This series is an update to the PUD hugepage support previously posted
at [0]. This patchset adds support for PUD hugepages at stage 2 a
feature that is useful on cores that have support for large sized TLB
mappings (e.g., 1GB for 4K granule).
This version fixes two bugs -
* Corrects stage 2 fault handling for unsupported hugepage sizes (new
patch 1/9). This is a long standing bug and needs backporting to
earlier kernels
* Ensures that multiple vcpus faulting on the same hugepage doesn't
hamper forward progress (patch 9)
The patches are based on dynamic IPA support which could lead to a
situation where the guest doesn't have PUD level. In this case, the
patches have been updated to fallback to the PTE level mappings at
stage 2.
The patches have been tested on AMD Seattle system with the following
hugepage sizes - 64K, 32M, 1G.
Thanks,
Punit
v6 -> v7
* Restrict thp check to exclude hugetlbfs pages - Patch 1
* Don't update PUD entry if there's no change - Patch 9
* Add check for PUD level in stage 2 - Patch 9
v5 -> v6
* Split Patch 1 to move out the refactoring of exec permissions on
page table entries.
* Patch 4 - Initialise p*dpp pointers in stage2_get_leaf_entry()
* Patch 5 - Trigger a BUG() in kvm_pud_pfn() on arm
v4 -> v5:
* Patch 1 - Drop helper stage2_should_exec() and refactor the
condition to decide if a page table entry should be marked
executable
* Patch 4-6 - Introduce stage2_get_leaf_entry() and use it in this and
latter patches
* Patch 7 - Use stage 2 accessors instead of using the page table
helpers directly
* Patch 7 - Add a note to update the PUD hugepage support when number
of levels of stage 2 tables differs from stage 1
v3 -> v4:
* Patch 1 and 7 - Don't put down hugepages pte if logging is enabled
* Patch 4-5 - Add PUD hugepage support for exec and access faults
* Patch 6 - PUD hugepage support for aging page table entries
v2 -> v3:
* Update vma_pagesize directly if THP [1/4]. Previsouly this was done
indirectly via hugetlb
* Added review tag [4/4]
v1 -> v2:
* Create helper to check if the page should have exec permission [1/4]
* Fix broken condition to detect THP hugepage [1/4]
* Fix in-correct hunk resulting from a rebase [4/4]
[0] https://www.spinics.net/lists/kvm-arm/msg32241.html
[1] https://www.spinics.net/lists/kvm-arm/msg32641.html
Punit Agrawal (9):
KVM: arm/arm64: Ensure only THP is candidate for adjustment
KVM: arm/arm64: Share common code in user_mem_abort()
KVM: arm/arm64: Re-factor setting the Stage 2 entry to exec on fault
KVM: arm/arm64: Introduce helpers to manipulate page table entries
KVM: arm64: Support dirty page tracking for PUD hugepages
KVM: arm64: Support PUD hugepage in stage2_is_exec()
KVM: arm64: Support handling access faults for PUD hugepages
KVM: arm64: Update age handlers to support PUD hugepages
KVM: arm64: Add support for creating PUD hugepages at stage 2
arch/arm/include/asm/kvm_mmu.h | 61 +++++
arch/arm64/include/asm/kvm_mmu.h | 48 ++++
arch/arm64/include/asm/pgtable-hwdef.h | 4 +
arch/arm64/include/asm/pgtable.h | 9 +
virt/kvm/arm/mmu.c | 324 +++++++++++++++++++------
5 files changed, 368 insertions(+), 78 deletions(-)
--
2.18.0
PageTransCompoundMap() returns true for hugetlbfs and THP
hugepages. This behaviour incorrectly leads to stage 2 faults for
unsupported hugepage sizes (e.g., 64K hugepage with 4K pages) to be
treated as THP faults.
Tighten the check to filter out hugetlbfs pages. This also leads to
consistently mapping all unsupported hugepage sizes as PTE level
entries at stage 2.
Signed-off-by: Punit Agrawal <[email protected]>
Cc: Christoffer Dall <[email protected]>
Cc: Marc Zyngier <[email protected]>
Cc: Suzuki Poulose <[email protected]>
Cc: [email protected] # v4.13+
---
virt/kvm/arm/mmu.c | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c
index 7e477b3cae5b..c23a1b323aad 100644
--- a/virt/kvm/arm/mmu.c
+++ b/virt/kvm/arm/mmu.c
@@ -1231,8 +1231,14 @@ static bool transparent_hugepage_adjust(kvm_pfn_t *pfnp, phys_addr_t *ipap)
{
kvm_pfn_t pfn = *pfnp;
gfn_t gfn = *ipap >> PAGE_SHIFT;
+ struct page *page = pfn_to_page(pfn);
- if (PageTransCompoundMap(pfn_to_page(pfn))) {
+ /*
+ * PageTransCompoungMap() returns true for THP and
+ * hugetlbfs. Make sure the adjustment is done only for THP
+ * pages.
+ */
+ if (!PageHuge(page) && PageTransCompoundMap(page)) {
unsigned long mask;
/*
* The address we faulted on is backed by a transparent huge
--
2.18.0
The code for operations such as marking the pfn as dirty, and
dcache/icache maintenance during stage 2 fault handling is duplicated
between normal pages and PMD hugepages.
Instead of creating another copy of the operations when we introduce
PUD hugepages, let's share them across the different pagesizes.
Signed-off-by: Punit Agrawal <[email protected]>
Cc: Suzuki K Poulose <[email protected]>
Cc: Christoffer Dall <[email protected]>
Cc: Marc Zyngier <[email protected]>
---
virt/kvm/arm/mmu.c | 45 +++++++++++++++++++++++++++++----------------
1 file changed, 29 insertions(+), 16 deletions(-)
diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c
index c23a1b323aad..5b76ee204000 100644
--- a/virt/kvm/arm/mmu.c
+++ b/virt/kvm/arm/mmu.c
@@ -1490,7 +1490,7 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
kvm_pfn_t pfn;
pgprot_t mem_type = PAGE_S2;
bool logging_active = memslot_is_logging(memslot);
- unsigned long flags = 0;
+ unsigned long vma_pagesize, flags = 0;
write_fault = kvm_is_write_fault(vcpu);
exec_fault = kvm_vcpu_trap_is_iabt(vcpu);
@@ -1510,10 +1510,17 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
return -EFAULT;
}
- if (vma_kernel_pagesize(vma) == PMD_SIZE && !logging_active) {
+ vma_pagesize = vma_kernel_pagesize(vma);
+ if (vma_pagesize == PMD_SIZE && !logging_active) {
hugetlb = true;
gfn = (fault_ipa & PMD_MASK) >> PAGE_SHIFT;
} else {
+ /*
+ * Fallback to PTE if it's not one of the Stage 2
+ * supported hugepage sizes
+ */
+ vma_pagesize = PAGE_SIZE;
+
/*
* Pages belonging to memslots that don't have the same
* alignment for userspace and IPA cannot be mapped using
@@ -1579,23 +1586,34 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
if (mmu_notifier_retry(kvm, mmu_seq))
goto out_unlock;
- if (!hugetlb && !force_pte)
+ if (!hugetlb && !force_pte) {
+ /*
+ * Only PMD_SIZE transparent hugepages(THP) are
+ * currently supported. This code will need to be
+ * updated to support other THP sizes.
+ */
hugetlb = transparent_hugepage_adjust(&pfn, &fault_ipa);
+ if (hugetlb)
+ vma_pagesize = PMD_SIZE;
+ }
+
+ if (writable)
+ kvm_set_pfn_dirty(pfn);
- if (hugetlb) {
+ if (fault_status != FSC_PERM)
+ clean_dcache_guest_page(pfn, vma_pagesize);
+
+ if (exec_fault)
+ invalidate_icache_guest_page(pfn, vma_pagesize);
+
+ if (hugetlb && vma_pagesize == PMD_SIZE) {
pmd_t new_pmd = pfn_pmd(pfn, mem_type);
new_pmd = pmd_mkhuge(new_pmd);
- if (writable) {
+ if (writable)
new_pmd = kvm_s2pmd_mkwrite(new_pmd);
- kvm_set_pfn_dirty(pfn);
- }
-
- if (fault_status != FSC_PERM)
- clean_dcache_guest_page(pfn, PMD_SIZE);
if (exec_fault) {
new_pmd = kvm_s2pmd_mkexec(new_pmd);
- invalidate_icache_guest_page(pfn, PMD_SIZE);
} else if (fault_status == FSC_PERM) {
/* Preserve execute if XN was already cleared */
if (stage2_is_exec(kvm, fault_ipa))
@@ -1608,16 +1626,11 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
if (writable) {
new_pte = kvm_s2pte_mkwrite(new_pte);
- kvm_set_pfn_dirty(pfn);
mark_page_dirty(kvm, gfn);
}
- if (fault_status != FSC_PERM)
- clean_dcache_guest_page(pfn, PAGE_SIZE);
-
if (exec_fault) {
new_pte = kvm_s2pte_mkexec(new_pte);
- invalidate_icache_guest_page(pfn, PAGE_SIZE);
} else if (fault_status == FSC_PERM) {
/* Preserve execute if XN was already cleared */
if (stage2_is_exec(kvm, fault_ipa))
--
2.18.0
Stage 2 fault handler marks a page as executable if it is handling an
execution fault or if it was a permission fault in which case the
executable bit needs to be preserved.
The logic to decide if the page should be marked executable is
duplicated for PMD and PTE entries. To avoid creating another copy
when support for PUD hugepages is introduced refactor the code to
share the checks needed to mark a page table entry as executable.
