>From all I can tell, this doesn't violate lock ordering as other places
call heap allocation functions from inside hd->mapping_lock.
Signed-off-by: Jan Beulich <jbeulich@xxxxxxxxxx>
--- 2009年01月27日.orig/xen/drivers/passthrough/amd/iommu_map.c 2009年01月29日 
11:33:26.000000000 +0100
+++ 2009年01月27日/xen/drivers/passthrough/amd/iommu_map.c 2009年01月30日 
08:43:35.000000000 +0100
@@ -567,6 +567,8 @@ int amd_iommu_sync_p2m(struct domain *d)
 if ( hd->p2m_synchronized )
 goto out;
 
+ spin_lock(&d->page_alloc_lock);
+
 page_list_for_each ( page, &d->page_list )
 {
 mfn = page_to_mfn(page);
@@ -579,6 +581,7 @@ int amd_iommu_sync_p2m(struct domain *d)
 
 if ( iommu_l2e == 0 )
 {
+ spin_unlock(&d->page_alloc_lock);
 amd_iov_error("Invalid IO pagetable entry gfn = %lx\n", gfn);
 spin_unlock_irqrestore(&hd->mapping_lock, flags);
 return -EFAULT;
@@ -587,6 +590,8 @@ int amd_iommu_sync_p2m(struct domain *d)
 set_iommu_l1e_present(iommu_l2e, gfn, (u64)mfn << PAGE_SHIFT, iw, ir);
 }
 
+ spin_unlock(&d->page_alloc_lock);
+
 hd->p2m_synchronized = 1;
 
 out:
_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel