Alle 21:09, luned=EC 29 dicembre 2003, Jason Lunz ha scritto:
> blaisorblade_spam@... said:
> > I have some doubts about the 2.6 host-skas patch. I've searched through
> > the list for earlier ports to 2.6 and I've seen someone (Bryan O'Sulliv=
> > <bos@...>) complaining:
> > "The big issue with the skas patches, though, appears to be that they
> > don't handle the UML kernel being killed, which results in the host
> > kernel's memory management being completely trashed."
> > Is this a real issue or not?
> I think so. I'm running 2.6.0 vanilla + Ingo's combo -A2 on my host,
> with 2.4.23 UMLs.
> I tried to do something in the uml involving lots of
> tmpfs usage, and the host system crashed.
If you use the Ingo's combo patch, maybe it's not the Skas mode itself.
There is a change in mm/memory.c which is fine for UML, but IMHO probably
hurts for the host kernel. Looking at the patch he posts, however, he didn'=
use this change, so beyond the mm/memory.c problem there must be another on=
To test that, you can either select the changes to that file away from the=
patch, or unapply (with -R option to patch) this patch, which contains=20
exactly that change:
Paolo Giarrusso, aka Blaisorblade
Linux Kernel 2.4.21/2.6.0-test on an i686; Linux registered user n. 292729