* Ingo Molnar <[EMAIL PROTECTED]> wrote:
> > Yes it does. It calls nonpaging_init_context() which calls
> > vmx_set_cr3() which promptly trashes address space of the VM that
> > previously ran on that vcpu (or, if there were none, logs a vmwrite
> > error).
>
> ok, i missed that. Nevertheless the problem of the nonatomic alloc
> remains. I guess a kvm_mmu_init() needs to be split into
> kvm_mmu_create() and kvm_mmu_setup()?
the patch below implements this fix. Lightly tested on 32-bit/VMX on
2.6.20-rc2-rt2 but seems to have done the trick.
Ingo
-------------------->
Subject: [patch] kvm: fix GFP_KERNEL allocation in atomic section in
kvm_dev_ioctl_create_vcpu()
From: Ingo Molnar <[EMAIL PROTECTED]>
fix an GFP_KERNEL allocation in atomic section:
kvm_dev_ioctl_create_vcpu() called kvm_mmu_init(), which calls
alloc_pages(), while holding the vcpu.
The fix is to set up the MMU state in two phases: kvm_mmu_create() and
kvm_mmu_setup().
(NOTE: free_vcpus does an kvm_mmu_destroy() call so there's no need
for any extra teardown branch on allocation/init failure here.)
Signed-off-by: Ingo Molnar <[EMAIL PROTECTED]>
---
drivers/kvm/kvm.h | 3 ++-
drivers/kvm/kvm_main.c | 10 ++++++----
drivers/kvm/mmu.c | 24 +++++++++---------------
3 files changed, 17 insertions(+), 20 deletions(-)
Index: linux/drivers/kvm/kvm.h
===================================================================
--- linux.orig/drivers/kvm/kvm.h
+++ linux/drivers/kvm/kvm.h
@@ -319,7 +319,8 @@ int kvm_init_arch(struct kvm_arch_ops *o
void kvm_exit_arch(void);
void kvm_mmu_destroy(struct kvm_vcpu *vcpu);
-int kvm_mmu_init(struct kvm_vcpu *vcpu);
+int kvm_mmu_create(struct kvm_vcpu *vcpu);
+int kvm_mmu_setup(struct kvm_vcpu *vcpu);
int kvm_mmu_reset_context(struct kvm_vcpu *vcpu);
void kvm_mmu_slot_remove_write_access(struct kvm *kvm, int slot);
Index: linux/drivers/kvm/kvm_main.c
===================================================================
--- linux.orig/drivers/kvm/kvm_main.c
+++ linux/drivers/kvm/kvm_main.c
@@ -522,12 +522,14 @@ static int kvm_dev_ioctl_create_vcpu(str
if (r < 0)
goto out_free_vcpus;
- kvm_arch_ops->vcpu_load(vcpu);
+ r = kvm_mmu_create(vcpu);
+ if (r < 0)
+ goto out_free_vcpus;
- r = kvm_arch_ops->vcpu_setup(vcpu);
+ kvm_arch_ops->vcpu_load(vcpu);
+ r = kvm_mmu_setup(vcpu);
if (r >= 0)
- r = kvm_mmu_init(vcpu);
-
+ r = kvm_arch_ops->vcpu_setup(vcpu);
vcpu_put(vcpu);
if (r < 0)
Index: linux/drivers/kvm/mmu.c
===================================================================
--- linux.orig/drivers/kvm/mmu.c
+++ linux/drivers/kvm/mmu.c
@@ -639,28 +639,22 @@ error_1:
return -ENOMEM;
}
-int kvm_mmu_init(struct kvm_vcpu *vcpu)
+int kvm_mmu_create(struct kvm_vcpu *vcpu)
{
- int r;
-
ASSERT(vcpu);
ASSERT(!VALID_PAGE(vcpu->mmu.root_hpa));
ASSERT(list_empty(&vcpu->free_pages));
- r = alloc_mmu_pages(vcpu);
- if (r)
- goto out;
-
- r = init_kvm_mmu(vcpu);
- if (r)
- goto out_free_pages;
+ return alloc_mmu_pages(vcpu);
+}
- return 0;
+int kvm_mmu_setup(struct kvm_vcpu *vcpu)
+{
+ ASSERT(vcpu);
+ ASSERT(!VALID_PAGE(vcpu->mmu.root_hpa));
+ ASSERT(!list_empty(&vcpu->free_pages));
-out_free_pages:
- free_mmu_pages(vcpu);
-out:
- return r;
+ return init_kvm_mmu(vcpu);
}
void kvm_mmu_destroy(struct kvm_vcpu *vcpu)
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
