From 6de4f3ada40b336522250a7832a0cc4de8856589 Mon Sep 17 00:00:00 2001 From: Avi Kivity Date: Sun, 31 May 2009 22:58:47 +0300 Subject: KVM: Cache pdptrs Instead of reloading the pdptrs on every entry and exit (vmcs writes on vmx, guest memory access on svm) extract them on demand. Signed-off-by: Avi Kivity --- arch/x86/kvm/paging_tmpl.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86/kvm/paging_tmpl.h') diff --git a/arch/x86/kvm/paging_tmpl.h b/arch/x86/kvm/paging_tmpl.h index 67785f635399..4cb1dbfd7c2a 100644 --- a/arch/x86/kvm/paging_tmpl.h +++ b/arch/x86/kvm/paging_tmpl.h @@ -131,7 +131,7 @@ walk: pte = vcpu->arch.cr3; #if PTTYPE == 64 if (!is_long_mode(vcpu)) { - pte = vcpu->arch.pdptrs[(addr >> 30) & 3]; + pte = kvm_pdptr_read(vcpu, (addr >> 30) & 3); if (!is_present_pte(pte)) goto not_present; --walker->level; -- cgit v1.2.3 From 43a3795a3a12425de31e25ce0ebc3bb41501cef7 Mon Sep 17 00:00:00 2001 From: Avi Kivity Date: Wed, 10 Jun 2009 14:12:05 +0300 Subject: KVM: MMU: Adjust pte accessors to explicitly indicate guest or shadow pte Since the guest and host ptes can have wildly different format, adjust the pte accessor names to indicate on which type of pte they operate on. No functional changes. Signed-off-by: Avi Kivity --- arch/x86/kvm/mmu.c | 16 ++++++++-------- arch/x86/kvm/mmu.h | 2 +- arch/x86/kvm/paging_tmpl.h | 22 +++++++++++----------- arch/x86/kvm/x86.c | 2 +- 4 files changed, 21 insertions(+), 21 deletions(-) (limited to 'arch/x86/kvm/paging_tmpl.h') diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c index 8f2cb29db2fe..a039e6bc21f7 100644 --- a/arch/x86/kvm/mmu.c +++ b/arch/x86/kvm/mmu.c @@ -240,12 +240,12 @@ static int is_writeble_pte(unsigned long pte) return pte & PT_WRITABLE_MASK; } -static int is_dirty_pte(unsigned long pte) +static int is_dirty_gpte(unsigned long pte) { return pte & PT_DIRTY_MASK; } -static int is_rmap_pte(u64 pte) +static int is_rmap_spte(u64 pte) { return is_shadow_present_pte(pte); } @@ -502,7 +502,7 @@ static int rmap_add(struct kvm_vcpu *vcpu, u64 *spte, gfn_t gfn, int lpage) unsigned long *rmapp; int i, count = 0; - if (!is_rmap_pte(*spte)) + if (!is_rmap_spte(*spte)) return count; gfn = unalias_gfn(vcpu->kvm, gfn); sp = page_header(__pa(spte)); @@ -567,7 +567,7 @@ static void rmap_remove(struct kvm *kvm, u64 *spte) unsigned long *rmapp; int i; - if (!is_rmap_pte(*spte)) + if (!is_rmap_spte(*spte)) return; sp = page_header(__pa(spte)); pfn = spte_to_pfn(*spte); @@ -1769,7 +1769,7 @@ static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *shadow_pte, __func__, *shadow_pte, pt_access, write_fault, user_fault, gfn); - if (is_rmap_pte(*shadow_pte)) { + if (is_rmap_spte(*shadow_pte)) { /* * If we overwrite a PTE page pointer with a 2MB PMD, unlink * the parent of the now unreachable PTE. @@ -1805,7 +1805,7 @@ static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *shadow_pte, page_header_update_slot(vcpu->kvm, shadow_pte, gfn); if (!was_rmapped) { rmap_count = rmap_add(vcpu, shadow_pte, gfn, largepage); - if (!is_rmap_pte(*shadow_pte)) + if (!is_rmap_spte(*shadow_pte)) kvm_release_pfn_clean(pfn); if (rmap_count > RMAP_RECYCLE_THRESHOLD) rmap_recycle(vcpu, gfn, largepage); @@ -1984,7 +1984,7 @@ static int mmu_alloc_roots(struct kvm_vcpu *vcpu) ASSERT(!VALID_PAGE(root)); if (vcpu->arch.mmu.root_level == PT32E_ROOT_LEVEL) { pdptr = kvm_pdptr_read(vcpu, i); - if (!is_present_pte(pdptr)) { + if (!is_present_gpte(pdptr)) { vcpu->arch.mmu.pae_root[i] = 0; continue; } @@ -2475,7 +2475,7 @@ static void mmu_guess_page_from_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa, if ((bytes == 4) && (gpa % 4 == 0)) memcpy((void *)&gpte, new, 4); } - if (!is_present_pte(gpte)) + if (!is_present_gpte(gpte)) return; gfn = (gpte & PT64_BASE_ADDR_MASK) >> PAGE_SHIFT; diff --git a/arch/x86/kvm/mmu.h b/arch/x86/kvm/mmu.h index 3494a2fb136e..016bf7183e9f 100644 --- a/arch/x86/kvm/mmu.h +++ b/arch/x86/kvm/mmu.h @@ -75,7 +75,7 @@ static inline int is_paging(struct kvm_vcpu *vcpu) return vcpu->arch.cr0 & X86_CR0_PG; } -static inline int is_present_pte(unsigned long pte) +static inline int is_present_gpte(unsigned long pte) { return pte & PT_PRESENT_MASK; } diff --git a/arch/x86/kvm/paging_tmpl.h b/arch/x86/kvm/paging_tmpl.h index 4cb1dbfd7c2a..238a193bbf5b 100644 --- a/arch/x86/kvm/paging_tmpl.h +++ b/arch/x86/kvm/paging_tmpl.h @@ -132,7 +132,7 @@ walk: #if PTTYPE == 64 if (!is_long_mode(vcpu)) { pte = kvm_pdptr_read(vcpu, (addr >> 30) & 3); - if (!is_present_pte(pte)) + if (!is_present_gpte(pte)) goto not_present; --walker->level; } @@ -155,7 +155,7 @@ walk: kvm_read_guest(vcpu->kvm, pte_gpa, &pte, sizeof(pte)); - if (!is_present_pte(pte)) + if (!is_present_gpte(pte)) goto not_present; rsvd_fault = is_rsvd_bits_set(vcpu, pte, walker->level); @@ -205,7 +205,7 @@ walk: --walker->level; } - if (write_fault && !is_dirty_pte(pte)) { + if (write_fault && !is_dirty_gpte(pte)) { bool ret; mark_page_dirty(vcpu->kvm, table_gfn); @@ -252,7 +252,7 @@ static void FNAME(update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *page, gpte = *(const pt_element_t *)pte; if (~gpte & (PT_PRESENT_MASK | PT_ACCESSED_MASK)) { - if (!is_present_pte(gpte)) + if (!is_present_gpte(gpte)) set_shadow_pte(spte, shadow_notrap_nonpresent_pte); return; } @@ -289,7 +289,7 @@ static u64 *FNAME(fetch)(struct kvm_vcpu *vcpu, gva_t addr, pt_element_t curr_pte; struct kvm_shadow_walk_iterator iterator; - if (!is_present_pte(gw->ptes[gw->level - 1])) + if (!is_present_gpte(gw->ptes[gw->level - 1])) return NULL; for_each_shadow_entry(vcpu, addr, iterator) { @@ -318,7 +318,7 @@ static u64 *FNAME(fetch)(struct kvm_vcpu *vcpu, gva_t addr, if (level == PT_DIRECTORY_LEVEL && gw->level == PT_DIRECTORY_LEVEL) { direct = 1; - if (!is_dirty_pte(gw->ptes[level - 1])) + if (!is_dirty_gpte(gw->ptes[level - 1])) access &= ~ACC_WRITE_MASK; table_gfn = gpte_to_gfn(gw->ptes[level - 1]); } else { @@ -489,7 +489,7 @@ static void FNAME(invlpg)(struct kvm_vcpu *vcpu, gva_t gva) if (kvm_read_guest_atomic(vcpu->kvm, pte_gpa, &gpte, sizeof(pt_element_t))) return; - if (is_present_pte(gpte) && (gpte & PT_ACCESSED_MASK)) { + if (is_present_gpte(gpte) && (gpte & PT_ACCESSED_MASK)) { if (mmu_topup_memory_caches(vcpu)) return; kvm_mmu_pte_write(vcpu, pte_gpa, (const u8 *)&gpte, @@ -536,7 +536,7 @@ static void FNAME(prefetch_page)(struct kvm_vcpu *vcpu, r = kvm_read_guest_atomic(vcpu->kvm, pte_gpa, pt, sizeof pt); pte_gpa += ARRAY_SIZE(pt) * sizeof(pt_element_t); for (j = 0; j < ARRAY_SIZE(pt); ++j) - if (r || is_present_pte(pt[j])) + if (r || is_present_gpte(pt[j])) sp->spt[i+j] = shadow_trap_nonpresent_pte; else sp->spt[i+j] = shadow_notrap_nonpresent_pte; @@ -574,12 +574,12 @@ static int FNAME(sync_page)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp) sizeof(pt_element_t))) return -EINVAL; - if (gpte_to_gfn(gpte) != gfn || !is_present_pte(gpte) || + if (gpte_to_gfn(gpte) != gfn || !is_present_gpte(gpte) || !