Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Subject: Re: [PATCH v14 19/30] LoongArch: KVM: Implement kvm mmu operations
From:   zhaotianrui <zhaotianrui@loongson.cn>
To:     Huacai Chen <chenhuacai@kernel.org>
Cc:     linux-kernel@vger.kernel.org, kvm@vger.kernel.org,
        Paolo Bonzini <pbonzini@redhat.com>,
        WANG Xuerui <kernel@xen0n.name>,
        Greg Kroah-Hartman <gregkh@linuxfoundation.org>,
        loongarch@lists.linux.dev, Jens Axboe <axboe@kernel.dk>,
        Mark Brown <broonie@kernel.org>,
        Alex Deucher <alexander.deucher@amd.com>,
        Oliver Upton <oliver.upton@linux.dev>, maobibo@loongson.cn,
        Xi Ruoyao <xry111@xry111.site>, tangyouling@loongson.cn
References: <20230619083255.3841777-1-zhaotianrui@loongson.cn>
 <20230619083255.3841777-20-zhaotianrui@loongson.cn>
 <CAAhV-H6RS59O0bG31HUe3nfLPXammLq0KXxXr-n_KqUgc1=naQ@mail.gmail.com>
 <8db3a563-d69a-2ab8-787c-a80e44f95e44@loongson.cn>
Message-ID: <5196349b-95dd-97ad-214a-0101e9e68cf9@loongson.cn>
Date:   Tue, 20 Jun 2023 16:14:39 +0800
User-Agent: Mozilla/5.0 (X11; Linux loongarch64; rv:68.0) Gecko/20100101
 Thunderbird/68.7.0
MIME-Version: 1.0
In-Reply-To: <8db3a563-d69a-2ab8-787c-a80e44f95e44@loongson.cn>
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit
Content-Language: en-US
Precedence: bulk


在 2023/6/20 下午2:59, zhaotianrui 写道:
>
> 在 2023/6/19 下午6:30, Huacai Chen 写道:
>> Hi, Tianrui,
>>
>> On Mon, Jun 19, 2023 at 4:33 PM Tianrui Zhao 
>> <zhaotianrui@loongson.cn> wrote:
>>> Implement LoongArch kvm mmu, it is used to switch gpa to hpa when
>>> guest exit because of address translation exception. This patch
>>> implement allocate gpa page table, search gpa from it and flush guest
>>> gpa in the table.
>>>
>>> Reviewed-by: Bibo Mao <maobibo@loongson.cn>
>>> Signed-off-by: Tianrui Zhao <zhaotianrui@loongson.cn>
>>> ---
>>>   arch/loongarch/kvm/mmu.c | 725 
>>> +++++++++++++++++++++++++++++++++++++++
>>>   1 file changed, 725 insertions(+)
>>>   create mode 100644 arch/loongarch/kvm/mmu.c
>>>
>>> diff --git a/arch/loongarch/kvm/mmu.c b/arch/loongarch/kvm/mmu.c
>>> new file mode 100644
>>> index 000000000000..d75446139546
>>> --- /dev/null
>>> +++ b/arch/loongarch/kvm/mmu.c
>>> @@ -0,0 +1,725 @@
>>> +// SPDX-License-Identifier: GPL-2.0
>>> +/*
>>> + * Copyright (C) 2020-2023 Loongson Technology Corporation Limited
>>> + */
>>> +
>>> +#include <linux/highmem.h>
>>> +#include <linux/page-flags.h>
>>> +#include <linux/kvm_host.h>
>>> +#include <linux/uaccess.h>
>>> +#include <asm/mmu_context.h>
>>> +#include <asm/pgalloc.h>
>>> +#include <asm/tlb.h>
>>> +
>>> +/*
>>> + * KVM_MMU_CACHE_MIN_PAGES is the number of GPA page table 
>>> translation levels
>>> + * for which pages need to be cached.
>>> + */
>>> +#define KVM_MMU_CACHE_MIN_PAGES (CONFIG_PGTABLE_LEVELS - 1)
>>> +
>>> +/**
>>> + * kvm_pgd_alloc() - Allocate and initialise a KVM GPA page directory.
>>> + *
>>> + * Allocate a blank KVM GPA page directory (PGD) for representing 
>>> guest physical
>>> + * to host physical page mappings.
>>> + *
>>> + * Returns:    Pointer to new KVM GPA page directory.
>>> + *             NULL on allocation failure.
