powerpc/iommu: Update the generic code to use dynamic iommu page sizes

This patch updates the generic iommu backend code to use the
it_page_shift field to determine the iommu page size instead of
using hardcoded values.
Signed-off-by: Alistair Popple <alistair@popple.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>

arch/powerpc/include/asm/iommu.h

@@ -35,17 +35,14 @@
 #define IOMMU_PAGE_MASK_4K       (~((1 << IOMMU_PAGE_SHIFT_4K) - 1))
 #define IOMMU_PAGE_ALIGN_4K(addr) _ALIGN_UP(addr, IOMMU_PAGE_SIZE_4K)
 
+#define IOMMU_PAGE_SIZE(tblptr) (ASM_CONST(1) << (tblptr)->it_page_shift)
+#define IOMMU_PAGE_MASK(tblptr) (~((1 << (tblptr)->it_page_shift) - 1))
+#define IOMMU_PAGE_ALIGN(addr, tblptr) _ALIGN_UP(addr, IOMMU_PAGE_SIZE(tblptr))
+
 /* Boot time flags */
 extern int iommu_is_off;
 extern int iommu_force_on;
 
-/* Pure 2^n version of get_order */
-static __inline__ __attribute_const__ int get_iommu_order(unsigned long size)
-{
-	return __ilog2((size - 1) >> IOMMU_PAGE_SHIFT_4K) + 1;
-}
-
-
 /*
  * IOMAP_MAX_ORDER defines the largest contiguous block
  * of dma space we can get.  IOMAP_MAX_ORDER = 13
@@ -82,6 +79,14 @@ struct iommu_table {
 #endif
 };
 
+/* Pure 2^n version of get_order */
+static inline __attribute_const__
+int get_iommu_order(unsigned long size, struct iommu_table *tbl)
+{
+	return __ilog2((size - 1) >> tbl->it_page_shift) + 1;
+}
+
+
 struct scatterlist;
 
 static inline void set_iommu_table_base(struct device *dev, void *base)

arch/powerpc/kernel/dma-iommu.c

@@ -83,10 +83,10 @@ static int dma_iommu_dma_supported(struct device *dev, u64 mask)
 		return 0;
 	}
 
-	if (tbl->it_offset > (mask >> IOMMU_PAGE_SHIFT_4K)) {
+	if (tbl->it_offset > (mask >> tbl->it_page_shift)) {
 		dev_info(dev, "Warning: IOMMU offset too big for device mask\n");
 		dev_info(dev, "mask: 0x%08llx, table offset: 0x%08lx\n",
-				mask, tbl->it_offset << IOMMU_PAGE_SHIFT_4K);
+				mask, tbl->it_offset << tbl->it_page_shift);
 		return 0;
 	} else
 		return 1;

arch/powerpc/kernel/iommu.c

@@ -251,14 +251,13 @@ static unsigned long iommu_range_alloc(struct device *dev,
 
 	if (dev)
 		boundary_size = ALIGN(dma_get_seg_boundary(dev) + 1,
-				      1 << IOMMU_PAGE_SHIFT_4K);
+				      1 << tbl->it_page_shift);
 	else
-		boundary_size = ALIGN(1UL << 32, 1 << IOMMU_PAGE_SHIFT_4K);
+		boundary_size = ALIGN(1UL << 32, 1 << tbl->it_page_shift);
 	/* 4GB boundary for iseries_hv_alloc and iseries_hv_map */
 
-	n = iommu_area_alloc(tbl->it_map, limit, start, npages,
-			tbl->it_offset, boundary_size >> IOMMU_PAGE_SHIFT_4K,
-			align_mask);
+	n = iommu_area_alloc(tbl->it_map, limit, start, npages, tbl->it_offset,
+			     boundary_size >> tbl->it_page_shift, align_mask);
 	if (n == -1) {
 		if (likely(pass == 0)) {
 			/* First try the pool from the start */
@@ -320,12 +319,12 @@ static dma_addr_t iommu_alloc(struct device *dev, struct iommu_table *tbl,
 		return DMA_ERROR_CODE;
 
 	entry += tbl->it_offset;	/* Offset into real TCE table */
-	ret = entry << IOMMU_PAGE_SHIFT_4K;	/* Set the return dma address */
+	ret = entry << tbl->it_page_shift;	/* Set the return dma address */
 