(gpte & PT_ACCESSED_MASK)) { u64 nonpresent; rmap_remove(vcpu->kvm, &sp->spt[i]); - if (is_present_pte(gpte)) + if (is_present_gpte(gpte)) nonpresent = shadow_trap_nonpresent_pte; else nonpresent = shadow_notrap_nonpresent_pte; @@ -590,7 +590,7 @@ static int FNAME(sync_page)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp) nr_present++; pte_access = sp->role.access & FNAME(gpte_access)(vcpu, gpte); set_spte(vcpu, &sp->spt[i], pte_access, 0, 0, - is_dirty_pte(gpte), 0, gfn, + is_dirty_gpte(gpte), 0, gfn, spte_to_pfn(sp->spt[i]), true, false); } diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c index 05cbe83c74e2..e877efa37620 100644 --- a/arch/x86/kvm/x86.c +++ b/arch/x86/kvm/x86.c @@ -237,7 +237,7 @@ int load_pdptrs(struct kvm_vcpu *vcpu, unsigned long cr3) goto out; } for (i = 0; i < ARRAY_SIZE(pdpte); ++i) { - if (is_present_pte(pdpte[i]) && + if (is_present_gpte(pdpte[i]) && (pdpte[i] & vcpu->arch.mmu.rsvd_bits_mask[0][2])) { ret = 0; goto out; -- cgit v1.2.3 From d555c333aa544b222fe077adcd5dfea024b2c913 Mon Sep 17 00:00:00 2001 From: Avi Kivity Date: Wed, 10 Jun 2009 14:24:23 +0300 Subject: KVM: MMU: s/shadow_pte/spte/ We use shadow_pte and spte inconsistently, switch to the shorter spelling. Rename set_shadow_pte() to __set_spte() to avoid a conflict with the existing set_spte(), and to indicate its lowlevelness. Signed-off-by: Avi Kivity --- arch/x86/kvm/mmu.c | 102 ++++++++++++++++++++++----------------------- arch/x86/kvm/paging_tmpl.h | 16 +++---- 2 files changed, 59 insertions(+), 59 deletions(-) (limited to 'arch/x86/kvm/paging_tmpl.h') diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c index a039e6bc21f7..d443a421ca3e 100644 --- a/arch/x86/kvm/mmu.c +++ b/arch/x86/kvm/mmu.c @@ -143,7 +143,7 @@ module_param(oos_shadow, bool, 0644); #define SHADOW_PT_INDEX(addr, level) PT64_INDEX(addr, level) struct kvm_rmap_desc { - u64 *shadow_ptes[RMAP_EXT]; + u64 *sptes[RMAP_EXT]; struct kvm_rmap_desc *more; }; @@ -262,7 +262,7 @@ static gfn_t pse36_gfn_delta(u32 gpte) return (gpte & PT32_DIR_PSE36_MASK) << shift; } -static void set_shadow_pte(u64 *sptep, u64 spte) +static void __set_spte(u64 *sptep, u64 spte) { #ifdef CONFIG_X86_64 set_64bit((unsigned long *)sptep, spte); @@ -514,23 +514,23 @@ static int rmap_add(struct kvm_vcpu *vcpu, u64 *spte, gfn_t gfn, int lpage) } else if (!(*rmapp & 1)) { rmap_printk("rmap_add: %p %llx 1->many\n", spte, *spte); desc = mmu_alloc_rmap_desc(vcpu); - desc->shadow_ptes[0] = (u64 *)*rmapp; - desc->shadow_ptes[1] = spte; + desc->sptes[0] = (u64 *)*rmapp; + desc->sptes[1] = spte; *rmapp = (unsigned long)desc | 1; } else { rmap_printk("rmap_add: %p %llx many->many\n", spte, *spte); desc = (struct kvm_rmap_desc *)(*rmapp & ~1ul); - while (desc->shadow_ptes[RMAP_EXT-1] && desc->more) { + while (desc->sptes[RMAP_EXT-1] && desc->more) { desc = desc->more; count += RMAP_EXT; } - if (desc->shadow_ptes[RMAP_EXT-1]) { + if (desc->sptes[RMAP_EXT-1]) { desc->more = mmu_alloc_rmap_desc(vcpu); desc = desc->more; } - for (i = 0; desc->shadow_ptes[i]; ++i) + for (i = 0; desc->sptes[i]; ++i) ; - desc->shadow_ptes[i] = spte; + desc->sptes[i] = spte; } return count; } @@ -542,14 +542,14 @@ static void rmap_desc_remove_entry(unsigned long *rmapp, { int j; - for (j = RMAP_EXT - 1; !desc->shadow_ptes[j] && j > i; --j) + for (j = RMAP_EXT - 1; !desc->sptes[j] && j > i; --j) ; - desc->shadow_ptes[i] = desc->shadow_ptes[j]; - desc->shadow_ptes[j] = NULL; + desc->sptes[i] = desc->sptes[j]; + desc->sptes[j] = NULL; if (j != 0) return; if (!prev_desc && !desc->more) - *rmapp = (unsigned long)desc->shadow_ptes[0]; + *rmapp = (unsigned long)desc->sptes[0]; else if (prev_desc) prev_desc->more = desc->more; @@ -594,8 +594,8 @@ static void rmap_remove(struct kvm *kvm, u64 *spte) desc = (struct kvm_rmap_desc *)(*rmapp & ~1ul); prev_desc = NULL; while (desc) { - for (i = 0; i < RMAP_EXT && desc->shadow_ptes[i]; ++i) - if (desc->shadow_ptes[i] == spte) { + for (i = 0; i < RMAP_EXT && desc->sptes[i]; ++i) + if (desc->sptes[i] == spte) { rmap_desc_remove_entry(rmapp, desc, i, prev_desc); @@ -626,10 +626,10 @@ static u64 *rmap_next(struct kvm *kvm, unsigned long *rmapp, u64 *spte) prev_desc = NULL; prev_spte = NULL; while (desc) { - for (i = 0; i < RMAP_EXT && desc->shadow_ptes[i]; ++i) { + for (i = 0; i < RMAP_EXT && desc->sptes[i]; ++i) { if (prev_spte == spte) - return desc->shadow_ptes[i]; - prev_spte = desc->shadow_ptes[i]; + return desc->sptes[i]; + prev_spte = desc->sptes[i]; } desc = desc->more; } @@ -651,7 +651,7 @@ static int rmap_write_protect(struct kvm *kvm, u64 gfn) BUG_ON(!(*spte & PT_PRESENT_MASK)); rmap_printk("rmap_write_protect: spte %p %llx\n", spte, *spte); if (is_writeble_pte(*spte)) { - set_shadow_pte(spte, *spte & ~PT_WRITABLE_MASK); + __set_spte(spte, *spte & ~PT_WRITABLE_MASK); write_protected = 1; } spte = rmap_next(kvm, rmapp, spte); @@ -675,7 +675,7 @@ static int rmap_write_protect(struct kvm *kvm, u64 gfn) if (is_writeble_pte(*spte)) { rmap_remove(kvm, spte); --kvm->stat.lpages; - set_shadow_pte(spte, shadow_trap_nonpresent_pte); + __set_spte(spte, shadow_trap_nonpresent_pte); spte = NULL; write_protected = 1; } @@ -694,7 +694,7 @@ static int kvm_unmap_rmapp(struct kvm *kvm, unsigned long *rmapp) BUG_ON(!(*spte & PT_PRESENT_MASK)); rmap_printk("kvm_rmap_unmap_hva: spte %p %llx\n", spte, *spte); rmap_remove(kvm, spte); - set_shadow_pte(spte, shadow_trap_nonpresent_pte); + __set_spte(spte, shadow_trap_nonpresent_pte); need_tlb_flush = 1; } return need_tlb_flush; @@ -1369,7 +1369,7 @@ static void kvm_mmu_unlink_parents(struct kvm *kvm, struct kvm_mmu_page *sp) } BUG_ON(!parent_pte); kvm_mmu_put_page(sp, parent_pte); - set_shadow_pte(parent_pte, shadow_trap_nonpresent_pte); + __set_spte(parent_pte, shadow_trap_nonpresent_pte); } } @@ -1517,7 +1517,7 @@ static void mmu_convert_notrap(struct kvm_mmu_page *sp) for (i = 0; i < PT64_ENT_PER_PAGE; ++i) { if (pt[i] == shadow_notrap_nonpresent_pte) - set_shadow_pte(&pt[i], shadow_trap_nonpresent_pte); + __set_spte(&pt[i], shadow_trap_nonpresent_pte); } } @@ -1683,7 +1683,7 @@ static int mmu_need_write_protect(struct kvm_vcpu *vcpu, gfn_t gfn, return 0; } -static int set_spte(struct kvm_vcpu *vcpu, u64 *shadow_pte, +static int set_spte(struct kvm_vcpu *vcpu, u64 *sptep, unsigned pte_access, int user_fault, int write_fault, int dirty, int largepage, gfn_t gfn, pfn_t pfn, bool speculative, @@ -1733,7 +1733,7 @@ static int set_spte(struct kvm_vcpu *vcpu, u64 *shadow_pte, * is responsibility of mmu_get_page / kvm_sync_page. * Same reasoning can be applied to dirty page accounting. */ - if (!can_unsync && is_writeble_pte(*shadow_pte)) + if (!can_unsync && is_writeble_pte(*sptep)) goto set_pte; if (mmu_need_write_protect(vcpu, gfn, can_unsync)) { @@ -1750,62 +1750,62 @@ static int set_spte(struct kvm_vcpu *vcpu, u64 *shadow_pte, mark_page_dirty(vcpu->kvm, gfn); set_pte: - set_shadow_pte(shadow_pte, spte); + __set_spte(sptep, spte); return ret; } -static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *shadow_pte, +static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *sptep, unsigned pt_access, unsigned pte_access, int user_fault, int write_fault, int dirty, int *ptwrite, int largepage, gfn_t gfn, pfn_t pfn, bool speculative) { int was_rmapped = 0; - int was_writeble = is_writeble_pte(*shadow_pte); + int was_writeble = is_writeble_pte(*sptep); int rmap_count; pgprintk("%s: spte %llx access %x write_fault %d" " user_fault %d gfn %lx\n", - __func__, *shadow_pte, pt_access, + __func__, *sptep, pt_access, write_fault, user_fault, gfn); - if (is_rmap_spte(*shadow_pte)) { + if (is_rmap_spte(*sptep)) { /* * If we overwrite a PTE page pointer with a 2MB PMD, unlink * the parent of the now unreachable PTE. */ - if (largepage && !is_large_pte(*shadow_pte)) { + if (largepage && !is_large_pte(*sptep)) { struct kvm_mmu_page *child; - u64 pte = *shadow_pte; + u64 pte = *sptep; child = page_header(pte & PT64_BASE_ADDR_MASK); - mmu_page_remove_parent_pte(child, shadow_pte); - } else if (pfn != spte_to_pfn(*shadow_pte)) { + mmu_page_remove_parent_pte(child, sptep); + } else if (pfn != spte_to_pfn(*sptep)) { pgprintk("hfn old %lx new %lx\n", - spte_to_pfn(*shadow_pte), pfn); - rmap_remove(vcpu->kvm, shadow_pte); + spte_to_pfn(*sptep), pfn); + rmap_remove(vcpu->kvm, sptep); } else was_rmapped = 1; } - if (set_spte(vcpu, shadow_pte, pte_access, user_fault, write_fault, + if (set_spte(vcpu, sptep, pte_access, user_fault, write_fault, dirty, largepage, gfn, pfn, speculative, true)) { if (write_fault) *ptwrite = 1; kvm_x86_ops->tlb_flush(vcpu); } - pgprintk("%s: setting spte %llx\n", __func__, *shadow_pte); + pgprintk("%s: setting spte %llx\n", __func__, *sptep); pgprintk("instantiating %s PTE (%s) at %ld (%llx) addr %p\n", - is_large_pte(*shadow_pte)? "2MB" : "4kB", - is_present_pte(*shadow_pte)?"RW":"R", gfn, - *shadow_pte, shadow_pte); - if (!was_rmapped && is_large_pte(*shadow_pte)) + is_large_pte(*sptep)? "2MB" : "4kB", + is_present_pte(*sptep)?"RW":"R", gfn, + *shadow_pte, sptep); + if (!was_rmapped && is_large_pte(*sptep)) ++vcpu->kvm->stat.lpages; - page_header_update_slot(vcpu->kvm, shadow_pte, gfn); + page_header_update_slot(vcpu->kvm, sptep, gfn); if (!was_rmapped) { - rmap_count = rmap_add(vcpu, shadow_pte, gfn, largepage); - if (!is_rmap_spte(*shadow_pte)) + rmap_count = rmap_add(vcpu, sptep, gfn, largepage); + if (!is_rmap_spte(*sptep)) kvm_release_pfn_clean(pfn); if (rmap_count > RMAP_RECYCLE_THRESHOLD) rmap_recycle(vcpu, gfn, largepage); @@ -1816,7 +1816,7 @@ static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *shadow_pte, kvm_release_pfn_clean(pfn); } if (speculative) { - vcpu->arch.last_pte_updated = shadow_pte; + vcpu->arch.last_pte_updated = sptep; vcpu->arch.last_pte_gfn = gfn; } } @@ -1854,10 +1854,10 @@ static int __direct_map(struct kvm_vcpu *vcpu, gpa_t v, int write, return -ENOMEM; } - set_shadow_pte(iterator.sptep, - __pa(sp->spt) - | PT_PRESENT_MASK | PT_WRITABLE_MASK - | shadow_user_mask | shadow_x_mask); + __set_spte(iterator.sptep, + __pa(sp->spt) + | PT_PRESENT_MASK | PT_WRITABLE_MASK + | shadow_user_mask | shadow_x_mask); } } return pt_write; @@ -2389,7 +2389,7 @@ static void mmu_pte_write_zap_pte(struct kvm_vcpu *vcpu, mmu_page_remove_parent_pte(child, spte); } } - set_shadow_pte(spte, shadow_trap_nonpresent_pte); + __set_spte(spte, shadow_trap_nonpresent_pte); if (is_large_pte(pte)) --vcpu->kvm->stat.lpages; } @@ -3125,7 +3125,7 @@ static int count_rmaps(struct kvm_vcpu *vcpu) d = (struct kvm_rmap_desc *)(*rmapp & ~1ul); while (d) { for (k = 0; k < RMAP_EXT; ++k) - if (d->shadow_ptes[k]) + if (d->sptes[k]) ++nmaps; else break; diff --git a/arch/x86/kvm/paging_tmpl.h b/arch/x86/kvm/paging_tmpl.h index 238a193bbf5b..322e8113aeea 100644 --- a/arch/x86/kvm/paging_tmpl.h +++ b/arch/x86/kvm/paging_tmpl.h @@ -253,7 +253,7 @@ static void FNAME(update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *page, gpte = *(const pt_element_t *)pte; if (~gpte & (PT_PRESENT_MASK | PT_ACCESSED_MASK)) { if (!is_present_gpte(gpte)) - set_shadow_pte(spte, shadow_notrap_nonpresent_pte); + __set_spte(spte, shadow_notrap_nonpresent_pte); return; } pgprintk("%s: gpte %llx spte %p\n", __func__, (u64)gpte, spte); @@ -311,7 +311,7 @@ static u64 *FNAME(fetch)(struct kvm_vcpu *vcpu, gva_t addr, if (is_large_pte(*sptep)) { rmap_remove(vcpu->kvm, sptep); - set_shadow_pte(sptep, shadow_trap_nonpresent_pte); + __set_spte(sptep, shadow_trap_nonpresent_pte); kvm_flush_remote_tlbs(vcpu->kvm); } @@ -369,7 +369,7 @@ static int FNAME(page_fault)(struct kvm_vcpu *vcpu, gva_t addr, int user_fault = error_code & PFERR_USER_MASK; int fetch_fault = error_code & PFERR_FETCH_MASK; struct guest_walker walker; - u64 *shadow_pte; + u64 *sptep; int write_pt = 0; int r; pfn_t pfn; @@ -422,11 +422,11 @@ static int FNAME(page_fault)(struct kvm_vcpu *vcpu, gva_t addr, if (mmu_notifier_retry(vcpu, mmu_seq)) goto out_unlock; kvm_mmu_free_some_pages(vcpu); - shadow_pte = FNAME(fetch)(vcpu, addr, &walker, user_fault, write_fault, - largepage, &write_pt, pfn); + sptep = FNAME(fetch)(vcpu, addr, &walker, user_fault, write_fault, + largepage, &write_pt, pfn); pgprintk("%s: shadow pte %p %llx ptwrite %d\n", __func__, - shadow_pte, *shadow_pte, write_pt); + sptep, *sptep, write_pt); if (!write_pt) vcpu->arch.last_pt_write_count = 0; /* reset fork detector */ @@ -472,7 +472,7 @@ static void FNAME(invlpg)(struct kvm_vcpu *vcpu, gva_t gva) --vcpu->kvm->stat.lpages; need_flush = 1; } - set_shadow_pte(sptep, shadow_trap_nonpresent_pte); + __set_spte(sptep, shadow_trap_nonpresent_pte); break; } @@ -583,7 +583,7 @@ static int FNAME(sync_page)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp) nonpresent = shadow_trap_nonpresent_pte; else nonpresent = shadow_notrap_nonpresent_pte; - set_shadow_pte(&sp->spt[i], nonpresent); + __set_spte(&sp->spt[i], nonpresent); continue; } -- cgit v1.2.3 From ec04b2604c3707a46db1d26d98f82b11d0844669 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Fri, 19 Jun 2009 15:16:23 +0200 Subject: KVM: Prepare memslot data structures for multiple hugepage sizes [avi: fix build on non-x86] Signed-off-by: Joerg Roedel Signed-off-by: Avi Kivity --- arch/ia64/include/asm/kvm_host.h | 3 +- arch/powerpc/include/asm/kvm_host.h | 3 +- arch/s390/include/asm/kvm_host.h | 6 +++- arch/x86/include/asm/kvm_host.h | 12 ++++---- arch/x86/kvm/mmu.c | 30 ++++++++++---------- arch/x86/kvm/paging_tmpl.h | 3 +- include/linux/kvm_host.h | 2 +- virt/kvm/kvm_main.c | 56 ++++++++++++++++++++++++++----------- 8 files changed, 73 insertions(+), 42 deletions(-) (limited to 'arch/x86/kvm/paging_tmpl.h') diff --git a/arch/ia64/include/asm/kvm_host.h b/arch/ia64/include/asm/kvm_host.h index 9cf1c4b1f92f..d9b6325a9328 100644 --- a/arch/ia64/include/asm/kvm_host.h +++ b/arch/ia64/include/asm/kvm_host.h @@ -235,7 +235,8 @@ struct kvm_vm_data { #define KVM_REQ_PTC_G 32 #define KVM_REQ_RESUME 33 -#define KVM_PAGES_PER_HPAGE 1 +#define KVM_NR_PAGE_SIZES 1 +#define KVM_PAGES_PER_HPAGE(x) 1 struct kvm; struct kvm_vcpu; diff --git a/arch/powerpc/include/asm/kvm_host.h b/arch/powerpc/include/asm/kvm_host.h index d4caa6127f55..c9c930ed11d7 100644 --- a/arch/powerpc/include/asm/kvm_host.h +++ b/arch/powerpc/include/asm/kvm_host.h @@ -34,7 +34,8 @@ #define KVM_COALESCED_MMIO_PAGE_OFFSET 1 /* We don't currently support large pages. */ -#define KVM_PAGES_PER_HPAGE (1UL << 31) +#define KVM_NR_PAGE_SIZES 1 +#define KVM_PAGES_PER_HPAGE(x) (1UL<<31) struct kvm; struct kvm_run; diff --git a/arch/s390/include/asm/kvm_host.h b/arch/s390/include/asm/kvm_host.h index 75535d4d7a05..78e07a622b45 100644 --- a/arch/s390/include/asm/kvm_host.h +++ b/arch/s390/include/asm/kvm_host.h @@ -40,7 +40,11 @@ struct sca_block { struct sca_entry cpu[64]; } __attribute__((packed)); -#define KVM_PAGES_PER_HPAGE 256 +#define KVM_NR_PAGE_SIZES 2 +#define KVM_HPAGE_SHIFT(x) (PAGE_SHIFT + ((x) - 1) * 8) +#define KVM_HPAGE_SIZE(x) (1UL << KVM_HPAGE_SHIFT(x)) +#define KVM_HPAGE_MASK(x) (~(KVM_HPAGE_SIZE(x) - 1)) +#define KVM_PAGES_PER_HPAGE(x) (KVM_HPAGE_SIZE(x) / PAGE_SIZE) #define CPUSTAT_HOST 0x80000000 #define CPUSTAT_WAIT 0x10000000 diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h index 19027ab20412..30b625d8e5f0 100644 --- a/arch/x86/include/asm/kvm_host.h +++ b/arch/x86/include/asm/kvm_host.h @@ -54,12 +54,12 @@ #define INVALID_PAGE (~(hpa_t)0) #define UNMAPPED_GVA (~(gpa_t)0) -/* shadow tables are PAE even on non-PAE hosts */ -#define KVM_HPAGE_SHIFT 21 -#define KVM_HPAGE_SIZE (1UL << KVM_HPAGE_SHIFT) -#define KVM_HPAGE_MASK (~(KVM_HPAGE_SIZE - 1)) - -#define KVM_PAGES_PER_HPAGE (KVM_HPAGE_SIZE / PAGE_SIZE) +/* KVM Hugepage definitions for x86 */ +#define KVM_NR_PAGE_SIZES 2 +#define KVM_HPAGE_SHIFT(x) (PAGE_SHIFT + (((x) - 1) * 9)) +#define KVM_HPAGE_SIZE(x) (1UL << KVM_HPAGE_SHIFT(x)) +#define KVM_HPAGE_MASK(x) (~(KVM_HPAGE_SIZE(x) - 1)) +#define KVM_PAGES_PER_HPAGE(x) (KVM_HPAGE_SIZE(x) / PAGE_SIZE) #define DE_VECTOR 0 #define DB_VECTOR 1 diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c index 12974de88aa5..b67585c1ef08 100644 --- a/arch/x86/kvm/mmu.c +++ b/arch/x86/kvm/mmu.c @@ -394,9 +394,9 @@ static int *slot_largepage_idx(gfn_t gfn, struct kvm_memory_slot *slot) { unsigned long idx; - idx = (gfn / KVM_PAGES_PER_HPAGE) - - (slot->base_gfn / KVM_PAGES_PER_HPAGE); - return &slot->lpage_info[idx].write_count; + idx = (gfn / KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL)) - + (slot->base_gfn / KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL)); + return &slot->lpage_info[0][idx].write_count; } static void account_shadowed(struct kvm *kvm, gfn_t gfn) @@ -485,10 +485,10 @@ static unsigned long *gfn_to_rmap(struct kvm *kvm, gfn_t gfn, int lpage) if (!lpage) return &slot->rmap[gfn - slot->base_gfn]; - idx = (gfn / KVM_PAGES_PER_HPAGE) - - (slot->base_gfn / KVM_PAGES_PER_HPAGE); + idx = (gfn / KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL)) - + (slot->base_gfn / KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL)); - return &slot->lpage_info[idx].rmap_pde; + return &slot->lpage_info[0][idx].rmap_pde; } /* @@ -731,11 +731,11 @@ static int kvm_handle_hva(struct kvm *kvm, unsigned long hva, end = start + (memslot->npages << PAGE_SHIFT); if (hva >= start && hva < end) { gfn_t gfn_offset = (hva - start) >> PAGE_SHIFT; + int idx = gfn_offset / + KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL); retval |= handler(kvm, &memslot->rmap[gfn_offset]); retval |= handler(kvm, - &memslot->lpage_info[ - gfn_offset / - KVM_PAGES_PER_HPAGE].rmap_pde); + &memslot->lpage_info[0][idx].rmap_pde); } } @@ -1876,8 +1876,9 @@ static int nonpaging_map(struct kvm_vcpu *vcpu, gva_t v, int write, gfn_t gfn) pfn_t pfn; unsigned long mmu_seq; - if (is_largepage_backed(vcpu, gfn & ~(KVM_PAGES_PER_HPAGE-1))) { - gfn &= ~(KVM_PAGES_PER_HPAGE-1); + if (is_largepage_backed(vcpu, gfn & + ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1))) { + gfn &= ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1); largepage = 1; } @@ -2082,8 +2083,9 @@ static int tdp_page_fault(struct kvm_vcpu *vcpu, gva_t gpa, if (r) return r; - if (is_largepage_backed(vcpu, gfn & ~(KVM_PAGES_PER_HPAGE-1))) { - gfn &= ~(KVM_PAGES_PER_HPAGE-1); + if (is_largepage_backed(vcpu, gfn & + ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1))) { + gfn &= ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1); largepage = 1; } mmu_seq = vcpu->kvm->mmu_notifier_seq; @@ -2485,7 +2487,7 @@ static void mmu_guess_page_from_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa, gfn = (gpte & PT64_BASE_ADDR_MASK) >> PAGE_SHIFT; if (is_large_pte(gpte) && is_largepage_backed(vcpu, gfn)) { - gfn &= ~(KVM_PAGES_PER_HPAGE-1); + gfn &= ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1); vcpu->arch.update_pte.largepage = 1; } vcpu->arch.update_pte.mmu_seq = vcpu->kvm->mmu_notifier_seq; diff --git a/arch/x86/kvm/paging_tmpl.h b/arch/x86/kvm/paging_tmpl.h index 322e8113aeea..53e129cec5fd 100644 --- a/arch/x86/kvm/paging_tmpl.h +++ b/arch/x86/kvm/paging_tmpl.h @@ -401,7 +401,8 @@ static int FNAME(page_fault)(struct kvm_vcpu *vcpu, gva_t addr, if (walker.level == PT_DIRECTORY_LEVEL) { gfn_t large_gfn; - large_gfn = walker.gfn & ~(KVM_PAGES_PER_HPAGE-1); + large_gfn = walker.gfn & + ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1); if (is_largepage_backed(vcpu, large_gfn)) { walker.gfn = large_gfn; largepage = 1; diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h index 6988858dc56e..06af936a250a 100644 --- a/include/linux/kvm_host.h +++ b/include/linux/kvm_host.h @@ -103,7 +103,7 @@ struct kvm_memory_slot { struct { unsigned long rmap_pde; int write_count; - } *lpage_info; + } *lpage_info[KVM_NR_PAGE_SIZES - 1]; unsigned long userspace_addr; int user_alloc; }; diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c index 1da8072d61b1..8361662e7e0a 100644 --- a/virt/kvm/kvm_main.c +++ b/virt/kvm/kvm_main.c @@ -1001,19 +1001,25 @@ out: static void kvm_free_physmem_slot(struct kvm_memory_slot *free, struct kvm_memory_slot *dont) { + int i; + if (!dont || free->rmap != dont->rmap) vfree(free->rmap); if (!dont || free->dirty_bitmap != dont->dirty_bitmap) vfree(free->dirty_bitmap); - if (!dont || free->lpage_info != dont->lpage_info) - vfree(free->lpage_info); + + for (i = 0; i < KVM_NR_PAGE_SIZES - 1; ++i) { + if (!dont || free->lpage_info[i] != dont->lpage_info[i]) { + vfree(free->lpage_info[i]); + free->lpage_info[i] = NULL; + } + } free->npages = 0; free->dirty_bitmap = NULL; free->rmap = NULL; - free->lpage_info = NULL; } void kvm_free_physmem(struct kvm *kvm) @@ -1087,7 +1093,8 @@ int __kvm_set_memory_region(struct kvm *kvm, int r; gfn_t base_gfn; unsigned long npages, ugfn; - unsigned long largepages, i; + int lpages; + unsigned long i, j; struct kvm_memory_slot *memslot; struct kvm_memory_slot old, new; @@ -1161,33 +1168,48 @@ int __kvm_set_memory_region(struct kvm *kvm, else new.userspace_addr = 0; } - if (npages && !new.lpage_info) { - largepages = 1 + (base_gfn + npages - 1) / KVM_PAGES_PER_HPAGE; - largepages -= base_gfn / KVM_PAGES_PER_HPAGE; + if (!npages) + goto skip_lpage; - new.lpage_info = vmalloc(largepages * sizeof(*new.lpage_info)); + for (i = 0; i < KVM_NR_PAGE_SIZES - 1; ++i) { + int level = i + 2; - if (!new.lpage_info) + /* Avoid unused variable warning if no large pages */ + (void)level; + + if (new.lpage_info[i]) + continue; + + lpages = 1 + (base_gfn + npages - 1) / + KVM_PAGES_PER_HPAGE(level); + lpages -= base_gfn / KVM_PAGES_PER_HPAGE(level); + + new.lpage_info[i] = vmalloc(lpages * sizeof(*new.lpage_info[i])); + + if (!new.lpage_info[i]) goto out_free; - memset(new.lpage_info, 0, largepages * sizeof(*new.lpage_info)); + memset(new.lpage_info[i], 0, + lpages * sizeof(*new.lpage_info[i])); - if (base_gfn % KVM_PAGES_PER_HPAGE) - new.lpage_info[0].write_count = 1; - if ((base_gfn+npages) % KVM_PAGES_PER_HPAGE) - new.lpage_info[largepages-1].write_count = 1; + if (base_gfn % KVM_PAGES_PER_HPAGE(level)) + new.lpage_info[i][0].write_count = 1; + if ((base_gfn+npages) % KVM_PAGES_PER_HPAGE(level)) + new.lpage_info[i][lpages - 1].write_count = 1; ugfn = new.userspace_addr >> PAGE_SHIFT; /* * If the gfn and userspace address are not aligned wrt each * other, or if explicitly asked to, disable large page * support for this slot */ - if ((base_gfn ^ ugfn) & (KVM_PAGES_PER_HPAGE - 1) || + if ((base_gfn ^ ugfn) & (KVM_PAGES_PER_HPAGE(level) - 1) || !largepages_enabled) - for (i = 0; i < largepages; ++i) - new.lpage_info[i].write_count = 1; + for (j = 0; j < lpages; ++j) + new.lpage_info[i][j].write_count = 1; } +skip_lpage: + /* Allocate page dirty bitmap if needed */ if ((new.flags & KVM_MEM_LOG_DIRTY_PAGES) && !new.dirty_bitmap) { unsigned dirty_bytes = ALIGN(npages, BITS_PER_LONG) / 8; -- cgit v1.2.3 From 07420171593908406c3a59d6f884d426a921a5ea Mon Sep 17 00:00:00 2001 From: Avi Kivity Date: Mon, 6 Jul 2009 12:21:32 +0300 Subject: KVM: MMU: Trace guest pagetable walker Signed-off-by: Avi Kivity --- arch/x86/kvm/mmu.c | 3 ++ arch/x86/kvm/mmutrace.h | 117 +++++++++++++++++++++++++++++++++++++++++++++ arch/x86/kvm/paging_tmpl.h | 11 +++-- 3 files changed, 128 insertions(+), 3 deletions(-) create mode 100644 arch/x86/kvm/mmutrace.h (limited to 'arch/x86/kvm/paging_tmpl.h') diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c index b67585c1ef08..c0dda6447b9f 100644 --- a/arch/x86/kvm/mmu.c +++ b/arch/x86/kvm/mmu.c @@ -140,6 +140,9 @@ module_param(oos_shadow, bool, 0644); #define ACC_USER_MASK PT_USER_MASK #define ACC_ALL (ACC_EXEC_MASK | ACC_WRITE_MASK | ACC_USER_MASK) +#define CREATE_TRACE_POINTS +#include "mmutrace.h" + #define SHADOW_PT_INDEX(addr, level) PT64_INDEX(addr, level) struct kvm_rmap_desc { diff --git a/arch/x86/kvm/mmutrace.h b/arch/x86/kvm/mmutrace.h new file mode 100644 index 000000000000..1367f82717d1 --- /dev/null +++ b/arch/x86/kvm/mmutrace.h @@ -0,0 +1,117 @@ +#if !defined(_TRACE_KVMMMU_H) || defined(TRACE_HEADER_MULTI_READ) +#define _TRACE_KVMMMU_H + +#include + +#undef TRACE_SYSTEM +#define TRACE_SYSTEM kvmmmu +#define TRACE_INCLUDE_PATH . +#define TRACE_INCLUDE_FILE mmutrace + +#define kvm_mmu_trace_pferr_flags \ + { PFERR_PRESENT_MASK, "P" }, \ + { PFERR_WRITE_MASK, "W" }, \ + { PFERR_USER_MASK, "U" }, \ + { PFERR_RSVD_MASK, "RSVD" }, \ + { PFERR_FETCH_MASK, "F" } + +/* + * A pagetable walk has started + */ +TRACE_EVENT( + kvm_mmu_pagetable_walk, + TP_PROTO(u64 addr, int write_fault, int user_fault, int fetch_fault), + TP_ARGS(addr, write_fault, user_fault, fetch_fault), + + TP_STRUCT__entry( + __field(__u64, addr) + __field(__u32, pferr) + ), + + TP_fast_assign( + __entry->addr = addr; + __entry->pferr = (!!write_fault << 1) | (!!user_fault << 2) + | (!!fetch_fault << 4); + ), + + TP_printk("addr %llx pferr %x %s", __entry->addr, __entry->pferr, + __print_flags(__entry->pferr, "|", kvm_mmu_trace_pferr_flags)) +); + + +/* We just walked a paging element */ +TRACE_EVENT( + kvm_mmu_paging_element, + TP_PROTO(u64 pte, int level), + TP_ARGS(pte, level), + + TP_STRUCT__entry( + __field(__u64, pte) + __field(__u32, level) + ), + + TP_fast_assign( + __entry->pte = pte; + __entry->level = level; + ), + + TP_printk("pte %llx level %u", __entry->pte, __entry->level) +); + +/* We set a pte accessed bit */ +TRACE_EVENT( + kvm_mmu_set_accessed_bit, + TP_PROTO(unsigned long table_gfn, unsigned index, unsigned size), + TP_ARGS(table_gfn, index, size), + + TP_STRUCT__entry( + __field(__u64, gpa) + ), + + TP_fast_assign( + __entry->gpa = ((u64)table_gfn << PAGE_SHIFT) + + index * size; + ), + + TP_printk("gpa %llx", __entry->gpa) +); + +/* We set a pte dirty bit */ +TRACE_EVENT( + kvm_mmu_set_dirty_bit, + TP_PROTO(unsigned long table_gfn, unsigned index, unsigned size), + TP_ARGS(table_gfn, index, size), + + TP_STRUCT__entry( + __field(__u64, gpa) + ), + + TP_fast_assign( + __entry->gpa = ((u64)table_gfn << PAGE_SHIFT) + + index * size; + ), + + TP_printk("gpa %llx", __entry->gpa) +); + +TRACE_EVENT( + kvm_mmu_walker_error, + TP_PROTO(u32 pferr), + TP_ARGS(pferr), + + TP_STRUCT__entry( + __field(__u32, pferr) + ), + + TP_fast_assign( + __entry->pferr = pferr; + ), + + TP_printk("pferr %x %s", __entry->pferr, + __print_flags(__entry->pferr, "|", kvm_mmu_trace_pferr_flags)) +); + +#endif /* _TRACE_KVMMMU_H */ + +/* This part must be outside protection */ +#include diff --git a/arch/x86/kvm/paging_tmpl.h b/arch/x86/kvm/paging_tmpl.h index 53e129cec5fd..36ac6d70a847 100644 --- a/arch/x86/kvm/paging_tmpl.h +++ b/arch/x86/kvm/paging_tmpl.h @@ -125,13 +125,15 @@ static int FNAME(walk_addr)(struct guest_walker *walker, gpa_t pte_gpa; int rsvd_fault = 0; - pgprintk("%s: addr %lx\n", __func__, addr); + trace_kvm_mmu_pagetable_walk(addr, write_fault, user_fault, + fetch_fault); walk: walker->level = vcpu->arch.mmu.root_level; pte = vcpu->arch.cr3; #if PTTYPE == 64 if (!is_long_mode(vcpu)) { pte = kvm_pdptr_read(vcpu, (addr >> 30) & 3); + trace_kvm_mmu_paging_element(pte, walker->level); if (!is_present_gpte(pte)) goto not_present; --walker->level; @@ -150,10 +152,9 @@ walk: pte_gpa += index * sizeof(pt_element_t); walker->table_gfn[walker->level - 1] = table_gfn; walker->pte_gpa[walker->level - 1] = pte_gpa; - pgprintk("%s: table_gfn[%d] %lx\n", __func__, - walker->level - 1, table_gfn); kvm_read_guest(vcpu->kvm, pte_gpa, &pte, sizeof(pte)); + trace_kvm_mmu_paging_element(pte, walker->level); if (!is_present_gpte(pte)) goto not_present; @@ -175,6 +176,8 @@ walk: #endif if (!(pte & PT_ACCESSED_MASK)) { + trace_kvm_mmu_set_accessed_bit(table_gfn, index, + sizeof(pte)); mark_page_dirty(vcpu->kvm, table_gfn); if (FNAME(cmpxchg_gpte)(vcpu->kvm, table_gfn, index, pte, pte|PT_ACCESSED_MASK)) @@ -208,6 +211,7 @@ walk: if (write_fault && !is_dirty_gpte(pte)) { bool ret; + trace_kvm_mmu_set_dirty_bit(table_gfn, index, sizeof(pte)); mark_page_dirty(vcpu->kvm, table_gfn); ret = FNAME(cmpxchg_gpte)(vcpu->kvm, table_gfn, index, pte, pte|PT_DIRTY_MASK); @@ -239,6 +243,7 @@ err: walker->error_code |= PFERR_FETCH_MASK; if (rsvd_fault) walker->error_code |= PFERR_RSVD_MASK; + trace_kvm_mmu_walker_error(walker->error_code); return 0; } -- cgit v1.2.3 From d25797b24c0ff2efc2b2fabaebb0ec0cafc0d3e3 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Mon, 27 Jul 2009 16:30:43 +0200 Subject: KVM: MMU: rename is_largepage_backed to mapping_level With the new name and the corresponding backend changes this function can now support multiple hugepage sizes. Signed-off-by: Joerg Roedel Signed-off-by: Avi Kivity --- arch/x86/kvm/mmu.c | 100 ++++++++++++++++++++++++++++++--------------- arch/x86/kvm/paging_tmpl.h | 4 +- 2 files changed, 69 insertions(+), 35 deletions(-) (limited to 'arch/x86/kvm/paging_tmpl.h') diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c index b93ad2c79c15..c707936b2414 100644 --- a/arch/x86/kvm/mmu.c +++ b/arch/x86/kvm/mmu.c @@ -393,37 +393,52 @@ static void mmu_free_rmap_desc(struct kvm_rmap_desc *rd) * Return the pointer to the largepage write count for a given * gfn, handling slots that are not large page aligned. */ -static int *slot_largepage_idx(gfn_t gfn, struct kvm_memory_slot *slot) +static int *slot_largepage_idx(gfn_t gfn, + struct kvm_memory_slot *slot, + int level) { unsigned long idx; - idx = (gfn / KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL)) - - (slot->base_gfn / KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL)); - return &slot->lpage_info[0][idx].write_count; + idx = (gfn / KVM_PAGES_PER_HPAGE(level)) - + (slot->base_gfn / KVM_PAGES_PER_HPAGE(level)); + return &slot->lpage_info[level - 2][idx].write_count; } static void account_shadowed(struct kvm *kvm, gfn_t gfn) { + struct kvm_memory_slot *slot; int *write_count; + int i; gfn = unalias_gfn(kvm, gfn); - write_count = slot_largepage_idx(gfn, - gfn_to_memslot_unaliased(kvm, gfn)); - *write_count += 1; + + slot = gfn_to_memslot_unaliased(kvm, gfn); + for (i = PT_DIRECTORY_LEVEL; + i < PT_PAGE_TABLE_LEVEL + KVM_NR_PAGE_SIZES; ++i) { + write_count = slot_largepage_idx(gfn, slot, i); + *write_count += 1; + } } static void unaccount_shadowed(struct kvm *kvm, gfn_t gfn) { + struct kvm_memory_slot *slot; int *write_count; + int i; gfn = unalias_gfn(kvm, gfn); - write_count = slot_largepage_idx(gfn, - gfn_to_memslot_unaliased(kvm, gfn)); - *write_count -= 1; - WARN_ON(*write_count < 0); + for (i = PT_DIRECTORY_LEVEL; + i < PT_PAGE_TABLE_LEVEL + KVM_NR_PAGE_SIZES; ++i) { + slot = gfn_to_memslot_unaliased(kvm, gfn); + write_count = slot_largepage_idx(gfn, slot, i); + *write_count -= 1; + WARN_ON(*write_count < 0); + } } -static int has_wrprotected_page(struct kvm *kvm, gfn_t gfn) +static int has_wrprotected_page(struct kvm *kvm, + gfn_t gfn, + int level) { struct kvm_memory_slot *slot; int *largepage_idx; @@ -431,47 +446,67 @@ static int has_wrprotected_page(struct kvm *kvm, gfn_t gfn) gfn = unalias_gfn(kvm, gfn); slot = gfn_to_memslot_unaliased(kvm, gfn); if (slot) { - largepage_idx = slot_largepage_idx(gfn, slot); + largepage_idx = slot_largepage_idx(gfn, slot, level); return *largepage_idx; } return 1; } -static int host_largepage_backed(struct kvm *kvm, gfn_t gfn) +static int host_mapping_level(struct kvm *kvm, gfn_t gfn) { + unsigned long page_size = PAGE_SIZE; struct vm_area_struct *vma; unsigned long addr; - int ret = 0; + int i, ret = 0; addr = gfn_to_hva(kvm, gfn); if (kvm_is_error_hva(addr)) - return ret; + return page_size; down_read(¤t->mm->mmap_sem); vma = find_vma(current->mm, addr); - if (vma && is_vm_hugetlb_page(vma)) - ret = 1; + if (!vma) + goto out; + + page_size = vma_kernel_pagesize(vma); + +out: up_read(¤t->mm->mmap_sem); + for (i = PT_PAGE_TABLE_LEVEL; + i < (PT_PAGE_TABLE_LEVEL + KVM_NR_PAGE_SIZES); ++i) { + if (page_size >= KVM_HPAGE_SIZE(i)) + ret = i; + else + break; + } + return ret; } -static int is_largepage_backed(struct kvm_vcpu *vcpu, gfn_t large_gfn) +static int mapping_level(struct kvm_vcpu *vcpu, gfn_t large_gfn) { struct kvm_memory_slot *slot; - - if (has_wrprotected_page(vcpu->kvm, large_gfn)) - return 0; - - if (!host_largepage_backed(vcpu->kvm, large_gfn)) - return 0; + int host_level; + int level = PT_PAGE_TABLE_LEVEL; slot = gfn_to_memslot(vcpu->kvm, large_gfn); if (slot && slot->dirty_bitmap) - return 0; + return PT_PAGE_TABLE_LEVEL; - return 1; + host_level = host_mapping_level(vcpu->kvm, large_gfn); + + if (host_level == PT_PAGE_TABLE_LEVEL) + return host_level; + + for (level = PT_DIRECTORY_LEVEL; level <= host_level; ++level) { + + if (has_wrprotected_page(vcpu->kvm, large_gfn, level)) + break; + } + + return level - 1; } /* @@ -1733,7 +1768,7 @@ static int set_spte(struct kvm_vcpu *vcpu, u64 *sptep, if ((pte_access & ACC_WRITE_MASK) || (write_fault && !is_write_protection(vcpu) && !user_fault)) { - if (largepage && has_wrprotected_page(vcpu->kvm, gfn)) { + if (largepage && has_wrprotected_page(vcpu->kvm, gfn, 1)) { ret = 1; spte = shadow_trap_nonpresent_pte; goto set_pte; @@ -1884,8 +1919,7 @@ static int nonpaging_map(struct kvm_vcpu *vcpu, gva_t v, int write, gfn_t gfn) pfn_t pfn; unsigned long mmu_seq; - if (is_largepage_backed(vcpu, gfn & - ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1))) { + if (mapping_level(vcpu, gfn) == PT_DIRECTORY_LEVEL) { gfn &= ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1); largepage = 1; } @@ -2091,8 +2125,7 @@ static int tdp_page_fault(struct kvm_vcpu *vcpu, gva_t gpa, if (r) return r; - if (is_largepage_backed(vcpu, gfn & - ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1))) { + if (mapping_level(vcpu, gfn) == PT_DIRECTORY_LEVEL) { gfn &= ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1); largepage = 1; } @@ -2494,7 +2527,8 @@ static void mmu_guess_page_from_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa, return; gfn = (gpte & PT64_BASE_ADDR_MASK) >> PAGE_SHIFT; - if (is_large_pte(gpte) && is_largepage_backed(vcpu, gfn)) { + if (is_large_pte(gpte) && + (mapping_level(vcpu, gfn) == PT_DIRECTORY_LEVEL)) { gfn &= ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1); vcpu->arch.update_pte.largepage = 1; } diff --git a/arch/x86/kvm/paging_tmpl.h b/arch/x86/kvm/paging_tmpl.h index 36ac6d70a847..44f0346578cb 100644 --- a/arch/x86/kvm/paging_tmpl.h +++ b/arch/x86/kvm/paging_tmpl.h @@ -407,8 +407,8 @@ static int FNAME(page_fault)(struct kvm_vcpu *vcpu, gva_t addr, if (walker.level == PT_DIRECTORY_LEVEL) { gfn_t large_gfn; large_gfn = walker.gfn & - ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1); - if (is_largepage_backed(vcpu, large_gfn)) { + ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1); + if (mapping_level(vcpu, large_gfn) == PT_DIRECTORY_LEVEL) { walker.gfn = large_gfn; largepage = 1; } -- cgit v1.2.3 From 852e3c19ac64b7c3912e8efe42d3ce090ebc0161 Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Mon, 27 Jul 2009 16:30:44 +0200 Subject: KVM: MMU: make direct mapping paths aware of mapping levels Signed-off-by: Joerg Roedel Signed-off-by: Avi Kivity --- arch/x86/include/asm/kvm_host.h | 2 +- arch/x86/kvm/mmu.c | 83 ++++++++++++++++++++++++----------------- arch/x86/kvm/paging_tmpl.h | 6 +-- 3 files changed, 53 insertions(+), 38 deletions(-) (limited to 'arch/x86/kvm/paging_tmpl.h') diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h index e210b218df44..e09dc26d96bd 100644 --- a/arch/x86/include/asm/kvm_host.h +++ b/arch/x86/include/asm/kvm_host.h @@ -315,7 +315,7 @@ struct kvm_vcpu_arch { struct { gfn_t gfn; /* presumed gfn during guest pte update */ pfn_t pfn; /* pfn corresponding to that gfn */ - int largepage; + int level; unsigned long mmu_seq; } update_pte; diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c index c707936b2414..110c224ed1fb 100644 --- a/arch/x86/kvm/mmu.c +++ b/arch/x86/kvm/mmu.c @@ -257,7 +257,7 @@ static int is_last_spte(u64 pte, int level) { if (level == PT_PAGE_TABLE_LEVEL) return 1; - if (level == PT_DIRECTORY_LEVEL && is_large_pte(pte)) + if (is_large_pte(pte)) return 1; return 0; } @@ -753,7 +753,7 @@ static int kvm_unmap_rmapp(struct kvm *kvm, unsigned long *rmapp) static int kvm_handle_hva(struct kvm *kvm, unsigned long hva, int (*handler)(struct kvm *kvm, unsigned long *rmapp)) { - int i; + int i, j; int retval = 0; /* @@ -772,11 +772,15 @@ static int kvm_handle_hva(struct kvm *kvm, unsigned long hva, end = start + (memslot->npages << PAGE_SHIFT); if (hva >= start && hva < end) { gfn_t gfn_offset = (hva - start) >> PAGE_SHIFT; - int idx = gfn_offset / - KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL); + retval |= handler(kvm, &memslot->rmap[gfn_offset]); - retval |= handler(kvm, - &memslot->lpage_info[0][idx].rmap_pde); + + for (j = 0; j < KVM_NR_PAGE_SIZES - 1; ++j) { + int idx = gfn_offset; + idx /= KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL + j); + retval |= handler(kvm, + &memslot->lpage_info[j][idx].rmap_pde); + } } } @@ -814,12 +818,15 @@ static int kvm_age_rmapp(struct kvm *kvm, unsigned long *rmapp) #define RMAP_RECYCLE_THRESHOLD 1000 -static void rmap_recycle(struct kvm_vcpu *vcpu, gfn_t gfn, int lpage) +static void rmap_recycle(struct kvm_vcpu *vcpu, u64 *spte, gfn_t gfn) { unsigned long *rmapp; + struct kvm_mmu_page *sp; + + sp = page_header(__pa(spte)); gfn = unalias_gfn(vcpu->kvm, gfn); - rmapp = gfn_to_rmap(vcpu->kvm, gfn, lpage); + rmapp = gfn_to_rmap(vcpu->kvm, gfn, sp->role.level); kvm_unmap_rmapp(vcpu->kvm, rmapp); kvm_flush_remote_tlbs(vcpu->kvm); @@ -1734,7 +1741,7 @@ static int mmu_need_write_protect(struct kvm_vcpu *vcpu, gfn_t gfn, static int set_spte(struct kvm_vcpu *vcpu, u64 *sptep, unsigned pte_access, int user_fault, - int write_fault, int dirty, int largepage, + int write_fault, int dirty, int level, gfn_t gfn, pfn_t pfn, bool speculative, bool can_unsync) { @@ -1757,7 +1764,7 @@ static int set_spte(struct kvm_vcpu *vcpu, u64 *sptep, spte |= shadow_nx_mask; if (pte_access & ACC_USER_MASK) spte |= shadow_user_mask; - if (largepage) + if (level > PT_PAGE_TABLE_LEVEL) spte |= PT_PAGE_SIZE_MASK; if (tdp_enabled) spte |= kvm_x86_ops->get_mt_mask(vcpu, gfn, @@ -1768,7 +1775,8 @@ static int set_spte(struct kvm_vcpu *vcpu, u64 *sptep, if ((pte_access & ACC_WRITE_MASK) || (write_fault && !is_write_protection(vcpu) && !user_fault)) { - if (largepage && has_wrprotected_page(vcpu->kvm, gfn, 1)) { + if (level > PT_PAGE_TABLE_LEVEL && + has_wrprotected_page(vcpu->kvm, gfn, level)) { ret = 1; spte = shadow_trap_nonpresent_pte; goto set_pte; @@ -1806,7 +1814,7 @@ set_pte: static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *sptep, unsigned pt_access, unsigned pte_access, int user_fault, int write_fault, int dirty, - int *ptwrite, int largepage, gfn_t gfn, + int *ptwrite, int level, gfn_t gfn, pfn_t pfn, bool speculative) { int was_rmapped = 0; @@ -1823,7 +1831,8 @@ static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *sptep, * If we overwrite a PTE page pointer with a 2MB PMD, unlink * the parent of the now unreachable PTE. */ - if (largepage && !is_large_pte(*sptep)) { + if (level > PT_PAGE_TABLE_LEVEL && + !is_large_pte(*sptep)) { struct kvm_mmu_page *child; u64 pte = *sptep; @@ -1836,8 +1845,9 @@ static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *sptep, } else was_rmapped = 1; } + if (set_spte(vcpu, sptep, pte_access, user_fault, write_fault, - dirty, largepage, gfn, pfn, speculative, true)) { + dirty, level, gfn, pfn, speculative, true)) { if (write_fault) *ptwrite = 1; kvm_x86_ops->tlb_flush(vcpu); @@ -1857,7 +1867,7 @@ static void mmu_set_spte(struct kvm_vcpu *vcpu, u64 *sptep, if (!is_rmap_spte(*sptep)) kvm_release_pfn_clean(pfn); if (rmap_count > RMAP_RECYCLE_THRESHOLD) - rmap_recycle(vcpu, gfn, largepage); + rmap_recycle(vcpu, sptep, gfn); } else { if (was_writeble) kvm_release_pfn_dirty(pfn); @@ -1875,7 +1885,7 @@ static void nonpaging_new_cr3(struct kvm_vcpu *vcpu) } static int __direct_map(struct kvm_vcpu *vcpu, gpa_t v, int write, - int largepage, gfn_t gfn, pfn_t pfn) + int level, gfn_t gfn, pfn_t pfn) { struct kvm_shadow_walk_iterator iterator; struct kvm_mmu_page *sp; @@ -1883,11 +1893,10 @@ static int __direct_map(struct kvm_vcpu *vcpu, gpa_t v, int write, gfn_t pseudo_gfn; for_each_shadow_entry(vcpu, (u64)gfn << PAGE_SHIFT, iterator) { - if (iterator.level == PT_PAGE_TABLE_LEVEL - || (largepage && iterator.level == PT_DIRECTORY_LEVEL)) { + if (iterator.level == level) { mmu_set_spte(vcpu, iterator.sptep, ACC_ALL, ACC_ALL, 0, write, 1, &pt_write, - largepage, gfn, pfn, false); + level, gfn, pfn, false); ++vcpu->stat.pf_fixed; break; } @@ -1915,14 +1924,20 @@ static int __direct_map(struct kvm_vcpu *vcpu, gpa_t v, int write, static int nonpaging_map(struct kvm_vcpu *vcpu, gva_t v, int write, gfn_t gfn) { int r; - int largepage = 0; + int level; pfn_t pfn; unsigned long mmu_seq; - if (mapping_level(vcpu, gfn) == PT_DIRECTORY_LEVEL) { - gfn &= ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1); - largepage = 1; - } + level = mapping_level(vcpu, gfn); + + /* + * This path builds a PAE pagetable - so we can map 2mb pages at + * maximum. Therefore check if the level is larger than that. + */ + if (level > PT_DIRECTORY_LEVEL) + level = PT_DIRECTORY_LEVEL; + + gfn &= ~(KVM_PAGES_PER_HPAGE(level) - 1); mmu_seq = vcpu->kvm->mmu_notifier_seq; smp_rmb(); @@ -1938,7 +1953,7 @@ static int nonpaging_map(struct kvm_vcpu *vcpu, gva_t v, int write, gfn_t gfn) if (mmu_notifier_retry(vcpu, mmu_seq)) goto out_unlock; kvm_mmu_free_some_pages(vcpu); - r = __direct_map(vcpu, v, write, largepage, gfn, pfn); + r = __direct_map(vcpu, v, write, level, gfn, pfn); spin_unlock(&vcpu->kvm->mmu_lock); @@ -2114,7 +2129,7 @@ static int tdp_page_fault(struct kvm_vcpu *vcpu, gva_t gpa, { pfn_t pfn; int r; - int largepage = 0; + int level; gfn_t gfn = gpa >> PAGE_SHIFT; unsigned long mmu_seq; @@ -2125,10 +2140,10 @@ static int tdp_page_fault(struct kvm_vcpu *vcpu, gva_t gpa, if (r) return r; - if (mapping_level(vcpu, gfn) == PT_DIRECTORY_LEVEL) { - gfn &= ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1); - largepage = 1; - } + level = mapping_level(vcpu, gfn); + + gfn &= ~(KVM_PAGES_PER_HPAGE(level) - 1); + mmu_seq = vcpu->kvm->mmu_notifier_seq; smp_rmb(); pfn = gfn_to_pfn(vcpu->kvm, gfn); @@ -2141,7 +2156,7 @@ static int tdp_page_fault(struct kvm_vcpu *vcpu, gva_t gpa, goto out_unlock; kvm_mmu_free_some_pages(vcpu); r = __direct_map(vcpu, gpa, error_code & PFERR_WRITE_MASK, - largepage, gfn, pfn); + level, gfn, pfn); spin_unlock(&vcpu->kvm->mmu_lock); return r; @@ -2448,7 +2463,7 @@ static void mmu_pte_write_new_pte(struct kvm_vcpu *vcpu, const void *new) { if (sp->role.level != PT_PAGE_TABLE_LEVEL) { - if (!vcpu->arch.update_pte.largepage || + if (vcpu->arch.update_pte.level == PT_PAGE_TABLE_LEVEL || sp->role.glevels == PT32_ROOT_LEVEL) { ++vcpu->kvm->stat.mmu_pde_zapped; return; @@ -2498,7 +2513,7 @@ static void mmu_guess_page_from_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa, u64 gpte = 0; pfn_t pfn; - vcpu->arch.update_pte.largepage = 0; + vcpu->arch.update_pte.level = PT_PAGE_TABLE_LEVEL; if (bytes != 4 && bytes != 8) return; @@ -2530,7 +2545,7 @@ static void mmu_guess_page_from_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa, if (is_large_pte(gpte) && (mapping_level(vcpu, gfn) == PT_DIRECTORY_LEVEL)) { gfn &= ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1); - vcpu->arch.update_pte.largepage = 1; + vcpu->arch.update_pte.level = PT_DIRECTORY_LEVEL; } vcpu->arch.update_pte.mmu_seq = vcpu->kvm->mmu_notifier_seq; smp_rmb(); diff --git a/arch/x86/kvm/paging_tmpl.h b/arch/x86/kvm/paging_tmpl.h index 44f0346578cb..b167f0d57b54 100644 --- a/arch/x86/kvm/paging_tmpl.h +++ b/arch/x86/kvm/paging_tmpl.h @@ -253,7 +253,7 @@ static void FNAME(update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *page, pt_element_t gpte; unsigned pte_access; pfn_t pfn; - int largepage = vcpu->arch.update_pte.largepage; + int level = vcpu->arch.update_pte.level; gpte = *(const pt_element_t *)pte; if (~gpte & (PT_PRESENT_MASK | PT_ACCESSED_MASK)) { @@ -272,7 +272,7 @@ static void FNAME(update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *page, return; kvm_get_pfn(pfn); mmu_set_spte(vcpu, spte, page->role.access, pte_access, 0, 0, - gpte & PT_DIRTY_MASK, NULL, largepage, + gpte & PT_DIRTY_MASK, NULL, level, gpte_to_gfn(gpte), pfn, true); } @@ -306,7 +306,7 @@ static u64 *FNAME(fetch)(struct kvm_vcpu *vcpu, gva_t addr, gw->pte_access & access, user_fault, write_fault, gw->ptes[gw->level-1] & PT_DIRTY_MASK, - ptwrite, largepage, + ptwrite, level, gw->gfn, pfn, false); break; } -- cgit v1.2.3 From e04da980c35d75fa050ba4009ad99025432d8d7d Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Mon, 27 Jul 2009 16:30:45 +0200 Subject: KVM: MMU: make page walker aware of mapping levels The page walker may be used with nested paging too when accessing mmio areas. Make it support the additional page-level too. [ Marcelo: fix reserved bit check for 1gb pte ] Signed-off-by: Joerg Roedel Signed-off-by: Marcelo Tosatti Signed-off-by: Avi Kivity --- arch/x86/kvm/mmu.c | 17 ++++++++++++++- arch/x86/kvm/paging_tmpl.h | 52 +++++++++++++++++++++++++--------------------- 2 files changed, 44 insertions(+), 25 deletions(-) (limited to 'arch/x86/kvm/paging_tmpl.h') diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c index 110c224ed1fb..09ab6433bf1d 100644 --- a/arch/x86/kvm/mmu.c +++ b/arch/x86/kvm/mmu.c @@ -108,6 +108,9 @@ module_param(oos_shadow, bool, 0644); #define PT32_LEVEL_MASK(level) \ (((1ULL << PT32_LEVEL_BITS) - 1) << PT32_LEVEL_SHIFT(level)) +#define PT32_LVL_OFFSET_MASK(level) \ + (PT32_BASE_ADDR_MASK & ((1ULL << (PAGE_SHIFT + (((level) - 1) \ + * PT32_LEVEL_BITS))) - 1)) #define PT32_INDEX(address, level)\ (((address) >> PT32_LEVEL_SHIFT(level)) & ((1 << PT32_LEVEL_BITS) - 1)) @@ -116,10 +119,19 @@ module_param(oos_shadow, bool, 0644); #define PT64_BASE_ADDR_MASK (((1ULL << 52) - 1) & ~(u64)(PAGE_SIZE-1)) #define PT64_DIR_BASE_ADDR_MASK \ (PT64_BASE_ADDR_MASK & ~((1ULL << (PAGE_SHIFT + PT64_LEVEL_BITS)) - 1)) +#define PT64_LVL_ADDR_MASK(level) \ + (PT64_BASE_ADDR_MASK & ~((1ULL << (PAGE_SHIFT + (((level) - 1) \ + * PT64_LEVEL_BITS))) - 1)) +#define PT64_LVL_OFFSET_MASK(level) \ + (PT64_BASE_ADDR_MASK & ((1ULL << (PAGE_SHIFT + (((level) - 1) \ + * PT64_LEVEL_BITS))) - 1)) #define PT32_BASE_ADDR_MASK PAGE_MASK #define PT32_DIR_BASE_ADDR_MASK \ (PAGE_MASK & ~((1ULL << (PAGE_SHIFT + PT32_LEVEL_BITS)) - 1)) +#define PT32_LVL_ADDR_MASK(level) \ + (PAGE_MASK & ~((1ULL << (PAGE_SHIFT + (((level) - 1) \ + * PT32_LEVEL_BITS))) - 1)) #define PT64_PERM_MASK (PT_PRESENT_MASK | PT_WRITABLE_MASK | PT_USER_MASK \ | PT64_NX_MASK) @@ -130,6 +142,7 @@ module_param(oos_shadow, bool, 0644); #define PFERR_RSVD_MASK (1U << 3) #define PFERR_FETCH_MASK (1U << 4) +#define PT_PDPE_LEVEL 3 #define PT_DIRECTORY_LEVEL 2 #define PT_PAGE_TABLE_LEVEL 1 @@ -2273,7 +2286,9 @@ static void reset_rsvds_bits_mask(struct kvm_vcpu *vcpu, int level) context->rsvd_bits_mask[0][0] = exb_bit_rsvd | rsvd_bits(maxphyaddr, 51); context->rsvd_bits_mask[1][3] = context->rsvd_bits_mask[0][3]; - context->rsvd_bits_mask[1][2] = context->rsvd_bits_mask[0][2]; + context->rsvd_bits_mask[1][2] = exb_bit_rsvd | + rsvd_bits(maxphyaddr, 51) | + rsvd_bits(13, 29); context->rsvd_bits_mask[1][1] = exb_bit_rsvd | rsvd_bits(maxphyaddr, 51) | rsvd_bits(13, 20); /* large page */ diff --git a/arch/x86/kvm/paging_tmpl.h b/arch/x86/kvm/paging_tmpl.h index b167f0d57b54..578276e34bd9 100644 --- a/arch/x86/kvm/paging_tmpl.h +++ b/arch/x86/kvm/paging_tmpl.h @@ -27,7 +27,8 @@ #define guest_walker guest_walker64 #define FNAME(name) paging##64_##name #define PT_BASE_ADDR_MASK PT64_BASE_ADDR_MASK - #define PT_DIR_BASE_ADDR_MASK PT64_DIR_BASE_ADDR_MASK + #define PT_LVL_ADDR_MASK(lvl) PT64_LVL_ADDR_MASK(lvl) + #define PT_LVL_OFFSET_MASK(lvl) PT64_LVL_OFFSET_MASK(lvl) #define PT_INDEX(addr, level) PT64_INDEX(addr, level) #define PT_LEVEL_MASK(level) PT64_LEVEL_MASK(level) #define PT_LEVEL_BITS PT64_LEVEL_BITS @@ -43,7 +44,8 @@ #define guest_walker guest_walker32 #define FNAME(name) paging##32_##name #define PT_BASE_ADDR_MASK PT32_BASE_ADDR_MASK - #define PT_DIR_BASE_ADDR_MASK PT32_DIR_BASE_ADDR_MASK + #define PT_LVL_ADDR_MASK(lvl) PT32_LVL_ADDR_MASK(lvl) + #define PT_LVL_OFFSET_MASK(lvl) PT32_LVL_OFFSET_MASK(lvl) #define PT_INDEX(addr, level) PT32_INDEX(addr, level) #define PT_LEVEL_MASK(level) PT32_LEVEL_MASK(level) #define PT_LEVEL_BITS PT32_LEVEL_BITS @@ -53,8 +55,8 @@ #error Invalid PTTYPE value #endif -#define gpte_to_gfn FNAME(gpte_to_gfn) -#define gpte_to_gfn_pde FNAME(gpte_to_gfn_pde) +#define gpte_to_gfn_lvl FNAME(gpte_to_gfn_lvl) +#define gpte_to_gfn(pte) gpte_to_gfn_lvl((pte), PT_PAGE_TABLE_LEVEL) /* * The guest_walker structure emulates the behavior of the hardware page @@ -71,14 +73,9 @@ struct guest_walker { u32 error_code; }; -static gfn_t gpte_to_gfn(pt_element_t gpte) +static gfn_t gpte_to_gfn_lvl(pt_element_t gpte, int lvl) { - return (gpte & PT_BASE_ADDR_MASK) >> PAGE_SHIFT; -} - -static gfn_t gpte_to_gfn_pde(pt_element_t gpte) -{ - return (gpte & PT_DIR_BASE_ADDR_MASK) >> PAGE_SHIFT; + return (gpte & PT_LVL_ADDR_MASK(lvl)) >> PAGE_SHIFT; } static bool FNAME(cmpxchg_gpte)(struct