could get no where with the pi.

I have managed to use a usb serial port conveter to communicate with a Garmin Etex GPS unit on both the pi and the laptop succesfully.

Chaining two usb hubs together doesn't work either.

I have got quite fustrated trying to get things to work with the pi.

You also have to buy, or have a lot of spare stuff to get the pi up and running working keyboard, mouse sd card, blue tooth dongle, wifi adapter,power supply, usb hub, telly, e.t.c.
Posts: 180
Joined: Sun Nov 20, 2011 5:43 pm
by ferrymanr » Thu Aug 16, 2012 11:23 pm
Latest updates definite and very welcome improvement. I was unable to get the Soundblaster Audigy 2NX USB card to work at all but now it is working on record and playback without error and with full control. Just occasional timeout on network when soundblaster is working hard. Previously I could only just get away with simultaneous audio record and network access using a small sound dongle based on CM119 chip (but with lousy audio). The update also seems to have increased reliability using a powered USB hub on one USB port for keyboard and mouse whilst using the other USB port for the audio device.
Thanks to all who have worked on this.
Posts: 61
Joined: Fri Mar 16, 2012 11:09 pm
by abishur » Fri Aug 17, 2012 2:50 am
zog wrote:I agree USB stability is a major problem. I have debian wheezy running on an old i386 laptop and on the raspberrypi. I bought numerous bluetooth devices to try to get blue tooth running on the pi.
All the blue tooth devices worked on the old laptop.


Just to make a point here, debian wheezy x86 doesn't equate to debian wheezy ARM anymore than Windows 7 x86 equates to Windows 7 x64. These are completely different architectures. Just because something has a driver and works on one architecture doesn't mean it will work on the other.

This isn't to suggest that there isn't a USB problem in general, that would be asinine. I am saying that your particular issue is probably due more to a lack of appropriate drivers (just like we had to work out the drivers for the wireless adapters.) ;-)
Dear forum: Play nice ;-)
User avatar
Posts: 4478
Joined: Thu Jul 28, 2011 4:10 am
Location: USA
by blavery » Fri Aug 17, 2012 5:13 am
... Add another 34 minutes testing a few USB devices.
So far: it's hard to make sense. Some stuff works, some doesn't.
But this version performs differently than before.
Seems to baulk at various things on 7-port hub (4+4). That 5370 wifi likes only to sit right on the pi port (not hub), despite the rated max current figure.
Still playing here.
Posts: 95
Joined: Sun Jul 01, 2012 2:57 am
Location: QLD. Australia
by zog » Fri Aug 17, 2012 10:23 am
Hi again,
I have tried to plug all the devices again under the latest raspbian version, with the same results, that most of them malfunctioned as before. I get a lot of time out errors with the blue tooth stuff for example. I don't know much about the linux kernel, so I am going to ask some stupid questions I hope some one can answer.

1) Is the pi to slow to handle some modern usb devices ?

2) Is it possible to increase the priority of a USB drivers especially the under lying USB HC drivers and if so how would I do this ?

3) I read that some one had over clocked their pi to 900 Mhz and if so would this help ?

4) Would recompiling the kernel with different USB settings help solve the problems ?

5) Is there an underlying problem with the USB chip sets on the pi - which other devices use the pi chips sets and do they also have problems with USB ?

N.B I have printed some documents out on how to write linux drivers which I am going to go away and study.

I desperately want the ras pi to work well as it's low power consumption, and it's io capabilities as well it's SD card interface are very attractive. However without a good USB system to work many of the projects I have envisaged using the pi for are not possible.

It's not all bad news however

I have had success with the free pascal language as I have managed to port code from Delphi in Win32 to Lazarus running in linux armhf with very little bother which I am extreamely pleased about.
Posts: 180
Joined: Sun Nov 20, 2011 5:43 pm
by RaTTuS » Fri Aug 17, 2012 10:34 am
tail -f /var/log/syslog
and plug the things in one by one
record the error messages and then post again.
How To ask Questions :- http://www.catb.org/esr/faqs/smart-questions.html
WARNING - some parts of this post may be erroneous YMMV

1QC43qbL5FySu2Pi51vGqKqxy3UiJgukSX
Covfefe
User avatar
Posts: 9183
Joined: Tue Nov 29, 2011 11:12 am
Location: North West UK
by samsamsam » Fri Aug 17, 2012 11:40 am
dom wrote:We updated the synopsys driver to a newer version (2.94) last night.
Thanks to ddv2005 for tracking down a bug that had been breaking the newer version of the driver.

