Google Git
Sign in
fuchsia / fuchsia / refs/heads/main / . / zircon / kernel / vm / vm_mapping.cc
blob: 5d26ddf30b1bafd60a31205e1254d2de8ba93813 [file] [log] [blame]
// Copyright 2016 The Fuchsia Authors
//
// Use of this source code is governed by a MIT-style
// license that can be found in the LICENSE file or at
// https://opensource.org/licenses/MIT
#include <align.h>
#include <assert.h>
#include <inttypes.h>
#include <lib/counters.h>
#include <lib/fit/defer.h>
#include <trace.h>
#include <zircon/errors.h>
#include <zircon/types.h>
#include <fbl/alloc_checker.h>
#include <ktl/algorithm.h>
#include <ktl/iterator.h>
#include <ktl/move.h>
#include <vm/fault.h>
#include <vm/physmap.h>
#include <vm/vm.h>
#include <vm/vm_address_region.h>
#include <vm/vm_aspace.h>
#include <vm/vm_object.h>
#include <vm/vm_object_paged.h>
#include <vm/vm_object_physical.h>
#include "vm/vm_address_region.h"
#include "vm_priv.h"
#include <ktl/enforce.h>
#define LOCAL_TRACE VM_GLOBAL_TRACE(0)
namespace {
KCOUNTER(vm_mapping_attribution_queries, "vm.attributed_pages.mapping.queries")
KCOUNTER(vm_mapping_attribution_cache_hits, "vm.attributed_pages.mapping.cache_hits")
KCOUNTER(vm_mapping_attribution_cache_misses, "vm.attributed_pages.mapping.cache_misses")
KCOUNTER(vm_mappings_merged, "vm.aspace.mapping.merged_neighbors")
KCOUNTER(vm_mappings_protect_no_write, "vm.aspace.mapping.protect_without_write")
} // namespace
VmMapping::VmMapping(VmAddressRegion& parent, vaddr_t base, size_t size, uint32_t vmar_flags,
fbl::RefPtr<VmObject> vmo, uint64_t vmo_offset,
MappingProtectionRanges&& ranges, Mergeable mergeable)
: VmAddressRegionOrMapping(base, size, vmar_flags, parent.aspace_.get(), &parent, true),
mergeable_(mergeable),
object_(ktl::move(vmo)),
object_offset_(vmo_offset),
protection_ranges_(ktl::move(ranges)) {
LTRACEF("%p aspace %p base %#" PRIxPTR " size %#zx offset %#" PRIx64 "\n", this, aspace_.get(),
base_, size_, vmo_offset);
}
VmMapping::VmMapping(VmAddressRegion& parent, vaddr_t base, size_t size, uint32_t vmar_flags,
fbl::RefPtr<VmObject> vmo, uint64_t vmo_offset, uint arch_mmu_flags,
Mergeable mergeable)
: VmMapping(parent, base, size, vmar_flags, vmo, vmo_offset,
MappingProtectionRanges(arch_mmu_flags), mergeable) {}
VmMapping::~VmMapping() {
canary_.Assert();
LTRACEF("%p aspace %p base %#" PRIxPTR " size %#zx\n", this, aspace_.get(), base_, size_);
}
fbl::RefPtr<VmObject> VmMapping::vmo() const {
Guard<CriticalMutex> guard{lock()};
return vmo_locked();
}
VmMapping::AttributionCounts VmMapping::AllocatedPagesLocked() {
canary_.Assert();
if (state_ != LifeCycleState::ALIVE) {
return AttributionCounts{};
}
vm_mapping_attribution_queries.Add(1);
if (!object_->is_paged()) {
return object_->AttributedPagesInRange(object_offset_locked(), size_);
}
// If |object_| is a VmObjectPaged, check if the previously cached value still holds.
auto object_paged = static_cast<VmObjectPaged*>(object_.get());
uint64_t vmo_gen_count = object_paged->GetHierarchyGenerationCount();
uint64_t mapping_gen_count = GetMappingGenerationCountLocked();
// Return the cached page count if the mapping's generation count and the vmo's generation count
// have not changed.
if (cached_page_attribution_.mapping_generation_count == mapping_gen_count &&
cached_page_attribution_.vmo_generation_count == vmo_gen_count) {
vm_mapping_attribution_cache_hits.Add(1);
return cached_page_attribution_.page_counts;
}
vm_mapping_attribution_cache_misses.Add(1);
AttributionCounts page_counts =
object_paged->AttributedPagesInRange(object_offset_locked(), size_);
DEBUG_ASSERT(cached_page_attribution_.mapping_generation_count != mapping_gen_count ||
cached_page_attribution_.vmo_generation_count != vmo_gen_count);
cached_page_attribution_.mapping_generation_count = mapping_gen_count;
cached_page_attribution_.vmo_generation_count = vmo_gen_count;
cached_page_attribution_.page_counts = page_counts;
return page_counts;
}
void VmMapping::DumpLocked(uint depth, bool verbose) const {
canary_.Assert();
for (uint i = 0; i < depth; ++i) {
printf(" ");
}
char vmo_name[32];
object_->get_name(vmo_name, sizeof(vmo_name));
printf("map %p [%#" PRIxPTR " %#" PRIxPTR "] sz %#zx state %d mergeable %s\n", this, base_,
base_ + size_ - 1, size_, (int)state_, mergeable_ == Mergeable::YES ? "true" : "false");
EnumerateProtectionRangesLocked(base_, size_, [depth](vaddr_t base, size_t len, uint mmu_flags) {
for (uint i = 0; i < depth + 1; ++i) {
printf(" ");
}
printf(" [%#" PRIxPTR " %#" PRIxPTR "] mmufl %#x\n", base, base + len - 1, mmu_flags);
return ZX_ERR_NEXT;
});
for (uint i = 0; i < depth + 1; ++i) {
printf(" ");
}
AttributionCounts page_counts = object_->AttributedPagesInRange(object_offset_locked(), size_);
printf("vmo %p/k%" PRIu64 " off %#" PRIx64 " pages (%zu/%zu) ref %d '%s'\n", object_.get(),
object_->user_id(), object_offset_locked(), page_counts.uncompressed,
page_counts.compressed, ref_count_debug(), vmo_name);
if (verbose) {
object_->Dump(depth + 1, false);
}
}
zx_status_t VmMapping::Protect(vaddr_t base, size_t size, uint new_arch_mmu_flags) {
canary_.Assert();
LTRACEF("%p %#" PRIxPTR " %#x %#x\n", this, base_, flags_, new_arch_mmu_flags);
if (!IS_PAGE_ALIGNED(base)) {
return ZX_ERR_INVALID_ARGS;
}
size = ROUNDUP(size, PAGE_SIZE);
Guard<CriticalMutex> guard{lock()};
if (state_ != LifeCycleState::ALIVE) {
return ZX_ERR_BAD_STATE;
}
if (size == 0 || !is_in_range_locked(base, size)) {
return ZX_ERR_INVALID_ARGS;
}
// Do not allow changing caching.
if (new_arch_mmu_flags & ARCH_MMU_FLAG_CACHE_MASK) {
return ZX_ERR_INVALID_ARGS;
}
if (!is_valid_mapping_flags(new_arch_mmu_flags)) {
return ZX_ERR_ACCESS_DENIED;
}
return ProtectLocked(base, size, new_arch_mmu_flags);
}
// static
zx_status_t VmMapping::ProtectOrUnmap(const fbl::RefPtr<VmAspace>& aspace, vaddr_t base,
size_t size, uint new_arch_mmu_flags) {
// This can never be used to set a WRITE permission since it does not ask the underlying VMO to
// perform the copy-on-write step. The underlying VMO might also support dirty tracking, which
// requires write permission faults in order to track pages as dirty when written.
ASSERT(!(new_arch_mmu_flags & ARCH_MMU_FLAG_PERM_WRITE));
// If not removing all permissions do the protect, otherwise skip straight to unmapping the entire
// region.
if ((new_arch_mmu_flags & ARCH_MMU_FLAG_PERM_RWX_MASK) != 0) {
zx_status_t status = aspace->arch_aspace().Protect(base, size / PAGE_SIZE, new_arch_mmu_flags,
aspace->EnlargeArchUnmap()
? ArchVmAspace::EnlargeOperation::Yes
: ArchVmAspace::EnlargeOperation::No);
// If the unmap failed and we are allowed to unmap extra portions of the aspace then fall
// through and unmap, otherwise return with whatever the status is.
if (likely(status == ZX_OK) || !aspace->EnlargeArchUnmap()) {
return status;
}
}
return aspace->arch_aspace().Unmap(base, size / PAGE_SIZE, aspace->EnlargeArchUnmap(), nullptr);
}
zx_status_t VmMapping::ProtectLocked(vaddr_t base, size_t size, uint new_arch_mmu_flags) {
// Assert a few things that should already have been checked by the caller.