 	/* Put the TCEs in the HW table */
 	build_fail = ppc_md.tce_build(tbl, entry, npages,
-				(unsigned long)page & IOMMU_PAGE_MASK_4K,
-				direction, attrs);
+				      (unsigned long)page &
+				      IOMMU_PAGE_MASK(tbl), direction, attrs);
 
 	/* ppc_md.tce_build() only returns non-zero for transient errors.
 	 * Clean up the table bitmap in this case and return
@@ -352,7 +351,7 @@ static bool iommu_free_check(struct iommu_table *tbl, dma_addr_t dma_addr,
 {
 	unsigned long entry, free_entry;
 
-	entry = dma_addr >> IOMMU_PAGE_SHIFT_4K;
+	entry = dma_addr >> tbl->it_page_shift;
 	free_entry = entry - tbl->it_offset;
 
 	if (((free_entry + npages) > tbl->it_size) ||
@@ -401,7 +400,7 @@ static void __iommu_free(struct iommu_table *tbl, dma_addr_t dma_addr,
 	unsigned long flags;
 	struct iommu_pool *pool;
 
-	entry = dma_addr >> IOMMU_PAGE_SHIFT_4K;
+	entry = dma_addr >> tbl->it_page_shift;
 	free_entry = entry - tbl->it_offset;
 
 	pool = get_pool(tbl, free_entry);
@@ -468,13 +467,13 @@ int iommu_map_sg(struct device *dev, struct iommu_table *tbl,
 		}
 		/* Allocate iommu entries for that segment */
 		vaddr = (unsigned long) sg_virt(s);
-		npages = iommu_num_pages(vaddr, slen, IOMMU_PAGE_SIZE_4K);
+		npages = iommu_num_pages(vaddr, slen, IOMMU_PAGE_SIZE(tbl));
 		align = 0;
-		if (IOMMU_PAGE_SHIFT_4K < PAGE_SHIFT && slen >= PAGE_SIZE &&
+		if (tbl->it_page_shift < PAGE_SHIFT && slen >= PAGE_SIZE &&
 		    (vaddr & ~PAGE_MASK) == 0)
-			align = PAGE_SHIFT - IOMMU_PAGE_SHIFT_4K;
+			align = PAGE_SHIFT - tbl->it_page_shift;
 		entry = iommu_range_alloc(dev, tbl, npages, &handle,
-					  mask >> IOMMU_PAGE_SHIFT_4K, align);
+					  mask >> tbl->it_page_shift, align);
 
 		DBG("  - vaddr: %lx, size: %lx\n", vaddr, slen);
 
@@ -489,16 +488,16 @@ int iommu_map_sg(struct device *dev, struct iommu_table *tbl,
 
 		/* Convert entry to a dma_addr_t */
 		entry += tbl->it_offset;
-		dma_addr = entry << IOMMU_PAGE_SHIFT_4K;
-		dma_addr |= (s->offset & ~IOMMU_PAGE_MASK_4K);
+		dma_addr = entry << tbl->it_page_shift;
+		dma_addr |= (s->offset & ~IOMMU_PAGE_MASK(tbl));
 
 		DBG("  - %lu pages, entry: %lx, dma_addr: %lx\n",
 			    npages, entry, dma_addr);
 
 		/* Insert into HW table */
 		build_fail = ppc_md.tce_build(tbl, entry, npages,
-					vaddr & IOMMU_PAGE_MASK_4K,
-					direction, attrs);
+					      vaddr & IOMMU_PAGE_MASK(tbl),
+					      direction, attrs);
 		if(unlikely(build_fail))
 			goto failure;
 
@@ -559,9 +558,9 @@ int iommu_map_sg(struct device *dev, struct iommu_table *tbl,
 		if (s->dma_length != 0) {
 			unsigned long vaddr, npages;
 