Whilst no specific bug we are aware of is fixed by this update (sorry, lost packets and limited number of periodic endpoints are still present), it does improve iperf performance by nearly 10% and seems to handle locking more cleanly, so it could help some miscellaneous issues.

We are still treating the lost packets as the most urgent issue, but it's proving tricky to find.
We're also tidying up the FIQ patch and the limited number of periodic endpoints patch, so they can be enabled with a command line parameter for wider testing.


Is there somewhere patch for "No host channel available for periodic transfer" which can be applied for new version of the dcw_otg driver?
Posts: 36
Joined: Fri Aug 17, 2012 11:36 am
by kermitas » Fri Aug 17, 2012 12:18 pm
blavery wrote:... Add another 34 minutes testing a few USB devices.
So far: it's hard to make sense. Some stuff works, some doesn't.
But this version performs differently than before.
Seems to baulk at various things on 7-port hub (4+4). That 5370 wifi likes only to sit right on the pi port (not hub), despite the rated max current figure.
Still playing here.


Did you try to boot from SD card but whole rootfs keep on pendrive and then add to cmdline.txt sdhci-bcm2708.missing_status=0 sdhci-bcm2708.sync_after_dma=0?
Posts: 108
Joined: Thu Jan 26, 2012 11:49 am
by obcd » Fri Aug 17, 2012 2:23 pm
@zog:
1. Currently they say the problem come from interrupt latency being 2 high.
So, you could indeed say that the Pi is 2 slow.
As admitting that isn't very good for sales, it will probably never happen.
2. The problem with Usb is that changing the priority is difficult. When an usb device has data, the application that uses it needs to process it before new data can arrive. So, it's not only the driver that needs a higher priority. Somehow, it should be possible to process some interrupts faster, when there is no data available.
3. It won't make much difference. It's not the solution.
4. It could help a little. In my case, my self compiled kernel wasn't having wifi issues.
The one from the getting started had my wifi failing under heavy traffic. I had some config settings different, but have no idea which one made the difference.
5. The roku 2 media player uses the same broadcom Soc. Most people will only use it's usb host port to connect a flash drive to play media. So, it's unknown how it would perform when multiple usb clients are connected. It probably isn't having driver support for many different usb clients. It will support the Storage Class so that it can be used to connect memory sticks.
Posts: 908
Joined: Sun Jul 29, 2012 9:06 pm
by JonB » Fri Aug 17, 2012 2:43 pm
May I refer you all to this thread: http://www.raspberrypi.org/phpBB3/viewtopic.php?f=28&t=14805

Is it the same issue that you are discussing?
Posts: 220
Joined: Tue Nov 29, 2011 9:26 pm
by JonB » Fri Aug 17, 2012 5:08 pm
Update : Have done rpi update and set what I think are the correct config.txt values, but the Lapdock's (powered) hub keeps disconnecting.
Posts: 220
Joined: Tue Nov 29, 2011 9:26 pm
by obcd » Fri Aug 17, 2012 8:18 pm
There is a threat somewhere around here about people connecting a lapdock.
I believe it works if you disable the pi on board ethernet. Someone even posted a script that does it.
And yes, likely it's related to our elephant I am afraid.
Gordon (gsh) is working on the issue. He will reduce the number of interrupts generated by the usb controller (He will handle those that have no data with a fast assembly routine if I understand well.)It's called like a greek priest... Kanunic approach or something. Since you will start to study linux drivers, maybe you will find something about it. I have no idea how things are working out at the moment.
Posts: 908
Joined: Sun Jul 29, 2012 9:06 pm
by JonB » Fri Aug 17, 2012 8:49 pm
Disable the Ethernet? Would not be able to go online then. Well, not without a wifi adapter anyway. Kind of defeats the point of having Ethernet really...but don't get me wrong - thanks for the info, I'll look into it!
Posts: 220
Joined: Tue Nov 29, 2011 9:26 pm
by itimpi » Fri Aug 17, 2012 9:53 pm
I am using a Lapdock and do not have to disable the ethernet! I regularly plug either a 2.5" HD or a 16GB USB stick into the Lapdock USB ports and that also works fine with the Pi.
Posts: 1067
Joined: Sun Sep 25, 2011 11:44 am
Location: Potters Bar, United Kingdom
by jamesh » Sat Aug 18, 2012 3:39 am
obcd wrote:@zog:
1. Currently they say the problem come from interrupt latency being 2 high.
So, you could indeed say that the Pi is 2 slow.
As admitting that isn't very good for sales, it will probably never happen.


