FC2 x86_64 SMP broken.

[ByteEnable]

On Thursday 20 May 2004 21:41, Sam Varshavchik wrote:
> Warren Togami writes:
> > Sam Varshavchik wrote:
> >> Add me to the list of those whose FC2 kernel crashes and burns during
> >> boot time.  Bug 123860.  At least I can limp along on the UP kernel.
> >>
> >> Well, perhaps not.  In FC1, I can crash the 2.4 kernel, SMP or UP,
> >> anytime I want.
> >>
> >> Let's see if I can pronounce FC2 to be completely and totally useless.
> >
> > This is an unfair flame to make and simply goes to far in simply
> > dismissing an entire operating system.  This kind of comment is totally
> > unhelpful for these reasons:
> >
> > 1) You provide ZERO information about your hardware.
>
> I submitted a complete traceback to bugzilla.  Perhaps you missed that. 
> The kernel crashed somewhere in its memory handler
> (cache_grow/kmem_cache_all/cache_alloc_refill) which later down the line
> was invoked by the ext3 driver.  Does not look like hardware-specific to
> me.
>
> > 2) You are not searching for a solution, you only declare Fedora to be
> > crap.
>
> Actually I've been searching for a solution for about three months now.
>
> > 3) SMP x86_64 works for MANY other people, including me.  This is
> > clearly not indicative of a useless operating system.
> >
> > This kind of post makes developers who put their blood, sweat and tears
> > into trying their best to simply ignore you.  Can we instead try to work
> > toward a solution together, and constructively?
> >
> > http://people.redhat.com/arjanv/2.6/
> > As a first step, can you try the newer test kernels from Arjan's
> > repository?
>
> I most certainly did.  The 2.6.6 kernel crashed in exactly the same manner.
>
> You know, I try to help when I can.  This is similar to an incident that
> happened in the early 2.4 series, and I was actually was able to find and
> fix a boner in the apic interrupt handler.  Unfortunately, things were much
> simpler back in those days.  Now, these kinds of things are beyond my mere
> mortal powers to fix.
>
> And if it was only this crash it wouldn't be too bad.  But having a bunch
> of SCSI modules pulled out, with hardly a mention in any README or
> RELEASE-NOTES is really not NICE.

He must have realized that he didn't have memory DIMM plugged in 
correctly...hehe....

Byte