DEBUG_ASSERT(size != 0 && IS_PAGE_ALIGNED(base) && IS_PAGE_ALIGNED(size));
DEBUG_ASSERT(!(new_arch_mmu_flags & ARCH_MMU_FLAG_CACHE_MASK));
DEBUG_ASSERT(is_valid_mapping_flags(new_arch_mmu_flags));
DEBUG_ASSERT(object_);
// grab the lock for the vmo
Guard<CriticalMutex> guard{object_->lock()};
// Persist our current caching mode. Every protect region will have the same caching mode so we
// can acquire this from any region.
new_arch_mmu_flags |= (protection_ranges_.FirstRegionMmuFlags() & ARCH_MMU_FLAG_CACHE_MASK);
// This will get called by UpdateProtectionRange below for every existing unique protection range
// that gets changed and allows us to fine tune the protect action based on the previous flags.
auto protect_callback = [new_arch_mmu_flags, this](vaddr_t base, size_t size,
uint old_arch_mmu_flags) {
// Perform an early return if the new and old flags are the same, as there's nothing to be done.
if (new_arch_mmu_flags == old_arch_mmu_flags) {
return;
}
uint flags = new_arch_mmu_flags;
// Check if the new flags have the write permission. This is problematic as we cannot just
// change any existing hardware mappings to have the write permission, as any individual mapping
// may be the result of a read fault and still need to have a copy-on-write step performed. This
// could also map a dirty tracked VMO which requires write permission faults to track pages as
// dirty when written.
if (new_arch_mmu_flags & ARCH_MMU_FLAG_PERM_WRITE) {
// Whatever happens, we're not going to be protecting the arch aspace to have write mappings,
// so this has to be a user aspace so that we can lazily take write faults in the future.
ASSERT(aspace_->is_user() || aspace_->is_guest_physical());
flags &= ~ARCH_MMU_FLAG_PERM_WRITE;
vm_mappings_protect_no_write.Add(1);
// If the new flags without write permission are the same as the old flags, then skip the
// protect step since it will be a no-op.
if (flags == old_arch_mmu_flags) {
return;
}
}
zx_status_t status = ProtectOrUnmap(aspace_, base, size, flags);
// If the protect failed then we do not have sufficient information left to rollback in order to
// return an error, nor can we claim success, so require the protect to have succeeded to
// continue.
ASSERT(status == ZX_OK);
};
zx_status_t status = protection_ranges_.UpdateProtectionRange(
base_, size_, base, size, new_arch_mmu_flags, protect_callback);
ASSERT(status == ZX_OK || status == ZX_ERR_NO_MEMORY);
return status;
}
zx_status_t VmMapping::Unmap(vaddr_t base, size_t size) {
LTRACEF("%p %#" PRIxPTR " %zu\n", this, base, size);
if (!IS_PAGE_ALIGNED(base)) {
return ZX_ERR_INVALID_ARGS;
}
size = ROUNDUP(size, PAGE_SIZE);
fbl::RefPtr<VmAspace> aspace(aspace_);
if (!aspace) {
return ZX_ERR_BAD_STATE;
}
Guard<CriticalMutex> guard{lock()};
if (state_ != LifeCycleState::ALIVE) {
return ZX_ERR_BAD_STATE;
}
if (size == 0 || !is_in_range_locked(base, size)) {
return ZX_ERR_INVALID_ARGS;
}
// If we're unmapping everything, destroy this mapping
if (base == base_ && size == size_) {
return DestroyLocked();
}
return UnmapLocked(base, size);
}
zx_status_t VmMapping::UnmapLocked(vaddr_t base, size_t size) {
canary_.Assert();
DEBUG_ASSERT(size != 0 && IS_PAGE_ALIGNED(size) && IS_PAGE_ALIGNED(base));
DEBUG_ASSERT(base >= base_ && base - base_ < size_);
DEBUG_ASSERT(size_ - (base - base_) >= size);
DEBUG_ASSERT(parent_);
if (state_ != LifeCycleState::ALIVE) {
return ZX_ERR_BAD_STATE;
}
AssertHeld(parent_->lock_ref());
// Should never be unmapping everything, otherwise should destroy.
DEBUG_ASSERT(base != base_ || size != size_);
LTRACEF("%p\n", this);
// Grab the lock for the vmo. This is acquired here so that it is held continuously over both the
// architectural unmap and the set_size_locked call.
DEBUG_ASSERT(object_);
Guard<CriticalMutex> guard{object_->lock()};
// Check if unmapping from one of the ends
if (base_ == base || base + size == base_ + size_) {
LTRACEF("unmapping base %#lx size %#zx\n", base, size);
zx_status_t status =
aspace_->arch_aspace().Unmap(base, size / PAGE_SIZE, aspace_->EnlargeArchUnmap(), nullptr);
if (status != ZX_OK) {
return status;
}
if (base_ == base) {
DEBUG_ASSERT(size != size_);
// First remove any protect regions we will no longer need.
protection_ranges_.DiscardBelow(base_ + size);
// We need to remove ourselves from tree before updating base_, since base_ is the tree key.
// In this case, size_ must also be updated outside of the tree to prevent overlapping ranges
// when subtree_state_ is updated on insertion.
AssertHeld(parent_->lock_ref());
fbl::RefPtr<VmAddressRegionOrMapping> ref(parent_->subregions_.RemoveRegion(this));
base_ += size;
object_offset_ += size;
set_size_locked(size_ - size);
parent_->subregions_.InsertRegion(ktl::move(ref));
} else {
// Resize the protection range, will also cause it to discard any protection ranges that are
// outside the new size.
protection_ranges_.DiscardAbove(base);
// In this case, size_ can be changed while in the tree, since it will not overlap other
// regions when subtree_state_ is updated by set_size_locked.
set_size_locked(size_ - size);
}
return ZX_OK;
}
// We're unmapping from the center, so we need to split the mapping
DEBUG_ASSERT(parent_->state_ == LifeCycleState::ALIVE);
const uint64_t vmo_offset = object_offset_ + (base + size) - base_;
const vaddr_t new_base = base + size;
const size_t new_size = (base_ + size_) - new_base;
// Split off any protection information for the new mapping.
MappingProtectionRanges new_protect = protection_ranges_.SplitAt(new_base);
fbl::AllocChecker ac;
fbl::RefPtr<VmMapping> mapping(
fbl::AdoptRef(new (&ac) VmMapping(*parent_, new_base, new_size, flags_, object_, vmo_offset,
ktl::move(new_protect), Mergeable::YES)));
if (!ac.check()) {
return ZX_ERR_NO_MEMORY;
}
// Unmap the middle segment
LTRACEF("unmapping base %#lx size %#zx\n", base, size);
zx_status_t status =
aspace_->arch_aspace().Unmap(base, size / PAGE_SIZE, aspace_->EnlargeArchUnmap(), nullptr);
if (status != ZX_OK) {
return status;
}
// Turn us into the left half
protection_ranges_.DiscardAbove(base);
// Reduce size_ in tree and update subtree_state_.
set_size_locked(base - base_);
AssertHeld(mapping->lock_ref());
mapping->assert_object_lock();
mapping->ActivateLocked();
// Release the VMO lock and then apply any memory priority to the new mapping portion.
guard.Release();
status = mapping->SetMemoryPriorityLocked(memory_priority_);
DEBUG_ASSERT(status == ZX_OK);
return ZX_OK;
}
bool VmMapping::ObjectRangeToVaddrRange(uint64_t offset, uint64_t len, vaddr_t* base,
uint64_t* virtual_len) const {
DEBUG_ASSERT(IS_PAGE_ALIGNED(offset));
DEBUG_ASSERT(IS_PAGE_ALIGNED(len));
DEBUG_ASSERT(base);
DEBUG_ASSERT(virtual_len);
// Zero sized ranges are considered to have no overlap.
if (len == 0) {
*base = 0;
*virtual_len = 0;
return false;
}
// compute the intersection of the passed in vmo range and our mapping
uint64_t offset_new;
if (!GetIntersect(object_offset_locked_object(), static_cast<uint64_t>(size_locked_object()),
offset, len, &offset_new, virtual_len)) {
return false;
}
DEBUG_ASSERT(*virtual_len > 0 && *virtual_len <= SIZE_MAX);
DEBUG_ASSERT(offset_new >= object_offset_locked_object());
LTRACEF("intersection offset %#" PRIx64 ", len %#" PRIx64 "\n", offset_new, *virtual_len);
// make sure the base + offset is within our address space
// should be, according to the range stored in base_ + size_
bool overflowed =
add_overflow(base_locked_object(), offset_new - object_offset_locked_object(), base);
ASSERT(!overflowed);
// make sure we're only operating within our window
ASSERT(*base >= base_locked_object());
ASSERT((*base + *virtual_len - 1) <= (base_locked_object() + size_locked_object() - 1));
return true;
}
void VmMapping::AspaceUnmapLockedObject(uint64_t offset, uint64_t len) const {
canary_.Assert();
// NOTE: must be acquired with the vmo lock held, but doesn't need to take
// the address space lock, since it will not manipulate its location in the
// vmar tree. However, it must be held in the ALIVE state across this call.