-			vaddr = s->dma_address & IOMMU_PAGE_MASK_4K;
+			vaddr = s->dma_address & IOMMU_PAGE_MASK(tbl);
 			npages = iommu_num_pages(s->dma_address, s->dma_length,
-						 IOMMU_PAGE_SIZE_4K);
+						 IOMMU_PAGE_SIZE(tbl));
 			__iommu_free(tbl, vaddr, npages);
 			s->dma_address = DMA_ERROR_CODE;
 			s->dma_length = 0;
@@ -592,7 +591,7 @@ void iommu_unmap_sg(struct iommu_table *tbl, struct scatterlist *sglist,
 		if (sg->dma_length == 0)
 			break;
 		npages = iommu_num_pages(dma_handle, sg->dma_length,
-					 IOMMU_PAGE_SIZE_4K);
+					 IOMMU_PAGE_SIZE(tbl));
 		__iommu_free(tbl, dma_handle, npages);
 		sg = sg_next(sg);
 	}
@@ -676,7 +675,7 @@ struct iommu_table *iommu_init_table(struct iommu_table *tbl, int nid)
 		set_bit(0, tbl->it_map);
 
 	/* We only split the IOMMU table if we have 1GB or more of space */
-	if ((tbl->it_size << IOMMU_PAGE_SHIFT_4K) >= (1UL * 1024 * 1024 * 1024))
+	if ((tbl->it_size << tbl->it_page_shift) >= (1UL * 1024 * 1024 * 1024))
 		tbl->nr_pools = IOMMU_NR_POOLS;
 	else
 		tbl->nr_pools = 1;
@@ -768,16 +767,16 @@ dma_addr_t iommu_map_page(struct device *dev, struct iommu_table *tbl,
 
 	vaddr = page_address(page) + offset;
 	uaddr = (unsigned long)vaddr;
-	npages = iommu_num_pages(uaddr, size, IOMMU_PAGE_SIZE_4K);
+	npages = iommu_num_pages(uaddr, size, IOMMU_PAGE_SIZE(tbl));
 
 	if (tbl) {
 		align = 0;
-		if (IOMMU_PAGE_SHIFT_4K < PAGE_SHIFT && size >= PAGE_SIZE &&
+		if (tbl->it_page_shift < PAGE_SHIFT && size >= PAGE_SIZE &&
 		    ((unsigned long)vaddr & ~PAGE_MASK) == 0)
-			align = PAGE_SHIFT - IOMMU_PAGE_SHIFT_4K;
+			align = PAGE_SHIFT - tbl->it_page_shift;
 
 		dma_handle = iommu_alloc(dev, tbl, vaddr, npages, direction,
-					 mask >> IOMMU_PAGE_SHIFT_4K, align,
+					 mask >> tbl->it_page_shift, align,
 					 attrs);
 		if (dma_handle == DMA_ERROR_CODE) {
 			if (printk_ratelimit())  {
@@ -786,7 +785,7 @@ dma_addr_t iommu_map_page(struct device *dev, struct iommu_table *tbl,
 					 npages);
 			}
 		} else
-			dma_handle |= (uaddr & ~IOMMU_PAGE_MASK_4K);
+			dma_handle |= (uaddr & ~IOMMU_PAGE_MASK(tbl));
 	}
 
 	return dma_handle;
@@ -801,7 +800,8 @@ void iommu_unmap_page(struct iommu_table *tbl, dma_addr_t dma_handle,
 	BUG_ON(direction == DMA_NONE);
 
 	if (tbl) {
-		npages = iommu_num_pages(dma_handle, size, IOMMU_PAGE_SIZE_4K);
+		npages = iommu_num_pages(dma_handle, size,
+					 IOMMU_PAGE_SIZE(tbl));
 		iommu_free(tbl, dma_handle, npages);
 	}
 }
@@ -845,10 +845,10 @@ void *iommu_alloc_coherent(struct device *dev, struct iommu_table *tbl,
 	memset(ret, 0, size);
 
 	/* Set up tces to cover the allocated range */
-	nio_pages = size >> IOMMU_PAGE_SHIFT_4K;
-	io_order = get_iommu_order(size);
+	nio_pages = size >> tbl->it_page_shift;
+	io_order = get_iommu_order(size, tbl);
 	mapping = iommu_alloc(dev, tbl, ret, nio_pages, DMA_BIDIRECTIONAL,
-			      mask >> IOMMU_PAGE_SHIFT_4K, io_order, NULL);
+			      mask >> tbl->it_page_shift, io_order, NULL);
 	if (mapping == DMA_ERROR_CODE) {
 		free_pages((unsigned long)ret, order);
 		return NULL;
@@ -864,7 +864,7 @@ void iommu_free_coherent(struct iommu_table *tbl, size_t size,
 		unsigned int nio_pages;
 
 		size = PAGE_ALIGN(size);
-		nio_pages = size >> IOMMU_PAGE_SHIFT_4K;
+		nio_pages = size >> tbl->it_page_shift;
 		iommu_free(tbl, dma_handle, nio_pages);
 		size = PAGE_ALIGN(size);
 		free_pages((unsigned long)vaddr, get_order(size));
@@ -935,10 +935,10 @@ int iommu_tce_clear_param_check(struct iommu_table *tbl,
 	if (tce_value)
 		return -EINVAL;
 
-	if (ioba & ~IOMMU_PAGE_MASK_4K)
+	if (ioba & ~IOMMU_PAGE_MASK(tbl))
 		return -EINVAL;
 
-	ioba >>= IOMMU_PAGE_SHIFT_4K;
+	ioba >>= tbl->it_page_shift;
 	if (ioba < tbl->it_offset)
 		return -EINVAL;
 
@@ -955,13 +955,13 @@ int iommu_tce_put_param_check(struct iommu_table *tbl,
 	if (!(tce & (TCE_PCI_WRITE | TCE_PCI_READ)))
 		return -EINVAL;
 
-	if (tce & ~(IOMMU_PAGE_MASK_4K | TCE_PCI_WRITE | TCE_PCI_READ))
+	if (tce & ~(IOMMU_PAGE_MASK(tbl) | TCE_PCI_WRITE | TCE_PCI_READ))
 		return -EINVAL;
 
-	if (ioba & ~IOMMU_PAGE_MASK_4K)
+	if (ioba & ~IOMMU_PAGE_MASK(tbl))
 		return -EINVAL;
 
-	ioba >>= IOMMU_PAGE_SHIFT_4K;
+	ioba >>= tbl->it_page_shift;
 	if (ioba < tbl->it_offset)
 		return -EINVAL;
 
@@ -1037,7 +1037,7 @@ int iommu_tce_build(struct iommu_table *tbl, unsigned long entry,
 
 	/* if (unlikely(ret))
 		pr_err("iommu_tce: %s failed on hwaddr=%lx ioba=%lx kva=%lx ret=%d\n",
-				__func__, hwaddr, entry << IOMMU_PAGE_SHIFT_4K,
+			__func__, hwaddr, entry << IOMMU_PAGE_SHIFT(tbl),
 				hwaddr, ret); */
 
 	return ret;
@@ -1049,14 +1049,14 @@ int iommu_put_tce_user_mode(struct iommu_table *tbl, unsigned long entry,
 {
 	int ret;
 	struct page *page = NULL;
-	unsigned long hwaddr, offset = tce & IOMMU_PAGE_MASK_4K & ~PAGE_MASK;
+	unsigned long hwaddr, offset = tce & IOMMU_PAGE_MASK(tbl) & ~PAGE_MASK;
 	enum dma_data_direction direction = iommu_tce_direction(tce);
 
 	ret = get_user_pages_fast(tce & PAGE_MASK, 1,
 			direction != DMA_TO_DEVICE, &page);
 	if (unlikely(ret != 1)) {
 		/* pr_err("iommu_tce: get_user_pages_fast failed tce=%lx ioba=%lx ret=%d\n",
-				tce, entry << IOMMU_PAGE_SHIFT_4K, ret); */
+				tce, entry << IOMMU_PAGE_SHIFT(tbl), ret); */
 		return -EFAULT;
 	}
 	hwaddr = (unsigned long) page_address(page) + offset;
@@ -1067,7 +1067,7 @@ int iommu_put_tce_user_mode(struct iommu_table *tbl, unsigned long entry,
 
 	if (ret < 0)
 		pr_err("iommu_tce: %s failed ioba=%lx, tce=%lx, ret=%d\n",
-			__func__, entry << IOMMU_PAGE_SHIFT_4K, tce, ret);
+			__func__, entry << tbl->it_page_shift, tce, ret);
 
 	return ret;
 }
@@ -1127,6 +1127,12 @@ int iommu_add_device(struct device *dev)
 	pr_debug("iommu_tce: adding %s to iommu group %d\n",
 			dev_name(dev), iommu_group_id(tbl->it_group));
 
