On Mon, Oct 03, 2005 at 04:01:23PM -0400, Alan Stern wrote:
> On Mon, 3 Oct 2005, Yedidyah Bar-David wrote:
> 
> > > The device that works with everything except 2.6.14-rc3 is the card 
> > > reader?
> > 
> > Yes.
> 
> It certainly would be worthwhile getting some debugging logs for 
> comparison on that.  2.6.14-rc2 and 2.6.14-rc3 don't differ by very much.

OK, I'll make some tomorrow. It's too late now for me :-(

> 
> > The pen drive.
> > 
> > > I was unable to download them (error 404).
> > 
> > Sorry, my mistake - it's in
> > <http://www.cs.tau.ac.il/~didi/kernA.log>
> > <http://www.cs.tau.ac.il/~didi/kernB.log>
> > 
> > I can also report that it works with 2.6.8-rc1 and does not work with
> > 2.6.8-rc2. Which means the regression happened between these two. Can
> > anyone tell me how to see individual patches between them? Or I came to
> > the finest point in my binary search and should start looking at the
> > code?
> 
> At this point you probably want to examine the code.  Or at least the 
> portions that changed between -rc1 and -rc2.

That's what I intend to do now (tomorrow).

> 
> However I'm not completely convinced you'll find anything.  Those "Medium 
> not present" messages in your log should never occur with a flash device.  
> They indicate a hardware or firmware problem.
> 
> It might turn out to be something trivial, like your B machine doesn't 
> provide sufficient voltage or current on the power connection of the USB 
> bus.

Fine, but why does it work with 2.6.8-rc1?
Maybe, for example, it is indeed a hardware bug, which didn't manifest
itself in 2.6.8-rc1 because there was somewhere a larger delay that let
the bad hardware manage to do its job, and rc2 shortened that delay.
Wouldn't it make sense to make the delay larger back?

As I said in the first message, BTW, I do know for sure that the
hardware is bad. But because it always worked for me well, and still
does on some versions, I thought the only problem with it is in a
specific part of the memory, not in other parts like the firmware. If
you are right, and this indeed should never happen with good hardware,
then
1. I am much releaved, and might put the machine to production even
without solving this, as I do not particularly care about the specific
pen - only used it for testing,
2. It's still a question whether we want linux to support bad hardware
if possible.

> 
> Also keep in mind that sometimes failures occur at random.  You might get 
> fooled into thinking a particular kernel release is at fault, simply 
> because by pure chance the previous release ran okay in your test.

I do not understand how that can be. I tried several times to plug it
in and out in both versions, and it always worked with rc1 and never
with rc2.
-- 
Didi



-------------------------------------------------------
This SF.Net email is sponsored by:
Power Architecture Resource Center: Free content, downloads, discussions,
and more. http://solutions.newsforge.com/ibmarch.tmpl
_______________________________________________
[email protected]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-users

Reply via email to