//
// Avoids a race with DestroyLocked() since it removes ourself from the VMO's
// mapping list with the VMO lock held before dropping this state to DEAD. The
// VMO cant call back to us once we're out of their list.
DEBUG_ASSERT(get_state_locked_object() == LifeCycleState::ALIVE);
// |object_| itself is not accessed in this method, and we do not hold the correct lock for it,
// but we know the object_->lock() is held and so therefore object_ is valid and will not be
// modified. Therefore it's correct to read object_ here for the purposes of an assert, but cannot
// be expressed nicely with regular annotations.
[&]() TA_NO_THREAD_SAFETY_ANALYSIS { DEBUG_ASSERT(object_); }();
LTRACEF("region %p obj_offset %#" PRIx64 " size %zu, offset %#" PRIx64 " len %#" PRIx64 "\n",
this, object_offset_locked_object(), size_, offset, len);
// If we're currently faulting and are responsible for the vmo code to be calling
// back to us, detect the recursion and abort here.
// The specific path we're avoiding is if the VMO calls back into us during
// a VmCowPages::LookupCursor call via AspaceUnmapLockedObject(). If we set this flag we're short
// circuiting the unmap operation so that we don't do extra work.
if (unlikely(currently_faulting_)) {
LTRACEF("recursing to ourself, abort\n");
return;
}
// See if there's an intersect.
vaddr_t base;
uint64_t new_len;
if (!ObjectRangeToVaddrRange(offset, len, &base, &new_len)) {
return;
}
// If this is a kernel mapping then we should not be removing mappings out of the arch aspace,
// unless this mapping has explicitly opted out of this check.
DEBUG_ASSERT(aspace_->is_user() || aspace_->is_guest_physical() ||
flags_ & VMAR_FLAG_DEBUG_DYNAMIC_KERNEL_MAPPING);
zx_status_t status =
aspace_->arch_aspace().Unmap(base, new_len / PAGE_SIZE, aspace_->EnlargeArchUnmap(), nullptr);
ASSERT(status == ZX_OK);
}
void VmMapping::AspaceRemoveWriteLockedObject(uint64_t offset, uint64_t len) const {
LTRACEF("region %p obj_offset %#" PRIx64 " size %zu, offset %#" PRIx64 " len %#" PRIx64 "\n",
this, object_offset_, size_, offset, len);
canary_.Assert();
// NOTE: must be acquired with the vmo lock held, but doesn't need to take
// the address space lock, since it will not manipulate its location in the
// vmar tree. However, it must be held in the ALIVE state across this call.
//
// Avoids a race with DestroyLocked() since it removes ourself from the VMO's
// mapping list with the VMO lock held before dropping this state to DEAD. The
// VMO cant call back to us once we're out of their list.
DEBUG_ASSERT(get_state_locked_object() == LifeCycleState::ALIVE);
// |object_| itself is not accessed in this method, and we do not hold the correct lock for it,
// but we know the object_->lock() is held and so therefore object_ is valid and will not be
// modified. Therefore it's correct to read object_ here for the purposes of an assert, but cannot
// be expressed nicely with regular annotations.
[&]() TA_NO_THREAD_SAFETY_ANALYSIS { DEBUG_ASSERT(object_); }();
// If this doesn't support writing then nothing to be done, as we know we have no write mappings.
if (!(flags_ & VMAR_FLAG_CAN_MAP_WRITE)) {
return;
}
// See if there's an intersect.
vaddr_t base;
uint64_t new_len;
if (!ObjectRangeToVaddrRange(offset, len, &base, &new_len)) {
return;
}
// If this is a kernel mapping then we should not be modify mappings in the arch aspace,
// unless this mapping has explicitly opted out of this check.
DEBUG_ASSERT(aspace_->is_user() || aspace_->is_guest_physical() ||
flags_ & VMAR_FLAG_DEBUG_DYNAMIC_KERNEL_MAPPING);
zx_status_t status = ProtectRangesLockedObject().EnumerateProtectionRanges(
base_locked_object(), size_locked_object(), base, new_len,
[this](vaddr_t region_base, size_t region_len, uint mmu_flags) {
// If this range doesn't currently support being writable then we can skip.
if (!(mmu_flags & ARCH_MMU_FLAG_PERM_WRITE)) {
return ZX_ERR_NEXT;
}
// Build new mmu flags without writing.
mmu_flags &= ~(ARCH_MMU_FLAG_PERM_WRITE);
zx_status_t result = ProtectOrUnmap(aspace_, region_base, region_len, mmu_flags);
if (result == ZX_OK) {
return ZX_ERR_NEXT;
}
return result;
});
ASSERT(status == ZX_OK);
}
void VmMapping::AspaceDebugUnpinLockedObject(uint64_t offset, uint64_t len) const {
LTRACEF("region %p obj_offset %#" PRIx64 " size %zu, offset %#" PRIx64 " len %#" PRIx64 "\n",
this, object_offset_, size_, offset, len);
canary_.Assert();
// NOTE: must be acquired with the vmo lock held, but doesn't need to take
// the address space lock, since it will not manipulate its location in the
// vmar tree. However, it must be held in the ALIVE state across this call.
//
// Avoids a race with DestroyLocked() since it removes ourself from the VMO's
// mapping list with the VMO lock held before dropping this state to DEAD. The
// VMO cant call back to us once we're out of their list.
DEBUG_ASSERT(get_state_locked_object() == LifeCycleState::ALIVE);
// See if there's an intersect.
vaddr_t base;
uint64_t new_len;
if (!ObjectRangeToVaddrRange(offset, len, &base, &new_len)) {
return;
}
// This unpin is not allowed for kernel mappings, unless the mapping has specifically opted out of
// this debug check due to it performing its own dynamic management.
DEBUG_ASSERT(aspace_->is_user() || aspace_->is_guest_physical() ||
flags_ & VMAR_FLAG_DEBUG_DYNAMIC_KERNEL_MAPPING);
}
namespace {
template <size_t NumPages>
class VmMappingCoalescer {
public:
VmMappingCoalescer(VmMapping* mapping, vaddr_t base, uint mmu_flags,
ArchVmAspace::ExistingEntryAction existing_entry_action)
TA_REQ(mapping->lock());
~VmMappingCoalescer();
// Add a page to the mapping run. If this fails, the VmMappingCoalescer is
// no longer valid.
zx_status_t Append(vaddr_t vaddr, paddr_t paddr) {
AssertHeld(mapping_->lock_ref());
DEBUG_ASSERT(!aborted_);
// If this isn't the expected vaddr, flush the run we have first.
if (!can_append(vaddr)) {
zx_status_t status = Flush();
if (status != ZX_OK) {
return status;
}
base_ = vaddr;
}
phys_[count_] = paddr;
++count_;
return ZX_OK;
}
zx_status_t AppendOrAdjustMapping(vaddr_t vaddr, paddr_t paddr, uint mmu_flags) {
AssertHeld(mapping_->lock_ref());
DEBUG_ASSERT(!aborted_);
// If this isn't the expected vaddr or mmu_flags have changed, flush the run we have first.
if (!can_append(vaddr) || mmu_flags != mmu_flags_) {
zx_status_t status = Flush();
if (status != ZX_OK) {
return status;
}
base_ = vaddr;
mmu_flags_ = mmu_flags;
}
// Query aspace and adjust the mapping if there is already a page mapped here.
zx::result result = mapping_->AdjustMapping(vaddr, paddr, mmu_flags_);
if (result.is_error()) {
return result.status_value();
}
// If page was already mapped, don't add it to the mapping run.
if (result.value()) {
return ZX_OK;
}
phys_[count_] = paddr;
++count_;
return ZX_OK;
}
// How much space remains in the phys_ array, starting from vaddr, that can be used to
// opportunistically map additional pages.
size_t ExtraPageCapacityFrom(vaddr_t vaddr) {
// vaddr must be appendable & the coalescer can't be empty.
return (can_append(vaddr) && count_ != 0) ? NumPages - count_ : 0;
}
// Functions for the user to manually manage the pages array. It is up to the user to manage the
// page count and ensure the coalescer doesn't overflow, maintains the correct page count and that
// the pages are contiguous.
paddr_t* GetNextPageSlot() { return &phys_[count_]; }
void IncrementCount(size_t i) { count_ += i; }
// Submit any outstanding mappings to the MMU. If this fails, the
// VmMappingCoalescer is no longer valid.
zx_status_t Flush();
// Drop the current outstanding mappings without sending them to the MMU.