>>> + */
>>> +pgd_t *kvm_pgd_alloc(void)
>>> +{
>>> +       pgd_t *pgd;
>>> +
>>> +       pgd = (pgd_t *)__get_free_pages(GFP_KERNEL, 0);
>>> +       if (pgd)
>>> +               pgd_init((void *)pgd);
>>> +
>>> +       return pgd;
>>> +}
>>> +
>>> +/**
>>> + * kvm_walk_pgd() - Walk page table with optional allocation.
>>> + * @pgd:       Page directory pointer.
>>> + * @addr:      Address to index page table using.
>>> + * @cache:     MMU page cache to allocate new page tables from, or 
>>> NULL.
>>> + *
>>> + * Walk the page tables pointed to by @pgd to find the PTE 
>>> corresponding to the
>>> + * address @addr. If page tables don't exist for @addr, they will 
>>> be created
>>> + * from the MMU cache if @cache is not NULL.
>>> + *
>>> + * Returns:    Pointer to pte_t corresponding to @addr.
>>> + *             NULL if a page table doesn't exist for @addr and 
>>> !@cache.
>>> + *             NULL if a page table allocation failed.
>>> + */
>>> +static pte_t *kvm_walk_pgd(pgd_t *pgd, struct kvm_mmu_memory_cache 
>>> *cache,
>>> +                               unsigned long addr)
>>> +{
>>> +       p4d_t *p4d;
>>> +       pud_t *pud;
>>> +       pmd_t *pmd;
>>> +
>>> +       pgd += pgd_index(addr);
>>> +       if (pgd_none(*pgd)) {
>>> +               /* Not used yet */
>>> +               BUG();
>>> +               return NULL;
>>> +       }
>>> +       p4d = p4d_offset(pgd, addr);
>>> +       pud = pud_offset(p4d, addr);
>>> +       if (pud_none(*pud)) {
>>> +               pmd_t *new_pmd;
>>> +
>>> +               if (!cache)
>>> +                       return NULL;
>>> +               new_pmd = kvm_mmu_memory_cache_alloc(cache);
>>> +               pmd_init((void *)new_pmd);
>>> +               pud_populate(NULL, pud, new_pmd);
>>> +       }
>>> +       pmd = pmd_offset(pud, addr);
>>> +       if (pmd_none(*pmd)) {
>>> +               pte_t *new_pte;
>>> +
>>> +               if (!cache)
>>> +                       return NULL;
>>> +               new_pte = kvm_mmu_memory_cache_alloc(cache);
>>> +               clear_page(new_pte);
>>> +               pmd_populate_kernel(NULL, pmd, new_pte);
>>> +       }
>>> +       return pte_offset_kernel(pmd, addr);
>>> +}
>>> +
>>> +/* Caller must hold kvm->mm_lock */
>>> +static pte_t *kvm_pte_for_gpa(struct kvm *kvm,
>>> +                               struct kvm_mmu_memory_cache *cache,
>>> +                               unsigned long addr)
>>> +{
>>> +       return kvm_walk_pgd(kvm->arch.gpa_mm.pgd, cache, addr);
>>> +}
>>> +
>>> +/*
>>> + * level2_flush_{pte,pmd,pud,pgd,pt}.
>>> + * Flush a range of guest physical address space from the VM's GPA 
>>> page tables.
>>> + */
>> What's the meaning of level2 in this patch? It looks strange and I
>> haven't seen others use them, maybe we can use some better names.
>>
>> Huacai
> level2 means the address translation of gpa to hpa. As for level1, it 
> means gva to gpa. The same as stage-2 in arm64 KVM.
>
> Thanks
> Tianrui Zhao
And there are some other archs KVM mmu name, such as gstage_mmu in 
riscv, tdp(two dimension paging)_mmu in x86, stage2_mmu in arm64, so 
what do you think of level2_mmu in our LoongArch KVM?