Signed-off-by: Punit Agrawal <[email protected]>
Reviewed-by: Suzuki K Poulose <[email protected]>
Cc: Christoffer Dall <[email protected]>
Cc: Marc Zyngier <[email protected]>
---
virt/kvm/arm/mmu.c | 28 +++++++++++++++-------------
1 file changed, 15 insertions(+), 13 deletions(-)
diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c
index 5b76ee204000..ec64d21c6571 100644
--- a/virt/kvm/arm/mmu.c
+++ b/virt/kvm/arm/mmu.c
@@ -1481,7 +1481,8 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
unsigned long fault_status)
{
int ret;
- bool write_fault, exec_fault, writable, hugetlb = false, force_pte = false;
+ bool write_fault, writable, hugetlb = false, force_pte = false;
+ bool exec_fault, needs_exec;
unsigned long mmu_seq;
gfn_t gfn = fault_ipa >> PAGE_SHIFT;
struct kvm *kvm = vcpu->kvm;
@@ -1606,19 +1607,25 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
if (exec_fault)
invalidate_icache_guest_page(pfn, vma_pagesize);
+ /*
+ * If we took an execution fault we have made the
+ * icache/dcache coherent above and should now let the s2
+ * mapping be executable.
+ *
+ * Write faults (!exec_fault && FSC_PERM) are orthogonal to
+ * execute permissions, and we preserve whatever we have.
+ */
+ needs_exec = exec_fault ||
+ (fault_status == FSC_PERM && stage2_is_exec(kvm, fault_ipa));
+
if (hugetlb && vma_pagesize == PMD_SIZE) {
pmd_t new_pmd = pfn_pmd(pfn, mem_type);
new_pmd = pmd_mkhuge(new_pmd);
if (writable)
new_pmd = kvm_s2pmd_mkwrite(new_pmd);
- if (exec_fault) {
+ if (needs_exec)
new_pmd = kvm_s2pmd_mkexec(new_pmd);
- } else if (fault_status == FSC_PERM) {
- /* Preserve execute if XN was already cleared */
- if (stage2_is_exec(kvm, fault_ipa))
- new_pmd = kvm_s2pmd_mkexec(new_pmd);
- }
ret = stage2_set_pmd_huge(kvm, memcache, fault_ipa, &new_pmd);
} else {
@@ -1629,13 +1636,8 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
mark_page_dirty(kvm, gfn);
}
- if (exec_fault) {
+ if (needs_exec)
new_pte = kvm_s2pte_mkexec(new_pte);
- } else if (fault_status == FSC_PERM) {
- /* Preserve execute if XN was already cleared */
- if (stage2_is_exec(kvm, fault_ipa))
- new_pte = kvm_s2pte_mkexec(new_pte);
- }
ret = stage2_set_pte(kvm, memcache, fault_ipa, &new_pte, flags);
}
--
2.18.0
Introduce helpers to abstract architectural handling of the conversion
of pfn to page table entries and marking a PMD page table entry as a
block entry.
The helpers are introduced in preparation for supporting PUD hugepages
at stage 2 - which are supported on arm64 but do not exist on arm.
Signed-off-by: Punit Agrawal <[email protected]>
Reviewed-by: Suzuki K Poulose <[email protected]>
Acked-by: Christoffer Dall <[email protected]>
Cc: Marc Zyngier <[email protected]>
Cc: Russell King <[email protected]>
Cc: Catalin Marinas <[email protected]>
Cc: Will Deacon <[email protected]>
---
arch/arm/include/asm/kvm_mmu.h | 5 +++++
arch/arm64/include/asm/kvm_mmu.h | 5 +++++
virt/kvm/arm/mmu.c | 14 ++++++++------
3 files changed, 18 insertions(+), 6 deletions(-)
diff --git a/arch/arm/include/asm/kvm_mmu.h b/arch/arm/include/asm/kvm_mmu.h
index 5ad1a54f98dc..e77212e53e77 100644
--- a/arch/arm/include/asm/kvm_mmu.h
+++ b/arch/arm/include/asm/kvm_mmu.h
@@ -82,6 +82,11 @@ void kvm_clear_hyp_idmap(void);
#define kvm_mk_pud(pmdp) __pud(__pa(pmdp) | PMD_TYPE_TABLE)
#define kvm_mk_pgd(pudp) ({ BUILD_BUG(); 0; })
+#define kvm_pfn_pte(pfn, prot) pfn_pte(pfn, prot)
+#define kvm_pfn_pmd(pfn, prot) pfn_pmd(pfn, prot)
+
+#define kvm_pmd_mkhuge(pmd) pmd_mkhuge(pmd)
+
static inline pte_t kvm_s2pte_mkwrite(pte_t pte)
{
pte_val(pte) |= L_PTE_S2_RDWR;
diff --git a/arch/arm64/include/asm/kvm_mmu.h b/arch/arm64/include/asm/kvm_mmu.h
index 77b1af9e64db..baabea0cbb66 100644
--- a/arch/arm64/include/asm/kvm_mmu.h
+++ b/arch/arm64/include/asm/kvm_mmu.h
@@ -184,6 +184,11 @@ void kvm_clear_hyp_idmap(void);
#define kvm_mk_pgd(pudp) \
__pgd(__phys_to_pgd_val(__pa(pudp)) | PUD_TYPE_TABLE)
+#define kvm_pfn_pte(pfn, prot) pfn_pte(pfn, prot)
+#define kvm_pfn_pmd(pfn, prot) pfn_pmd(pfn, prot)
+
+#define kvm_pmd_mkhuge(pmd) pmd_mkhuge(pmd)
+
static inline pte_t kvm_s2pte_mkwrite(pte_t pte)
{
pte_val(pte) |= PTE_S2_RDWR;
diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c
index ec64d21c6571..21079eb5bc15 100644
--- a/virt/kvm/arm/mmu.c
+++ b/virt/kvm/arm/mmu.c
@@ -607,7 +607,7 @@ static void create_hyp_pte_mappings(pmd_t *pmd, unsigned long start,
addr = start;
do {
pte = pte_offset_kernel(pmd, addr);
- kvm_set_pte(pte, pfn_pte(pfn, prot));
+ kvm_set_pte(pte, kvm_pfn_pte(pfn, prot));
get_page(virt_to_page(pte));
pfn++;
} while (addr += PAGE_SIZE, addr != end);
@@ -1202,7 +1202,7 @@ int kvm_phys_addr_ioremap(struct kvm *kvm, phys_addr_t guest_ipa,
pfn = __phys_to_pfn(pa);
for (addr = guest_ipa; addr < end; addr += PAGE_SIZE) {
- pte_t pte = pfn_pte(pfn, PAGE_S2_DEVICE);
+ pte_t pte = kvm_pfn_pte(pfn, PAGE_S2_DEVICE);
if (writable)
pte = kvm_s2pte_mkwrite(pte);
@@ -1619,8 +1619,10 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
(fault_status == FSC_PERM && stage2_is_exec(kvm, fault_ipa));
if (hugetlb && vma_pagesize == PMD_SIZE) {
- pmd_t new_pmd = pfn_pmd(pfn, mem_type);
- new_pmd = pmd_mkhuge(new_pmd);
+ pmd_t new_pmd = kvm_pfn_pmd(pfn, mem_type);
+
+ new_pmd = kvm_pmd_mkhuge(new_pmd);
+
if (writable)
new_pmd = kvm_s2pmd_mkwrite(new_pmd);
@@ -1629,7 +1631,7 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
ret = stage2_set_pmd_huge(kvm, memcache, fault_ipa, &new_pmd);
} else {
- pte_t new_pte = pfn_pte(pfn, mem_type);
+ pte_t new_pte = kvm_pfn_pte(pfn, mem_type);
if (writable) {
new_pte = kvm_s2pte_mkwrite(new_pte);
@@ -1886,7 +1888,7 @@ void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte)
* just like a translation fault and clean the cache to the PoC.
*/
clean_dcache_guest_page(pfn, PAGE_SIZE);
- stage2_pte = pfn_pte(pfn, PAGE_S2);
+ stage2_pte = kvm_pfn_pte(pfn, PAGE_S2);
handle_hva_to_gpa(kvm, hva, end, &kvm_set_spte_handler, &stage2_pte);
}
--
2.18.0
In preparation for creating PUD hugepages at stage 2, add support for
write protecting PUD hugepages when they are encountered. Write
protecting guest tables is used to track dirty pages when migrating
VMs.
Also, provide trivial implementations of required kvm_s2pud_* helpers
to allow sharing of code with arm32.
Signed-off-by: Punit Agrawal <[email protected]>
Reviewed-by: Christoffer Dall <[email protected]>
Reviewed-by: Suzuki K Poulose <[email protected]>
Cc: Marc Zyngier <[email protected]>
Cc: Russell King <[email protected]>
Cc: Catalin Marinas <[email protected]>
Cc: Will Deacon <[email protected]>
---
arch/arm/include/asm/kvm_mmu.h | 15 +++++++++++++++
arch/arm64/include/asm/kvm_mmu.h | 10 ++++++++++
virt/kvm/arm/mmu.c | 11 +++++++----
3 files changed, 32 insertions(+), 4 deletions(-)
diff --git a/arch/arm/include/asm/kvm_mmu.h b/arch/arm/include/asm/kvm_mmu.h
index e77212e53e77..9ec09f4cc284 100644
--- a/arch/arm/include/asm/kvm_mmu.h
+++ b/arch/arm/include/asm/kvm_mmu.h
@@ -87,6 +87,21 @@ void kvm_clear_hyp_idmap(void);
#define kvm_pmd_mkhuge(pmd) pmd_mkhuge(pmd)
+/*
+ * The following kvm_*pud*() functions are provided strictly to allow
+ * sharing code with arm64. They should never be called in practice.
+ */
+static inline void kvm_set_s2pud_readonly(pud_t *pud)
+{
+ BUG();
+}
+
+static inline bool kvm_s2pud_readonly(pud_t *pud)
+{
+ BUG();
+ return false;
+}
+
static inline pte_t kvm_s2pte_mkwrite(pte_t pte)
{
pte_val(pte) |= L_PTE_S2_RDWR;
diff --git a/arch/arm64/include/asm/kvm_mmu.h b/arch/arm64/include/asm/kvm_mmu.h
index baabea0cbb66..3cc342177474 100644
--- a/arch/arm64/include/asm/kvm_mmu.h
+++ b/arch/arm64/include/asm/kvm_mmu.h
@@ -251,6 +251,16 @@ static inline bool kvm_s2pmd_exec(pmd_t *pmdp)
return !(READ_ONCE(pmd_val(*pmdp)) & PMD_S2_XN);
}
+static inline void kvm_set_s2pud_readonly(pud_t *pudp)
+{
+ kvm_set_s2pte_readonly((pte_t *)pudp);
+}
+
+static inline bool kvm_s2pud_readonly(pud_t *pudp)
+{
+ return kvm_s2pte_readonly((pte_t *)pudp);
+}
+
#define hyp_pte_table_empty(ptep) kvm_page_empty(ptep)
#ifdef __PAGETABLE_PMD_FOLDED
diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c
index 21079eb5bc15..9c48f2ca6583 100644
--- a/virt/kvm/arm/mmu.c
+++ b/virt/kvm/arm/mmu.c
@@ -1347,9 +1347,12 @@ static void stage2_wp_puds(struct kvm *kvm, pgd_t *pgd,
do {
next = stage2_pud_addr_end(kvm, addr, end);
if (!stage2_pud_none(kvm, *pud)) {
- /* TODO:PUD not supported, revisit later if supported */
- BUG_ON(stage2_pud_huge(kvm, *pud));
- stage2_wp_pmds(kvm, pud, addr, next);
+ if (stage2_pud_huge(kvm, *pud)) {
+ if (!kvm_s2pud_readonly(pud))
+ kvm_set_s2pud_readonly(pud);
+ } else {
+ stage2_wp_pmds(kvm, pud, addr, next);
+ }
}
} while (pud++, addr = next, addr != end);
}
@@ -1392,7 +1395,7 @@ static void stage2_wp_range(struct kvm *kvm, phys_addr_t addr, phys_addr_t end)
*
* Called to start logging dirty pages after memory region
* KVM_MEM_LOG_DIRTY_PAGES operation is called. After this function returns
- * all present PMD and PTEs are write protected in the memory region.
+ * all present PUD, PMD and PTEs are write protected in the memory region.
* Afterwards read of dirty page log can be called.
*
* Acquires kvm_mmu_lock. Called with kvm->slots_lock mutex acquired,
--
2.18.0
In preparation for creating larger hugepages at Stage 2, extend the
access fault handling at Stage 2 to support PUD hugepages when
encountered.
Provide trivial helpers for arm32 to allow sharing of code.
Signed-off-by: Punit Agrawal <[email protected]>
Reviewed-by: Suzuki K Poulose <[email protected]>
Cc: Christoffer Dall <[email protected]>
Cc: Marc Zyngier <[email protected]>
Cc: Russell King <[email protected]>
Cc: Catalin Marinas <[email protected]>
Cc: Will Deacon <[email protected]>
---
arch/arm/include/asm/kvm_mmu.h | 9 +++++++++
arch/arm64/include/asm/kvm_mmu.h | 7 +++++++
arch/arm64/include/asm/pgtable.h | 6 ++++++
virt/kvm/arm/mmu.c | 22 +++++++++++-----------
4 files changed, 33 insertions(+), 11 deletions(-)
diff --git a/arch/arm/include/asm/kvm_mmu.h b/arch/arm/include/asm/kvm_mmu.h
index 26a2ab05b3f6..95b34aad0dc8 100644
--- a/arch/arm/include/asm/kvm_mmu.h
+++ b/arch/arm/include/asm/kvm_mmu.h
@@ -85,6 +85,9 @@ void kvm_clear_hyp_idmap(void);
#define kvm_pfn_pte(pfn, prot) pfn_pte(pfn, prot)
#define kvm_pfn_pmd(pfn, prot) pfn_pmd(pfn, prot)
+#define kvm_pud_pfn(pud) ({ BUG(); 0; })
+
+
#define kvm_pmd_mkhuge(pmd) pmd_mkhuge(pmd)
/*
@@ -108,6 +111,12 @@ static inline bool kvm_s2pud_exec(pud_t *pud)
return false;
}
+static inline pud_t kvm_s2pud_mkyoung(pud_t pud)
+{
+ BUG();
+ return pud;
+}
+
static inline pte_t kvm_s2pte_mkwrite(pte_t pte)
{
pte_val(pte) |= L_PTE_S2_RDWR;
diff --git a/arch/arm64/include/asm/kvm_mmu.h b/arch/arm64/include/asm/kvm_mmu.h
index c06ef3be8ca9..b93e5167728f 100644
--- a/arch/arm64/include/asm/kvm_mmu.h
+++ b/arch/arm64/include/asm/kvm_mmu.h
@@ -187,6 +187,8 @@ void kvm_clear_hyp_idmap(void);
#define kvm_pfn_pte(pfn, prot) pfn_pte(pfn, prot)
#define kvm_pfn_pmd(pfn, prot) pfn_pmd(pfn, prot)
+#define kvm_pud_pfn(pud) pud_pfn(pud)
+
#define kvm_pmd_mkhuge(pmd) pmd_mkhuge(pmd)
static inline pte_t kvm_s2pte_mkwrite(pte_t pte)
@@ -266,6 +268,11 @@ static inline bool kvm_s2pud_exec(pud_t *pudp)
return !(READ_ONCE(pud_val(*pudp)) & PUD_S2_XN);
}
+static inline pud_t kvm_s2pud_mkyoung(pud_t pud)
+{
+ return pud_mkyoung(pud);
+}
+
#define hyp_pte_table_empty(ptep) kvm_page_empty(ptep)
#ifdef __PAGETABLE_PMD_FOLDED
diff --git a/arch/arm64/include/asm/pgtable.h b/arch/arm64/include/asm/pgtable.h
index 1bdeca8918a6..a64a5c35beb1 100644
--- a/arch/arm64/include/asm/pgtable.h
+++ b/arch/arm64/include/asm/pgtable.h
@@ -314,6 +314,11 @@ static inline pte_t pud_pte(pud_t pud)
return __pte(pud_val(pud));
}
+static inline pud_t pte_pud(pte_t pte)
+{
+ return __pud(pte_val(pte));
+}
+
static inline pmd_t pud_pmd(pud_t pud)
{
return __pmd(pud_val(pud));
@@ -380,6 +385,7 @@ static inline int pmd_protnone(pmd_t pmd)
#define pfn_pmd(pfn,prot) __pmd(__phys_to_pmd_val((phys_addr_t)(pfn) << PAGE_SHIFT) | pgprot_val(prot))
#define mk_pmd(page,prot) pfn_pmd(page_to_pfn(page),prot)
+#define pud_mkyoung(pud) pte_pud(pte_mkyoung(pud_pte(pud)))
#define pud_write(pud) pte_write(pud_pte(pud))
#define __pud_to_phys(pud) __pte_to_phys(pud_pte(pud))
diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c
index 5fd1eae7d964..e2487e5fff37 100644
--- a/virt/kvm/arm/mmu.c
+++ b/virt/kvm/arm/mmu.c
@@ -1706,6 +1706,7 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
*/
static void handle_access_fault(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa)
{
+ pud_t *pud;
pmd_t *pmd;
pte_t *pte;
kvm_pfn_t pfn;
@@ -1715,24 +1716,23 @@ static void handle_access_fault(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa)
spin_lock(&vcpu->kvm->mmu_lock);
- pmd = stage2_get_pmd(vcpu->kvm, NULL, fault_ipa);
- if (!pmd || pmd_none(*pmd)) /* Nothing there */
+ if (!stage2_get_leaf_entry(vcpu->kvm, fault_ipa, &pud, &pmd, &pte))
goto out;
- if (pmd_thp_or_huge(*pmd)) { /* THP, HugeTLB */
+ if (pud) { /* HugeTLB */
+ *pud = kvm_s2pud_mkyoung(*pud);
+ pfn = kvm_pud_pfn(*pud);
+ pfn_valid = true;
+ } else if (pmd) { /* THP, HugeTLB */
*pmd = pmd_mkyoung(*pmd);
pfn = pmd_pfn(*pmd);
pfn_valid = true;
- goto out;
+ } else {
+ *pte = pte_mkyoung(*pte); /* Just a page... */
+ pfn = pte_pfn(*pte);
+ pfn_valid = true;
}
- pte = pte_offset_kernel(pmd, fault_ipa);
- if (pte_none(*pte)) /* Nothing there either */
- goto out;
-
- *pte = pte_mkyoung(*pte); /* Just a page... */
- pfn = pte_pfn(*pte);
- pfn_valid = true;
out:
spin_unlock(&vcpu->kvm->mmu_lock);
if (pfn_valid)
--
2.18.0
KVM only supports PMD hugepages at stage 2. Now that the various page
handling routines are updated, extend the stage 2 fault handling to
map in PUD hugepages.
Addition of PUD hugepage support enables additional page sizes (e.g.,
1G with 4K granule) which can be useful on cores that support mapping
larger block sizes in the TLB entries.
