Skip to content

Coalesce memory regions to increase performance #24

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
neonbyte1 wants to merge 1 commit into
base: main
Choose a base branch
from
+31 −5
Open

Coalesce memory regions to increase performance #24

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
36 changes: 31 additions & 5 deletions memflow-kmod/vmtools.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -124,11 +124,34 @@ static int get_sorted_memslots(struct kvm_memslots *slots, int max_slots, vm_mem

sort(slots_out, slot_count, sizeof(*slots_out), memslot_compare, NULL);

//TODO: coalesce nearby slots

return slot_count;
}

static int coalesce_sorted_memslots(vm_memslot_t *slots, int slot_count)
{
int write_idx = 0;
int read_idx;

if (slot_count <= 1)
return slot_count;

for (read_idx = 1; read_idx < slot_count; read_idx++) {
vm_memslot_t *cur = &slots[write_idx];
vm_memslot_t *next = &slots[read_idx];

if (cur->base + cur->map_size == next->base && cur->host_base + cur->map_size == next->host_base) {
cur->map_size += next->map_size;

continue;
}

if (++write_idx != read_idx)
slots[write_idx] = *next;
}

return ++write_idx;
}

static int get_vm_info(struct kvm *kvm, vm_info_t __user *user_info)
{
vm_info_t kernel_info;
Expand Down Expand Up @@ -170,6 +193,8 @@ static int get_vm_info(struct kvm *kvm, vm_info_t __user *user_info)
if ((slot_count = get_sorted_memslots(kvm_memslots(kvm), kernel_info.slot_count, memslot_map)) == -1)
goto free_slots;

slot_count = coalesce_sorted_memslots(memslot_map, slot_count);

kernel_info.userspace_pid = kvm->userspace_pid;
kernel_info.slot_count = slot_count;

Expand Down Expand Up @@ -483,7 +508,7 @@ static int do_map_vm(struct kvm *kvm, vm_map_info_t __user *user_info)

if (copy_from_user(&user_info_copied, user_info, sizeof(vm_map_info_t)))
goto free_alloc;

priv->vm_map_info = user_info_copied;
priv->vm_map_info.slots = priv->map_slots;

Expand All @@ -510,7 +535,7 @@ static int do_map_vm(struct kvm *kvm, vm_map_info_t __user *user_info)
priv->vm_map_info.slot_count = memslot_count;

mmap_write_lock(other_mm);

// Once we hold mmap_sem, the slots won't be freed so there is no purpose to hold the locks
mutex_unlock(&kvm->lock);
mutex_unlock(&kvm->slots_lock);
Expand All @@ -526,6 +551,8 @@ static int do_map_vm(struct kvm *kvm, vm_map_info_t __user *user_info)
// Now remap all unique mappings
remap_vmas(priv, other_task);

priv->vm_map_info.slot_count = coalesce_sorted_memslots(priv->vm_map_info.slots, priv->vm_map_info.slot_count);

if (!priv->mapped_vma_count)
goto release_file;

Expand Down Expand Up @@ -574,4 +601,3 @@ static int memflow_vm_mapped_release(struct inode *inode, struct file *filp)
vfree(filp->private_data);
return 0;
}