I'm a abit confused by the statement - we've said that the Interrupt latency is too high - correct, but that's not because the Pi is too slow (its a 700Mhz device and USB works on much slower machines than that). It does appear to be a driver problem (SD card I/F?) somewhere making the latency high.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Please direct all questions to the forum, I do not do support via PM.
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 17223
Joined: Sat Jul 30, 2011 7:41 pm
by Lob0426 » Sat Aug 18, 2012 6:12 am
The script "disable-ethernet" is a fix for using a USB WiFi adapter with the Lapdock. If you are not using a USB WiFi, with a Lapdock, then do not use the script. Some powered hubs seem to have the same problems with WiFi as the Lapdock. The script may fix them too.
If you leave the onboard Ethernet working, with a WiFi adapter attached, you end up with a whole bunch of USB errors. That is what the script fixes.
You can enable and disable the onboard Ethernet adapter with that script. Sometimes you just have to get it from the horses mouth. Lapdock discussion:
viewtopic.php?f=63&t=6747
Script discussion:
viewtopic.php?p=127151#p127151

As to the Elephant in the Room:
:arrow: " Gracy Get my GUN, There's an Elephant in the ROOM "! :shock:

There are indeed some USB problems, But I really do not see them! Lapdock works just fine with a USB HDD and a USB WiFi. Of course I am not transfering large files with the setup. I think the largest so far was like 75MB and no problems. It works fine for a device designed to be a programming tools for education. It, and the software, was not designed to be anything else. There is software out there that makes it do other things. But even so, there are a good number of people out there trying to fix the USB problems.
I know that the performance of the Raspberry Pi has greatly improved since it was released. And it sounds like there is more software fixes being worked on.
I think overall, "Thank you" is the thing to say!
512MB version 2.0 as WordPress Server
Motorola Lapdock with Pi2B
Modded Rev 1.0 with pin headers at USB

http://rich1.dyndns.tv/
(RS)Allied ships old stock to reward its Customers for long wait!
User avatar
Posts: 2156
Joined: Fri Aug 05, 2011 4:30 pm
Location: Susanville CA.
by pluggy » Sat Aug 18, 2012 6:56 am
I went as far as soldering bypass shunts on the polyfuses and powering the Pi through the GPIO pins to try and get one of my 4 webcams to work reliably. Powered hubs tend to cause as many problems as they solve. They all work for a short period before croaking with segmentation faults and dmesg full of error messages. I'd rather they did the USB and Ethernet properly with some extra silicon (and mainstream drivers) and charge a bit more for it. They sold out to get the price down in my opinion.

I fear people might judge Linux by the Pi, and having used Linux for 10+ years and now use it exclusively on all my computers, I know Linux works, its the Pi that doesn't.
Don't judge Linux by the Pi.......
I must not tread on too many sacred cows......
User avatar
Posts: 3636
Joined: Thu May 31, 2012 3:52 pm
Location: Barnoldswick, Lancashire,UK
by jamesh » Sat Aug 18, 2012 7:40 am
pluggy wrote:I went as far as soldering bypass shunts on the polyfuses and powering the Pi through the GPIO pins to try and get one of my 4 webcams to work reliably. Powered hubs tend to cause as many problems as they solve. They all work for a short period before croaking with segmentation faults and dmesg full of error messages. I'd rather they did the USB and Ethernet properly with some extra silicon (and mainstream drivers) and charge a bit more for it. They sold out to get the price down in my opinion.

I fear people might judge Linux by the Pi, and having used Linux for 10+ years and now use it exclusively on all my computers, I know Linux works, its the Pi that doesn't.


Not true at all. This almost certainly isn't a silicon problem, it's a driver problem. And Silicon problems (were they to exist) are not the Foundations issue, but the chip suppliers. As far as the Raspi foundation are concerned, when the board was designed and tested, it all worked, so there was no reason to believe any more money needed to be spend on different components.

And for most people the Pi does work. Stop being overly dramatic. Your particular use case might not work yet, but a lot of others do.
Principal Software Engineer at Raspberry Pi (Trading) Ltd.
Please direct all questions to the forum, I do not do support via PM.
Raspberry Pi Engineer & Forum Moderator
Raspberry Pi Engineer & Forum Moderator
Posts: 17223
Joined: Sat Jul 30, 2011 7:41 pm