Subject: Re: NetBSD SQA process
To: NetBSD Kernel Technical Discussion List <>
From: Greg A. Woods <>
List: tech-kern
Date: 04/19/1999 16:59:27
[ On Monday, April 19, 1999 at 18:35:59 (+0200), I. Souvatzis wrote: ]
> Subject: Re: NetBSD SQA process
> On Mon, Apr 19, 1999 at 12:31:02PM -0400, Greg A. Woods wrote:
> > [ On Monday, April 19, 1999 at 10:25:50 (+0200), I. Souvatzis wrote: ]
> > > Subject: Re: NetBSD SQA process
> > >
> > > I know of (admittedly broken) code (involving signal trampolines) that crashes
> > > NetBSD/AMiga on a 68060 cpu (at least on Rev. 1 CPUs). The compiler has been 
> > > fixed to emit correct code, but the kernel problem (if this is can be fixed
> > > at all) hasn't been found.
> > > 
> > > So, I'm sure crashme will eventually crash my machine, and it won't tell me
> > > anything new by doing so...
> > 
> > It won't tell you anything new if, and only if it *only* crashes your
> > machine because of that one known bug!  ;-)
> how would I tell the difference?

If the symptoms were identical to those caused by the "known" bug, then
obviously you wouldn't be able to.  However I'm guessing that all
possible crashes won't have the same symptoms as this bug.

BTW, I definitely do not condone adding "make regression-tests" to the
normal build, and it looks like the current behaviour of "regress" being
the default target in all of the relevant regress/.../Makefiles might
even be considered a bug -- I certainly didn't expect it.  Until I
peeked at that stuff for the first time today I had expected "make
build" to simply build any necessary test programs and harnesses,
etc. and that "make regression-tests" was explicitly required to
actually run the tests.  I certainly would have complained if one of
these tests had caused a panic during "make build"!  I definitely want
to know when I'm running tests so that I can prepare to do something
with the results!

							Greg A. Woods

+1 416 218-0098      VE3TCP      <>      <robohack!woods>
Planix, Inc. <>; Secrets of the Weird <>