RGBCube/serenity
1
Fork 0
mirror of https://github.com/RGBCube/serenity synced 2025-05-31 06:58:11 +00:00
Code Issues Activity Actions

Kernel: Don't COW volatile VM objects

Browse source 
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.
This commit is contained in:
Andreas Kling 2021-07-25 02:27:43 +02:00
parent 297c0748f0
commit 5fb91e2e84
1 changed files with 14 additions and 7 deletions
Download patch file Download diff file Expand all files Collapse all files

21
Kernel/VM/AnonymousVMObject.cpp
View file

@ -18,17 +18,21 @@ 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 = 0;
if (is_volatile()) {
// NOTE: If this object is currently volatile, we don't own any committed pages.
} else {
new_cow_pages_needed = page_count();
}
size_t new_cow_pages_needed = page_count();
dbgln_if(COMMIT_DEBUG, "Cloning {:p}, need {} committed cow pages", this, new_cow_pages_needed);
@ -205,11 +209,14 @@ KResult AnonymousVMObject::set_volatile(bool is_volatile, bool& was_purged)
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;