[Bug,37262] VNC access to KVM VM is unusable

Commit Message

Xiao Guangrong June 17, 2011, 2:25 a.m. UTC

On 06/17/2011 04:39 AM, [e-mail] wrote:
> https://bugzilla.kernel.org/show_bug.cgi?id=37262
> 
> 
> 
> 
> 
> --- Comment #43 from Steve <[e-mail]>  2011-06-16 20:39:32 ---
> Solution of this bug is very simple:
> 
> arch/x86/kvm/mmu.c:
> 
> static bool mapping_level_dirty_bitmap(struct kvm_vcpu *vcpu, gfn_t large_gfn)
> {
> return (gfn_to_memslot_dirty_bitmap(vcpu, large_gfn, true) == NULL)
> ? true : false;
> }
> 
> Only opposite condition.
> Please commit when it's possible.
> 

Yes, it is the reason, thanks for your time and great job, i am so
sorry, this is caused by my mistake :-(

And this is the fix patch from you and i did little cleanup

From: Steve <[e-mail]>
Subject: [PATCH] KVM: MMU: fix opposite condition in mapping_level_dirty_bitmap

The condition is opposite, it always maps huge page for the dirty tracked page

Reported-by: Steve <[e-mail]>
Signed-off-by: Steve <[e-mail]>
---
 arch/x86/kvm/mmu.c |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

Comments

Avi Kivity June 19, 2011, 8:50 a.m. UTC | #1

On 06/17/2011 05:25 AM, Xiao Guangrong wrote:
> From: Steve<[e-mail]>
> Subject: [PATCH] KVM: MMU: fix opposite condition in mapping_level_dirty_bitmap
>
> The condition is opposite, it always maps huge page for the dirty tracked page
>

Thanks, applied and queued.

Patch

diff --git a/arch/x86/kvm/mmu.c b/arch/x86/kvm/mmu.c
index 19ebb76..15afa1e 100644
--- a/arch/x86/kvm/mmu.c
+++ b/arch/x86/kvm/mmu.c
@@ -549,7 +549,7 @@  gfn_to_memslot_dirty_bitmap(struct kvm_vcpu *vcpu, gfn_t gfn,
 
 static bool mapping_level_dirty_bitmap(struct kvm_vcpu *vcpu, gfn_t large_gfn)
 {
-	return gfn_to_memslot_dirty_bitmap(vcpu, large_gfn, true);
+	return !gfn_to_memslot_dirty_bitmap(vcpu, large_gfn, true);
 }
 
 static int mapping_level(struct kvm_vcpu *vcpu, gfn_t large_gfn)