Thanks
Tianrui Zhao
>>
>>> +static int level2_flush_pte(pmd_t *pmd, unsigned long addr, 
>>> unsigned long end)
>>> +{
>>> +       pte_t *pte;
>>> +       unsigned long next, start;
>>> +       int ret;
>>> +
>>> +       ret = 0;
>>> +       start = addr;
>>> +       pte = pte_offset_kernel(pmd, addr);
>>> +       do {
>>> +               next = addr + PAGE_SIZE;
>>> +               if (!pte_present(*pte))
>>> +                       continue;
>>> +
>>> +               set_pte(pte, __pte(0));
>>> +               ret = 1;
>>> +       } while (pte++, addr = next, addr != end);
>>> +
>>> +       if (start + PMD_SIZE == end) {
>>> +               pte = pte_offset_kernel(pmd, 0);
>>> +               pmd_clear(pmd);
>>> +               pte_free_kernel(NULL, pte);
>>> +       }
>>> +       return ret;
>>> +}
>>> +
>>> +static int level2_flush_pmd(pud_t *pud, unsigned long addr, 
>>> unsigned long end)
>>> +{
>>> +       pmd_t *pmd;
>>> +       unsigned long next, start;
>>> +       int ret;
>>> +
>>> +       ret = 0;
>>> +       start = addr;
>>> +       pmd = pmd_offset(pud, addr);
>>> +       do {
>>> +               next = pmd_addr_end(addr, end);
>>> +               if (!pmd_present(*pmd))
>>> +                       continue;
>>> +
>>> +               ret |= level2_flush_pte(pmd, addr, next);
>>> +       } while (pmd++, addr = next, addr != end);
>>> +
>>> +       if (start + PUD_SIZE == end) {
>>> +               pmd = pmd_offset(pud, 0);
>>> +               pud_clear(pud);
>>> +               pmd_free(NULL, pmd);
>>> +       }
>>> +       return ret;
>>> +}
>>> +
>>> +static int level2_flush_pud(pgd_t *pgd, unsigned long addr, 
>>> unsigned long end)
>>> +{
>>> +       p4d_t *p4d;
>>> +       pud_t *pud;
>>> +       unsigned long next, start;
>>> +       int ret;
>>> +
>>> +       ret = 0;
>>> +       start = addr;
>>> +       p4d = p4d_offset(pgd, addr);
>>> +       pud = pud_offset(p4d, addr);
>>> +       do {
>>> +               next = pud_addr_end(addr, end);
>>> +               if (!pud_present(*pud))
>>> +                       continue;
>>> +
>>> +               ret |= level2_flush_pmd(pud, addr, next);
>>> +       } while (pud++, addr = next, addr != end);
>>> +
>>> +       if (start + PGDIR_SIZE == end) {
>>> +               pud = pud_offset(p4d, 0);
>>> +               pgd_clear(pgd);
>>> +               pud_free(NULL, pud);
>>> +       }
>>> +       return ret;
>>> +}
>>> +
>>> +static int level2_flush_pgd(pgd_t *pgd, unsigned long addr, 
>>> unsigned long end)
>>> +{
>>> +       unsigned long next;
>>> +       int ret;
>>> +
>>> +       ret = 0;
>>> +       if (addr > end - 1)
>>> +               return ret;
>>> +       pgd = pgd + pgd_index(addr);
>>> +       do {
>>> +               next = pgd_addr_end(addr, end);
>>> +               if (!pgd_present(*pgd))
>>> +                       continue;
>>> +
>>> +               ret |= level2_flush_pud(pgd, addr, next);
>>> +       }  while (pgd++, addr = next, addr != end);
>>> +
>>> +       return ret;
>>> +}
>>> +
>>> +/**
>>> + * level2_flush_range() - Flush a range of guest physical addresses.
>>> + * @kvm:       KVM pointer.
>>> + * @start_gfn: Guest frame number of first page in GPA range to flush.
>>> + * @end_gfn:   Guest frame number of last page in GPA range to flush.
>>> + *
>>> + * Flushes a range of GPA mappings from the GPA page tables.
>>> + *
>>> + * The caller must hold the @kvm->mmu_lock spinlock.
>>> + *
>>> + * Returns:    Whether its safe to remove the top level page 
>>> directory because
>>> + *             all lower levels have been removed.
>>> + */
>>> +static bool level2_flush_range(struct kvm *kvm, gfn_t start_gfn, 
>>> gfn_t end_gfn)
>>> +{
>>> +       return level2_flush_pgd(kvm->arch.gpa_mm.pgd, start_gfn << 
>>> PAGE_SHIFT,
>>> +                               end_gfn << PAGE_SHIFT);
>>> +}
>>> +
>>> +typedef int (*level2_pte_ops)(void *pte);
>>> +/*
>>> + * level2_mkclean_pte
>>> + * Mark a range of guest physical address space clean (writes 
>>> fault) in the VM's
>>> + * GPA page table to allow dirty page tracking.