// After this call, the VmMappingCoalescer is no longer valid.
void Abort() { aborted_ = true; }
private:
// Vaddr can be appended if it's the next free slot and the coalescer isn't full.
bool can_append(vaddr_t vaddr) {
return count_ < ktl::size(phys_) && vaddr == base_ + count_ * PAGE_SIZE;
}
DISALLOW_COPY_ASSIGN_AND_MOVE(VmMappingCoalescer);
VmMapping* mapping_;
vaddr_t base_;
paddr_t phys_[NumPages];
size_t count_;
bool aborted_;
uint mmu_flags_;
const ArchVmAspace::ExistingEntryAction existing_entry_action_;
};
template <size_t NumPages>
VmMappingCoalescer<NumPages>::VmMappingCoalescer(
VmMapping* mapping, vaddr_t base, uint mmu_flags,
ArchVmAspace::ExistingEntryAction existing_entry_action)
: mapping_(mapping),
base_(base),
count_(0),
aborted_(false),
mmu_flags_(mmu_flags),
existing_entry_action_(existing_entry_action) {}
template <size_t NumPages>
VmMappingCoalescer<NumPages>::~VmMappingCoalescer() {
// Make sure we've flushed or aborted
DEBUG_ASSERT(count_ == 0 || aborted_);
}
template <size_t NumPages>
zx_status_t VmMappingCoalescer<NumPages>::Flush() {
AssertHeld(mapping_->lock_ref());
if (count_ == 0) {
return ZX_OK;
}
// Assert that we're not accidentally mapping the zero page writable. Unless called from a kernel
// aspace, as the zero page can be mapped writeable from the kernel aspace in mexec.
DEBUG_ASSERT(
!(mmu_flags_ & ARCH_MMU_FLAG_PERM_WRITE) ||
ktl::all_of(phys_, &phys_[count_], [](paddr_t p) { return p != vm_get_zero_page_paddr(); }) ||
!mapping_->aspace()->is_user());
if (mmu_flags_ & ARCH_MMU_FLAG_PERM_RWX_MASK) {
size_t mapped;
zx_status_t ret = mapping_->aspace()->arch_aspace().Map(base_, phys_, count_, mmu_flags_,
existing_entry_action_, &mapped);
if (ret != ZX_OK) {
TRACEF("error %d mapping %zu pages starting at va %#" PRIxPTR "\n", ret, count_, base_);
aborted_ = true;
return ret;
}
DEBUG_ASSERT_MSG(mapped == count_, "mapped %zu, count %zu\n", mapped, count_);
}
base_ += count_ * PAGE_SIZE;
count_ = 0;
return ZX_OK;
}
} // namespace
zx_status_t VmMapping::MapRange(size_t offset, size_t len, bool commit, bool ignore_existing) {
Guard<CriticalMutex> aspace_guard{lock()};
canary_.Assert();
len = ROUNDUP(len, PAGE_SIZE);
if (len == 0) {
return ZX_ERR_INVALID_ARGS;
}
if (state_ != LifeCycleState::ALIVE) {
return ZX_ERR_BAD_STATE;
}
LTRACEF("region %p, offset %#zx, size %#zx, commit %d\n", this, offset, len, commit);
DEBUG_ASSERT(object_);
if (!IS_PAGE_ALIGNED(offset) || !is_in_range_locked(base_ + offset, len)) {
return ZX_ERR_INVALID_ARGS;
}
// If this is a kernel mapping then validate that all pages being mapped are currently pinned,
// ensuring that they cannot be taken away for any reason, unless the mapping has specifically
// opted out of this debug check due to it performing its own dynamic management.
DEBUG_ASSERT(aspace_->is_user() || aspace_->is_guest_physical() ||
(flags_ & VMAR_FLAG_DEBUG_DYNAMIC_KERNEL_MAPPING) ||
object_->DebugIsRangePinned(object_offset_locked() + offset, len));
// grab the lock for the vmo
Guard<CriticalMutex> object_guard{object_->lock()};
// Cache whether the object is dirty tracked, we need to know this when computing mmu flags later.
const bool dirty_tracked = object_->is_dirty_tracked_locked();
// set the currently faulting flag for any recursive calls the vmo may make back into us.
DEBUG_ASSERT(!currently_faulting_);
currently_faulting_ = true;
auto cleanup = fit::defer([&]() {
assert_object_lock();
currently_faulting_ = false;
});
// Trim our range to the current VMO size. Our mapping might exceed the VMO in the case where the
// VMO is resizable, and this should not be considered an error.
len = TrimmedObjectRangeLocked(offset, len);
if (len == 0) {
return ZX_OK;
}
// The region to map could have multiple different current arch mmu flags, so we need to iterate
// over them to ensure we install mappings with the correct permissions.
return EnumerateProtectionRangesLocked(
base_ + offset, len,
[this, commit, dirty_tracked, ignore_existing](vaddr_t base, size_t len, uint mmu_flags) {
AssertHeld(lock_ref());
// Remove the write permission if this maps a vmo that supports dirty tracking, in order to
// trigger write permission faults when writes occur, enabling us to track when pages are
// dirtied.
if (dirty_tracked) {
mmu_flags &= ~ARCH_MMU_FLAG_PERM_WRITE;
}
// In the scenario where we are committing, and calling RequireOwnedPage, we are supposed to
// pass in a non-null LazyPageRequest. Technically we could get away with not passing in a
// PageRequest since:
// * Only internal kernel VMOs will have the 'commit' flag passed in for their mappings
// * Only pager backed VMOs or VMOs that support delayed memory allocations need to fill
// out a PageRequest
// * Internal kernel VMOs are never pager backed or have the delayed memory allocation flag
// set.
// However, should these assumptions ever get violated it's better to catch this gracefully
// than have RequireOwnedPage error/crash internally, and it costs nothing to create and
// pass in.
__UNINITIALIZED LazyPageRequest page_request;
VmMappingCoalescer<16> coalescer(this, base, mmu_flags,
ignore_existing
? ArchVmAspace::ExistingEntryAction::Skip
: ArchVmAspace::ExistingEntryAction::Error);
const uint64_t vmo_offset = object_offset_locked() + (base - base_);
if (likely(object_->is_paged())) {
VmObjectPaged* object = static_cast<VmObjectPaged*>(object_.get());
AssertHeld(object->lock_ref());
const bool writing = mmu_flags & ARCH_MMU_FLAG_PERM_WRITE;
__UNINITIALIZED auto cursor = object->GetLookupCursorLocked(vmo_offset, len);
if (cursor.is_error()) {
return cursor.error_value();
}
// Do not consider pages touched when mapping in, if they are actually touched they will
// get an accessed bit set in the hardware.
cursor->DisableMarkAccessed();
AssertHeld(cursor->lock_ref());
for (uint64_t off = 0; off < len; off += PAGE_SIZE) {
vm_page_t* page = nullptr;
if (commit) {
__UNINITIALIZED zx::result<VmCowPages::LookupCursor::RequireResult> result =
cursor->RequireOwnedPage(writing, 1, &page_request);
if (result.is_error()) {
zx_status_t status = result.error_value();
// As per the comment above page_request definition, there should never be commit
// + pager backed VMO and so we should never end up with a PageRequest needing to be
// waited on.
ASSERT(status != ZX_ERR_SHOULD_WAIT);
// fail when we can't commit every requested page
coalescer.Abort();
return status;
}
page = result->page;
} else {
// Not committing so get a page if one exists. This increments the cursor, returning
// nullptr if no page.
page = cursor->MaybePage(writing);
// This page was not present and if we are in a run of absent pages we would like to
// efficiently skip them, instead of querying each virtual address individually. Due
// to the assumptions of the cursor, we cannot call SkipMissingPages if we had just
// requested the last page in the range of the cursor.
if (!page && off + PAGE_SIZE < len) {
// Increment |off| for the any pages we skip and let the original page from
// MaybePage get incremented on the way around the loop before the range gets
// checked.
off += cursor->SkipMissingPages() * PAGE_SIZE;
}
}
if (page) {
zx_status_t status = coalescer.Append(base + off, page->paddr());
if (status != ZX_OK) {
return status;
}
}
}
} else {
VmObjectPhysical* object = static_cast<VmObjectPhysical*>(object_.get());
AssertHeld(object->lock_ref());
// Physical VMOs are always allocated and contiguous, just need to get the paddr.
paddr_t phys_base = 0;
zx_status_t status = object->LookupContiguousLocked(vmo_offset, len, &phys_base);
ASSERT(status == ZX_OK);
for (size_t offset = 0; offset < len; offset += PAGE_SIZE) {
status = coalescer.Append(base + offset, phys_base + offset);
if (status != ZX_OK) {
return status;
}
}
}
return coalescer.Flush();
});
}
zx_status_t VmMapping::DecommitRange(size_t offset, size_t len) {
canary_.Assert();
LTRACEF("%p [%#zx+%#zx], offset %#zx, len %#zx\n", this, base_, size_, offset, len);
Guard<CriticalMutex> guard{lock()};
if (state_ != LifeCycleState::ALIVE) {
return ZX_ERR_BAD_STATE;
}
if (offset + len < offset || offset + len > size_) {
return ZX_ERR_OUT_OF_RANGE;
}
// VmObject::DecommitRange will typically call back into our instance's
// VmMapping::AspaceUnmapLockedObject.
return object_->DecommitRange(object_offset_locked() + offset, len);
}
zx_status_t VmMapping::DestroyLocked() {
canary_.Assert();
LTRACEF("%p\n", this);
// Take a reference to ourself, so that we do not get destructed after
// dropping our last reference in this method (e.g. when calling
// subregions_.erase below).
fbl::RefPtr<VmMapping> self(this);
// If this is the last_fault_ then clear it before removing from the VMAR tree. Even if this
// destroy fails, it's always safe to clear last_fault_, so we preference doing it upfront for
// clarity.
if (aspace_->last_fault_ == this) {
aspace_->last_fault_ = nullptr;
}
// The vDSO code mapping can never be unmapped, not even
// by VMAR destruction (except for process exit, of course).