Signed-off-by: Punit Agrawal <[email protected]>
Cc: Christoffer Dall <[email protected]>
Cc: Marc Zyngier <[email protected]>
Cc: Russell King <[email protected]>
Cc: Catalin Marinas <[email protected]>
Cc: Will Deacon <[email protected]>
---
arch/arm/include/asm/kvm_mmu.h | 20 +++++
arch/arm64/include/asm/kvm_mmu.h | 16 ++++
arch/arm64/include/asm/pgtable-hwdef.h | 2 +
arch/arm64/include/asm/pgtable.h | 2 +
virt/kvm/arm/mmu.c | 108 +++++++++++++++++++++++--
5 files changed, 143 insertions(+), 5 deletions(-)
diff --git a/arch/arm/include/asm/kvm_mmu.h b/arch/arm/include/asm/kvm_mmu.h
index a42b9505c9a7..a8e86b926ee0 100644
--- a/arch/arm/include/asm/kvm_mmu.h
+++ b/arch/arm/include/asm/kvm_mmu.h
@@ -84,11 +84,14 @@ void kvm_clear_hyp_idmap(void);
#define kvm_pfn_pte(pfn, prot) pfn_pte(pfn, prot)
#define kvm_pfn_pmd(pfn, prot) pfn_pmd(pfn, prot)
+#define kvm_pfn_pud(pfn, prot) (__pud(0))
#define kvm_pud_pfn(pud) ({ BUG(); 0; })
#define kvm_pmd_mkhuge(pmd) pmd_mkhuge(pmd)
+/* No support for pud hugepages */
+#define kvm_pud_mkhuge(pud) (pud)
/*
* The following kvm_*pud*() functions are provided strictly to allow
@@ -105,6 +108,23 @@ static inline bool kvm_s2pud_readonly(pud_t *pud)
return false;
}
+static inline void kvm_set_pud(pud_t *pud, pud_t new_pud)
+{
+ BUG();
+}
+
+static inline pud_t kvm_s2pud_mkwrite(pud_t pud)
+{
+ BUG();
+ return pud;
+}
+
+static inline pud_t kvm_s2pud_mkexec(pud_t pud)
+{
+ BUG();
+ return pud;
+}
+
static inline bool kvm_s2pud_exec(pud_t *pud)
{
BUG();
diff --git a/arch/arm64/include/asm/kvm_mmu.h b/arch/arm64/include/asm/kvm_mmu.h
index 3baf72705dcc..b4e9c2cceecb 100644
--- a/arch/arm64/include/asm/kvm_mmu.h
+++ b/arch/arm64/include/asm/kvm_mmu.h
@@ -184,12 +184,16 @@ void kvm_clear_hyp_idmap(void);
#define kvm_mk_pgd(pudp) \
__pgd(__phys_to_pgd_val(__pa(pudp)) | PUD_TYPE_TABLE)
+#define kvm_set_pud(pudp, pud) set_pud(pudp, pud)
+
#define kvm_pfn_pte(pfn, prot) pfn_pte(pfn, prot)
#define kvm_pfn_pmd(pfn, prot) pfn_pmd(pfn, prot)
+#define kvm_pfn_pud(pfn, prot) pfn_pud(pfn, prot)
#define kvm_pud_pfn(pud) pud_pfn(pud)
#define kvm_pmd_mkhuge(pmd) pmd_mkhuge(pmd)
+#define kvm_pud_mkhuge(pud) pud_mkhuge(pud)
static inline pte_t kvm_s2pte_mkwrite(pte_t pte)
{
@@ -203,6 +207,12 @@ static inline pmd_t kvm_s2pmd_mkwrite(pmd_t pmd)
return pmd;
}
+static inline pud_t kvm_s2pud_mkwrite(pud_t pud)
+{
+ pud_val(pud) |= PUD_S2_RDWR;
+ return pud;
+}
+
static inline pte_t kvm_s2pte_mkexec(pte_t pte)
{
pte_val(pte) &= ~PTE_S2_XN;
@@ -215,6 +225,12 @@ static inline pmd_t kvm_s2pmd_mkexec(pmd_t pmd)
return pmd;
}
+static inline pud_t kvm_s2pud_mkexec(pud_t pud)
+{
+ pud_val(pud) &= ~PUD_S2_XN;
+ return pud;
+}
+
static inline void kvm_set_s2pte_readonly(pte_t *ptep)
{
pteval_t old_pteval, pteval;
diff --git a/arch/arm64/include/asm/pgtable-hwdef.h b/arch/arm64/include/asm/pgtable-hwdef.h
index 10ae592b78b8..e327665e94d1 100644
--- a/arch/arm64/include/asm/pgtable-hwdef.h
+++ b/arch/arm64/include/asm/pgtable-hwdef.h
@@ -193,6 +193,8 @@
#define PMD_S2_RDWR (_AT(pmdval_t, 3) << 6) /* HAP[2:1] */
#define PMD_S2_XN (_AT(pmdval_t, 2) << 53) /* XN[1:0] */
+#define PUD_S2_RDONLY (_AT(pudval_t, 1) << 6) /* HAP[2:1] */
+#define PUD_S2_RDWR (_AT(pudval_t, 3) << 6) /* HAP[2:1] */
#define PUD_S2_XN (_AT(pudval_t, 2) << 53) /* XN[1:0] */
/*
diff --git a/arch/arm64/include/asm/pgtable.h b/arch/arm64/include/asm/pgtable.h
index 4d9476e420d9..0afc34f94ff5 100644
--- a/arch/arm64/include/asm/pgtable.h
+++ b/arch/arm64/include/asm/pgtable.h
@@ -389,6 +389,8 @@ static inline int pmd_protnone(pmd_t pmd)
#define pud_mkyoung(pud) pte_pud(pte_mkyoung(pud_pte(pud)))
#define pud_write(pud) pte_write(pud_pte(pud))
+#define pud_mkhuge(pud) (__pud(pud_val(pud) & ~PUD_TABLE_BIT))
+
#define __pud_to_phys(pud) __pte_to_phys(pud_pte(pud))
#define __phys_to_pud_val(phys) __phys_to_pte_val(phys)
#define pud_pfn(pud) ((__pud_to_phys(pud) & PUD_MASK) >> PAGE_SHIFT)
diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c
index 3ff7ebb262d2..5b8163537bc2 100644
--- a/virt/kvm/arm/mmu.c
+++ b/virt/kvm/arm/mmu.c
@@ -115,6 +115,25 @@ static void stage2_dissolve_pmd(struct kvm *kvm, phys_addr_t addr, pmd_t *pmd)
put_page(virt_to_page(pmd));
}
+/**
+ * stage2_dissolve_pud() - clear and flush huge PUD entry
+ * @kvm: pointer to kvm structure.
+ * @addr: IPA
+ * @pud: pud pointer for IPA
+ *
+ * Function clears a PUD entry, flushes addr 1st and 2nd stage TLBs. Marks all
+ * pages in the range dirty.
+ */
+static void stage2_dissolve_pud(struct kvm *kvm, phys_addr_t addr, pud_t *pudp)
+{
+ if (!stage2_pud_huge(kvm, *pudp))
+ return;
+
+ stage2_pud_clear(kvm, pudp);
+ kvm_tlb_flush_vmid_ipa(kvm, addr);
+ put_page(virt_to_page(pudp));
+}
+
static int mmu_topup_memory_cache(struct kvm_mmu_memory_cache *cache,
int min, int max)
{
@@ -1022,7 +1041,7 @@ static pmd_t *stage2_get_pmd(struct kvm *kvm, struct kvm_mmu_memory_cache *cache
pmd_t *pmd;
pud = stage2_get_pud(kvm, cache, addr);
- if (!pud)
+ if (!pud || stage2_pud_huge(kvm, *pud))
return NULL;
if (stage2_pud_none(kvm, *pud)) {
@@ -1083,6 +1102,36 @@ static int stage2_set_pmd_huge(struct kvm *kvm, struct kvm_mmu_memory_cache
return 0;
}
+static int stage2_set_pud_huge(struct kvm *kvm, struct kvm_mmu_memory_cache *cache,
+ phys_addr_t addr, const pud_t *new_pudp)
+{
+ pud_t *pudp, old_pud;
+
+ pudp = stage2_get_pud(kvm, cache, addr);
+ VM_BUG_ON(!pudp);
+
+ old_pud = *pudp;
+
+ /*
+ * A large number of vcpus faulting on the same stage 2 entry,
+ * can lead to a refault due to the
+ * stage2_pud_clear()/tlb_flush(). Skip updating the page
+ * tables if there is no change.
+ */
+ if (pud_val(old_pud) == pud_val(*new_pudp))
+ return 0;
+
+ if (stage2_pud_present(kvm, old_pud)) {
+ stage2_pud_clear(kvm, pudp);
+ kvm_tlb_flush_vmid_ipa(kvm, addr);
+ } else {
+ get_page(virt_to_page(pudp));
+ }
+
+ kvm_set_pud(pudp, *new_pudp);
+ return 0;
+}
+
/*
* stage2_get_leaf_entry - walk the stage2 VM page tables and return
* true if a valid and present leaf-entry is found. A pointer to the
@@ -1149,6 +1198,7 @@ static int stage2_set_pte(struct kvm *kvm, struct kvm_mmu_memory_cache *cache,
phys_addr_t addr, const pte_t *new_pte,
unsigned long flags)
{
+ pud_t *pud;
pmd_t *pmd;
pte_t *pte, old_pte;
bool iomap = flags & KVM_S2PTE_FLAG_IS_IOMAP;
@@ -1157,7 +1207,31 @@ static int stage2_set_pte(struct kvm *kvm, struct kvm_mmu_memory_cache *cache,
VM_BUG_ON(logging_active && !cache);
/* Create stage-2 page table mapping - Levels 0 and 1 */
- pmd = stage2_get_pmd(kvm, cache, addr);
+ pud = stage2_get_pud(kvm, cache, addr);
+ if (!pud) {
+ /*
+ * Ignore calls from kvm_set_spte_hva for unallocated
+ * address ranges.
+ */
+ return 0;
+ }
+
+ /*
+ * While dirty page logging - dissolve huge PUD, then continue
+ * on to allocate page.
+ */
+ if (logging_active)
+ stage2_dissolve_pud(kvm, addr, pud);
+
+ if (stage2_pud_none(kvm, *pud)) {
+ if (!cache)
+ return 0; /* ignore calls from kvm_set_spte_hva */
+ pmd = mmu_memory_cache_alloc(cache);
+ stage2_pud_populate(kvm, pud, pmd);
+ get_page(virt_to_page(pud));
+ }
+
+ pmd = stage2_pmd_offset(kvm, pud, addr);
if (!pmd) {
/*
* Ignore calls from kvm_set_spte_hva for unallocated
@@ -1563,9 +1637,12 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
}
vma_pagesize = vma_kernel_pagesize(vma);
- if (vma_pagesize == PMD_SIZE && !logging_active) {
+ if ((vma_pagesize == PMD_SIZE || vma_pagesize == PUD_SIZE) &&
+ !logging_active) {
+ struct hstate *h = hstate_vma(vma);
+
hugetlb = true;
- gfn = (fault_ipa & PMD_MASK) >> PAGE_SHIFT;
+ gfn = (fault_ipa & huge_page_mask(h)) >> PAGE_SHIFT;
} else {
/*
* Fallback to PTE if it's not one of the Stage 2
@@ -1669,7 +1746,28 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
needs_exec = exec_fault ||
(fault_status == FSC_PERM && stage2_is_exec(kvm, fault_ipa));
- if (hugetlb && vma_pagesize == PMD_SIZE) {
+ if (hugetlb && vma_pagesize == PUD_SIZE) {
+ /*
+ * Assuming that PUD level always exists at Stage 2 -
+ * this is true for 4k pages with 40 bits IPA
+ * currently supported.