>>> + */
>>> +static int level2_mkclean_pte(void *pte)
>>> +{
>>> +       pte_t val;
>>> +
>>> +       val = *(pte_t *)pte;
>>> +       if (pte_dirty(val)) {
>>> +               *(pte_t *)pte = pte_mkclean(val);
>>> +               return 1;
>>> +       }
>>> +       return 0;
>>> +}
>>> +
>>> +static int level2_ptw_pte(pmd_t *pmd, unsigned long addr, unsigned 
>>> long end,
>>> +                       level2_pte_ops func)
>>> +{
>>> +       pte_t *pte;
>>> +       unsigned long next;
>>> +       int ret;
>>> +
>>> +       ret = 0;
>>> +       pte = pte_offset_kernel(pmd, addr);
>>> +       do {
>>> +               next = addr + PAGE_SIZE;
>>> +               if (!pte_present(*pte))
>>> +                       continue;
>>> +
>>> +               ret |= func(pte);
>>> +       } while (pte++, addr = next, addr != end);
>>> +
>>> +       return ret;
>>> +}
>>> +
>>> +static int level2_ptw_pmd(pud_t *pud, unsigned long addr, unsigned 
>>> long end,
>>> +                       level2_pte_ops func)
>>> +{
>>> +       pmd_t *pmd;
>>> +       unsigned long next;
>>> +       int ret;
>>> +
>>> +       ret = 0;
>>> +       pmd = pmd_offset(pud, addr);
>>> +       do {
>>> +               next = pmd_addr_end(addr, end);
>>> +               if (!pmd_present(*pmd))
>>> +                       continue;
>>> +
>>> +               ret |= level2_ptw_pte(pmd, addr, next, func);
>>> +       } while (pmd++, addr = next, addr != end);
>>> +
>>> +       return ret;
>>> +}
>>> +
>>> +static int level2_ptw_pud(pgd_t *pgd, unsigned long addr, unsigned 
>>> long end,
>>> +                       level2_pte_ops func)
>>> +{
>>> +       p4d_t *p4d;
>>> +       pud_t *pud;
>>> +       unsigned long next;
>>> +       int ret;
>>> +
>>> +       ret = 0;
>>> +       p4d = p4d_offset(pgd, addr);
>>> +       pud = pud_offset(p4d, addr);
>>> +       do {
>>> +               next = pud_addr_end(addr, end);
>>> +               if (!pud_present(*pud))
>>> +                       continue;
>>> +
>>> +               ret |= level2_ptw_pmd(pud, addr, next, func);
>>> +       } while (pud++, addr = next, addr != end);
>>> +
>>> +       return ret;
>>> +}
>>> +
>>> +static int level2_ptw_pgd(pgd_t *pgd, unsigned long addr, unsigned 
>>> long end,
>>> +                       level2_pte_ops func)
>>> +{
>>> +       unsigned long next;
>>> +       int ret;
>>> +
>>> +       ret = 0;
>>> +       if (addr > end - 1)
>>> +               return ret;
>>> +       pgd = pgd + pgd_index(addr);
>>> +       do {
>>> +               next = pgd_addr_end(addr, end);
>>> +               if (!pgd_present(*pgd))
>>> +                       continue;
>>> +
>>> +               ret |= level2_ptw_pud(pgd, addr, next, func);
>>> +       }  while (pgd++, addr = next, addr != end);
>>> +
>>> +       return ret;
>>> +}
>>> +
>>> +/*
>>> + * kvm_mkclean_gpa_pt() - Make a range of guest physical addresses 
>>> clean.
>>> + * @kvm:       KVM pointer.
>>> + * @start_gfn: Guest frame number of first page in GPA range to flush.
>>> + * @end_gfn:   Guest frame number of last page in GPA range to flush.
>>> + *
>>> + * Make a range of GPA mappings clean so that guest writes will 
>>> fault and
>>> + * trigger dirty page logging.
>>> + *
>>> + * The caller must hold the @kvm->mmu_lock spinlock.
>>> + *
>>> + * Returns:    Whether any GPA mappings were modified, which would 
>>> require
>>> + *             derived mappings (GVA page tables & TLB enties) to be
>>> + *             invalidated.
>>> + */
>>> +static int kvm_mkclean_gpa_pt(struct kvm *kvm, gfn_t start_gfn, 
>>> gfn_t end_gfn)
>>> +{
>>> +       return level2_ptw_pgd(kvm->arch.gpa_mm.pgd, start_gfn << 
>>> PAGE_SHIFT,
>>> +                               end_gfn << PAGE_SHIFT, 
>>> level2_mkclean_pte);
>>> +}
>>> +
>>> +/*
>>> + * kvm_arch_mmu_enable_log_dirty_pt_masked() - write protect dirty 
>>> pages
>>> + * @kvm:       The KVM pointer
>>> + * @slot:      The memory slot associated with mask
>>> + * @gfn_offset:        The gfn offset in memory slot
>>> + * @mask:      The mask of dirty pages at offset 'gfn_offset' in 
>>> this memory
>>> + *             slot to be write protected
>>> + *
>>> + * Walks bits set in mask write protects the associated pte's. 