// TODO(mcgrathr): Turn this into a policy-driven process-fatal case
// at some point. teisenbe@ wants to eventually make zx_vmar_destroy
// never fail.
if (aspace_->vdso_code_mapping_ == self) {
return ZX_ERR_ACCESS_DENIED;
}
// Remove any priority.
zx_status_t status = SetMemoryPriorityLocked(MemoryPriority::DEFAULT);
DEBUG_ASSERT(status == ZX_OK);
// grab the object lock to unmap and remove ourselves from its list.
{
Guard<CriticalMutex> guard{object_->lock()};
// Perform unmap holding the object lock to prevent mappings being modified in between.
status = aspace_->arch_aspace().Unmap(base_, size_ / PAGE_SIZE, aspace_->EnlargeArchUnmap(),
nullptr);
if (status != ZX_OK) {
return status;
}
protection_ranges_.clear();
object_->RemoveMappingLocked(this);
// Detach the region from the parent.
if (parent_) {
AssertHeld(parent_->lock_ref());
DEBUG_ASSERT(this->in_subregion_tree());
parent_->subregions_.RemoveRegion(this);
}
// The size may only be set to zero when not in the subregion tree.
set_size_locked(0);
}
// Clear the cached attribution count.
// The generation count should already have been incremented by UnmapLocked above.
cached_page_attribution_ = {};
// detach from any object we have mapped. Note that we are holding the aspace_->lock() so we
// will not race with other threads calling vmo()
object_.reset();
// mark ourself as dead
parent_ = nullptr;
state_ = LifeCycleState::DEAD;
return ZX_OK;
}
zx::result<bool> VmMapping::AdjustMapping(vaddr_t va, paddr_t pa, uint mmu_flags) {
// Using the `Upgrade` action for `Map` accounts for existing mappings, so
// when we are using it, strip out the 2 cases in this method (Protect and Unmap+Map).
//
// TODO(https://issues.fuchsia.dev/issues/42182886)
// When `ENABLE_PAGE_FAULT_UPGRADE` is always true, remove this method.
if constexpr (ENABLE_PAGE_FAULT_UPGRADE) {
return zx::ok(false);
}
// see if something is mapped here now
// this may happen if we are one of multiple threads racing on a single address
uint page_flags;
paddr_t mapped_pa;
zx_status_t err = aspace_->arch_aspace().Query(va, &mapped_pa, &page_flags);
if (err >= 0) {
LTRACEF("queried va, page at pa %#" PRIxPTR ", flags %#x is already there\n", mapped_pa,
page_flags);
if (mapped_pa == pa) {
// Faulting on a mapping that is the correct page could happen for a few reasons
// 1. Permission are incorrect and this fault is a write fault for a read only mapping.
// 2. Fault was caused by (1), but we were racing with another fault and the mapping is
// already fixed.
// 3. Some other error, such as an access flag missing on arm, caused this fault
// Of these three scenarios (1) is overwhelmingly the most common, and requires us to protect
// the page with the new permissions. In the scenario of (2) we could fast return and not
// perform the potentially expensive protect, but this scenario is quite rare and requires a
// multi-thread race on causing and handling the fault. (3) should also be highly uncommon as
// access faults would normally be handled by a separate fault handler, nevertheless we should
// still resolve such faults here, which requires calling protect.
// Given that (2) is rare and hard to distinguish from (3) we simply always call protect to
// ensure the fault is resolved.
// assert that we're not accidentally marking the zero page writable
DEBUG_ASSERT((mapped_pa != vm_get_zero_page_paddr()) ||
!(mmu_flags & ARCH_MMU_FLAG_PERM_WRITE));
// same page, different permission
zx_status_t status =
aspace_->arch_aspace().Protect(va, 1, mmu_flags, ArchVmAspace::EnlargeOperation::Yes);
if (unlikely(status != ZX_OK)) {
// ZX_ERR_NO_MEMORY is the only legitimate reason for Protect to fail.
ASSERT_MSG(status == ZX_ERR_NO_MEMORY, "Unexpected failure from protect: %d\n", status);
TRACEF("failed to modify permissions on existing mapping\n");
return zx::error(status);
}
return zx::ok(true);
} else {
// some other page is mapped there already
LTRACEF("thread %s faulted on va %#" PRIxPTR ", different page was present\n",
Thread::Current::Get()->name(), va);
// assert that we're not accidentally marking the zero page writable
DEBUG_ASSERT((pa != vm_get_zero_page_paddr()) || !(mmu_flags & ARCH_MMU_FLAG_PERM_WRITE));
// unmap the old one
zx_status_t status =
aspace_->arch_aspace().Unmap(va, 1, aspace_->EnlargeArchUnmap(), nullptr);
if (status != ZX_OK) {
ASSERT_MSG(status == ZX_ERR_NO_MEMORY, "Unexpected failure from unmap: %d\n", status);
TRACEF("failed to remove old mapping before replacing\n");
return zx::error(status);
}
}
}
return zx::ok(false);
}
zx_status_t VmMapping::PageFaultLocked(vaddr_t va, const uint pf_flags,
LazyPageRequest* page_request) {
VM_KTRACE_DURATION(
2, "VmMapping::PageFault",
("user_id", KTRACE_ANNOTATED_VALUE(AssertHeld(lock_ref()), object_->user_id())),
("va", ktrace::Pointer{va}));
canary_.Assert();
DEBUG_ASSERT(is_in_range_locked(va, 1));
va = ROUNDDOWN(va, PAGE_SIZE);
uint64_t vmo_offset = va - base_ + object_offset_locked();
[[maybe_unused]] char pf_string[5];
LTRACEF("%p va %#" PRIxPTR " vmo_offset %#" PRIx64 ", pf_flags %#x (%s)\n", this, va, vmo_offset,
pf_flags, vmm_pf_flags_to_string(pf_flags, pf_string));
// Need to look up the mmu flags for this virtual address, as well as how large a region those
// flags are for so we can cap the extra mappings we create.
MappingProtectionRanges::FlagsRange range =
ProtectRangesLocked().FlagsRangeAtAddr(base_, size_, va);
// Build the mmu flags we need to have based on the page fault. This strategy of building the
// flags and then comparing all at once allows the compiler to provide much better code gen.
uint needed_mmu_flags = 0;
if (pf_flags & VMM_PF_FLAG_USER) {
needed_mmu_flags |= ARCH_MMU_FLAG_PERM_USER;
}
const bool write = pf_flags & VMM_PF_FLAG_WRITE;
if (write) {
needed_mmu_flags |= ARCH_MMU_FLAG_PERM_WRITE;
} else {
needed_mmu_flags |= ARCH_MMU_FLAG_PERM_READ;
}
if (pf_flags & VMM_PF_FLAG_INSTRUCTION) {
needed_mmu_flags |= ARCH_MMU_FLAG_PERM_EXECUTE;
}
// Check that all the needed flags are present.
if (unlikely((range.mmu_flags & needed_mmu_flags) != needed_mmu_flags)) {
if ((pf_flags & VMM_PF_FLAG_USER) && !(range.mmu_flags & ARCH_MMU_FLAG_PERM_USER)) {
// user page fault on non user mapped region
LTRACEF("permission failure: user fault on non user region\n");
}
if ((pf_flags & VMM_PF_FLAG_WRITE) && !(range.mmu_flags & ARCH_MMU_FLAG_PERM_WRITE)) {
// write to a non-writeable region
LTRACEF("permission failure: write fault on non-writable region\n");
}
if (!(pf_flags & VMM_PF_FLAG_WRITE) && !(range.mmu_flags & ARCH_MMU_FLAG_PERM_READ)) {
// read to a non-readable region
LTRACEF("permission failure: read fault on non-readable region\n");
}
if ((pf_flags & VMM_PF_FLAG_INSTRUCTION) && !(range.mmu_flags & ARCH_MMU_FLAG_PERM_EXECUTE)) {
// instruction fetch from a no execute region
LTRACEF("permission failure: execute fault on no execute region\n");
}
return ZX_ERR_ACCESS_DENIED;
}
// grab the lock for the vmo
Guard<CriticalMutex> guard{object_->lock()};
// set the currently faulting flag for any recursive calls the vmo may make back into us
// The specific path we're avoiding is if the VMO calls back into us during
// a VmCowPages::LookupCursor call via AspaceUnmapLockedObject(). Since we're responsible for
// that page, signal to ourself to skip the unmap operation.
