My build of kernel version 2.6.28.4 now works as well as 2.6.22.9, or in other words gets through the kernel boot fine and segfaults in an init script some place. There's a big annoying warning at the beginning about mtrrs not being set up, but the kernel silences that warning if it can tell it's running under kvm so I don't think it's very serious. This version of the kernel takes a while to verify that various timers are consistent, but even if it looks like it hung it does keep going.
The other kernel I've tried, 2.6.25.1, will panic when it tries to dereference a NULL pointer. I believe this is actually a kernel bug present all the way up to 2.6.30 and I've filed http://bugzilla.kernel.org/show_bug.cgi?id=13888 about it if you're curious. The real question now is why this particular bug is being tickled under M5 but not on real machines. Gabe Gabe Black wrote: > changeset 92c3973dd98c in /z/repo/m5 > details: http://repo.m5sim.org/m5?cmd=changeset;node=92c3973dd98c > description: > IDE: Configure the IDE control to reflect the initial value of the > command register. > > diffstat: > > 1 file changed, 3 insertions(+) > src/dev/ide_ctrl.cc | 3 +++ > > diffs (13 lines): > > diff -r 4c5671ecceda -r 92c3973dd98c src/dev/ide_ctrl.cc > --- a/src/dev/ide_ctrl.cc Sun Aug 02 08:39:29 2009 -0700 > +++ b/src/dev/ide_ctrl.cc Sun Aug 02 18:01:09 2009 -0700 > @@ -107,6 +107,9 @@ > } > primary.select(false); > secondary.select(false); > + > + ioEnabled = (config.command & htole(PCI_CMD_IOSE)); > + bmEnabled = (config.command & htole(PCI_CMD_BME)); > } > > bool > _______________________________________________ > m5-dev mailing list > [email protected] > http://m5sim.org/mailman/listinfo/m5-dev > _______________________________________________ m5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/m5-dev