+ *
+ * When using 64k pages, 40bits of IPA results in
+ * using only 2-levels at Stage 2. Overlooking this
+ * problem for now as a PUD hugepage with 64k pages is
+ * too big (4TB) to be practical.
+ */
+ pud_t new_pud = kvm_pfn_pud(pfn, mem_type);
+
+ new_pud = kvm_pud_mkhuge(new_pud);
+ if (writable)
+ new_pud = kvm_s2pud_mkwrite(new_pud);
+
+ if (needs_exec)
+ new_pud = kvm_s2pud_mkexec(new_pud);
+
+ ret = stage2_set_pud_huge(kvm, memcache, fault_ipa, &new_pud);
+ } else if (hugetlb && vma_pagesize == PMD_SIZE) {
pmd_t new_pmd = kvm_pfn_pmd(pfn, mem_type);
new_pmd = kvm_pmd_mkhuge(new_pmd);
--
2.18.0
In preparation for creating PUD hugepages at stage 2, add support for
detecting execute permissions on PUD page table entries. Faults due to
lack of execute permissions on page table entries is used to perform
i-cache invalidation on first execute.
Provide trivial implementations of arm32 helpers to allow sharing of
code.
Signed-off-by: Punit Agrawal <[email protected]>
Reviewed-by: Suzuki K Poulose <[email protected]>
Cc: Christoffer Dall <[email protected]>
Cc: Marc Zyngier <[email protected]>
Cc: Russell King <[email protected]>
Cc: Catalin Marinas <[email protected]>
Cc: Will Deacon <[email protected]>
---
arch/arm/include/asm/kvm_mmu.h | 6 +++
arch/arm64/include/asm/kvm_mmu.h | 5 +++
arch/arm64/include/asm/pgtable-hwdef.h | 2 +
virt/kvm/arm/mmu.c | 53 +++++++++++++++++++++++---
4 files changed, 61 insertions(+), 5 deletions(-)
diff --git a/arch/arm/include/asm/kvm_mmu.h b/arch/arm/include/asm/kvm_mmu.h
index 9ec09f4cc284..26a2ab05b3f6 100644
--- a/arch/arm/include/asm/kvm_mmu.h
+++ b/arch/arm/include/asm/kvm_mmu.h
@@ -102,6 +102,12 @@ static inline bool kvm_s2pud_readonly(pud_t *pud)
return false;
}
+static inline bool kvm_s2pud_exec(pud_t *pud)
+{
+ BUG();
+ return false;
+}
+
static inline pte_t kvm_s2pte_mkwrite(pte_t pte)
{
pte_val(pte) |= L_PTE_S2_RDWR;
diff --git a/arch/arm64/include/asm/kvm_mmu.h b/arch/arm64/include/asm/kvm_mmu.h
index 3cc342177474..c06ef3be8ca9 100644
--- a/arch/arm64/include/asm/kvm_mmu.h
+++ b/arch/arm64/include/asm/kvm_mmu.h
@@ -261,6 +261,11 @@ static inline bool kvm_s2pud_readonly(pud_t *pudp)
return kvm_s2pte_readonly((pte_t *)pudp);
}
+static inline bool kvm_s2pud_exec(pud_t *pudp)
+{
+ return !(READ_ONCE(pud_val(*pudp)) & PUD_S2_XN);
+}
+
#define hyp_pte_table_empty(ptep) kvm_page_empty(ptep)
#ifdef __PAGETABLE_PMD_FOLDED
diff --git a/arch/arm64/include/asm/pgtable-hwdef.h b/arch/arm64/include/asm/pgtable-hwdef.h
index fd208eac9f2a..10ae592b78b8 100644
--- a/arch/arm64/include/asm/pgtable-hwdef.h
+++ b/arch/arm64/include/asm/pgtable-hwdef.h
@@ -193,6 +193,8 @@
#define PMD_S2_RDWR (_AT(pmdval_t, 3) << 6) /* HAP[2:1] */
#define PMD_S2_XN (_AT(pmdval_t, 2) << 53) /* XN[1:0] */
+#define PUD_S2_XN (_AT(pudval_t, 2) << 53) /* XN[1:0] */
+
/*
* Memory Attribute override for Stage-2 (MemAttr[3:0])
*/
diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c
index 9c48f2ca6583..5fd1eae7d964 100644
--- a/virt/kvm/arm/mmu.c
+++ b/virt/kvm/arm/mmu.c
@@ -1083,23 +1083,66 @@ static int stage2_set_pmd_huge(struct kvm *kvm, struct kvm_mmu_memory_cache
return 0;
}
-static bool stage2_is_exec(struct kvm *kvm, phys_addr_t addr)
+/*
+ * stage2_get_leaf_entry - walk the stage2 VM page tables and return
+ * true if a valid and present leaf-entry is found. A pointer to the
+ * leaf-entry is returned in the appropriate level variable - pudpp,
+ * pmdpp, ptepp.
+ */
+static bool stage2_get_leaf_entry(struct kvm *kvm, phys_addr_t addr,
+ pud_t **pudpp, pmd_t **pmdpp, pte_t **ptepp)
{
+ pud_t *pudp;
pmd_t *pmdp;
pte_t *ptep;
- pmdp = stage2_get_pmd(kvm, NULL, addr);
+ *pudpp = NULL;
+ *pmdpp = NULL;
+ *ptepp = NULL;
+
+ pudp = stage2_get_pud(kvm, NULL, addr);
+ if (!pudp || stage2_pud_none(kvm, *pudp) || !stage2_pud_present(kvm, *pudp))
+ return false;
+
+ if (stage2_pud_huge(kvm, *pudp)) {
+ *pudpp = pudp;
+ return true;
+ }
+
+ pmdp = stage2_pmd_offset(kvm, pudp, addr);
if (!pmdp || pmd_none(*pmdp) || !pmd_present(*pmdp))
return false;
- if (pmd_thp_or_huge(*pmdp))
- return kvm_s2pmd_exec(pmdp);
+ if (pmd_thp_or_huge(*pmdp)) {
+ *pmdpp = pmdp;
+ return true;
+ }
ptep = pte_offset_kernel(pmdp, addr);
if (!ptep || pte_none(*ptep) || !pte_present(*ptep))
return false;
- return kvm_s2pte_exec(ptep);
+ *ptepp = ptep;
+ return true;
+}
+
+static bool stage2_is_exec(struct kvm *kvm, phys_addr_t addr)
+{
+ pud_t *pudp;
+ pmd_t *pmdp;
+ pte_t *ptep;
+ bool found;
+
+ found = stage2_get_leaf_entry(kvm, addr, &pudp, &pmdp, &ptep);
+ if (!found)
+ return false;
+
+ if (pudp)
+ return kvm_s2pud_exec(pudp);
+ else if (pmdp)
+ return kvm_s2pmd_exec(pmdp);
+ else
+ return kvm_s2pte_exec(ptep);
}
static int stage2_set_pte(struct kvm *kvm, struct kvm_mmu_memory_cache *cache,
--
2.18.0
In preparation for creating larger hugepages at Stage 2, add support
to the age handling notifiers for PUD hugepages when encountered.
Provide trivial helpers for arm32 to allow sharing code.
