Merge branch 'for-4.5/block-dax' into for-4.5/libnvdimm

4 files changed, 233 insertions, 9 deletions

diff --git a/drivers/nvdimm/core.c b/drivers/nvdimm/core.c
index 82c49bb870555..2e2832b83c939 100644
--- a/drivers/nvdimm/core.c
+++ b/drivers/nvdimm/core.c
@@ -11,6 +11,7 @@
  * General Public License for more details.
  */
 #include <linux/libnvdimm.h>
+#include <linux/badblocks.h>
 #include <linux/export.h>
 #include <linux/module.h>
 #include <linux/blkdev.h>
@@ -325,6 +326,7 @@ struct nvdimm_bus *__nvdimm_bus_register(struct device *parent,
 	if (!nvdimm_bus)
 		return NULL;
 	INIT_LIST_HEAD(&nvdimm_bus->list);
+	INIT_LIST_HEAD(&nvdimm_bus->poison_list);
 	init_waitqueue_head(&nvdimm_bus->probe_wait);
 	nvdimm_bus->id = ida_simple_get(&nd_ida, 0, 0, GFP_KERNEL);
 	mutex_init(&nvdimm_bus->reconfig_mutex);
@@ -359,6 +361,172 @@ struct nvdimm_bus *__nvdimm_bus_register(struct device *parent,
 }
 EXPORT_SYMBOL_GPL(__nvdimm_bus_register);
 