DEBUG_ASSERT(!currently_faulting_);
currently_faulting_ = true;
auto cleanup = fit::defer([&]() {
assert_object_lock();
currently_faulting_ = false;
});
// Determine how far to the end of the page table so we do not cause extra allocations.
const uint64_t next_pt_base = ArchVmAspace::NextUserPageTableOffset(va);
// Find the minimum between the size of this protection range and the end of the page table.
const uint64_t max_map = ktl::min(next_pt_base, range.region_top);
// Convert this into a number of pages, limited by the max lookup window.
static constexpr uint64_t kMaxPages = 16;
uint64_t max_out_pages = ktl::min((max_map - va) / PAGE_SIZE, kMaxPages);
DEBUG_ASSERT(max_out_pages > 0);
const uint64_t vmo_size = object_->size_locked();
if (vmo_offset >= vmo_size) {
return ZX_ERR_OUT_OF_RANGE;
}
// Trim out pages to the limit of the VMO.
max_out_pages = ktl::min(max_out_pages, (vmo_size - vmo_offset) / PAGE_SIZE);
__UNINITIALIZED VmMappingCoalescer<kMaxPages> coalescer(
this, va, range.mmu_flags,
ENABLE_PAGE_FAULT_UPGRADE ? ArchVmAspace::ExistingEntryAction::Upgrade
: ArchVmAspace::ExistingEntryAction::Skip);
if (likely(object_->is_paged())) {
VmObjectPaged* object = static_cast<VmObjectPaged*>(object_.get());
AssertHeld(object->lock_ref());
// If this is a write fault and the VMO supports dirty tracking, only lookup 1 page. The pages
// will also be marked dirty for a write, which we only want for the current page. We could
// optimize this to lookup following pages here too and map them in, however we would have to
// not mark them dirty during the lookup, and map them in without write permissions here, so
// that we can take a permission fault on a write to update their dirty tracking later. Instead,
// we can keep things simple by just looking up 1 page.
// TODO(rashaeqbal): Revisit this decision if there are performance issues.
if (write && object->is_dirty_tracked_locked()) {
max_out_pages = 1;
}
// fault in or grab existing pages.
__UNINITIALIZED auto cursor =
object->GetLookupCursorLocked(vmo_offset, max_out_pages * PAGE_SIZE);
if (cursor.is_error()) {
return cursor.error_value();
}
// Do not consider pages touched when mapping in, if they are actually touched they will
// get an accessed bit set in the hardware.
cursor->DisableMarkAccessed();
AssertHeld(cursor->lock_ref());
__UNINITIALIZED
zx::result<VmCowPages::LookupCursor::RequireResult> result =
cursor->RequirePage(write, 1, page_request);
if (result.is_error()) {
return result.error_value();
}
// We looked up in order to write. Mark as modified.
if (write) {
DEBUG_ASSERT(result->writable);
object->mark_modified_locked();
}
// if we read faulted, and lookup didn't say that this is always writable, then we map or modify
// the page without any write permissions. This ensures we will fault again if a write is
// attempted so we can potentially replace this page with a copy or a new one, or update the
// page's dirty state.
if (!write && !result->writable) {
// we read faulted, so only map with read permissions
range.mmu_flags &= ~ARCH_MMU_FLAG_PERM_WRITE;
}
zx_status_t status =
coalescer.AppendOrAdjustMapping(va, result->page->paddr(), range.mmu_flags);
if (status != ZX_OK) {
return status;
}
// Check how much space the coalescer has for faulting additional pages.
size_t extra_pages = coalescer.ExtraPageCapacityFrom(va + PAGE_SIZE);
extra_pages = ktl::min(extra_pages, max_out_pages - 1);
// Acquire any additional pages, but only if they already exist as the user has not actually
// attempted to use these pages yet.
if (extra_pages > 0) {
size_t num_pages = cursor->IfExistPages(result->writable, static_cast<uint>(extra_pages),
coalescer.GetNextPageSlot());
coalescer.IncrementCount(num_pages);
}
} else {
VmObjectPhysical* object = static_cast<VmObjectPhysical*>(object_.get());
AssertHeld(object->lock_ref());
// Already validated the size, and since physical VMOs are always allocated, and not resizable,
// we know we can always retrieve the maximum number of pages without failure.
uint64_t len = max_out_pages * PAGE_SIZE;
paddr_t phys_base = 0;
zx_status_t status = object->LookupContiguousLocked(vmo_offset, len, &phys_base);
ASSERT(status == ZX_OK);
status = coalescer.AppendOrAdjustMapping(va, phys_base, range.mmu_flags);
if (status != ZX_OK) {
return status;
}
// Extrapolate the pages from the base address.
for (size_t offset = PAGE_SIZE; offset < len; offset += PAGE_SIZE) {
status = coalescer.Append(va + offset, phys_base + offset);
if (status != ZX_OK) {
return status;
}
}
}
VM_KTRACE_DURATION(2, "map_page", ("va", ktrace::Pointer{va}), ("pf_flags", pf_flags));
return coalescer.Flush();
}
void VmMapping::ActivateLocked() {
DEBUG_ASSERT(state_ == LifeCycleState::NOT_READY);
DEBUG_ASSERT(parent_);
state_ = LifeCycleState::ALIVE;
object_->AddMappingLocked(this);
// Now that we have added a mapping to the VMO it's cache policy becomes fixed, and we can read it
// and augment our arch_mmu_flags.
uint32_t cache_policy = object_->GetMappingCachePolicyLocked();
uint arch_mmu_flags = protection_ranges_.FirstRegionMmuFlags();
if ((arch_mmu_flags & ARCH_MMU_FLAG_CACHE_MASK) != cache_policy) {
// Warn in the event that we somehow receive a VMO that has a cache
// policy set while also holding cache policy flags within the arch
// flags. The only path that should be able to achieve this is if
// something in the kernel maps into their aspace incorrectly.
if ((arch_mmu_flags & ARCH_MMU_FLAG_CACHE_MASK) != 0) {
TRACEF(
"warning: mapping has conflicting cache policies: vmo %#02x "
"arch_mmu_flags %#02x.\n",
cache_policy, arch_mmu_flags & ARCH_MMU_FLAG_CACHE_MASK);
// Clear the existing cache policy and use the new one.
arch_mmu_flags &= ~ARCH_MMU_FLAG_CACHE_MASK;
}
// If we are changing the cache policy then this can only happen if this is a new mapping region
// and not a new mapping occurring as a result of an unmap split. In the case of a new mapping
// region we know there cannot yet be any protection ranges.
DEBUG_ASSERT(protection_ranges_.IsSingleRegion());
arch_mmu_flags |= cache_policy;
protection_ranges_.SetFirstRegionMmuFlags(arch_mmu_flags);
}
AssertHeld(parent_->lock_ref());
parent_->subregions_.InsertRegion(fbl::RefPtr<VmAddressRegionOrMapping>(this));
}
void VmMapping::Activate() {
Guard<CriticalMutex> guard{object_->lock()};
ActivateLocked();
}
void VmMapping::TryMergeRightNeighborLocked(VmMapping* right_candidate) {
AssertHeld(right_candidate->lock_ref());
// This code is tolerant of many 'miss calls' if mappings aren't mergeable or are not neighbours
// etc, but the caller should not be attempting to merge if these mappings are not actually from
// the same vmar parent. Doing so indicates something structurally wrong with the hierarchy.
DEBUG_ASSERT(parent_ == right_candidate->parent_);
// Should not be able to have the same parent yet have gotten a different memory priority.
DEBUG_ASSERT(memory_priority_ == right_candidate->memory_priority_);
// These tests are intended to be ordered such that we fail as fast as possible. As such testing
// for mergeability, which we commonly expect to succeed and not fail, is done last.
// Need to refer to the same object.
if (object_.get() != right_candidate->object_.get()) {
return;
}
// Aspace and VMO ranges need to be contiguous. Validate that the right candidate is actually to
// the right in addition to checking that base+size lines up for single scenario where base_+size_
// can overflow and becomes zero.
if (base_ + size_ != right_candidate->base_ || right_candidate->base_ < base_) {
return;
}
if (object_offset_locked() + size_ != right_candidate->object_offset_locked()) {
return;
}
// All flags need to be consistent.
if (flags_ != right_candidate->flags_) {
return;
}
// Although we can combine the protect_region_list_rest_ of the two mappings, we require that they
// be of the same cacheability, as this is an assumption that mapping has a single cacheability
// type. Since all protection regions have the same cacheability we can check any arbitrary one in
// each of the mappings. Note that this check is technically redundant, since a VMO can only have
// one kind of cacheability and we already know this is the same VMO, but some extra paranoia here
// does not hurt.
if ((ProtectRangesLocked().FirstRegionMmuFlags() & ARCH_MMU_FLAG_CACHE_MASK) !=
(right_candidate->ProtectRangesLocked().FirstRegionMmuFlags() & ARCH_MMU_FLAG_CACHE_MASK)) {
return;
}
// Only merge live mappings.
if (state_ != LifeCycleState::ALIVE || right_candidate->state_ != LifeCycleState::ALIVE) {
return;
}
// Both need to be mergeable.
if (mergeable_ == Mergeable::NO || right_candidate->mergeable_ == Mergeable::NO) {
return;
}
// Destroy / DestroyLocked perform a lot more cleanup than we want, we just need to clear out a
// few things from right_candidate and then mark it as dead, as we do not want to clear out any
// arch page table mappings etc.