kvm *kvm, @@ -189,18 +186,24 @@ walk: walker->ptes[walker->level - 1] = pte; - if (walker->level == PT_PAGE_TABLE_LEVEL) { - walker->gfn = gpte_to_gfn(pte); - break; - } - - if (walker->level == PT_DIRECTORY_LEVEL - && (pte & PT_PAGE_SIZE_MASK) - && (PTTYPE == 64 || is_pse(vcpu))) { - walker->gfn = gpte_to_gfn_pde(pte); - walker->gfn += PT_INDEX(addr, PT_PAGE_TABLE_LEVEL); - if (PTTYPE == 32 && is_cpuid_PSE36()) + if ((walker->level == PT_PAGE_TABLE_LEVEL) || + ((walker->level == PT_DIRECTORY_LEVEL) && + (pte & PT_PAGE_SIZE_MASK) && + (PTTYPE == 64 || is_pse(vcpu))) || + ((walker->level == PT_PDPE_LEVEL) && + (pte & PT_PAGE_SIZE_MASK) && + is_long_mode(vcpu))) { + int lvl = walker->level; + + walker->gfn = gpte_to_gfn_lvl(pte, lvl); + walker->gfn += (addr & PT_LVL_OFFSET_MASK(lvl)) + >> PAGE_SHIFT; + + if (PTTYPE == 32 && + walker->level == PT_DIRECTORY_LEVEL && + is_cpuid_PSE36()) walker->gfn += pse36_gfn_delta(pte); + break; } @@ -609,9 +612,10 @@ static int FNAME(sync_page)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp) #undef PT_BASE_ADDR_MASK #undef PT_INDEX #undef PT_LEVEL_MASK -#undef PT_DIR_BASE_ADDR_MASK +#undef PT_LVL_ADDR_MASK +#undef PT_LVL_OFFSET_MASK #undef PT_LEVEL_BITS #undef PT_MAX_FULL_LEVELS #undef gpte_to_gfn -#undef gpte_to_gfn_pde +#undef gpte_to_gfn_lvl #undef CMPXCHG -- cgit v1.2.3 From 7e4e4056f72da51c5dede48515df0ecd20eaf8ca Mon Sep 17 00:00:00 2001 From: Joerg Roedel Date: Mon, 27 Jul 2009 16:30:46 +0200 Subject: KVM: MMU: shadow support for 1gb pages This patch adds support for shadow paging to the 1gb page table code in KVM. With this code the guest can use 1gb pages even if the host does not support them. [ Marcelo: fix shadow page collision on pmd level if a guest 1gb page is mapped with 4kb ptes on host level ] Signed-off-by: Joerg Roedel Signed-off-by: Marcelo Tosatti Signed-off-by: Avi Kivity --- arch/x86/include/asm/kvm_host.h | 1 - arch/x86/kvm/mmu.c | 14 ++------------ arch/x86/kvm/paging_tmpl.h | 43 +++++++++++++++++++---------------------- 3 files changed, 22 insertions(+), 36 deletions(-) (limited to 'arch/x86/kvm/paging_tmpl.h') diff --git a/arch/x86/include/asm/kvm_host.h b/arch/x86/include/asm/kvm_host.h index e09dc26d96bd..c9fb2bc13a81 100644 --- a/arch/x86/include/asm/kvm_host.h +++ b/arch/x86/include/asm/kvm_host.h @@ -315,7 +315,6 @@ struct kvm_vcpu_arch { struct { gfn_t gfn; /* presumed gfn during guest pte update */ pfn_t pfn; /* pfn corresponding to that gfn */ - int level; unsigned long mmu_seq; } update_pte; diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c index 09ab6433bf1d..1249c12e1d5c 100644 --- a/arch/x86/kvm/mmu.c +++ b/arch/x86/kvm/mmu.c @@ -2478,11 +2478,8 @@ static void mmu_pte_write_new_pte(struct kvm_vcpu *vcpu, const void *new) { if (sp->role.level != PT_PAGE_TABLE_LEVEL) { - if (vcpu->arch.update_pte.level == PT_PAGE_TABLE_LEVEL || - sp->role.glevels == PT32_ROOT_LEVEL) { - ++vcpu->kvm->stat.mmu_pde_zapped; - return; - } + ++vcpu->kvm->stat.mmu_pde_zapped; + return; } ++vcpu->kvm->stat.mmu_pte_updated; @@ -2528,8 +2525,6 @@ static void mmu_guess_page_from_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa, u64 gpte = 0; pfn_t pfn; - vcpu->arch.update_pte.level = PT_PAGE_TABLE_LEVEL; - if (bytes != 4 && bytes != 8) return; @@ -2557,11 +2552,6 @@ static void mmu_guess_page_from_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa, return; gfn = (gpte & PT64_BASE_ADDR_MASK) >> PAGE_SHIFT; - if (is_large_pte(gpte) && - (mapping_level(vcpu, gfn) == PT_DIRECTORY_LEVEL)) { - gfn &= ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1); - vcpu->arch.update_pte.level = PT_DIRECTORY_LEVEL; - } vcpu->arch.update_pte.mmu_seq = vcpu->kvm->mmu_notifier_seq; smp_rmb(); pfn = gfn_to_pfn(vcpu->kvm, gfn); diff --git a/arch/x86/kvm/paging_tmpl.h b/arch/x86/kvm/paging_tmpl.h index 578276e34bd9..d2fec9c12d22 100644 --- a/arch/x86/kvm/paging_tmpl.h +++ b/arch/x86/kvm/paging_tmpl.h @@ -256,7 +256,6 @@ static void FNAME(update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *page, pt_element_t gpte; unsigned pte_access; pfn_t pfn; - int level = vcpu->arch.update_pte.level; gpte = *(const pt_element_t *)pte; if (~gpte & (PT_PRESENT_MASK | PT_ACCESSED_MASK)) { @@ -275,7 +274,7 @@ static void FNAME(update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *page, return; kvm_get_pfn(pfn); mmu_set_spte(vcpu, spte, page->role.access, pte_access, 0, 0, - gpte & PT_DIRTY_MASK, NULL, level, + gpte & PT_DIRTY_MASK, NULL, PT_PAGE_TABLE_LEVEL, gpte_to_gfn(gpte), pfn, true); } @@ -284,7 +283,7 @@ static void FNAME(update_pte)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *page, */ static u64 *FNAME(fetch)(struct kvm_vcpu *vcpu, gva_t addr, struct guest_walker *gw, - int user_fault, int write_fault, int largepage, + int user_fault, int write_fault, int hlevel, int *ptwrite, pfn_t pfn) { unsigned access = gw->pt_access; @@ -303,8 +302,7 @@ static u64 *FNAME(fetch)(struct kvm_vcpu *vcpu, gva_t addr, for_each_shadow_entry(vcpu, addr, iterator) { level = iterator.level; sptep = iterator.sptep; - if (level == PT_PAGE_TABLE_LEVEL - || (largepage && level == PT_DIRECTORY_LEVEL)) { + if (iterator.level == hlevel) { mmu_set_spte(vcpu, sptep, access, gw->pte_access & access, user_fault, write_fault, @@ -323,12 +321,15 @@ static u64 *FNAME(fetch)(struct kvm_vcpu *vcpu, gva_t addr, kvm_flush_remote_tlbs(vcpu->kvm); } - if (level == PT_DIRECTORY_LEVEL - && gw->level == PT_DIRECTORY_LEVEL) { + if (level <= gw->level) { + int delta = level - gw->level + 1; direct = 1; - if (!is_dirty_gpte(gw->ptes[level - 1])) + if (!is_dirty_gpte(gw->ptes[level - delta])) access &= ~ACC_WRITE_MASK; - table_gfn = gpte_to_gfn(gw->ptes[level - 1]); + table_gfn = gpte_to_gfn(gw->ptes[level - delta]); + /* advance table_gfn when emulating 1gb pages with 4k */ + if (delta == 0) + table_gfn += PT_INDEX(addr, level); } else { direct = 0; table_gfn = gw->table_gfn[level - 2]; @@ -381,7 +382,7 @@ static int FNAME(page_fault)(struct kvm_vcpu *vcpu, gva_t addr, int write_pt = 0; int r; pfn_t pfn; - int largepage = 0; + int level = PT_PAGE_TABLE_LEVEL; unsigned long mmu_seq; pgprintk("%s: addr %lx err %x\n", __func__, addr, error_code); @@ -407,15 +408,11 @@ static int FNAME(page_fault)(struct kvm_vcpu *vcpu, gva_t addr, return 0; } - if (walker.level == PT_DIRECTORY_LEVEL) { - gfn_t large_gfn; - large_gfn = walker.gfn & - ~(KVM_PAGES_PER_HPAGE(PT_DIRECTORY_LEVEL) - 1); - if (mapping_level(vcpu, large_gfn) == PT_DIRECTORY_LEVEL) { - walker.gfn = large_gfn; - largepage = 1; - } + if (walker.level >= PT_DIRECTORY_LEVEL) { + level = min(walker.level, mapping_level(vcpu, walker.gfn)); + walker.gfn = walker.gfn & ~(KVM_PAGES_PER_HPAGE(level) - 1); } + mmu_seq = vcpu->kvm->mmu_notifier_seq; smp_rmb(); pfn = gfn_to_pfn(vcpu->kvm, walker.gfn); @@ -432,8 +429,7 @@ static int FNAME(page_fault)(struct kvm_vcpu *vcpu, gva_t addr, goto out_unlock; kvm_mmu_free_some_pages(vcpu); sptep = FNAME(fetch)(vcpu, addr, &walker, user_fault, write_fault, - largepage, &write_pt, pfn); - + level, &write_pt, pfn); pgprintk("%s: shadow pte %p %llx ptwrite %d\n", __func__, sptep, *sptep, write_pt); @@ -468,8 +464,9 @@ static void FNAME(invlpg)(struct kvm_vcpu *vcpu, gva_t gva) sptep = iterator.sptep; /* FIXME: properly handle invlpg on large guest pages */ - if (level == PT_PAGE_TABLE_LEVEL || - ((level == PT_DIRECTORY_LEVEL) && is_large_pte(*sptep))) { + if (level == PT_PAGE_TABLE_LEVEL || + ((level == PT_DIRECTORY_LEVEL && is_large_pte(*sptep))) || + ((level == PT_PDPE_LEVEL && is_large_pte(*sptep)))) { struct kvm_mmu_page *sp = page_header(__pa(sptep)); pte_gpa = (sp->gfn << PAGE_SHIFT); @@ -599,7 +596,7 @@ static int FNAME(sync_page)(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp) nr_present++; pte_access = sp->role.access & FNAME(gpte_access)(vcpu, gpte); set_spte(vcpu, &sp->spt[i], pte_access, 0, 0, - is_dirty_gpte(gpte), 0, gfn, + is_dirty_gpte(gpte), PT_PAGE_TABLE_LEVEL, gfn, spte_to_pfn(sp->spt[i]), true, false); } -- cgit v1.2.3