+static void set_badblock(struct badblocks *bb, sector_t s, int num)
+{
+	dev_dbg(bb->dev, "Found a poison range (0x%llx, 0x%llx)\n",
+			(u64) s * 512, (u64) num * 512);
+	/* this isn't an error as the hardware will still throw an exception */
+	if (badblocks_set(bb, s, num, 1))
+		dev_info_once(bb->dev, "%s: failed for sector %llx\n",
+				__func__, (u64) s);
+}
+
+/**
+ * __add_badblock_range() - Convert a physical address range to bad sectors
+ * @bb:		badblocks instance to populate
+ * @ns_offset:	namespace offset where the error range begins (in bytes)
+ * @len:	number of bytes of poison to be added
+ *
+ * This assumes that the range provided with (ns_offset, len) is within
+ * the bounds of physical addresses for this namespace, i.e. lies in the
+ * interval [ns_start, ns_start + ns_size)
+ */
+static void __add_badblock_range(struct badblocks *bb, u64 ns_offset, u64 len)
+{
+	const unsigned int sector_size = 512;
+	sector_t start_sector;
+	u64 num_sectors;
+	u32 rem;
+
+	start_sector = div_u64(ns_offset, sector_size);
+	num_sectors = div_u64_rem(len, sector_size, &rem);
+	if (rem)
+		num_sectors++;
+
+	if (unlikely(num_sectors > (u64)INT_MAX)) {
+		u64 remaining = num_sectors;
+		sector_t s = start_sector;
+
+		while (remaining) {
+			int done = min_t(u64, remaining, INT_MAX);
+
+			set_badblock(bb, s, done);
+			remaining -= done;
+			s += done;
+		}
+	} else
+		set_badblock(bb, start_sector, num_sectors);
+}
+
+/**
+ * nvdimm_namespace_add_poison() - Convert a list of poison ranges to badblocks
+ * @ndns:	the namespace containing poison ranges
+ * @bb:		badblocks instance to populate
+ * @offset:	offset at the start of the namespace before 'sector 0'
+ *
+ * The poison list generated during NFIT initialization may contain multiple,
+ * possibly overlapping ranges in the SPA (System Physical Address) space.
+ * Compare each of these ranges to the namespace currently being initialized,
+ * and add badblocks to the gendisk for all matching sub-ranges
+ */
+void nvdimm_namespace_add_poison(struct nd_namespace_common *ndns,
+		struct badblocks *bb, resource_size_t offset)
+{
+	struct nd_namespace_io *nsio = to_nd_namespace_io(&ndns->dev);
+	struct nd_region *nd_region = to_nd_region(ndns->dev.parent);
+	struct nvdimm_bus *nvdimm_bus;
+	struct list_head *poison_list;
+	u64 ns_start, ns_end, ns_size;
+	struct nd_poison *pl;
+
+	ns_size = nvdimm_namespace_capacity(ndns) - offset;
+	ns_start = nsio->res.start + offset;
+	ns_end = nsio->res.end;
+
+	nvdimm_bus = to_nvdimm_bus(nd_region->dev.parent);
+	poison_list = &nvdimm_bus->poison_list;
+	if (list_empty(poison_list))
+		return;
+
+	list_for_each_entry(pl, poison_list, list) {
+		u64 pl_end = pl->start + pl->length - 1;
+
+		/* Discard intervals with no intersection */
+		if (pl_end < ns_start)
+			continue;
+		if (pl->start > ns_end)
+			continue;
+		/* Deal with any overlap after start of the namespace */
+		if (pl->start >= ns_start) {
+			u64 start = pl->start;
+			u64 len;
+
+			if (pl_end <= ns_end)
+				len = pl->length;
+			else
+				len = ns_start + ns_size - pl->start;
+			__add_badblock_range(bb, start - ns_start, len);
+			continue;
+		}
+		/* Deal with overlap for poison starting before the namespace */
+		if (pl->start < ns_start) {
+			u64 len;
+
+			if (pl_end < ns_end)
+				len = pl->start + pl->length - ns_start;
+			else
+				len = ns_size;
+			__add_badblock_range(bb, 0, len);
+		}
+	}
+}
+EXPORT_SYMBOL_GPL(nvdimm_namespace_add_poison);
+
+static int __add_poison(struct nvdimm_bus *nvdimm_bus, u64 addr, u64 length)
+{
+	struct nd_poison *pl;
+
+	pl = kzalloc(sizeof(*pl), GFP_KERNEL);
+	if (!pl)
+		return -ENOMEM;
+
+	pl->start = addr;
+	pl->length = length;
+	list_add_tail(&pl->list, &nvdimm_bus->poison_list);
+
+	return 0;
+}
+
+int nvdimm_bus_add_poison(struct nvdimm_bus *nvdimm_bus, u64 addr, u64 length)
+{
+	struct nd_poison *pl;
+
+	if (list_empty(&nvdimm_bus->poison_list))
+		return __add_poison(nvdimm_bus, addr, length);
+
+	/*
+	 * There is a chance this is a duplicate, check for those first.
+	 * This will be the common case as ARS_STATUS returns all known
+	 * errors in the SPA space, and we can't query it per region
+	 */
+	list_for_each_entry(pl, &nvdimm_bus->poison_list, list)
+		if (pl->start == addr) {
+			/* If length has changed, update this list entry */
+			if (pl->length != length)
+				pl->length = length;
+			return 0;
+		}
+
+	/*
+	 * If not a duplicate or a simple length update, add the entry as is,
+	 * as any overlapping ranges will get resolved when the list is consumed
+	 * and converted to badblocks
+	 */
+	return __add_poison(nvdimm_bus, addr, length);
+}
+EXPORT_SYMBOL_GPL(nvdimm_bus_add_poison);
+
+static void free_poison_list(struct list_head *poison_list)
+{
+	struct nd_poison *pl, *next;
+
+	list_for_each_entry_safe(pl, next, poison_list, list) {
+		list_del(&pl->list);
+		kfree(pl);
+	}
+	list_del_init(poison_list);
+}
+
 static int child_unregister(struct device *dev, void *data)
 {
 	/*
@@ -385,6 +553,7 @@ void nvdimm_bus_unregister(struct nvdimm_bus *nvdimm_bus)
 
 	nd_synchronize();
 	device_for_each_child(&nvdimm_bus->dev, NULL, child_unregister);
+	free_poison_list(&nvdimm_bus->poison_list);
 	nvdimm_bus_destroy_ndctl(nvdimm_bus);
 