Signed-off-by: Punit Agrawal <[email protected]>
Reviewed-by: Suzuki K Poulose <[email protected]>
Cc: Christoffer Dall <[email protected]>
Cc: Marc Zyngier <[email protected]>
Cc: Russell King <[email protected]>
Cc: Catalin Marinas <[email protected]>
Cc: Will Deacon <[email protected]>
---
arch/arm/include/asm/kvm_mmu.h | 6 +++++
arch/arm64/include/asm/kvm_mmu.h | 5 ++++
arch/arm64/include/asm/pgtable.h | 1 +
virt/kvm/arm/mmu.c | 39 ++++++++++++++++----------------
4 files changed, 32 insertions(+), 19 deletions(-)
diff --git a/arch/arm/include/asm/kvm_mmu.h b/arch/arm/include/asm/kvm_mmu.h
index 95b34aad0dc8..a42b9505c9a7 100644
--- a/arch/arm/include/asm/kvm_mmu.h
+++ b/arch/arm/include/asm/kvm_mmu.h
@@ -117,6 +117,12 @@ static inline pud_t kvm_s2pud_mkyoung(pud_t pud)
return pud;
}
+static inline bool kvm_s2pud_young(pud_t pud)
+{
+ BUG();
+ return false;
+}
+
static inline pte_t kvm_s2pte_mkwrite(pte_t pte)
{
pte_val(pte) |= L_PTE_S2_RDWR;
diff --git a/arch/arm64/include/asm/kvm_mmu.h b/arch/arm64/include/asm/kvm_mmu.h
index b93e5167728f..3baf72705dcc 100644
--- a/arch/arm64/include/asm/kvm_mmu.h
+++ b/arch/arm64/include/asm/kvm_mmu.h
@@ -273,6 +273,11 @@ static inline pud_t kvm_s2pud_mkyoung(pud_t pud)
return pud_mkyoung(pud);
}
+static inline bool kvm_s2pud_young(pud_t pud)
+{
+ return pud_young(pud);
+}
+
#define hyp_pte_table_empty(ptep) kvm_page_empty(ptep)
#ifdef __PAGETABLE_PMD_FOLDED
diff --git a/arch/arm64/include/asm/pgtable.h b/arch/arm64/include/asm/pgtable.h
index a64a5c35beb1..4d9476e420d9 100644
--- a/arch/arm64/include/asm/pgtable.h
+++ b/arch/arm64/include/asm/pgtable.h
@@ -385,6 +385,7 @@ static inline int pmd_protnone(pmd_t pmd)
#define pfn_pmd(pfn,prot) __pmd(__phys_to_pmd_val((phys_addr_t)(pfn) << PAGE_SHIFT) | pgprot_val(prot))
#define mk_pmd(page,prot) pfn_pmd(page_to_pfn(page),prot)
+#define pud_young(pud) pte_young(pud_pte(pud))
#define pud_mkyoung(pud) pte_pud(pte_mkyoung(pud_pte(pud)))
#define pud_write(pud) pte_write(pud_pte(pud))
diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c
index e2487e5fff37..3ff7ebb262d2 100644
--- a/virt/kvm/arm/mmu.c
+++ b/virt/kvm/arm/mmu.c
@@ -1225,6 +1225,11 @@ static int stage2_pmdp_test_and_clear_young(pmd_t *pmd)
return stage2_ptep_test_and_clear_young((pte_t *)pmd);
}
+static int stage2_pudp_test_and_clear_young(pud_t *pud)
+{
+ return stage2_ptep_test_and_clear_young((pte_t *)pud);
+}
+
/**
* kvm_phys_addr_ioremap - map a device range to guest IPA
*
@@ -1940,42 +1945,38 @@ void kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte)
static int kvm_age_hva_handler(struct kvm *kvm, gpa_t gpa, u64 size, void *data)
{
+ pud_t *pud;
pmd_t *pmd;
pte_t *pte;
- WARN_ON(size != PAGE_SIZE && size != PMD_SIZE);
- pmd = stage2_get_pmd(kvm, NULL, gpa);
- if (!pmd || pmd_none(*pmd)) /* Nothing there */
+ WARN_ON(size != PAGE_SIZE && size != PMD_SIZE && size != PUD_SIZE);
+ if (!stage2_get_leaf_entry(kvm, gpa, &pud, &pmd, &pte))
return 0;
- if (pmd_thp_or_huge(*pmd)) /* THP, HugeTLB */
+ if (pud)
+ return stage2_pudp_test_and_clear_young(pud);
+ else if (pmd)
return stage2_pmdp_test_and_clear_young(pmd);
-
- pte = pte_offset_kernel(pmd, gpa);
- if (pte_none(*pte))
- return 0;
-
- return stage2_ptep_test_and_clear_young(pte);
+ else
+ return stage2_ptep_test_and_clear_young(pte);
}
static int kvm_test_age_hva_handler(struct kvm *kvm, gpa_t gpa, u64 size, void *data)
{
+ pud_t *pud;
pmd_t *pmd;
pte_t *pte;
- WARN_ON(size != PAGE_SIZE && size != PMD_SIZE);
- pmd = stage2_get_pmd(kvm, NULL, gpa);
- if (!pmd || pmd_none(*pmd)) /* Nothing there */
+ WARN_ON(size != PAGE_SIZE && size != PMD_SIZE && size != PUD_SIZE);
+ if (!stage2_get_leaf_entry(kvm, gpa, &pud, &pmd, &pte))
return 0;
- if (pmd_thp_or_huge(*pmd)) /* THP, HugeTLB */
+ if (pud)
+ return kvm_s2pud_young(*pud);
+ else if (pmd)
return pmd_young(*pmd);
-
- pte = pte_offset_kernel(pmd, gpa);
- if (!pte_none(*pte)) /* Just a page... */
+ else
return pte_young(*pte);
-
- return 0;
}
int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end)
--
2.18.0
Hi Punit,
On 09/24/2018 06:45 PM, Punit Agrawal wrote:
> PageTransCompoundMap() returns true for hugetlbfs and THP
> hugepages. This behaviour incorrectly leads to stage 2 faults for
> unsupported hugepage sizes (e.g., 64K hugepage with 4K pages) to be
> treated as THP faults.
>
> Tighten the check to filter out hugetlbfs pages. This also leads to
> consistently mapping all unsupported hugepage sizes as PTE level
> entries at stage 2.
>
> Signed-off-by: Punit Agrawal <[email protected]>
> Cc: Christoffer Dall <[email protected]>
> Cc: Marc Zyngier <[email protected]>
> Cc: Suzuki Poulose <[email protected]>
> Cc: [email protected] # v4.13+
> ---
> virt/kvm/arm/mmu.c | 8 +++++++-
> 1 file changed, 7 insertions(+), 1 deletion(-)
>
> diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c
> index 7e477b3cae5b..c23a1b323aad 100644
> --- a/virt/kvm/arm/mmu.c
> +++ b/virt/kvm/arm/mmu.c
> @@ -1231,8 +1231,14 @@ static bool transparent_hugepage_adjust(kvm_pfn_t *pfnp, phys_addr_t *ipap)
> {
> kvm_pfn_t pfn = *pfnp;
> gfn_t gfn = *ipap >> PAGE_SHIFT;
> + struct page *page = pfn_to_page(pfn);
>
> - if (PageTransCompoundMap(pfn_to_page(pfn))) {
> + /*
> + * PageTransCompoungMap() returns true for THP and
> + * hugetlbfs. Make sure the adjustment is done only for THP
> + * pages.
> + */
> + if (!PageHuge(page) && PageTransCompoundMap(page)) {
Reviewed-by: Suzuki K Poulose <[email protected]>
Hi Punit,
On 09/24/2018 06:45 PM, Punit Agrawal wrote:
> KVM only supports PMD hugepages at stage 2. Now that the various page
> handling routines are updated, extend the stage 2 fault handling to
> map in PUD hugepages.
>
> Addition of PUD hugepage support enables additional page sizes (e.g.,
> 1G with 4K granule) which can be useful on cores that support mapping
> larger block sizes in the TLB entries.
>
> Signed-off-by: Punit Agrawal <[email protected]>
> Cc: Christoffer Dall <[email protected]>
> Cc: Marc Zyngier <[email protected]>
> Cc: Russell King <[email protected]>
> Cc: Catalin Marinas <[email protected]>
> Cc: Will Deacon <[email protected]>
>
> diff --git a/arch/arm/include/asm/kvm_mmu.h b/arch/arm/include/asm/kvm_mmu.h
> index a42b9505c9a7..a8e86b926ee0 100644
> --- a/arch/arm/include/asm/kvm_mmu.h
> +++ b/arch/arm/include/asm/kvm_mmu.h
> @@ -84,11 +84,14 @@ void kvm_clear_hyp_idmap(void);
>
> #define kvm_pfn_pte(pfn, prot) pfn_pte(pfn, prot)
> #define kvm_pfn_pmd(pfn, prot) pfn_pmd(pfn, prot)
> +#define kvm_pfn_pud(pfn, prot) (__pud(0))
>
> #define kvm_pud_pfn(pud) ({ BUG(); 0; })
>
>
> #define kvm_pmd_mkhuge(pmd) pmd_mkhuge(pmd)
> +/* No support for pud hugepages */
> +#define kvm_pud_mkhuge(pud) (pud)
>
shouldn't this be BUG() like other PUD huge helpers for arm32 ?
> /*
> * The following kvm_*pud*() functions are provided strictly to allow
> @@ -105,6 +108,23 @@ static inline bool kvm_s2pud_readonly(pud_t *pud)
> return false;
> }
>
> +static inline void kvm_set_pud(pud_t *pud, pud_t new_pud)
> +{
> + BUG();
> +}
> +
> +static inline pud_t kvm_s2pud_mkwrite(pud_t pud)
> +{
> + BUG();
> + return pud;
> +}
> +
> +static inline pud_t kvm_s2pud_mkexec(pud_t pud)
> +{
> + BUG();
> + return pud;
> +}
> +
> diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c
> index 3ff7ebb262d2..5b8163537bc2 100644
> --- a/virt/kvm/arm/mmu.c
> +++ b/virt/kvm/arm/mmu.c
...
> @@ -1669,7 +1746,28 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
> needs_exec = exec_fault ||
> (fault_status == FSC_PERM && stage2_is_exec(kvm, fault_ipa));
>
> - if (hugetlb && vma_pagesize == PMD_SIZE) {
> + if (hugetlb && vma_pagesize == PUD_SIZE) {
> + /*
> + * Assuming that PUD level always exists at Stage 2 -
> + * this is true for 4k pages with 40 bits IPA
> + * currently supported.
> + *
> + * When using 64k pages, 40bits of IPA results in
> + * using only 2-levels at Stage 2. Overlooking this
> + * problem for now as a PUD hugepage with 64k pages is
> + * too big (4TB) to be practical.
> + */
> + pud_t new_pud = kvm_pfn_pud(pfn, mem_type);
Is this based on the Dynamic IPA series ? The cover letter seems
to suggest that it is. But I don't see the check to make sure we have
stage2 PUD level here before we go ahead and try PUD huge page at
stage2. Also the comment above seems outdated in that case.
> +
> + new_pud = kvm_pud_mkhuge(new_pud);
> + if (writable)
> + new_pud = kvm_s2pud_mkwrite(new_pud);
> +
> + if (needs_exec)
> + new_pud = kvm_s2pud_mkexec(new_pud);
> +
> + ret = stage2_set_pud_huge(kvm, memcache, fault_ipa, &new_pud);
> + } else if (hugetlb && vma_pagesize == PMD_SIZE) {
> pmd_t new_pmd = kvm_pfn_pmd(pfn, mem_type);
>
> new_pmd = kvm_pmd_mkhuge(new_pmd);
>
Suzuki
Suzuki K Poulose <[email protected]> writes:
> Hi Punit,
>
>
> On 09/24/2018 06:45 PM, Punit Agrawal wrote:
>> KVM only supports PMD hugepages at stage 2. Now that the various page
>> handling routines are updated, extend the stage 2 fault handling to
>> map in PUD hugepages.
>>
>> Addition of PUD hugepage support enables additional page sizes (e.g.,
>> 1G with 4K granule) which can be useful on cores that support mapping
>> larger block sizes in the TLB entries.
