(2010/04/12 2:15), Avi Kivity wrote:
On 04/09/2010 12:35 PM, Takuya Yoshikawa wrote:
Currently, x86 vmalloc()s a dirty bitmap every time when we swich
to the next dirty bitmap. To avoid this, we use the double buffering
technique: we also move the bitmaps to userspace, so that extra
bitmaps will not use the precious kernel resource.
This idea is based on Avi's suggestion.
Please fold this into the next patch. Introducing new data members
without their users is hard to follow.
OK.
#define KVM_MAX_VCPUS 64
#define KVM_MEMORY_SLOTS 32
/* memory slots that does not exposed to userspace */
diff --git a/include/linux/kvm_host.h b/include/linux/kvm_host.h
index dd6bcf4..07092d6 100644
--- a/include/linux/kvm_host.h
+++ b/include/linux/kvm_host.h
@@ -110,7 +110,13 @@ struct kvm_memory_slot {
unsigned long npages;
unsigned long flags;
unsigned long *rmap;
+#ifndef __KVM_HAVE_USER_DIRTYBITMAP
unsigned long *dirty_bitmap;
+#else
+ unsigned long __user *dirty_bitmap;
+ unsigned long __user *dirty_bitmap_old;
+ bool is_dirty;
+#endif
And, if we make set_user_bit() generic, we don't need the ifdefs.
OK, but we have one problem: ia64. I checked all architectures' dirty bitmap
implementations and thought generalizing this work is not so hard except for
ia64. It's already too different from other parts.
#ifdef CONFIG_IA64
unsigned long *dirty_bitmap;
#else
...
#endif
is acceptable?
--
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