+	if (PAGE_SIZE < IOMMU_PAGE_SIZE(tbl)) {
+		pr_err("iommu_tce: unsupported iommu page size.");
+		pr_err("%s has not been added\n", dev_name(dev));
+		return -EINVAL;
+	}
+
 	ret = iommu_group_add_device(tbl->it_group, dev);
 	if (ret < 0)
 		pr_err("iommu_tce: %s has not been added, ret=%d\n",

arch/powerpc/kernel/vio.c

@@ -518,16 +518,18 @@ static dma_addr_t vio_dma_iommu_map_page(struct device *dev, struct page *page,
                                          struct dma_attrs *attrs)
 {
 	struct vio_dev *viodev = to_vio_dev(dev);
+	struct iommu_table *tbl;
 	dma_addr_t ret = DMA_ERROR_CODE;
 
-	if (vio_cmo_alloc(viodev, roundup(size, IOMMU_PAGE_SIZE_4K))) {
+	tbl = get_iommu_table_base(dev);
+	if (vio_cmo_alloc(viodev, roundup(size, IOMMU_PAGE_SIZE(tbl)))) {
 		atomic_inc(&viodev->cmo.allocs_failed);
 		return ret;
 	}
 
 	ret = dma_iommu_ops.map_page(dev, page, offset, size, direction, attrs);
 	if (unlikely(dma_mapping_error(dev, ret))) {
-		vio_cmo_dealloc(viodev, roundup(size, IOMMU_PAGE_SIZE_4K));
+		vio_cmo_dealloc(viodev, roundup(size, IOMMU_PAGE_SIZE(tbl)));
 		atomic_inc(&viodev->cmo.allocs_failed);
 	}
 
@@ -540,10 +542,12 @@ static void vio_dma_iommu_unmap_page(struct device *dev, dma_addr_t dma_handle,
 				     struct dma_attrs *attrs)
 {
 	struct vio_dev *viodev = to_vio_dev(dev);
+	struct iommu_table *tbl;
 
+	tbl = get_iommu_table_base(dev);
 	dma_iommu_ops.unmap_page(dev, dma_handle, size, direction, attrs);
 
-	vio_cmo_dealloc(viodev, roundup(size, IOMMU_PAGE_SIZE_4K));
+	vio_cmo_dealloc(viodev, roundup(size, IOMMU_PAGE_SIZE(tbl)));
 }
 
 static int vio_dma_iommu_map_sg(struct device *dev, struct scatterlist *sglist,
@@ -551,12 +555,14 @@ static int vio_dma_iommu_map_sg(struct device *dev, struct scatterlist *sglist,
                                 struct dma_attrs *attrs)
 {
 	struct vio_dev *viodev = to_vio_dev(dev);
+	struct iommu_table *tbl;
 	struct scatterlist *sgl;
 	int ret, count = 0;
 	size_t alloc_size = 0;
 
+	tbl = get_iommu_table_base(dev);
 	for (sgl = sglist; count < nelems; count++, sgl++)
-		alloc_size += roundup(sgl->length, IOMMU_PAGE_SIZE_4K);
+		alloc_size += roundup(sgl->length, IOMMU_PAGE_SIZE(tbl));
 
 	if (vio_cmo_alloc(viodev, alloc_size)) {
 		atomic_inc(&viodev->cmo.allocs_failed);
@@ -572,7 +578,7 @@ static int vio_dma_iommu_map_sg(struct device *dev, struct scatterlist *sglist,
 	}
 
 	for (sgl = sglist, count = 0; count < ret; count++, sgl++)
-		alloc_size -= roundup(sgl->dma_length, IOMMU_PAGE_SIZE_4K);
+		alloc_size -= roundup(sgl->dma_length, IOMMU_PAGE_SIZE(tbl));
 	if (alloc_size)
 		vio_cmo_dealloc(viodev, alloc_size);
 
@@ -585,12 +591,14 @@ static void vio_dma_iommu_unmap_sg(struct device *dev,
 		struct dma_attrs *attrs)
 {
 	struct vio_dev *viodev = to_vio_dev(dev);
+	struct iommu_table *tbl;
 	struct scatterlist *sgl;
 	size_t alloc_size = 0;
 	int count = 0;
 
