[sclug] Recovering gracefully from lost interrupt

[Simon Huggins]

On Thu, Aug 07, 2003 at 12:54:59AM +0100, Will Dickson wrote:
> Simon Huggins wrote:
> >On Wed, Aug 06, 2003 at 12:19:00AM +0100, Will Dickson wrote:
> >>Simon Huggins wrote:
> >>Nope! Various other kernels, both custom and standard, various other
> >>distros (various versions of SuSE), all OK. This and the other
> >>hardware / driver oddities I mentioned earlier appear to be specific
> >>to debian.  I'd report it, except I can't work out what the most
> >>appropriate package is.
> >Or specific to the kernel you were running in Debian.
> Perhaps, although IIRC is was just 2.4.20; I can't remember whether it
> was Debian's precompiled one or a customised one. (In the latter case,
> I am familiar with the kernel compile process; I've been compiling
> customised kernels successfully for several years prior to this
> incident.)

Distribution kernels are rarely the same as a particular vanilla version
from kernel.org though.

> >What I mean is can you reproduce it:
> >	- running the Debian kernel on a different distribution
> >	  (which I'd expect you would be able to)
> >	- running the vanilla kernel of the same version on Debian
> >	- running the vanilla kernel of the same version on something
> >	  else
> >Obviously all with the same .config.
> Don't know; I don't have the time or the spare machine to test at this
> point.

Hmm, fair enough.

> >I'm only trying to work this out because you seem to be propagating FUD
> >that Debian eats data and this isn't true IME.
> I am not propagating FUD, and I find that remark insulting.

I'm sorry I didn't aim to insult you but rather to find out which part
was causing the corruption.  If you can't test it to find out where the
problem is then you're effectively sticking your finger in the air and
saying "It's all Debian's fault" but without really knowing why or
indeed if it is uniquely Debian's problem.

> I am trying to alert people to an extremely serious problem that I
> encountered, in case they do as well. It may well be that my
> configuration somehow got fubared, and that nobody else will ever see
> this problem, in which case great. And by happy chance I caught this
> one before it had any consequences beyond giving me a fright, so all's
> well that ends well.  *But* if I hadn't, it could have meant four
> years' worth of records going up in smoke, and the consequences for
> our business would have been fatal.

I can understand your motivation but I'm not sure you're really doing
anyone a service by saying this.  You would need to say "kernel version
blah" or "the debian patched kernel version blah" or "Debian's userspace
tools from the blah distribtion" are broken rather than implying that
all of Debian is.

> Debian per se may well not eat data, but my installation DID.

Sure, but then we don't know if the same kernel on a different
distribution would have had the same effects or if it was a hardware bug
triggered by that kernel or a bug triggered in a vanilla kernel or a
cosmic ray flippings bits in your CPU's cache ;)

-- 
 ,--huggie-at-earth-dot-li--------stuff-thing-stuff----------DF5CE2B4--.
_|        If a tree fell on a florist, would he make a sound?         |_
 |                                                                    |
 `- http://www.earth.li/~huggie/ - http://www.blackcatnetworks.co.uk/ -'