{
// Although it was safe to read size_ without holding the object lock, we need to acquire it to
// perform changes.
Guard<CriticalMutex> guard{AliasedLock, object_->lock(), right_candidate->object_->lock()};
// Attempt to merge the protection region lists first. This is done first as a node allocation
// might be needed, which could fail. If it fails we can still abort now without needing to roll
// back any changes.
zx_status_t status = protection_ranges_.MergeRightNeighbor(right_candidate->protection_ranges_,
right_candidate->base_);
if (status != ZX_OK) {
ASSERT(status == ZX_ERR_NO_MEMORY);
return;
}
right_candidate->object_->RemoveMappingLocked(right_candidate);
// Detach region from the parent, ensuring our caller is correctly holding a refptr.
DEBUG_ASSERT(right_candidate->in_subregion_tree());
DEBUG_ASSERT(right_candidate->ref_count_debug() > 1);
AssertHeld(parent_->lock_ref());
parent_->subregions_.RemoveRegion(right_candidate);
// The size of this mapping must be updated after removing the right candidate from the region
// tree to ensure correct re-validation of the subtree invariants. Failure to do so may trigger
// a consistency check, depending on the structure of related WAVLTree nodes.
set_size_locked(size_ + right_candidate->size_);
right_candidate->set_size_locked(0);
}
// Detach from the VMO. As we have removed ourselves as a mapping we also need to remove any
// priority we might have been propagating to that VMO. Additionally, we're destroying this
// mapping and so to not trip the assert in the destructor we can accomplish both goals by
// setting our priority back to the default.
right_candidate->SetMemoryPriorityLocked(MemoryPriority::DEFAULT);
right_candidate->object_.reset();
if (aspace_->last_fault_ == right_candidate) {
aspace_->last_fault_ = nullptr;
}
// Mark it as dead.
right_candidate->parent_ = nullptr;
right_candidate->state_ = LifeCycleState::DEAD;
vm_mappings_merged.Add(1);
}
void VmMapping::TryMergeNeighborsLocked() {
canary_.Assert();
// Check that this mapping is mergeable and is currently in the correct lifecycle state.
if (mergeable_ == Mergeable::NO || state_ != LifeCycleState::ALIVE) {
return;
}
// As a VmMapping if we we are alive we by definition have a parent.
DEBUG_ASSERT(parent_);
// We expect there to be a RefPtr to us held beyond the one for the wavl tree ensuring that we
// cannot trigger our own destructor should we remove ourselves from the hierarchy.
DEBUG_ASSERT(ref_count_debug() > 1);
// First consider merging any mapping on our right, into |this|.
AssertHeld(parent_->lock_ref());
auto right_candidate = parent_->subregions_.RightOf(this);
if (right_candidate.IsValid()) {
// Request mapping as a refptr as we need to hold a refptr across the try merge.
if (fbl::RefPtr<VmMapping> mapping = right_candidate->as_vm_mapping()) {
TryMergeRightNeighborLocked(mapping.get());
}
}
// Now attempt to merge |this| with any left neighbor.
AssertHeld(parent_->lock_ref());
auto left_candidate = parent_->subregions_.LeftOf(this);
if (!left_candidate.IsValid()) {
return;
}
if (auto mapping = left_candidate->as_vm_mapping()) {
// Attempt actual merge. If this succeeds then |this| is in the dead state, but that's fine as
// we are finished anyway.
AssertHeld(mapping->lock_ref());
mapping->TryMergeRightNeighborLocked(this);
}
}
void VmMapping::MarkMergeable(fbl::RefPtr<VmMapping>&& mapping) {
Guard<CriticalMutex> guard{mapping->lock()};
// Now that we have the lock check this mapping is still alive and we haven't raced with some
// kind of destruction.
if (mapping->state_ != LifeCycleState::ALIVE) {
return;
}
// Skip marking any vdso segments mergeable. Although there is currently only one vdso segment and
// so it would never actually get merged, marking it mergeable is technically incorrect.
if (mapping->aspace_->vdso_code_mapping_ == mapping) {
return;
}
mapping->mergeable_ = Mergeable::YES;
mapping->TryMergeNeighborsLocked();
}
zx_status_t VmMapping::SetMemoryPriorityLocked(VmAddressRegion::MemoryPriority priority) {
DEBUG_ASSERT(state_ == LifeCycleState::ALIVE);
if (priority == memory_priority_) {
return ZX_OK;
}
memory_priority_ = priority;
const bool is_high = priority == VmAddressRegion::MemoryPriority::HIGH;
aspace_->ChangeHighPriorityCountLocked(is_high ? 1 : -1);
Guard<CriticalMutex> guard{object_->lock()};
object_->ChangeHighPriorityCountLocked(is_high ? 1 : -1);
return ZX_OK;
}
void VmMapping::CommitHighMemoryPriority() {
fbl::RefPtr<VmObject> vmo;
uint64_t offset;
uint64_t len;
{
Guard<CriticalMutex> guard{lock()};
if (state_ != LifeCycleState::ALIVE || memory_priority_ != MemoryPriority::HIGH) {
return;
}
vmo = object_;
offset = object_offset_locked();
len = size_locked();
}
DEBUG_ASSERT(vmo);
vmo->CommitHighPriorityPages(offset, len);
// Ignore the return result of MapRange as this is just best effort.
MapRange(offset, len, false, true);
}
zx_status_t MappingProtectionRanges::EnumerateProtectionRanges(
vaddr_t mapping_base, size_t mapping_size, vaddr_t base, size_t size,
fit::inline_function<zx_status_t(vaddr_t region_base, size_t region_len, uint mmu_flags)>&&
func) const {
DEBUG_ASSERT(size > 0);
// Have a short circuit for the single protect region case to avoid wavl tree processing in the
// common case.
if (protect_region_list_rest_.is_empty()) {
zx_status_t result = func(base, size, first_region_arch_mmu_flags_);
if (result == ZX_ERR_NEXT || result == ZX_ERR_STOP) {
return ZX_OK;
}
return result;
}
// See comments in the loop that explain what next and current represent.
auto next = protect_region_list_rest_.upper_bound(base);
auto current = next;
current--;
const vaddr_t range_top = base + (size - 1);
do {
// The region starting from 'current' and ending at 'next' represents a single protection
// domain. We first work that, remembering that either of these could be an invalid node,
// meaning the start or end of the mapping respectively.
const vaddr_t protect_region_base = current.IsValid() ? current->region_start : mapping_base;
const vaddr_t protect_region_top =
next.IsValid() ? (next->region_start - 1) : (mapping_base + (mapping_size - 1));
// We should only be iterating nodes that are actually part of the requested range.
DEBUG_ASSERT(base <= protect_region_top);
DEBUG_ASSERT(range_top >= protect_region_base);
// The region found is of an entire protection block, and could extend outside the requested
// range, so trim if necessary.
const vaddr_t region_base = ktl::max(protect_region_base, base);
const size_t region_len = ktl::min(protect_region_top, range_top) - region_base + 1;
zx_status_t result =
func(region_base, region_len,
current.IsValid() ? current->arch_mmu_flags : first_region_arch_mmu_flags_);
if (result != ZX_ERR_NEXT) {
if (result == ZX_ERR_STOP) {
return ZX_OK;
}
return result;
}
// Move to the next block.
current = next;
next++;
// Continue looping as long we operating on nodes that overlap with the requested range.
} while (current.IsValid() && current->region_start <= range_top);
return ZX_OK;
}
template <typename F>
zx_status_t MappingProtectionRanges::UpdateProtectionRange(vaddr_t mapping_base,
size_t mapping_size, vaddr_t base,
size_t size, uint new_arch_mmu_flags,
F callback) {
// If we're changing the whole mapping, just make the change.
if (mapping_base == base && mapping_size == size) {
protect_region_list_rest_.clear();
callback(base, size, first_region_arch_mmu_flags_);
first_region_arch_mmu_flags_ = new_arch_mmu_flags;
return ZX_OK;
}
// Find the range of nodes that will need deleting.
auto first = protect_region_list_rest_.lower_bound(base);
auto last = protect_region_list_rest_.upper_bound(base + (size - 1));
// Work the flags in the regions before the first/last nodes. We need to cache these flags so that
// once we are inserting the new protection nodes, we do not insert nodes such that we would cause
// two regions to have the same flags (which would be redundant).
const uint start_carry_flags = FlagsForPreviousRegion(first);
const uint end_carry_flags = FlagsForPreviousRegion(last);
// Determine how many new nodes we are going to need so we can allocate up front. This ensures
// that after we have deleted nodes from the tree (and destroyed information) we do not have to
// do an allocation that might fail and leave us in an unrecoverable state. However, we would
// like to avoid actually performing allocations as far as possible, so do the following
// 1. Count how many nodes will be needed to represent the new protection range (after the nodes
// between first,last have been deleted. As a protection range has two points, a start and an
// end, the most nodes we can ever possibly need is two.
