Re: hard-to-reproduce bugs (was: plugged in thumb drive, which disconnected another USB drive)
On Mon, Mar 16, 2015, at 00:16, Mike Kupfer wrote:
> Chris Bannister wrote:
> > On Thu, Mar 12, 2015 at 09:04:51AM -0700, Mike Kupfer wrote:
> > > - Do I need to have a reproducible test case before opening a bug?
> >
> > If it is not reproducible, then how does one go about fixing it?
>
> I deal with hard-to-reproduce bugs in my day job fairly frequently.
> Sometimes I'm able to suggest refinements to the test case to make the
> problem easier to reproduce. Sometimes I can tell the submitter what
> additional information to gather the next time the problem occurs.
> Sometimes the bug report sits idle for awhile as I wait for additional
> reports and additional clues. In theory I can provide a special debug
> binary in the hopes of getting more information, though in practice I
> can't remember the last time I did this. Sometimes I just find where a
> particular message is generated and poke through the source code to see
> if any causes leap out at me.
>
> I don't mean to suggest that I expect Debian package maintainers to do
> all that. I asked because I've had mixed results asking support
> questions on this list, and I'm not sure what the cultural norms are in
> Debian for filing a formal bug report.
Fair enough. Anyway, in this specific case, it is likely caused by the
hardware layer: it really looks like what would happen if the IOMEGA
device was subject to a "sudden and short-term power loss/drop" for
whatever reason when you plugged the other USB device on the same hub.
I don't think it is worth reporting it as a kernel bug, unless you have
enough details to show it was bad behavior on the USB host side, or a
bug in the available power allocation done by the kernel before it
decides whether it can activate the new USB device endpoints or not.
Also, IMHO, you should get rid of that USB hub and connect important
devices directly to the motherboard USB ports, using a proper USB cable
(and finding a non-substandard USB cable might be more challenging than
expected).
I won't go into details unless people around here are very interested on
the specifics. But since it starts with "way too many USB devices and
cables do not implement the USB specification properly", and goes
downhill from there, it is not likely to be very helpful.
--
"One disk to rule them all, One disk to find them. One disk to bring
them all and in the darkness grind them. In the Land of Redmond
where the shadows lie." -- The Silicon Valley Tarot
Henrique de Moraes Holschuh <hmh@debian.org>
Reply to: