rminnich at lanl.gov
Thu Jun 3 14:48:00 CEST 2004
On Thu, 3 Jun 2004, Stefan Reinauer wrote:
> * ron minnich <rminnich at lanl.gov> [040603 17:23]:
> > > the current pirq code defines intel_irq_routing_table as const, thus
> > > making it potentially readonly.
> > should be const due to execute-in-place issues.
> What's the benefit in this case? we have dram long before this code is
> run, and the table is really small. Should it be copied to ram
> completely maybe?
the only reason is that in the execute in place case you can't write to
it. Leaving it as const will catch bugs involving people writing to it.
> Nope, parts of that is my fault. But I still think it is a good idea to
> go towards dynamic table creation.
yes. we're just not there.
More information about the coreboot