// 2. Of these new nodes we will need, work out how many we can reuse from deletion.
// 3. Allocate the remainder.
ktl::optional<ktl::unique_ptr<ProtectNode>> protect_nodes[2];
const uint total_nodes_needed = NodeAllocationsForRange(mapping_base, mapping_size, base, size,
first, last, new_arch_mmu_flags);
uint nodes_needed = total_nodes_needed;
// First see how many of the nodes we will be able to get by erasing and can reuse.
for (auto it = first; nodes_needed > 0 && it != last; it++) {
nodes_needed--;
}
// If there are any nodes_needed still, allocate them so that they are available.
uint nodes_available = 0;
// Allocate any remaining nodes_needed that we will not fulfill from deletions.
while (nodes_available < nodes_needed) {
fbl::AllocChecker ac;
ktl::unique_ptr<ProtectNode> new_node(ktl::make_unique<ProtectNode>(&ac));
if (!ac.check()) {
return ZX_ERR_NO_MEMORY;
}
protect_nodes[nodes_available++].emplace(ktl::move(new_node));
}
// Now that we have done all memory allocations and know that we cannot fail start the destructive
// part and erase any nodes in the the range as well as call the provided callback with the old
// data.
{
vaddr_t old_start = base;
uint old_flags = start_carry_flags;
while (first != last) {
// On the first iteration if the range is aligned to a node then we skip, since we do not want
// to do the callback for a zero sized range.
if (old_start != first->region_start) {
callback(old_start, first->region_start - old_start, old_flags);
}
old_start = first->region_start;
old_flags = first->arch_mmu_flags;
auto node = protect_region_list_rest_.erase(first++);
if (nodes_available < total_nodes_needed) {
protect_nodes[nodes_available++].emplace(ktl::move(node));
}
}
// If the range was not aligned to a node then process any remainder.
if (old_start <= base + (size - 1)) {
callback(old_start, base + size - old_start, old_flags);
}
}
// At this point we should now have all the nodes.
DEBUG_ASSERT(total_nodes_needed == nodes_available);
// Check if we are updating the implicit first node, which just involves changing
// first_region_arch_mmu_flags_, or if there's a protection change that requires a node insertion.
if (base == mapping_base) {
first_region_arch_mmu_flags_ = new_arch_mmu_flags;
} else if (start_carry_flags != new_arch_mmu_flags) {
ASSERT(nodes_available > 0);
auto node = ktl::move(protect_nodes[--nodes_available].value());
node->region_start = base;
node->arch_mmu_flags = new_arch_mmu_flags;
protect_region_list_rest_.insert(ktl::move(node));
}
// To create the end of the region we first check if there is a gap between the end of this region
// and the start of the next region. Additionally this needs to handle the case where there is no
// next node in the tree, and so we have to check against mapping limit of mapping_base +
// mapping_size.
const uint64_t next_region_start =
last.IsValid() ? last->region_start : (mapping_base + mapping_size);
if (next_region_start != base + size) {
// There is a gap to the next node so we need to make sure it keeps its old protection value,
// end_carry_flags. However, it could have ended up that these flags are what we are protecting
// to, in which case a new node isn't needed as we can just effectively merge the gap into this
// protection range.
if (end_carry_flags != new_arch_mmu_flags) {
ASSERT(nodes_available > 0);
auto node = ktl::move(protect_nodes[--nodes_available].value());
node->region_start = base + size;
node->arch_mmu_flags = end_carry_flags;
protect_region_list_rest_.insert(ktl::move(node));
// Since we are essentially moving forward a node that we previously deleted, to essentially
// shrink the previous protection range, we know that there is no merging needed with the next
// node.
DEBUG_ASSERT(!last.IsValid() || last->arch_mmu_flags != end_carry_flags);
}
} else if (last.IsValid() && last->arch_mmu_flags == new_arch_mmu_flags) {
// From the previous `if` block we know that if last.IsValid is true, then the end of the region
// being protected is last->region_start. If this next region happens to have the same flags as
// what we just protected, then we need to drop this node.
protect_region_list_rest_.erase(last);
}
// We should not have allocated more nodes than we needed, this indicates a bug in the calculation
// logic.
DEBUG_ASSERT(nodes_available == 0);
return ZX_OK;
}
uint MappingProtectionRanges::MmuFlagsForWavlRegion(vaddr_t vaddr) const {
DEBUG_ASSERT(!protect_region_list_rest_.is_empty());
auto it = --protect_region_list_rest_.upper_bound(vaddr);
if (it.IsValid()) {
DEBUG_ASSERT(it->region_start <= vaddr);
return it->arch_mmu_flags;
} else {
DEBUG_ASSERT(protect_region_list_rest_.begin()->region_start > vaddr);
return first_region_arch_mmu_flags_;
}
}
// Counts how many nodes would need to be allocated for a protection range. This calculation is
// based of whether there are actually changes in the protection type that require a node to be
// added.
uint MappingProtectionRanges::NodeAllocationsForRange(vaddr_t mapping_base, size_t mapping_size,
vaddr_t base, size_t size,
RegionList::iterator removal_start,
RegionList::iterator removal_end,
uint new_mmu_flags) const {
uint nodes_needed = 0;
// Check if we will need a node at the start. if base==base_ then we will just be changing the
// first_region_arch_mmu_flags_, otherwise we need a node if we're actually causing a protection
// change.
if (base != mapping_base && FlagsForPreviousRegion(removal_start) != new_mmu_flags) {
nodes_needed++;
}
// The node for the end of the region is needed under two conditions
// 1. There will be a non-zero gap between the end of our new region and the start of the next
// existing region.
// 2. This non-zero sized gap is of a different protection type.
const uint64_t next_region_start =
removal_end.IsValid() ? removal_end->region_start : (mapping_base + mapping_size);
if (next_region_start != base + size && FlagsForPreviousRegion(removal_end) != new_mmu_flags) {
nodes_needed++;
}
return nodes_needed;
}
zx_status_t MappingProtectionRanges::MergeRightNeighbor(MappingProtectionRanges& right,
vaddr_t merge_addr) {
// We need to insert a node if the protection type of the end of the left mapping is not the
// same as the protection type of the start of the right mapping.
if (FlagsForPreviousRegion(protect_region_list_rest_.end()) !=
right.first_region_arch_mmu_flags_) {
fbl::AllocChecker ac;
ktl::unique_ptr<ProtectNode> region =
ktl::make_unique<ProtectNode>(&ac, merge_addr, right.first_region_arch_mmu_flags_);
if (!ac.check()) {
// No state has changed yet, so even though we do not forward up an error it is safe to just
// not merge.
TRACEF("Aborted region merge due to out of memory\n");
return ZX_ERR_NO_MEMORY;
}
protect_region_list_rest_.insert(ktl::move(region));
}
// Carry over any remaining regions.
while (!right.protect_region_list_rest_.is_empty()) {
protect_region_list_rest_.insert(right.protect_region_list_rest_.pop_front());
}
return ZX_OK;
}
MappingProtectionRanges MappingProtectionRanges::SplitAt(vaddr_t split) {
// Determine the mmu flags the right most mapping would start at.
auto right_nodes = protect_region_list_rest_.upper_bound(split);
const uint right_mmu_flags = FlagsForPreviousRegion(right_nodes);
MappingProtectionRanges ranges(right_mmu_flags);
// Move any protect regions into the right half.
while (right_nodes != protect_region_list_rest_.end()) {
ranges.protect_region_list_rest_.insert(protect_region_list_rest_.erase(right_nodes++));
}
return ranges;
}
void MappingProtectionRanges::DiscardBelow(vaddr_t addr) {
auto last = protect_region_list_rest_.upper_bound(addr);
while (protect_region_list_rest_.begin() != last) {
first_region_arch_mmu_flags_ = protect_region_list_rest_.pop_front()->arch_mmu_flags;
}
}
void MappingProtectionRanges::DiscardAbove(vaddr_t addr) {
for (auto it = protect_region_list_rest_.lower_bound(addr);
it != protect_region_list_rest_.end();) {
protect_region_list_rest_.erase(it++);
}
}
bool MappingProtectionRanges::DebugNodesWithinRange(vaddr_t mapping_base, size_t mapping_size) {
if (protect_region_list_rest_.is_empty()) {
return true;
}
if (protect_region_list_rest_.begin()->region_start < mapping_base) {
return false;
}
if ((--protect_region_list_rest_.end())->region_start >= mapping_base + mapping_size) {
return false;
}
return true;
}