>>
>> Signed-off-by: Punit Agrawal <[email protected]>
>> Cc: Christoffer Dall <[email protected]>
>> Cc: Marc Zyngier <[email protected]>
>> Cc: Russell King <[email protected]>
>> Cc: Catalin Marinas <[email protected]>
>> Cc: Will Deacon <[email protected]>
>
>
>>
>> diff --git a/arch/arm/include/asm/kvm_mmu.h b/arch/arm/include/asm/kvm_mmu.h
>> index a42b9505c9a7..a8e86b926ee0 100644
>> --- a/arch/arm/include/asm/kvm_mmu.h
>> +++ b/arch/arm/include/asm/kvm_mmu.h
>> @@ -84,11 +84,14 @@ void kvm_clear_hyp_idmap(void);
>> #define kvm_pfn_pte(pfn, prot) pfn_pte(pfn, prot)
>> #define kvm_pfn_pmd(pfn, prot) pfn_pmd(pfn, prot)
>> +#define kvm_pfn_pud(pfn, prot) (__pud(0))
>> #define kvm_pud_pfn(pud) ({ BUG(); 0; })
>> #define kvm_pmd_mkhuge(pmd) pmd_mkhuge(pmd)
>> +/* No support for pud hugepages */
>> +#define kvm_pud_mkhuge(pud) (pud)
>>
>
> shouldn't this be BUG() like other PUD huge helpers for arm32 ?
>
>> /*
>> * The following kvm_*pud*() functions are provided strictly to allow
>> @@ -105,6 +108,23 @@ static inline bool kvm_s2pud_readonly(pud_t *pud)
>> return false;
>> }
>> +static inline void kvm_set_pud(pud_t *pud, pud_t new_pud)
>> +{
>> + BUG();
>> +}
>> +
>> +static inline pud_t kvm_s2pud_mkwrite(pud_t pud)
>> +{
>> + BUG();
>> + return pud;
>> +}
>> +
>> +static inline pud_t kvm_s2pud_mkexec(pud_t pud)
>> +{
>> + BUG();
>> + return pud;
>> +}
>> +
>
>
>> diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c
>> index 3ff7ebb262d2..5b8163537bc2 100644
>> --- a/virt/kvm/arm/mmu.c
>> +++ b/virt/kvm/arm/mmu.c
>
> ...
>
>
>> @@ -1669,7 +1746,28 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
>> needs_exec = exec_fault ||
>> (fault_status == FSC_PERM && stage2_is_exec(kvm, fault_ipa));
>> - if (hugetlb && vma_pagesize == PMD_SIZE) {
>> + if (hugetlb && vma_pagesize == PUD_SIZE) {
>> + /*
>> + * Assuming that PUD level always exists at Stage 2 -
>> + * this is true for 4k pages with 40 bits IPA
>> + * currently supported.
>> + *
>> + * When using 64k pages, 40bits of IPA results in
>> + * using only 2-levels at Stage 2. Overlooking this
>> + * problem for now as a PUD hugepage with 64k pages is
>> + * too big (4TB) to be practical.
>> + */
>> + pud_t new_pud = kvm_pfn_pud(pfn, mem_type);
>
> Is this based on the Dynamic IPA series ? The cover letter seems
> to suggest that it is. But I don't see the check to make sure we have
> stage2 PUD level here before we go ahead and try PUD huge page at
> stage2. Also the comment above seems outdated in that case.
It is indeed based on the Dynamic IPA series but I seem to have lost the
actual changes introducing the checks for PUD level. Let me fix that up
and post an update.
Sorry for the noise.
Punit
>
>> +
>> + new_pud = kvm_pud_mkhuge(new_pud);
>> + if (writable)
>> + new_pud = kvm_s2pud_mkwrite(new_pud);
>> +
>> + if (needs_exec)
>> + new_pud = kvm_s2pud_mkexec(new_pud);
>> +
>> + ret = stage2_set_pud_huge(kvm, memcache, fault_ipa, &new_pud);
>> + } else if (hugetlb && vma_pagesize == PMD_SIZE) {
>> pmd_t new_pmd = kvm_pfn_pmd(pfn, mem_type);
>> new_pmd = kvm_pmd_mkhuge(new_pmd);
>>
>
>
> Suzuki
KVM only supports PMD hugepages at stage 2. Now that the various page
handling routines are updated, extend the stage 2 fault handling to
map in PUD hugepages.
Addition of PUD hugepage support enables additional page sizes (e.g.,
1G with 4K granule) which can be useful on cores that support mapping
larger block sizes in the TLB entries.
Signed-off-by: Punit Agrawal <[email protected]>
Cc: Christoffer Dall <[email protected]>
Cc: Marc Zyngier <[email protected]>
Cc: Russell King <[email protected]>
Cc: Catalin Marinas <[email protected]>
Cc: Will Deacon <[email protected]>
---
v7 -> v7.1
* Added arm helper kvm_stage2_has_pud()
* Added check for PUD level present at stage 2
* Dropped redundant comment
* Fixed up kvm_pud_mkhuge() to complain on arm
arch/arm/include/asm/kvm_mmu.h | 20 +++++
arch/arm/include/asm/stage2_pgtable.h | 5 ++
arch/arm64/include/asm/kvm_mmu.h | 16 ++++
arch/arm64/include/asm/pgtable-hwdef.h | 2 +
arch/arm64/include/asm/pgtable.h | 2 +
virt/kvm/arm/mmu.c | 106 +++++++++++++++++++++++--
6 files changed, 145 insertions(+), 6 deletions(-)
diff --git a/arch/arm/include/asm/kvm_mmu.h b/arch/arm/include/asm/kvm_mmu.h
index a42b9505c9a7..da5f078ae68c 100644
--- a/arch/arm/include/asm/kvm_mmu.h
+++ b/arch/arm/include/asm/kvm_mmu.h
@@ -84,11 +84,14 @@ void kvm_clear_hyp_idmap(void);
#define kvm_pfn_pte(pfn, prot) pfn_pte(pfn, prot)
#define kvm_pfn_pmd(pfn, prot) pfn_pmd(pfn, prot)
+#define kvm_pfn_pud(pfn, prot) (__pud(0))
#define kvm_pud_pfn(pud) ({ BUG(); 0; })
#define kvm_pmd_mkhuge(pmd) pmd_mkhuge(pmd)
+/* No support for pud hugepages */
+#define kvm_pud_mkhuge(pud) ( {BUG(); pud; })
/*
* The following kvm_*pud*() functions are provided strictly to allow
@@ -105,6 +108,23 @@ static inline bool kvm_s2pud_readonly(pud_t *pud)
return false;
}
+static inline void kvm_set_pud(pud_t *pud, pud_t new_pud)
+{
+ BUG();
+}
+
+static inline pud_t kvm_s2pud_mkwrite(pud_t pud)
+{
+ BUG();
+ return pud;
+}
+
+static inline pud_t kvm_s2pud_mkexec(pud_t pud)
+{
+ BUG();
+ return pud;
+}
+
static inline bool kvm_s2pud_exec(pud_t *pud)
{
BUG();
diff --git a/arch/arm/include/asm/stage2_pgtable.h b/arch/arm/include/asm/stage2_pgtable.h
index f6a7ea805232..ec1567d9eb4b 100644
--- a/arch/arm/include/asm/stage2_pgtable.h
+++ b/arch/arm/include/asm/stage2_pgtable.h
@@ -68,4 +68,9 @@ stage2_pmd_addr_end(struct kvm *kvm, phys_addr_t addr, phys_addr_t end)
#define stage2_pmd_table_empty(kvm, pmdp) kvm_page_empty(pmdp)
#define stage2_pud_table_empty(kvm, pudp) false
+static inline bool kvm_stage2_has_pud(struct kvm *kvm)
+{
+ return KVM_VTCR_SL0 == VTCR_SL_L1;
+}
+
#endif /* __ARM_S2_PGTABLE_H_ */
diff --git a/arch/arm64/include/asm/kvm_mmu.h b/arch/arm64/include/asm/kvm_mmu.h
index 3baf72705dcc..b4e9c2cceecb 100644
--- a/arch/arm64/include/asm/kvm_mmu.h
+++ b/arch/arm64/include/asm/kvm_mmu.h
@@ -184,12 +184,16 @@ void kvm_clear_hyp_idmap(void);
#define kvm_mk_pgd(pudp) \
__pgd(__phys_to_pgd_val(__pa(pudp)) | PUD_TYPE_TABLE)
+#define kvm_set_pud(pudp, pud) set_pud(pudp, pud)
+
#define kvm_pfn_pte(pfn, prot) pfn_pte(pfn, prot)
#define kvm_pfn_pmd(pfn, prot) pfn_pmd(pfn, prot)
+#define kvm_pfn_pud(pfn, prot) pfn_pud(pfn, prot)
#define kvm_pud_pfn(pud) pud_pfn(pud)
#define kvm_pmd_mkhuge(pmd) pmd_mkhuge(pmd)
+#define kvm_pud_mkhuge(pud) pud_mkhuge(pud)
static inline pte_t kvm_s2pte_mkwrite(pte_t pte)
{
@@ -203,6 +207,12 @@ static inline pmd_t kvm_s2pmd_mkwrite(pmd_t pmd)
return pmd;
}
+static inline pud_t kvm_s2pud_mkwrite(pud_t pud)
+{
+ pud_val(pud) |= PUD_S2_RDWR;
+ return pud;
+}
+
static inline pte_t kvm_s2pte_mkexec(pte_t pte)
{
pte_val(pte) &= ~PTE_S2_XN;
@@ -215,6 +225,12 @@ static inline pmd_t kvm_s2pmd_mkexec(pmd_t pmd)
return pmd;
}
+static inline pud_t kvm_s2pud_mkexec(pud_t pud)
+{
+ pud_val(pud) &= ~PUD_S2_XN;
+ return pud;
+}
+
static inline void kvm_set_s2pte_readonly(pte_t *ptep)
{
pteval_t old_pteval, pteval;
diff --git a/arch/arm64/include/asm/pgtable-hwdef.h b/arch/arm64/include/asm/pgtable-hwdef.h
index 10ae592b78b8..e327665e94d1 100644
--- a/arch/arm64/include/asm/pgtable-hwdef.h
+++ b/arch/arm64/include/asm/pgtable-hwdef.h
@@ -193,6 +193,8 @@
#define PMD_S2_RDWR (_AT(pmdval_t, 3) << 6) /* HAP[2:1] */
#define PMD_S2_XN (_AT(pmdval_t, 2) << 53) /* XN[1:0] */
+#define PUD_S2_RDONLY (_AT(pudval_t, 1) << 6) /* HAP[2:1] */
+#define PUD_S2_RDWR (_AT(pudval_t, 3) << 6) /* HAP[2:1] */
#define PUD_S2_XN (_AT(pudval_t, 2) << 53) /* XN[1:0] */
/*
diff --git a/arch/arm64/include/asm/pgtable.h b/arch/arm64/include/asm/pgtable.h
index 4d9476e420d9..0afc34f94ff5 100644
--- a/arch/arm64/include/asm/pgtable.h
+++ b/arch/arm64/include/asm/pgtable.h
@@ -389,6 +389,8 @@ static inline int pmd_protnone(pmd_t pmd)
#define pud_mkyoung(pud) pte_pud(pte_mkyoung(pud_pte(pud)))
#define pud_write(pud) pte_write(pud_pte(pud))
+#define pud_mkhuge(pud) (__pud(pud_val(pud) & ~PUD_TABLE_BIT))
+
#define __pud_to_phys(pud) __pte_to_phys(pud_pte(pud))
#define __phys_to_pud_val(phys) __phys_to_pte_val(phys)
#define pud_pfn(pud) ((__pud_to_phys(pud) & PUD_MASK) >> PAGE_SHIFT)
diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c
index 3ff7ebb262d2..b46cc4f714bc 100644
--- a/virt/kvm/arm/mmu.c
+++ b/virt/kvm/arm/mmu.c
@@ -115,6 +115,25 @@ static void stage2_dissolve_pmd(struct kvm *kvm, phys_addr_t addr, pmd_t *pmd)
put_page(virt_to_page(pmd));
}
+/**
+ * stage2_dissolve_pud() - clear and flush huge PUD entry
+ * @kvm: pointer to kvm structure.
