ladybird/Kernel/VM/AnonymousVMObject.cpp
Andreas Kling 5fb91e2e84 Kernel: Don't COW volatile VM objects
If a purgeable VM object is in the "volatile" state when we're asked
to make a COW clone of it, make life simpler by simply "purging"
the cloned object right away.

This effectively means that a fork()'ed child process will discover
its purgeable+volatile regions to be empty if/when it tries making
them non-volatile.
2021-07-25 17:28:05 +02:00

389 lines
13 KiB
C++

/*
* Copyright (c) 2018-2021, Andreas Kling <kling@serenityos.org>
*
* SPDX-License-Identifier: BSD-2-Clause
*/
#include <Kernel/Arch/x86/SmapDisabler.h>
#include <Kernel/Debug.h>
#include <Kernel/Process.h>
#include <Kernel/VM/AnonymousVMObject.h>
#include <Kernel/VM/MemoryManager.h>
#include <Kernel/VM/PhysicalPage.h>
namespace Kernel {
RefPtr<VMObject> AnonymousVMObject::try_clone()
{
// We need to acquire our lock so we copy a sane state
ScopedSpinLock lock(m_lock);
if (is_purgeable() && is_volatile()) {
// If this object is purgeable+volatile, create a new zero-filled purgeable+volatile
// object, effectively "pre-purging" it in the child process.
auto clone = try_create_purgeable_with_size(size(), AllocationStrategy::None);
if (!clone)
return {};
clone->m_volatile = true;
return clone;
}
// We're the parent. Since we're about to become COW we need to
// commit the number of pages that we need to potentially allocate
// so that the parent is still guaranteed to be able to have all
// non-volatile memory available.
size_t new_cow_pages_needed = page_count();
dbgln_if(COMMIT_DEBUG, "Cloning {:p}, need {} committed cow pages", this, new_cow_pages_needed);
if (!MM.commit_user_physical_pages(new_cow_pages_needed))
return {};
// Create or replace the committed cow pages. When cloning a previously
// cloned vmobject, we want to essentially "fork", leaving us and the
// new clone with one set of shared committed cow pages, and the original
// one would keep the one it still has. This ensures that the original
// one and this one, as well as the clone have sufficient resources
// to cow all pages as needed
m_shared_committed_cow_pages = try_create<CommittedCowPages>(new_cow_pages_needed);
if (!m_shared_committed_cow_pages) {
MM.uncommit_user_physical_pages(new_cow_pages_needed);
return {};
}
// Both original and clone become COW. So create a COW map for ourselves
// or reset all pages to be copied again if we were previously cloned
ensure_or_reset_cow_map();
// FIXME: If this allocation fails, we need to rollback all changes.
return adopt_ref_if_nonnull(new (nothrow) AnonymousVMObject(*this));
}
RefPtr<AnonymousVMObject> AnonymousVMObject::try_create_with_size(size_t size, AllocationStrategy commit)
{
if (commit == AllocationStrategy::Reserve || commit == AllocationStrategy::AllocateNow) {
// We need to attempt to commit before actually creating the object
if (!MM.commit_user_physical_pages(ceil_div(size, static_cast<size_t>(PAGE_SIZE))))
return {};
}
return adopt_ref_if_nonnull(new (nothrow) AnonymousVMObject(size, commit));
}
RefPtr<AnonymousVMObject> AnonymousVMObject::try_create_purgeable_with_size(size_t size, AllocationStrategy commit)
{
if (commit == AllocationStrategy::Reserve || commit == AllocationStrategy::AllocateNow) {
// We need to attempt to commit before actually creating the object
if (!MM.commit_user_physical_pages(ceil_div(size, static_cast<size_t>(PAGE_SIZE))))
return {};
}
auto vmobject = adopt_ref_if_nonnull(new (nothrow) AnonymousVMObject(size, commit));
if (!vmobject)
return {};
vmobject->m_purgeable = true;
return vmobject;
}
RefPtr<AnonymousVMObject> AnonymousVMObject::try_create_with_physical_pages(Span<NonnullRefPtr<PhysicalPage>> physical_pages)
{
return adopt_ref_if_nonnull(new (nothrow) AnonymousVMObject(physical_pages));
}
RefPtr<AnonymousVMObject> AnonymousVMObject::try_create_for_physical_range(PhysicalAddress paddr, size_t size)
{
if (paddr.offset(size) < paddr) {
dbgln("Shenanigans! try_create_for_physical_range({}, {}) would wrap around", paddr, size);
return nullptr;
}
return adopt_ref_if_nonnull(new (nothrow) AnonymousVMObject(paddr, size));
}
AnonymousVMObject::AnonymousVMObject(size_t size, AllocationStrategy strategy)
: VMObject(size)
, m_unused_committed_pages(strategy == AllocationStrategy::Reserve ? page_count() : 0)
{
if (strategy == AllocationStrategy::AllocateNow) {
// Allocate all pages right now. We know we can get all because we committed the amount needed
for (size_t i = 0; i < page_count(); ++i)
physical_pages()[i] = MM.allocate_committed_user_physical_page(MemoryManager::ShouldZeroFill::Yes);
} else {
auto& initial_page = (strategy == AllocationStrategy::Reserve) ? MM.lazy_committed_page() : MM.shared_zero_page();
for (size_t i = 0; i < page_count(); ++i)
physical_pages()[i] = initial_page;
}
}
AnonymousVMObject::AnonymousVMObject(PhysicalAddress paddr, size_t size)
: VMObject(size)
{
VERIFY(paddr.page_base() == paddr);
for (size_t i = 0; i < page_count(); ++i)
physical_pages()[i] = PhysicalPage::create(paddr.offset(i * PAGE_SIZE), MayReturnToFreeList::No);
}
AnonymousVMObject::AnonymousVMObject(Span<NonnullRefPtr<PhysicalPage>> physical_pages)
: VMObject(physical_pages.size() * PAGE_SIZE)
{
for (size_t i = 0; i < physical_pages.size(); ++i) {
m_physical_pages[i] = physical_pages[i];
}
}
AnonymousVMObject::AnonymousVMObject(AnonymousVMObject const& other)
: VMObject(other)
, m_unused_committed_pages(other.m_unused_committed_pages)
, m_cow_map() // do *not* clone this
, m_shared_committed_cow_pages(other.m_shared_committed_cow_pages) // share the pool
{
// We can't really "copy" a spinlock. But we're holding it. Clear in the clone
VERIFY(other.m_lock.is_locked());
m_lock.initialize();
// The clone also becomes COW
ensure_or_reset_cow_map();
if (m_unused_committed_pages > 0) {
// The original vmobject didn't use up all committed pages. When
// cloning (fork) we will overcommit. For this purpose we drop all
// lazy-commit references and replace them with shared zero pages.
for (size_t i = 0; i < page_count(); i++) {
auto& phys_page = m_physical_pages[i];
if (phys_page && phys_page->is_lazy_committed_page()) {
phys_page = MM.shared_zero_page();
if (--m_unused_committed_pages == 0)
break;
}
}
VERIFY(m_unused_committed_pages == 0);
}
}
AnonymousVMObject::~AnonymousVMObject()
{
// Return any unused committed pages
if (m_unused_committed_pages > 0)
MM.uncommit_user_physical_pages(m_unused_committed_pages);
}
size_t AnonymousVMObject::purge()
{
ScopedSpinLock lock(m_lock);
if (!is_purgeable() || !is_volatile())
return 0;
size_t total_pages_purged = 0;
for (auto& page : m_physical_pages) {
VERIFY(page);
if (page->is_shared_zero_page())
continue;
page = MM.shared_zero_page();
++total_pages_purged;
}
m_was_purged = true;
for_each_region([](Region& region) {
region.remap();
});
return total_pages_purged;
}
KResult AnonymousVMObject::set_volatile(bool is_volatile, bool& was_purged)
{
VERIFY(is_purgeable());
ScopedSpinLock locker(m_lock);
was_purged = m_was_purged;
if (m_volatile == is_volatile)
return KSuccess;
if (is_volatile) {
// When a VMObject is made volatile, it gives up all of its committed memory.
// Any physical pages already allocated remain in the VMObject for now, but the kernel is free to take them at any moment.
for (auto& page : m_physical_pages) {
if (page && page->is_lazy_committed_page())
page = MM.shared_zero_page();
}
if (m_unused_committed_pages) {
MM.uncommit_user_physical_pages(m_unused_committed_pages);
m_unused_committed_pages = 0;
}
m_shared_committed_cow_pages = nullptr;
m_volatile = true;
m_was_purged = false;
return KSuccess;
}
// When a VMObject is made non-volatile, we try to commit however many pages are not currently available.
// If that fails, we return false to indicate that memory allocation failed.
size_t committed_pages_needed = 0;
for (auto& page : m_physical_pages) {
VERIFY(page);
if (page->is_shared_zero_page())
++committed_pages_needed;
}
if (!committed_pages_needed) {
m_volatile = false;
return KSuccess;
}
if (!MM.commit_user_physical_pages(committed_pages_needed))
return ENOMEM;
m_unused_committed_pages = committed_pages_needed;
for (auto& page : m_physical_pages) {
if (page->is_shared_zero_page())
page = MM.lazy_committed_page();
}
m_volatile = false;
m_was_purged = false;
return KSuccess;
}
NonnullRefPtr<PhysicalPage> AnonymousVMObject::allocate_committed_page(Badge<Region>)
{
{
ScopedSpinLock lock(m_lock);
VERIFY(m_unused_committed_pages > 0);
--m_unused_committed_pages;
}
return MM.allocate_committed_user_physical_page(MemoryManager::ShouldZeroFill::Yes);
}
Bitmap& AnonymousVMObject::ensure_cow_map()
{
if (m_cow_map.is_null())
m_cow_map = Bitmap { page_count(), true };
return m_cow_map;
}
void AnonymousVMObject::ensure_or_reset_cow_map()
{
if (m_cow_map.is_null())
ensure_cow_map();
else
m_cow_map.fill(true);
}
bool AnonymousVMObject::should_cow(size_t page_index, bool is_shared) const
{
auto& page = physical_pages()[page_index];
if (page && (page->is_shared_zero_page() || page->is_lazy_committed_page()))
return true;
if (is_shared)
return false;
return !m_cow_map.is_null() && m_cow_map.get(page_index);
}
void AnonymousVMObject::set_should_cow(size_t page_index, bool cow)
{
ensure_cow_map().set(page_index, cow);
}
size_t AnonymousVMObject::cow_pages() const
{
if (m_cow_map.is_null())
return 0;
return m_cow_map.count_slow(true);
}
PageFaultResponse AnonymousVMObject::handle_cow_fault(size_t page_index, VirtualAddress vaddr)
{
VERIFY_INTERRUPTS_DISABLED();
ScopedSpinLock lock(m_lock);
if (is_volatile()) {
// A COW fault in a volatile region? Userspace is writing to volatile memory, this is a bug. Crash.
dbgln("COW fault in volatile region, will crash.");
return PageFaultResponse::ShouldCrash;
}
auto& page_slot = physical_pages()[page_index];
bool have_committed = m_shared_committed_cow_pages;
if (page_slot->ref_count() == 1) {
dbgln_if(PAGE_FAULT_DEBUG, " >> It's a COW page but nobody is sharing it anymore. Remap r/w");
set_should_cow(page_index, false);
if (have_committed) {
if (m_shared_committed_cow_pages->return_one())
m_shared_committed_cow_pages = nullptr;
}
return PageFaultResponse::Continue;
}
RefPtr<PhysicalPage> page;
if (have_committed) {
dbgln_if(PAGE_FAULT_DEBUG, " >> It's a committed COW page and it's time to COW!");
page = m_shared_committed_cow_pages->allocate_one();
} else {
dbgln_if(PAGE_FAULT_DEBUG, " >> It's a COW page and it's time to COW!");
page = MM.allocate_user_physical_page(MemoryManager::ShouldZeroFill::No);
if (page.is_null()) {
dmesgln("MM: handle_cow_fault was unable to allocate a physical page");
return PageFaultResponse::OutOfMemory;
}
}
u8* dest_ptr = MM.quickmap_page(*page);
dbgln_if(PAGE_FAULT_DEBUG, " >> COW {} <- {}", page->paddr(), page_slot->paddr());
{
SmapDisabler disabler;
void* fault_at;
if (!safe_memcpy(dest_ptr, vaddr.as_ptr(), PAGE_SIZE, fault_at)) {
if ((u8*)fault_at >= dest_ptr && (u8*)fault_at <= dest_ptr + PAGE_SIZE)
dbgln(" >> COW: error copying page {}/{} to {}/{}: failed to write to page at {}",
page_slot->paddr(), vaddr, page->paddr(), VirtualAddress(dest_ptr), VirtualAddress(fault_at));
else if ((u8*)fault_at >= vaddr.as_ptr() && (u8*)fault_at <= vaddr.as_ptr() + PAGE_SIZE)
dbgln(" >> COW: error copying page {}/{} to {}/{}: failed to read from page at {}",
page_slot->paddr(), vaddr, page->paddr(), VirtualAddress(dest_ptr), VirtualAddress(fault_at));
else
VERIFY_NOT_REACHED();
}
}
page_slot = move(page);
MM.unquickmap_page();
set_should_cow(page_index, false);
return PageFaultResponse::Continue;
}
CommittedCowPages::CommittedCowPages(size_t committed_pages)
: m_committed_pages(committed_pages)
{
}
CommittedCowPages::~CommittedCowPages()
{
// Return unused committed pages
if (m_committed_pages > 0)
MM.uncommit_user_physical_pages(m_committed_pages);
}
NonnullRefPtr<PhysicalPage> CommittedCowPages::allocate_one()
{
VERIFY(m_committed_pages > 0);
m_committed_pages--;
return MM.allocate_committed_user_physical_page(MemoryManager::ShouldZeroFill::Yes);
}
bool CommittedCowPages::return_one()
{
VERIFY(m_committed_pages > 0);
m_committed_pages--;
MM.uncommit_user_physical_pages(1);
return m_committed_pages == 0;
}
}