On Tue, 25 Oct 2005, MAL wrote:

> Alan Stern wrote:
> > You know, it quite often turns out that instability in USB storage
> > devices, especially the "insert ATA disk in a USB enclosure" sort, is
> > caused not by problems in the host hardware or the system software.  It's
> > caused by defects and bad design of the USB (or USB-ATA) interface in the
> > device's electronics.
> 
> I can believe that, but the device works flawlessly time and again on a
> Windows host.

It's awfully difficult to learn exactly what's happening in situations
like this.  My own theory is that Linux pushes the hardware somewhat more
than Windows does (all within the limits set forth by the appropriate
Standards, of course), and manufacturers don't bother to verify that their
devices can work under such a load.  At most they test whether the devices
work with Windows -- sometimes not even that!


On Tue, 25 Oct 2005, Johan van den Berg wrote:

> I can fully second this... Using some drives on the lates kernel,
> different USB enclosures, and all of them give me some amount of issues
> every now and again.  At least 20% of the time I either get kernel
> panics, SCSI errors or simply filesystem corruptions.  I have really not
> had the time to do full debugging and log dumping and the like, mostly
> because I can't actually reproduce the errors easily.

I certainly can understand not having the time to do detailed debugging or 
other investigation.

SCSI errors are the natural result of hardware problems; there's not much
that can be done when the device stops responding to commands.  And
filesystem corruption follows hot on their heels, of course.

However you shouldn't get kernel panics, at least not with the latest 
kernels.  I'd like to think that the operating system is robust enough to 
deal gracefully with buggy hardware.  If you do come across another panic 
or oops on any system from 2.6.13 on up, please post the relevant 
information for the developers to look at.  We can't fix problems we don't 
know about!  :-)

> I know this is not really what you wanted to hear, but there it is. It
> feels as though USB2 on linux 2.6 is faulty.  I know the reasonings well
> why these drives function properly in Windows, and I know that you guys
> are trying your best to handle faulty hardware, but it doesn't change
> the user's perception that USB on linux 2.6 is buggy.

If we had a good test case to work with, we could try implementing an 
option to (hopefully) improve stability at the cost of reduced throughput.  

Alternatively, you can try using the ub driver in place of usb-storage.  
This is true for MAL as well.  I don't really know how it compares in 
terms of stability -- you'll have to try it and see for yourselves -- but 
it's definitely reduced throughput.  You can find ub in the kernel 
configuration under Device Drivers -> Block devices -> Low Performance USB 
Block driver.

Alan Stern



-------------------------------------------------------
This SF.Net email is sponsored by the JBoss Inc.
Get Certified Today * Register for a JBoss Training Course
Free Certification Exam for All Training Attendees Through End of 2005
Visit http://www.jboss.com/services/certification for more information
_______________________________________________
[email protected]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-users

Reply via email to