iommu/vt-d: Clean up size handling for intel_iommu_unmap()
We have this horrid API where iommu_unmap() can unmap more than it's asked
to, if the IOVA in question happens to be mapped with a large page.
Instead of propagating this nonsense to the point where we end up returning
the page order from dma_pte_clear_range(), let's just do it once and adjust
the 'size' parameter accordingly.
Augment pfn_to_dma_pte() to return the level at which the PTE was found,
which will also be useful later if we end up changing the API for
iommu_iova_to_phys() to behave the same way as is being discussed upstream.
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
diff --git a/drivers/iommu/intel-iommu.c b/drivers/iommu/intel-iommu.c
index 484d669..6472bf1 100644
--- a/drivers/iommu/intel-iommu.c
+++ b/drivers/iommu/intel-iommu.c
@@ -784,7 +784,7 @@
}
static struct dma_pte *pfn_to_dma_pte(struct dmar_domain *domain,
- unsigned long pfn, int target_level)
+ unsigned long pfn, int *target_level)
{
int addr_width = agaw_to_width(domain->agaw) - VTD_PAGE_SHIFT;
struct dma_pte *parent, *pte = NULL;
@@ -799,14 +799,14 @@
parent = domain->pgd;
- while (level > 0) {
+ while (1) {
void *tmp_page;
offset = pfn_level_offset(pfn, level);
pte = &parent[offset];
- if (!target_level && (dma_pte_superpage(pte) || !dma_pte_present(pte)))
+ if (!*target_level && (dma_pte_superpage(pte) || !dma_pte_present(pte)))
break;
- if (level == target_level)
+ if (level == *target_level)
break;
if (!dma_pte_present(pte)) {
@@ -827,10 +827,16 @@
domain_flush_cache(domain, pte, sizeof(*pte));
}
}
+ if (level == 1)
+ break;
+
parent = phys_to_virt(dma_pte_addr(pte));
level--;
}
+ if (!*target_level)
+ *target_level = level;
+
return pte;
}
@@ -868,7 +874,7 @@
}
/* clear last level pte, a tlb flush should be followed */
-static int dma_pte_clear_range(struct dmar_domain *domain,
+static void dma_pte_clear_range(struct dmar_domain *domain,
unsigned long start_pfn,
unsigned long last_pfn)
{
@@ -898,8 +904,6 @@
(void *)pte - (void *)first_pte);
} while (start_pfn && start_pfn <= last_pfn);
-
- return min_t(int, (large_page - 1) * 9, MAX_AGAW_PFN_WIDTH);
}
static void dma_pte_free_level(struct dmar_domain *domain, int level,
@@ -1832,7 +1836,7 @@
if (!pte) {
largepage_lvl = hardware_largepage_caps(domain, iov_pfn, phys_pfn, sg_res);
- first_pte = pte = pfn_to_dma_pte(domain, iov_pfn, largepage_lvl);
+ first_pte = pte = pfn_to_dma_pte(domain, iov_pfn, &largepage_lvl);
if (!pte)
return -ENOMEM;
/* It is large page*/
@@ -4099,15 +4103,23 @@
unsigned long iova, size_t size)
{
struct dmar_domain *dmar_domain = domain->priv;
- int order;
+ int level = 0;
- order = dma_pte_clear_range(dmar_domain, iova >> VTD_PAGE_SHIFT,
+ /* Cope with horrid API which requires us to unmap more than the
+ size argument if it happens to be a large-page mapping. */
+ if (!pfn_to_dma_pte(dmar_domain, iova >> VTD_PAGE_SHIFT, &level))
+ BUG();
+
+ if (size < VTD_PAGE_SIZE << level_to_offset_bits(level))
+ size = VTD_PAGE_SIZE << level_to_offset_bits(level);
+
+ dma_pte_clear_range(dmar_domain, iova >> VTD_PAGE_SHIFT,
(iova + size - 1) >> VTD_PAGE_SHIFT);
if (dmar_domain->max_addr == iova + size)
dmar_domain->max_addr = iova;
- return PAGE_SIZE << order;
+ return size;
}
static phys_addr_t intel_iommu_iova_to_phys(struct iommu_domain *domain,
@@ -4115,9 +4127,10 @@
{
struct dmar_domain *dmar_domain = domain->priv;
struct dma_pte *pte;
+ int level = 0;
u64 phys = 0;
- pte = pfn_to_dma_pte(dmar_domain, iova >> VTD_PAGE_SHIFT, 0);
+ pte = pfn_to_dma_pte(dmar_domain, iova >> VTD_PAGE_SHIFT, &level);
if (pte)
phys = dma_pte_addr(pte);