+	tbl = get_iommu_table_base(dev);
 	for (sgl = sglist; count < nelems; count++, sgl++)
-		alloc_size += roundup(sgl->dma_length, IOMMU_PAGE_SIZE_4K);
+		alloc_size += roundup(sgl->dma_length, IOMMU_PAGE_SIZE(tbl));
 
 	dma_iommu_ops.unmap_sg(dev, sglist, nelems, direction, attrs);
 
@@ -706,11 +714,14 @@ static int vio_cmo_bus_probe(struct vio_dev *viodev)
 {
 	struct vio_cmo_dev_entry *dev_ent;
 	struct device *dev = &viodev->dev;
+	struct iommu_table *tbl;
 	struct vio_driver *viodrv = to_vio_driver(dev->driver);
 	unsigned long flags;
 	size_t size;
 	bool dma_capable = false;
 
+	tbl = get_iommu_table_base(dev);
+
 	/* A device requires entitlement if it has a DMA window property */
 	switch (viodev->family) {
 	case VDEVICE:
@@ -737,7 +748,7 @@ static int vio_cmo_bus_probe(struct vio_dev *viodev)
 		}
 
 		viodev->cmo.desired =
-			IOMMU_PAGE_ALIGN_4K(viodrv->get_desired_dma(viodev));
+			IOMMU_PAGE_ALIGN(viodrv->get_desired_dma(viodev), tbl);
 		if (viodev->cmo.desired < VIO_CMO_MIN_ENT)
 			viodev->cmo.desired = VIO_CMO_MIN_ENT;
 		size = VIO_CMO_MIN_ENT;

arch/powerpc/platforms/powernv/pci.c

@@ -762,8 +762,6 @@ static struct notifier_block tce_iommu_bus_nb = {
 
 static int __init tce_iommu_bus_notifier_init(void)
 {
-	BUILD_BUG_ON(PAGE_SIZE < IOMMU_PAGE_SIZE_4K);
-
 	bus_register_notifier(&pci_bus_type, &tce_iommu_bus_nb);
 	return 0;
 }

drivers/net/ethernet/ibm/ibmveth.c

@@ -1276,31 +1276,34 @@ static unsigned long ibmveth_get_desired_dma(struct vio_dev *vdev)
 {
 	struct net_device *netdev = dev_get_drvdata(&vdev->dev);
 	struct ibmveth_adapter *adapter;
+	struct iommu_table *tbl;
 	unsigned long ret;
 	int i;
 	int rxqentries = 1;
 
+	tbl = get_iommu_table_base(&vdev->dev);
+
 	/* netdev inits at probe time along with the structures we need below*/
 	if (netdev == NULL)
-		return IOMMU_PAGE_ALIGN_4K(IBMVETH_IO_ENTITLEMENT_DEFAULT);
+		return IOMMU_PAGE_ALIGN(IBMVETH_IO_ENTITLEMENT_DEFAULT, tbl);
 
 	adapter = netdev_priv(netdev);
 
 	ret = IBMVETH_BUFF_LIST_SIZE + IBMVETH_FILT_LIST_SIZE;
-	ret += IOMMU_PAGE_ALIGN_4K(netdev->mtu);
+	ret += IOMMU_PAGE_ALIGN(netdev->mtu, tbl);
 
 	for (i = 0; i < IBMVETH_NUM_BUFF_POOLS; i++) {
 		/* add the size of the active receive buffers */
 		if (adapter->rx_buff_pool[i].active)
 			ret +=
 			    adapter->rx_buff_pool[i].size *
-			    IOMMU_PAGE_ALIGN_4K(adapter->rx_buff_pool[i].
-			            buff_size);
+			    IOMMU_PAGE_ALIGN(adapter->rx_buff_pool[i].
+					     buff_size, tbl);
 		rxqentries += adapter->rx_buff_pool[i].size;
 	}
 	/* add the size of the receive queue entries */
-	ret += IOMMU_PAGE_ALIGN_4K(
-		rxqentries * sizeof(struct ibmveth_rx_q_entry));
+	ret += IOMMU_PAGE_ALIGN(
+		rxqentries * sizeof(struct ibmveth_rx_q_entry), tbl);
 
 	return ret;
 }