+ * @addr: IPA
+ * @pud: pud pointer for IPA
+ *
+ * Function clears a PUD entry, flushes addr 1st and 2nd stage TLBs. Marks all
+ * pages in the range dirty.
+ */
+static void stage2_dissolve_pud(struct kvm *kvm, phys_addr_t addr, pud_t *pudp)
+{
+ if (!stage2_pud_huge(kvm, *pudp))
+ return;
+
+ stage2_pud_clear(kvm, pudp);
+ kvm_tlb_flush_vmid_ipa(kvm, addr);
+ put_page(virt_to_page(pudp));
+}
+
static int mmu_topup_memory_cache(struct kvm_mmu_memory_cache *cache,
int min, int max)
{
@@ -1022,7 +1041,7 @@ static pmd_t *stage2_get_pmd(struct kvm *kvm, struct kvm_mmu_memory_cache *cache
pmd_t *pmd;
pud = stage2_get_pud(kvm, cache, addr);
- if (!pud)
+ if (!pud || stage2_pud_huge(kvm, *pud))
return NULL;
if (stage2_pud_none(kvm, *pud)) {
@@ -1083,6 +1102,36 @@ static int stage2_set_pmd_huge(struct kvm *kvm, struct kvm_mmu_memory_cache
return 0;
}
+static int stage2_set_pud_huge(struct kvm *kvm, struct kvm_mmu_memory_cache *cache,
+ phys_addr_t addr, const pud_t *new_pudp)
+{
+ pud_t *pudp, old_pud;
+
+ pudp = stage2_get_pud(kvm, cache, addr);
+ VM_BUG_ON(!pudp);
+
+ old_pud = *pudp;
+
+ /*
+ * A large number of vcpus faulting on the same stage 2 entry,
+ * can lead to a refault due to the
+ * stage2_pud_clear()/tlb_flush(). Skip updating the page
+ * tables if there is no change.
+ */
+ if (pud_val(old_pud) == pud_val(*new_pudp))
+ return 0;
+
+ if (stage2_pud_present(kvm, old_pud)) {
+ stage2_pud_clear(kvm, pudp);
+ kvm_tlb_flush_vmid_ipa(kvm, addr);
+ } else {
+ get_page(virt_to_page(pudp));
+ }
+
+ kvm_set_pud(pudp, *new_pudp);
+ return 0;
+}
+
/*
* stage2_get_leaf_entry - walk the stage2 VM page tables and return
* true if a valid and present leaf-entry is found. A pointer to the
@@ -1149,6 +1198,7 @@ static int stage2_set_pte(struct kvm *kvm, struct kvm_mmu_memory_cache *cache,
phys_addr_t addr, const pte_t *new_pte,
unsigned long flags)
{
+ pud_t *pud;
pmd_t *pmd;
pte_t *pte, old_pte;
bool iomap = flags & KVM_S2PTE_FLAG_IS_IOMAP;
@@ -1157,7 +1207,31 @@ static int stage2_set_pte(struct kvm *kvm, struct kvm_mmu_memory_cache *cache,
VM_BUG_ON(logging_active && !cache);
/* Create stage-2 page table mapping - Levels 0 and 1 */
- pmd = stage2_get_pmd(kvm, cache, addr);
+ pud = stage2_get_pud(kvm, cache, addr);
+ if (!pud) {
+ /*
+ * Ignore calls from kvm_set_spte_hva for unallocated
+ * address ranges.
+ */
+ return 0;
+ }
+
+ /*
+ * While dirty page logging - dissolve huge PUD, then continue
+ * on to allocate page.
+ */
+ if (logging_active)
+ stage2_dissolve_pud(kvm, addr, pud);
+
+ if (stage2_pud_none(kvm, *pud)) {
+ if (!cache)
+ return 0; /* ignore calls from kvm_set_spte_hva */
+ pmd = mmu_memory_cache_alloc(cache);
+ stage2_pud_populate(kvm, pud, pmd);
+ get_page(virt_to_page(pud));
+ }
+
+ pmd = stage2_pmd_offset(kvm, pud, addr);
if (!pmd) {
/*
* Ignore calls from kvm_set_spte_hva for unallocated
@@ -1563,13 +1637,22 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
}
vma_pagesize = vma_kernel_pagesize(vma);
- if (vma_pagesize == PMD_SIZE && !logging_active) {
+ /*
+ * PUD level may not exist for a VM but PMD is guaranteed to
+ * exist.
+ */
+ if ((vma_pagesize == PMD_SIZE ||
+ (vma_pagesize == PUD_SIZE && kvm_stage2_has_pud(kvm))) &&
+ !logging_active) {
+ struct hstate *h = hstate_vma(vma);
+
hugetlb = true;
- gfn = (fault_ipa & PMD_MASK) >> PAGE_SHIFT;
+ gfn = (fault_ipa & huge_page_mask(h)) >> PAGE_SHIFT;
} else {
/*
* Fallback to PTE if it's not one of the Stage 2
- * supported hugepage sizes
+ * supported hugepage sizes or the corresponding level
+ * doesn't exist
*/
vma_pagesize = PAGE_SIZE;
@@ -1669,7 +1752,18 @@ static int user_mem_abort(struct kvm_vcpu *vcpu, phys_addr_t fault_ipa,
needs_exec = exec_fault ||
(fault_status == FSC_PERM && stage2_is_exec(kvm, fault_ipa));
- if (hugetlb && vma_pagesize == PMD_SIZE) {
+ if (hugetlb && vma_pagesize == PUD_SIZE) {
+ pud_t new_pud = kvm_pfn_pud(pfn, mem_type);
+
+ new_pud = kvm_pud_mkhuge(new_pud);
+ if (writable)
+ new_pud = kvm_s2pud_mkwrite(new_pud);
+
+ if (needs_exec)
+ new_pud = kvm_s2pud_mkexec(new_pud);
+
+ ret = stage2_set_pud_huge(kvm, memcache, fault_ipa, &new_pud);
+ } else if (hugetlb && vma_pagesize == PMD_SIZE) {
pmd_t new_pmd = kvm_pfn_pmd(pfn, mem_type);
new_pmd = kvm_pmd_mkhuge(new_pmd);
--
2.18.0
Punit Agrawal <[email protected]> writes:
> KVM only supports PMD hugepages at stage 2. Now that the various page
> handling routines are updated, extend the stage 2 fault handling to
> map in PUD hugepages.
>
> Addition of PUD hugepage support enables additional page sizes (e.g.,
> 1G with 4K granule) which can be useful on cores that support mapping
> larger block sizes in the TLB entries.
>
> Signed-off-by: Punit Agrawal <[email protected]>
> Cc: Christoffer Dall <[email protected]>
> Cc: Marc Zyngier <[email protected]>
> Cc: Russell King <[email protected]>
> Cc: Catalin Marinas <[email protected]>
> Cc: Will Deacon <[email protected]>
> ---
>
> v7 -> v7.1
>
> * Added arm helper kvm_stage2_has_pud()
> * Added check for PUD level present at stage 2
> * Dropped redundant comment
> * Fixed up kvm_pud_mkhuge() to complain on arm
>
> arch/arm/include/asm/kvm_mmu.h | 20 +++++
> arch/arm/include/asm/stage2_pgtable.h | 5 ++
> arch/arm64/include/asm/kvm_mmu.h | 16 ++++
> arch/arm64/include/asm/pgtable-hwdef.h | 2 +
> arch/arm64/include/asm/pgtable.h | 2 +
> virt/kvm/arm/mmu.c | 106 +++++++++++++++++++++++--
> 6 files changed, 145 insertions(+), 6 deletions(-)
>
[...]
> diff --git a/arch/arm/include/asm/stage2_pgtable.h b/arch/arm/include/asm/stage2_pgtable.h
> index f6a7ea805232..ec1567d9eb4b 100644
> --- a/arch/arm/include/asm/stage2_pgtable.h
> +++ b/arch/arm/include/asm/stage2_pgtable.h
> @@ -68,4 +68,9 @@ stage2_pmd_addr_end(struct kvm *kvm, phys_addr_t addr, phys_addr_t end)
> #define stage2_pmd_table_empty(kvm, pmdp) kvm_page_empty(pmdp)
> #define stage2_pud_table_empty(kvm, pudp) false
>
> +static inline bool kvm_stage2_has_pud(struct kvm *kvm)
> +{
> + return KVM_VTCR_SL0 == VTCR_SL_L1;
> +}
> +
Turns out this isn't quite the right check. On arm32, the maximum number
of supported levels is 3 with LPAE - effectively the helper should
always return false.
I've updated the check locally to key off of CONFIG_PGTABLE_LEVELS. I'll
post these patches later today.
Thanks,
Punit