On Sun, Jan 21, 2018 at 09:29:18AM +0000, Mark Cave-Ayland wrote:
On 20/01/18 15:51, Segher Boessenkool wrote:
On Sat, Jan 20, 2018 at 09:38:21AM -0500, Jd Lyons via OpenBIOS wrote:
--- a/arch/ppc/qemu/init.c +++ b/arch/ppc/qemu/init.c @@ -550,6 +550,18 @@ static const struct cpudef ppc_defs[] = { .tlb_size = 0x80, .initfn = cpu_g4_init, },
- .iu_version = 0x80030000,
- .name = "PowerPC,G4",
- .icache_size = 0x8000,
- .dcache_size = 0x8000,
- .icache_sets = 0x80,
- .dcache_sets = 0x80,
- .icache_block_size = 0x20,
- .dcache_block_size = 0x20,
- .tlb_sets = 0x40,
- .tlb_size = 0x80,
- .initfn = cpu_g4_init,
+},
These numbers are all correct (I checked against the manual).
Should there be something for L2 as well?
Good question - I suspect you're the best person to know here ;) Could not having the L2 properties cause strange behaviour, or will the guest OS carry on fine without them present? (I presume this isn't a NACK?)
This isn't a NAK; the other CPUs are like that as well in the openbios code I guess, and some of those have L2.
L2 is described in a node _under_ the "cpu" node in the device tree. Linux does not care if it is absent; I don't know about the other OSes though. If everything works it is not a pressing concern ;-)
Segher