>>> Caller must
>>> + * acquire @kvm->mmu_lock.
>>> + */
>>> +void kvm_arch_mmu_enable_log_dirty_pt_masked(struct kvm *kvm,
>>> +               struct kvm_memory_slot *slot,
>>> +               gfn_t gfn_offset, unsigned long mask)
>>> +{
>>> +       gfn_t base_gfn = slot->base_gfn + gfn_offset;
>>> +       gfn_t start = base_gfn +  __ffs(mask);
>>> +       gfn_t end = base_gfn + __fls(mask) + 1;
>>> +
>>> +       kvm_mkclean_gpa_pt(kvm, start, end);
>>> +}
>>> +
>>> +void kvm_arch_commit_memory_region(struct kvm *kvm,
>>> +                                  struct kvm_memory_slot *old,
>>> +                                  const struct kvm_memory_slot *new,
>>> +                                  enum kvm_mr_change change)
>>> +{
>>> +       int needs_flush;
>>> +
>>> +       /*
>>> +        * If dirty page logging is enabled, write protect all pages 
>>> in the slot
>>> +        * ready for dirty logging.
>>> +        *
>>> +        * There is no need to do this in any of the following cases:
>>> +        * CREATE:      No dirty mappings will already exist.
>>> +        * MOVE/DELETE: The old mappings will already have been 
>>> cleaned up by
>>> +        *              kvm_arch_flush_shadow_memslot()
>>> +        */
>>> +       if (change == KVM_MR_FLAGS_ONLY &&
>>> +           (!(old->flags & KVM_MEM_LOG_DIRTY_PAGES) &&
>>> +            new->flags & KVM_MEM_LOG_DIRTY_PAGES)) {
>>> +               spin_lock(&kvm->mmu_lock);
>>> +               /* Write protect GPA page table entries */
>>> +               needs_flush = kvm_mkclean_gpa_pt(kvm, new->base_gfn,
>>> +                                       new->base_gfn + new->npages);
>>> +               if (needs_flush)
>>> +                       kvm_flush_remote_tlbs(kvm);
>>> +               spin_unlock(&kvm->mmu_lock);
>>> +       }
>>> +}
>>> +
>>> +void kvm_arch_flush_shadow_all(struct kvm *kvm)
>>> +{
>>> +       /* Flush whole GPA */
>>> +       level2_flush_range(kvm, 0, kvm->arch.gpa_size >> PAGE_SHIFT);
>>> +       /* Flush vpid for each vCPU individually */
>>> +       kvm_flush_remote_tlbs(kvm);
>>> +}
>>> +
>>> +void kvm_arch_flush_shadow_memslot(struct kvm *kvm,
>>> +               struct kvm_memory_slot *slot)
>>> +{
>>> +       int ret;
>>> +
>>> +       /*
>>> +        * The slot has been made invalid (ready for moving or 
>>> deletion), so we
>>> +        * need to ensure that it can no longer be accessed by any 
>>> guest vCPUs.
>>> +        */
>>> +       spin_lock(&kvm->mmu_lock);
>>> +       /* Flush slot from GPA */
>>> +       ret = level2_flush_range(kvm, slot->base_gfn,
>>> +                       slot->base_gfn + slot->npages);
>>> +       /* Let implementation do the rest */
>>> +       if (ret)
>>> +               kvm_flush_remote_tlbs(kvm);
>>> +       spin_unlock(&kvm->mmu_lock);
>>> +}
>>> +
>>> +void _kvm_destroy_mm(struct kvm *kvm)
>>> +{
>>> +       /* It should always be safe to remove after flushing the 
>>> whole range */
>>> +       level2_flush_range(kvm, 0, kvm->arch.gpa_size >> PAGE_SHIFT);
>>> +       pgd_free(NULL, kvm->arch.gpa_mm.pgd);
>>> +       kvm->arch.gpa_mm.pgd = NULL;
>>> +}
>>> +
>>> +/*
>>> + * Mark a range of guest physical address space old (all accesses 
>>> fault) in the
>>> + * VM's GPA page table to allow detection of commonly used pages.
>>> + */
>>> +static int level2_mkold_pte(void *pte)
>>> +{
>>> +       pte_t val;
>>> +
>>> +       val = *(pte_t *)pte;
>>> +       if (pte_young(val)) {
>>> +               *(pte_t *)pte = pte_mkold(val);
>>> +               return 1;
>>> +       }
>>> +       return 0;
>>> +}
>>> +
>>> +bool kvm_unmap_gfn_range(struct kvm *kvm, struct kvm_gfn_range *range)
>>> +{
>>> +       return level2_flush_range(kvm, range->start, range->end);
>>> +}
>>> +
>>> +bool kvm_set_spte_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
>>> +{
>>> +       gpa_t gpa = range->start << PAGE_SHIFT;
>>> +       pte_t hva_pte = range->pte;
>>> +       pte_t *ptep = kvm_pte_for_gpa(kvm, NULL, gpa);
>>> +       pte_t old_pte;
>>> +
>>> +       if (!ptep)
>>> +               return false;
>>> +
>>> +       /* Mapping may need adjusting depending on memslot flags */
>>> +       old_pte = *ptep;
>>> +       if (range->slot->flags & KVM_MEM_LOG_DIRTY_PAGES && 
>>> !pte_dirty(old_pte))
>>> +               hva_pte = pte_mkclean(hva_pte);
>>> +       else if (range->slot->flags & KVM_MEM_READONLY)
>>> +               hva_pte = pte_wrprotect(hva_pte);
>>> +
>>> +       set_pte(ptep, hva_pte);
>>> +
>>> +       /* Replacing an absent or old page doesn't need flushes */
>>> +       if (!pte_present(old_pte) || !pte_young(old_pte))
>>> +               return false;
>>> +
>>> +       /* Pages swapped, aged, moved, or cleaned require flushes */
>>> +       return !pte_present(hva_pte) ||
>>> +              !pte_young(hva_pte) ||
>>> +              pte_pfn(old_pte) != pte_pfn(hva_pte) ||
>>> +              (pte_dirty(old_pte) && !pte_dirty(hva_pte));
>>> +}
>>> +
>>> +bool kvm_age_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
>>> +{
>>> +       return level2_ptw_pgd(kvm->arch.gpa_mm.pgd, range->start << 
>>> PAGE_SHIFT,
>>> +                               range->end << PAGE_SHIFT, 
>>> level2_mkold_pte);
>>> +}
>>> +
>>> +bool kvm_test_age_gfn(struct kvm *kvm, struct kvm_gfn_range *range)
>>> +{
>>> +       gpa_t gpa = range->start << PAGE_SHIFT;
>>> +       pte_t *ptep = kvm_pte_for_gpa(kvm, NULL, gpa);
>>> +
>>> +       if (ptep && pte_present(*ptep) && pte_young(*ptep))
>>> +               return true;
>>> +
>>> +       return false;
>>> +}
>>> +
>>> +/**
>>> + * kvm_map_page_fast() - Fast path GPA fault handler.
>>> + * @vcpu:              vCPU pointer.
>>> + * @gpa:               Guest physical address of fault.
>>> + * @write:     Whether the fault was due to a write.
>>> + *
>>> + * Perform fast path GPA fault handling, doing all that can be done 
>>> without
>>> + * calling into KVM. This handles marking old pages young (for idle 
>>> page
>>> + * tracking), and dirtying of clean pages (for dirty page logging).
>>> + *
>>> + * Returns:    0 on success, in which case we can update derived 
>>> mappings and
>>> + *             resume guest execution.
>>> + *             -EFAULT on failure due to absent GPA mapping or 
>>> write to
>>> + *             read-only page, in which case KVM must be consulted.
>>> + */
>>> +static int kvm_map_page_fast(struct kvm_vcpu *vcpu, unsigned long gpa,
>>> +                                  bool write)
>>> +{
>>> +       struct kvm *kvm = vcpu->kvm;
>>> +       gfn_t gfn = gpa >> PAGE_SHIFT;
>>> +       pte_t *ptep;
>>> +       kvm_pfn_t pfn = 0;
>>> +       bool pfn_valid = false;
>>> +       int ret = 0;
>>> +
>>> +       spin_lock(&kvm->mmu_lock);
>>> +
>>> +       /* Fast path - just check GPA page table for an existing 
>>> entry */
>>> +       ptep = kvm_pte_for_gpa(kvm, NULL, gpa);
>>> +       if (!ptep || !pte_present(*ptep)) {
>>> +               ret = -EFAULT;
>>> +               goto out;
>>> +       }
>>> +
>>> +       /* Track access to pages marked old */
>>> +       if (!pte_young(*ptep)) {
>>> +               set_pte(ptep, pte_mkyoung(*ptep));
>>> +               pfn = pte_pfn(*ptep);
>>> +               pfn_valid = true;
>>> +               /* call kvm_set_pfn_accessed() after unlock */
>>> +       }
>>> +       if (write && !pte_dirty(*ptep)) {
>>> +               if (!pte_write(*ptep)) {
>>> +                       ret = -EFAULT;
>>> +                       goto out;
>>> +               }
>>> +
>>> +               /* Track dirtying of writeable pages */
>>> +               set_pte(ptep, pte_mkdirty(*ptep));
>>> +               pfn = pte_pfn(*ptep);
>>> +               mark_page_dirty(kvm, gfn);
>>> +               kvm_set_pfn_dirty(pfn);
>>> +       }
>>> +
>>> +out:
>>> +       spin_unlock(&kvm->mmu_lock);
>>> +       if (pfn_valid)
>>> +               kvm_set_pfn_accessed(pfn);
>>> +       return ret;
>>> +}
>>> +
>>> +/**
>>> + * kvm_map_page() - Map a guest physical page.
