diff options
| author | Bibo Mao <maobibo@loongson.cn> | 2023-12-19 10:48:27 +0800 |
|---|---|---|
| committer | Huacai Chen <chenhuacai@loongson.cn> | 2023-12-19 10:48:27 +0800 |
| commit | 7ab6fb505b2a7447c4a7237a12c59e3ad0c7298c (patch) | |
| tree | ef5aba12fcdba621d75b04e9027c54206bea7bd7 /arch/loongarch/include | |
| parent | ceb6a6f023fd3e8b07761ed900352ef574010bcb (diff) | |
LoongArch: KVM: Optimization for memslot hugepage checking
During shadow mmu page fault, there is checking for huge page for
specified memslot. Page fault is hot path, check logic can be done
when memslot is created. Here two flags are added for huge page
checking, KVM_MEM_HUGEPAGE_CAPABLE and KVM_MEM_HUGEPAGE_INCAPABLE.
Indeed for an optimized qemu, memslot for DRAM is always huge page
aligned. The flag is firstly checked during hot page fault path.
Now only huge page flag is supported, there is a long way for super
page support in LoongArch system. Since super page size is 64G for
16K pagesize and 1G for 4K pagesize, 64G physical address is rarely
used and LoongArch kernel needs support super page for 4K. Also memory
layout of LoongArch qemu VM should be 1G aligned.
Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>
Diffstat (limited to 'arch/loongarch/include')
| -rw-r--r-- | arch/loongarch/include/asm/kvm_host.h | 3 |
1 files changed, 3 insertions, 0 deletions
diff --git a/arch/loongarch/include/asm/kvm_host.h b/arch/loongarch/include/asm/kvm_host.h index 11328700d4fa..0e89db020481 100644 --- a/arch/loongarch/include/asm/kvm_host.h +++ b/arch/loongarch/include/asm/kvm_host.h @@ -45,7 +45,10 @@ struct kvm_vcpu_stat { u64 signal_exits; }; +#define KVM_MEM_HUGEPAGE_CAPABLE (1UL << 0) +#define KVM_MEM_HUGEPAGE_INCAPABLE (1UL << 1) struct kvm_arch_memory_slot { + unsigned long flags; }; struct kvm_context { |