On Tue, Dec 31, 2013 at 05:17:51PM +0200, Aaro Koskinen wrote:
> Hi,
>
> On Mon, Dec 30, 2013 at 10:33:54PM +0100, Aurelien Jarno wrote:
> > > switch (c->processor_id & PRID_REV_MASK) {
> > > case PRID_REV_LOONGSON2E:
> > > + c->cputype = CPU_LOONGSON2;
> > > + __cpu_name[cpu] = "ICT Loongson-2E";
> > > set_elf_platform(cpu, "loongson2e");
> > > break;
> > > case PRID_REV_LOONGSON2F:
> > > + c->cputype = CPU_LOONGSON2;
> > > + __cpu_name[cpu] = "ICT Loongson-2F";
> > > set_elf_platform(cpu, "loongson2f");
> >
> > I have mixed feelings about the Loongson-2 name change. On one side it's
> > clearly better to have 2E and 2F instead of 2 V0.2 and 2 V0.3, and it
> > should have been like that since the beginning. That said changing that
> > now is kind of breaking the userland. I know that it would break debian
> > installer support for example, though that should not be a real problem
> > as we ship the installer with a given kernel version. I don't know if
> > there are other usages that can cause a problem. Any opinion from
> > others?
>
> Changing it would break also GCC's -march=native detection. So NACK.
Indeed, good catch. That said it already looks for both Godson2 and
Loongson-2, so I guess such a patch is acceptable if someone updates GCC
and waits a few years before submitting it again.
--
Aurelien Jarno GPG: 1024D/F1BCDB73
aurelien@xxxxxxxxxxx http://www.aurel32.net
