[Gta04-owner] Modem reenummerating

[NeilBrown]

On Mon, 26 Aug 2013 23:16:42 +0200 EdorFaus <edorfaus at xepher.net> wrote:

> On 08/26/2013 12:47 PM, Radek Polak wrote:
> > On Sunday, August 25, 2013 01:22:00 PM Christ van Willegen wrote:
> >  > On Sat, Aug 24, 2013 at 11:24 PM, Lukas Märdian <luk at slyon.de> wrote:
> >  > > Am 24.08.2013 23:04, schrieb NeilBrown:
> >  > >> I'm starting to think the the re-enumeration could be an omap-usb
> >  > >> problem. I turned off a couple of power-saving features and the rate
> >  > >> at which my modem reenumerates went way down. Sometimes the the modem
> >  > >
> >  > > Hmm could this by any chance have something to do with the infamous
> >  > > Linux-USB problem, recently pointed out by Sarah Sharp?
> >  > >
> >  > > http://permalink.gmane.org/gmane.linux.usb.general/93288
> >  >
> >  > Well, it's worth a try!
> 
> > I just tried the patch and it does not help with our problem :(
> 
> Hm, maybe try an even longer delay? The 20ms was based on tests with a 
> specific Intel chip, so maybe ours is even slower?

If you read the rest of the email thread, it looks like it was a false alarm.
Sarah found an issue with the USB3 (xhci) driver which looked like a long-term
general USB problem and jumped the gun a little.  USB2 (ehci) works
sufficiently differently that the real cause of the USB3 problem isn't
relevant to USB2.  And '10ms' is the correct time to wait for USB2.

> 
> I guess one way to test that theory would be to set the delay very long, 
> like a whole second or so - something that might not be usable in 
> practice, but at least should be definitely long enough if that *is* the 
> problem.
> 
> 
> > Btw the reenumaration happens even without suspend/resume at all.
> 
> Oh, hm. That indeed suggests deeper issues.

Actually it doesn't.  The problem wasn't about system suspend/resume.  It was
related to runtime suspend/resume.  And that can certainly happen without the
system going into suspend.


> 
> Or... at least *other* issues - which might possibly be additional 
> instead of alternative? *shrug*
> 
> 
> Hey, there's an idea... I don't know if the external USB port is 
> connected in an equivalent way as the modem is (or if there's any other 
> internal USB devices we could use), but if that port is set to host mode 
> and a device is connected, is that device also reenumerated when the 
> modem is?

The  external and internal ports are very different.  Different PHY,
different USB engine, different driver.  About the only thing they have in
common is the 3 letters U S and B.
So a nice idea, but unfortunately not one that goes anywhere.

> If it is, it's unlikely to be a problem with the modem. If not, we could 
> try to figure out what the difference between them is?
> 
> Of course, if it *is* a timing problem, that test might not yield any 
> useful info, since that other device would have different timing, 
> perhaps enough so to never trigger reenumeration anyway...
> 
> -Frode

NeilBrown

> _______________________________________________
> Gta04-owner mailing list
> Gta04-owner at goldelico.com
> http://lists.goldelico.com/mailman/listinfo/gta04-owner

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 828 bytes
Desc: not available
URL: <http://lists.goldelico.com/pipermail/gta04-owner/attachments/20130827/eb474802/attachment.bin>