>>> + * @vcpu:              vCPU pointer.
>>> + * @gpa:               Guest physical address of fault.
>>> + * @write:     Whether the fault was due to a write.
>>> + *
>>> + * Handle GPA faults by creating a new GPA mapping (or updating an 
>>> existing
>>> + * one).
>>> + *
>>> + * This takes care of marking pages young or dirty (idle/dirty page 
>>> tracking),
>>> + * asking KVM for the corresponding PFN, and creating a mapping in 
>>> the GPA page
>>> + * tables. Derived mappings (GVA page tables and TLBs) must be 
>>> handled by the
>>> + * caller.
>>> + *
>>> + * Returns:    0 on success
>>> + *             -EFAULT if there is no memory region at @gpa or a 
>>> write was
>>> + *             attempted to a read-only memory region. This is 
>>> usually handled
>>> + *             as an MMIO access.
>>> + */
>>> +static int kvm_map_page(struct kvm_vcpu *vcpu, unsigned long gpa, 
>>> bool write)
>>> +{
>>> +       bool writeable;
>>> +       int srcu_idx, err = 0, retry_no = 0;
>>> +       unsigned long hva;
>>> +       unsigned long mmu_seq;
>>> +       unsigned long prot_bits;
>>> +       pte_t *ptep, new_pte;
>>> +       kvm_pfn_t pfn;
>>> +       gfn_t gfn = gpa >> PAGE_SHIFT;
>>> +       struct vm_area_struct *vma;
>>> +       struct kvm *kvm = vcpu->kvm;
>>> +       struct kvm_memory_slot *memslot;
>>> +       struct kvm_mmu_memory_cache *memcache = 
>>> &vcpu->arch.mmu_page_cache;
>>> +
>>> +       /* Try the fast path to handle old / clean pages */
>>> +       srcu_idx = srcu_read_lock(&kvm->srcu);
>>> +       err = kvm_map_page_fast(vcpu, gpa, write);
>>> +       if (!err)
>>> +               goto out;
>>> +
>>> +       memslot = gfn_to_memslot(kvm, gfn);
>>> +       hva = gfn_to_hva_memslot_prot(memslot, gfn, &writeable);
>>> +       if (kvm_is_error_hva(hva) || (write && !writeable))
>>> +               goto out;
>>> +
>>> +       mmap_read_lock(current->mm);
>>> +       vma = find_vma_intersection(current->mm, hva, hva + 1);
>>> +       if (unlikely(!vma)) {
>>> +               kvm_err("Failed to find VMA for hva 0x%lx\n", hva);
>>> +               mmap_read_unlock(current->mm);
>>> +               err = -EFAULT;
>>> +               goto out;
>>> +       }
>>> +       mmap_read_unlock(current->mm);
>>> +
>>> +       /* We need a minimum of cached pages ready for page table 
>>> creation */
>>> +       err = kvm_mmu_topup_memory_cache(memcache, 
>>> KVM_MMU_CACHE_MIN_PAGES);
>>> +       if (err)
>>> +               goto out;
>>> +
>>> +retry:
>>> +       /*
>>> +        * Used to check for invalidations in progress, of the pfn 
>>> that is
>>> +        * returned by pfn_to_pfn_prot below.
>>> +        */
>>> +       mmu_seq = kvm->mmu_invalidate_seq;
>>> +       /*
>>> +        * Ensure the read of mmu_invalidate_seq isn't reordered 
>>> with PTE reads in
>>> +        * gfn_to_pfn_prot() (which calls get_user_pages()), so that 
>>> we don't
>>> +        * risk the page we get a reference to getting unmapped 
>>> before we have a
>>> +        * chance to grab the mmu_lock without 
>>> mmu_invalidate_retry() noticing.
