英雄帖: LA264/364/664 target support in QEMU

[xen0n]

We have received message from the Loongson QEMU team that they wouldn't (have the bandwidth to) work on these targets at the moment, so volunteers are welcome.

LA664 support would be more tricky, because one would have to implement the new semantics of LoongArch v1.1. The sc.q is a known tricky one to start with.

The other smaller uarches would probably be easier or even trivial, but that's only the CPU side. The SoC peripherals can also be tricky to get right, but we can see if a virtualization-oriented approach (i.e. virt platforms) would work.

Please reply in the thread if you'd like to try!

[xry111]

Oops, I just hit this when I attempted to test my GCC 16B atomic changes on a 3C6000 VM from AOSC.

[xry111]

Just tried sc.q on that VM and it worked perfectly. I'll see if CPUCFG output is correct.

[xry111]

Just tried sc.q on that VM and it worked perfectly. I'll see if CPUCFG output is correct.

No, CPUCFG does not announce the LA1.1 instructions (though they are usable).

[jiegec]

We should take TCG and KVM cases separately.