Re: Debian testing/unstable on RiscPC ?
Am Donnerstag, 29. Januar 2004 00:24 schrieb Philip Blundell:
> Yes, partly. StrongARM RiscPCs have always had a bit of an identity
> crisis; the core advertises itself as ARMv4, and does implement that
> architecture, but the memory system won't cope with halfword accesses so
> those instructions are unusable. Effectively, it's reduced to an ARMv3M
> kind of level. The kernel used to have some hack to fudge the visible
> architecture value in this situation, but it might have gotten lost.
Kernel 2.6.0-rmk2 reports a correct set of features in /proc/cpuinfo: "swp
26bit fastmult", and does not report "half". It does, however, advertise the
processor architecture as "v4l". I should probably generate a patch.