By this patch, bitmap allocation is replaced with do_mmap() and
bitmap manipulation is replaced with *_user() functions.
Note that this does not change the APIs between kernel and user space.
To get more advantage from this hack, we need to add a new interface
for triggering the bitmap swith and getting the bitmap addresses: the
addresses is in user space and we can export them to qemu.
TODO:
1. We want to use copy_in_user() for 32bit case too.
Note that this is only for the compatibility issue: in the future,
we hope, qemu will not need to use this ioctl.
2. We have to implement test_bit_user() to avoid extra set_bit.
Signed-off-by: Takuya Yoshikawa <yoshikawa.takuya@xxxxxxxxxxxxx>
Signed-off-by: Fernando Luis Vazquez Cao <fernando@xxxxxxxxxxxxx>
---
arch/x86/kvm/x86.c | 118 +++++++++++++++++++++++++++++++++++++--------
include/linux/kvm_host.h | 4 ++
virt/kvm/kvm_main.c | 30 +++++++++++-
3 files changed, 130 insertions(+), 22 deletions(-)
diff --git a/arch/x86/kvm/x86.c b/arch/x86/kvm/x86.c
index 450ecfe..995b970 100644
--- a/arch/x86/kvm/x86.c
+++ b/arch/x86/kvm/x86.c
@@ -2642,16 +2642,99 @@ static int kvm_vm_ioctl_reinject(struct kvm *kvm,
return 0;
}
+int kvm_arch_create_dirty_bitmap(struct kvm_memory_slot *memslot)
+{
+ unsigned long user_addr1;
+ unsigned long user_addr2;
+ int dirty_bytes = kvm_dirty_bitmap_bytes(memslot);
+
+ down_write(¤t->mm->mmap_sem);
+ user_addr1 = do_mmap(NULL, 0, dirty_bytes,
+ PROT_READ | PROT_WRITE,
+ MAP_PRIVATE | MAP_ANONYMOUS, 0);
+ if (IS_ERR((void *)user_addr1)) {
+ up_write(¤t->mm->mmap_sem);
+ return PTR_ERR((void *)user_addr1);
+ }
+ user_addr2 = do_mmap(NULL, 0, dirty_bytes,
+ PROT_READ | PROT_WRITE,
+ MAP_PRIVATE | MAP_ANONYMOUS, 0);
+ if (IS_ERR((void *)user_addr2)) {
+ do_munmap(current->mm, user_addr1, dirty_bytes);
+ up_write(¤t->mm->mmap_sem);
+ return PTR_ERR((void *)user_addr2);
+ }
+ up_write(¤t->mm->mmap_sem);
+
+ memslot->dirty_bitmap = (unsigned long __user *)user_addr1;
+ memslot->dirty_bitmap_old = (unsigned long __user *)user_addr2;
+ clear_user(memslot->dirty_bitmap, dirty_bytes);
+ clear_user(memslot->dirty_bitmap_old, dirty_bytes);
+
+ return 0;
+}
+
+void kvm_arch_destroy_dirty_bitmap(struct kvm_memory_slot *memslot)
+{
+ int n = kvm_dirty_bitmap_bytes(memslot);
+
+ if (!memslot->dirty_bitmap)
+ return;
+
+ down_write(¤t->mm->mmap_sem);
+ do_munmap(current->mm, (unsigned long)memslot->dirty_bitmap, n);
+ do_munmap(current->mm, (unsigned long)memslot->dirty_bitmap_old, n);
+ up_write(¤t->mm->mmap_sem);
+
+ memslot->dirty_bitmap = NULL;
+ memslot->dirty_bitmap_old = NULL;
+}
+
+static int kvm_copy_dirty_bitmap(unsigned long __user *to,
+ const unsigned long __user *from, int n)
+{
+#ifdef CONFIG_X86_64
+ if (copy_in_user(to, from, n) < 0) {
+ printk(KERN_WARNING "%s: copy_in_user failed\n", __func__);
+ return -EFAULT;
+ }
+ return 0;
+#else
+ int ret = 0;
+ void *p = vmalloc(n);
+
+ if (!p) {
+ ret = -ENOMEM;
+ goto out;
+ }
+ if (copy_from_user(p, from, n) < 0) {
+ printk(KERN_WARNING "%s: copy_from_user failed\n", __func__);
+ ret = -EFAULT;
+ goto out_free;
+ }
+ if (copy_to_user(to, p, n) < 0) {
+ printk(KERN_WARNING "%s: copy_to_user failed\n", __func__);
+ ret = -EFAULT;
+ goto out_free;
+ }
+
+out_free:
+ vfree(p);
+out:
+ return ret;
+#endif
+}
+
/*
* Get (and clear) the dirty memory log for a memory slot.
*/
int kvm_vm_ioctl_get_dirty_log(struct kvm *kvm,
struct kvm_dirty_log *log)
{
- int r, n, i;
+ int r, n;
struct kvm_memory_slot *memslot;
- unsigned long is_dirty = 0;
- unsigned long *dirty_bitmap = NULL;
+ unsigned long __user *dirty_bitmap;
+ unsigned long __user *dirty_bitmap_old;
mutex_lock(&kvm->slots_lock);
@@ -2664,44 +2747,37 @@ int kvm_vm_ioctl_get_dirty_log(struct kvm *kvm,
if (!memslot->dirty_bitmap)
goto out;
- n = kvm_dirty_bitmap_bytes(memslot);
-
- r = -ENOMEM;
- dirty_bitmap = vmalloc(n);
- if (!dirty_bitmap)
- goto out;
- memset(dirty_bitmap, 0, n);
+ dirty_bitmap = memslot->dirty_bitmap;
+ dirty_bitmap_old = memslot->dirty_bitmap_old;
- for (i = 0; !is_dirty && i < n/sizeof(long); i++)
- is_dirty = memslot->dirty_bitmap[i];
+ n = kvm_dirty_bitmap_bytes(memslot);
+ clear_user(dirty_bitmap_old, n);
/* If nothing is dirty, don't bother messing with page tables. */
- if (is_dirty) {
+ if (memslot->is_dirty) {
struct kvm_memslots *slots, *old_slots;
spin_lock(&kvm->mmu_lock);
kvm_mmu_slot_remove_write_access(kvm, log->slot);
spin_unlock(&kvm->mmu_lock);
+ r = -ENOMEM;
slots = kzalloc(sizeof(struct kvm_memslots), GFP_KERNEL);
if (!slots)
- goto out_free;
+ goto out;
memcpy(slots, kvm->memslots, sizeof(struct kvm_memslots));
- slots->memslots[log->slot].dirty_bitmap = dirty_bitmap;
+ slots->memslots[log->slot].dirty_bitmap = dirty_bitmap_old;
+ slots->memslots[log->slot].dirty_bitmap_old = dirty_bitmap;
+ slots->memslots[log->slot].is_dirty = false;
old_slots = kvm->memslots;
rcu_assign_pointer(kvm->memslots, slots);
synchronize_srcu_expedited(&kvm->srcu);
- dirty_bitmap = old_slots->memslots[log->slot].dirty_bitmap;
kfree(old_slots);
}
- r = 0;
- if (copy_to_user(log->dirty_bitmap, dirty_bitmap, n))
- r = -EFAULT;
-out_free:
- vfree(dirty_bitmap);
+ r = kvm_copy_dirty_bitmap(log->dirty_bitmap, dirty_bitmap, n);
out:
mutex_unlock(&kvm->slots_lock);
return r;
diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index 07092d6..834812f 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -276,6 +276,10 @@ void kvm_arch_commit_memory_region(struct kvm *kvm,
struct kvm_userspace_memory_region *mem,
struct kvm_memory_slot old,
int user_alloc);
+#ifdef __KVM_HAVE_USER_DIRTYBITMAP
+int kvm_arch_create_dirty_bitmap(struct kvm_memory_slot *memslot);
+void kvm_arch_destroy_dirty_bitmap(struct kvm_memory_slot *memslot);
+#endif
void kvm_disable_largepages(void);
void kvm_arch_flush_shadow(struct kvm *kvm);
gfn_t unalias_gfn(struct kvm *kvm, gfn_t gfn);
diff --git a/virt/kvm/kvm_main.c b/virt/kvm/kvm_main.c
index f919bd1..038a677 100644
--- a/virt/kvm/kvm_main.c
+++ b/virt/kvm/kvm_main.c
@@ -433,8 +433,12 @@ out_err_nodisable:
static void kvm_destroy_dirty_bitmap(struct kvm_memory_slot *memslot)
{
+#ifdef __KVM_HAVE_USER_DIRTYBITMAP
+ kvm_arch_destroy_dirty_bitmap(memslot);
+#else
vfree(memslot->dirty_bitmap);
memslot->dirty_bitmap = NULL;
+#endif
}
/*
@@ -463,13 +467,26 @@ static void kvm_free_physmem_slot(struct kvm_memory_slot *free,
free->rmap = NULL;
}
+/*
+ * We don't munmap dirty bitmaps by ourselves in the case of vm destruction.
+ */
+static void kvm_pre_destroy_dirty_bitmap(struct kvm_memory_slot *memslot)
+{
+#ifdef __KVM_HAVE_USER_DIRTYBITMAP
+ memslot->dirty_bitmap = NULL;
+ memslot->dirty_bitmap_old = NULL;
+#endif
+}
+
void kvm_free_physmem(struct kvm *kvm)
{
int i;
struct kvm_memslots *slots = kvm->memslots;
- for (i = 0; i < slots->nmemslots; ++i)
+ for (i = 0; i < slots->nmemslots; ++i) {
+ kvm_pre_destroy_dirty_bitmap(&slots->memslots[i]);
kvm_free_physmem_slot(&slots->memslots[i], NULL);
+ }
kfree(kvm->memslots);
}
@@ -523,6 +540,9 @@ static int kvm_vm_release(struct inode *inode, struct file *filp)
static int kvm_create_dirty_bitmap(struct kvm_memory_slot *memslot)
{
+#ifdef __KVM_HAVE_USER_DIRTYBITMAP
+ return kvm_arch_create_dirty_bitmap(memslot);
+#else
int dirty_bytes = kvm_dirty_bitmap_bytes(memslot);
memslot->dirty_bitmap = vmalloc(dirty_bytes);
@@ -530,6 +550,7 @@ static int kvm_create_dirty_bitmap(struct kvm_memory_slot *memslot)
return -ENOMEM;
memset(memslot->dirty_bitmap, 0, dirty_bytes);
+#endif
return 0;
}
@@ -1197,9 +1218,16 @@ void mark_page_dirty(struct kvm *kvm, gfn_t gfn)
if (memslot && memslot->dirty_bitmap) {
unsigned long rel_gfn = gfn - memslot->base_gfn;
+#ifdef __KVM_HAVE_USER_DIRTYBITMAP
+ if (set_bit_user(rel_gfn, memslot->dirty_bitmap) < 0)
+ printk(KERN_WARNING "%s: set_bit_user failed\n", __func__);
+
+ memslot->is_dirty = true;
+#else
/* avoid RMW */
if (!generic_test_le_bit(rel_gfn, memslot->dirty_bitmap))
generic___set_le_bit(rel_gfn, memslot->dirty_bitmap);
+#endif
}
}
--
1.6.3.3
--
To unsubscribe from this list: send the line "unsubscribe kvm" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