>>> +        *
>>> +        * This smp_rmb() pairs with the effective smp_wmb() of the 
>>> combination
>>> +        * of the pte_unmap_unlock() after the PTE is zapped, and the
>>> +        * spin_lock() in 
>>> kvm_mmu_invalidate_invalidate_<page|range_end>() before
>>> +        * mmu_invalidate_seq is incremented.
>>> +        */
>>> +       smp_rmb();
>>> +
>>> +       /* Slow path - ask KVM core whether we can access this GPA */
>>> +       pfn = gfn_to_pfn_prot(kvm, gfn, write, &writeable);
>>> +       if (is_error_noslot_pfn(pfn)) {
>>> +               err = -EFAULT;
>>> +               goto out;
>>> +       }
>>> +
>>> +       spin_lock(&kvm->mmu_lock);
>>> +       /* Check if an invalidation has taken place since we got pfn */
>>> +       if (mmu_invalidate_retry(kvm, mmu_seq)) {
>>> +               /*
>>> +                * This can happen when mappings are changed 
>>> asynchronously, but
>>> +                * also synchronously if a COW is triggered by
>>> +                * gfn_to_pfn_prot().
>>> +                */
>>> +               spin_unlock(&kvm->mmu_lock);
>>> +               kvm_set_pfn_accessed(pfn);
>>> +               kvm_release_pfn_clean(pfn);
>>> +               if (retry_no > 100) {
>>> +                       retry_no = 0;
>>> +                       schedule();
>>> +               }
>>> +               retry_no++;
>>> +               goto retry;
>>> +       }
>>> +
>>> +       /*
>>> +        * For emulated devices such virtio device, actual cache 
>>> attribute is
>>> +        * determined by physical machine.
>>> +        * For pass through physical device, it should be uncachable
>>> +        */
>>> +       prot_bits = _PAGE_PRESENT | __READABLE;
>>> +       if (vma->vm_flags & (VM_IO | VM_PFNMAP))
>>> +               prot_bits |= _CACHE_SUC;
>>> +       else
>>> +               prot_bits |= _CACHE_CC;
>>> +
>>> +       if (writeable) {
>>> +               prot_bits |= _PAGE_WRITE;
>>> +               if (write) {
>>> +                       prot_bits |= __WRITEABLE;
>>> +                       mark_page_dirty(kvm, gfn);
>>> +                       kvm_set_pfn_dirty(pfn);
>>> +               }
>>> +       }
>>> +
>>> +       /* Ensure page tables are allocated */
>>> +       ptep = kvm_pte_for_gpa(kvm, memcache, gpa);
>>> +       new_pte = pfn_pte(pfn, __pgprot(prot_bits));
>>> +       set_pte(ptep, new_pte);
>>> +
>>> +       err = 0;
>>> +       spin_unlock(&kvm->mmu_lock);
>>> +       kvm_release_pfn_clean(pfn);
>>> +       kvm_set_pfn_accessed(pfn);
>>> +out:
>>> +       srcu_read_unlock(&kvm->srcu, srcu_idx);
>>> +       return err;
>>> +}
>>> +
>>> +int kvm_handle_mm_fault(struct kvm_vcpu *vcpu, unsigned long gpa, 
>>> bool write)
>>> +{
>>> +       int ret;
>>> +
>>> +       ret = kvm_map_page(vcpu, gpa, write);
>>> +       if (ret)
>>> +               return ret;
>>> +
>>> +       /* Invalidate this entry in the TLB */
>>> +       return kvm_flush_tlb_gpa(vcpu, gpa);
>>> +}
>>> +
>>> +void kvm_arch_sync_dirty_log(struct kvm *kvm, struct 
>>> kvm_memory_slot *memslot)
>>> +{
>>> +
>>> +}
>>> +
>>> +int kvm_arch_prepare_memory_region(struct kvm *kvm,
>>> +                                  const struct kvm_memory_slot *old,
>>> +                                  struct kvm_memory_slot *new,
>>> +                                  enum kvm_mr_change change)
>>> +{
>>> +       return 0;
>>> +}
>>> +
>>> +void kvm_arch_flush_remote_tlbs_memslot(struct kvm *kvm,
>>> +                                       const struct kvm_memory_slot 
>>> *memslot)
>>> +{
>>> +       kvm_flush_remote_tlbs(kvm);
>>> +}
>>> -- 
>>> 2.39.1
>>>
>>>