Improve test_bios_region_rw() by using clflush for properly ensuring caches are flushed

[mintsuki]

Also, move test_bios_region_rw() to unlock_region.c file and make it static.

[FlyGoat]

I'm not sure cflush is necessary as BIOS regions are usually writeback cached and it won't hurt to keep data in cache. It won't hurt either, so I'm happy to apply this change tho.

[mintsuki]

I'm not sure cflush is necessary as BIOS regions are usually writeback cached and it won't hurt to keep data in cache. It won't hurt either, so I'm happy to apply this change tho.

But what if the data is evicted, or what about APs? Then the BIOS and VBIOS would basically disappear, which doesn't sound quite right to me...

[FlyGoat]

I'm not sure cflush is necessary as BIOS regions are usually writeback cached and it won't hurt to keep data in cache. It won't hurt either, so I'm happy to apply this change tho.

But what if the data is evicted, or what about APs? Then the BIOS and VBIOS would basically disappear, which doesn't sound quite right to me...

Ah I see your point, so your concern is Cache is accepting writes, but underlying memory is not? This won't happen on Intel/AMD h/w as it seems like writing permission check is done at microcode level but it makes sense to test it this way.

[FlyGoat]

LGTM now, thanks a lot.