 	device_unregister(&nvdimm_bus->dev);
diff --git a/drivers/nvdimm/nd-core.h b/drivers/nvdimm/nd-core.h
index 3249c498892a8..1d1500f3d8b51 100644
--- a/drivers/nvdimm/nd-core.h
+++ b/drivers/nvdimm/nd-core.h
@@ -30,6 +30,7 @@ struct nvdimm_bus {
 	struct list_head list;
 	struct device dev;
 	int id, probe_active;
+	struct list_head poison_list;
 	struct mutex reconfig_mutex;
 };
 
diff --git a/drivers/nvdimm/nd.h b/drivers/nvdimm/nd.h
index e4e9f9ae0cc88..ba1633b9da31f 100644
--- a/drivers/nvdimm/nd.h
+++ b/drivers/nvdimm/nd.h
@@ -31,6 +31,12 @@ enum {
 	INT_LBASIZE_ALIGNMENT = 64,
 };
 
+struct nd_poison {
+	u64 start;
+	u64 length;
+	struct list_head list;
+};
+
 struct nvdimm_drvdata {
 	struct device *dev;
 	int nsindex_size;
@@ -256,6 +262,8 @@ int nvdimm_namespace_attach_btt(struct nd_namespace_common *ndns);
 int nvdimm_namespace_detach_btt(struct nd_namespace_common *ndns);
 const char *nvdimm_namespace_disk_name(struct nd_namespace_common *ndns,
 		char *name);
+void nvdimm_namespace_add_poison(struct nd_namespace_common *ndns,
+		struct badblocks *bb, resource_size_t offset);
 int nd_blk_region_init(struct nd_region *nd_region);
 void __nd_iostat_start(struct bio *bio, unsigned long *start);
 static inline bool nd_iostat_start(struct bio *bio, unsigned long *start)
diff --git a/drivers/nvdimm/pmem.c b/drivers/nvdimm/pmem.c
index ab689bca727d9..b493ff3fccb2d 100644
--- a/drivers/nvdimm/pmem.c
+++ b/drivers/nvdimm/pmem.c
@@ -23,6 +23,7 @@
 #include <linux/module.h>
 #include <linux/memory_hotplug.h>
 #include <linux/moduleparam.h>
+#include <linux/badblocks.h>
 #include <linux/vmalloc.h>
 #include <linux/slab.h>
 #include <linux/pmem.h>
@@ -41,11 +42,25 @@ struct pmem_device {
 	phys_addr_t		data_offset;
 	void __pmem		*virt_addr;
 	size_t			size;
+	struct badblocks	bb;
 };
 
 static int pmem_major;
 
-static void pmem_do_bvec(struct pmem_device *pmem, struct page *page,
+static bool is_bad_pmem(struct badblocks *bb, sector_t sector, unsigned int len)
+{
+	if (bb->count) {
+		sector_t first_bad;
+		int num_bad;
+
+		return !!badblocks_check(bb, sector, len / 512, &first_bad,
+				&num_bad);
+	}
+
+	return false;
+}
+
+static int pmem_do_bvec(struct pmem_device *pmem, struct page *page,
 			unsigned int len, unsigned int off, int rw,
 			sector_t sector)
 {
@@ -54,6 +69,8 @@ static void pmem_do_bvec(struct pmem_device *pmem, struct page *page,
 	void __pmem *pmem_addr = pmem->virt_addr + pmem_off;
 
 	if (rw == READ) {
+		if (unlikely(is_bad_pmem(&pmem->bb, sector, len)))
+			return -EIO;
 		memcpy_from_pmem(mem + off, pmem_addr, len);
 		flush_dcache_page(page);
 	} else {
@@ -62,10 +79,12 @@ static void pmem_do_bvec(struct pmem_device *pmem, struct page *page,
 	}
 
 	kunmap_atomic(mem);
+	return 0;
 }
 
 static blk_qc_t pmem_make_request(struct request_queue *q, struct bio *bio)
 {
+	int rc = 0;
 	bool do_acct;
 	unsigned long start;
 	struct bio_vec bvec;
@@ -74,9 +93,15 @@ static blk_qc_t pmem_make_request(struct request_queue *q, struct bio *bio)
 	struct pmem_device *pmem = bdev->bd_disk->private_data;
 
 	do_acct = nd_iostat_start(bio, &start);
-	bio_for_each_segment(bvec, bio, iter)
-		pmem_do_bvec(pmem, bvec.bv_page, bvec.bv_len, bvec.bv_offset,
-				bio_data_dir(bio), iter.bi_sector);
+	bio_for_each_segment(bvec, bio, iter) {
+		rc = pmem_do_bvec(pmem, bvec.bv_page, bvec.bv_len,
+				bvec.bv_offset, bio_data_dir(bio),
+				iter.bi_sector);
+		if (rc) {
+			bio->bi_error = rc;
+			break;
+		}
+	}
 	if (do_acct)
 		nd_iostat_end(bio, start);
 
@@ -91,13 +116,22 @@ static int pmem_rw_page(struct block_device *bdev, sector_t sector,
 		       struct page *page, int rw)
 {
 	struct pmem_device *pmem = bdev->bd_disk->private_data;
+	int rc;
 
-	pmem_do_bvec(pmem, page, PAGE_CACHE_SIZE, 0, rw, sector);
+	rc = pmem_do_bvec(pmem, page, PAGE_CACHE_SIZE, 0, rw, sector);
 	if (rw & WRITE)
 		wmb_pmem();
-	page_endio(page, rw & WRITE, 0);
 
-	return 0;
+	/*
+	 * The ->rw_page interface is subtle and tricky.  The core
+	 * retries on any error, so we can only invoke page_endio() in
+	 * the successful completion case.  Otherwise, we'll see crashes
+	 * caused by double completion.
+	 */
+	if (rc == 0)
+		page_endio(page, rw & WRITE, 0);
+
+	return rc;
 }
 
 static long pmem_direct_access(struct block_device *bdev, sector_t sector,
@@ -195,7 +229,12 @@ static int pmem_attach_disk(struct device *dev,
 	disk->driverfs_dev = dev;
 	set_capacity(disk, (pmem->size - pmem->data_offset) / 512);
 	pmem->pmem_disk = disk;
+	devm_exit_badblocks(dev, &pmem->bb);
+	if (devm_init_badblocks(dev, &pmem->bb))
+		return -ENOMEM;
+	nvdimm_namespace_add_poison(ndns, &pmem->bb, pmem->data_offset);
 
+	disk->bb = &pmem->bb;
 	add_disk(disk);
 	revalidate_disk(disk);
 
@@ -212,9 +251,13 @@ static int pmem_rw_bytes(struct nd_namespace_common *ndns,
 		return -EFAULT;
 	}
 
-	if (rw == READ)
+	if (rw == READ) {
+		unsigned int sz_align = ALIGN(size + (offset & (512 - 1)), 512);
+
+		if (unlikely(is_bad_pmem(&pmem->bb, offset / 512, sz_align)))
+			return -EIO;
 		memcpy_from_pmem(buf, pmem->virt_addr + offset, size);
-	else {
+	} else {
 		memcpy_to_pmem(pmem->virt_addr + offset, buf, size);
 		wmb_pmem();
 	}
@@ -377,6 +420,9 @@ static int nd_pmem_probe(struct device *dev)
 	pmem->ndns = ndns;
 	dev_set_drvdata(dev, pmem);
 	ndns->rw_bytes = pmem_rw_bytes;
+	if (devm_init_badblocks(dev, &pmem->bb))
+		return -ENOMEM;
+	nvdimm_namespace_add_poison(ndns, &pmem->bb, 0);
 
 	if (is_nd_btt(dev))
 		return nvdimm_namespace_attach_btt(ndns);