Steven,
> diff --git a/arch/mips/kernel/proc.c b/arch/mips/kernel/proc.c
> index 5542817..2c18317 100644
> --- a/arch/mips/kernel/proc.c
> +++ b/arch/mips/kernel/proc.c
> @@ -64,6 +64,7 @@ static int show_cpuinfo(struct seq_file *m, void *v)
> cpu_data[n].watch_reg_masks[i]);
> seq_printf(m, "]\n");
> }
> + seq_printf(m, "microMIPS\t\t: %s\n", cpu_has_mmips ? "yes" : "no");
> seq_printf(m, "ASEs implemented\t:%s%s%s%s%s%s\n",
> cpu_has_mips16 ? " mips16" : "",
> cpu_has_mdmx ? " mdmx" : "",
I'm sorry I didn't notice that before, but it got lost in the load of
formatting changes. I think this should really be reported along the
other ASEs in the next line. Please move it down there, the pattern is
obvious (and then the MCU ASE may be added too with a simple follow-up
change).
However what I would find useful here and what I think is not reported
anywhere else and otherwise tricky (though possible) to track down is
whether the kernel itself has been built as a microMIPS or standard MIPS
binary. Perhaps you could reuse your entry above for that purpose -- I
suppose no tool has relied on this /proc/cpuinfo entry to make any
decisions so far, so it may be the right moment now to get it standardised
somehow.
I'll try to get back to the rest of the microMIPS review soon, I welcome
your submission very warmly, but that's a substantial change and I'm
really running out of time, sorry about that.
Maciej
|