1 files changed, 473 insertions, 183 deletions

diff --git a/fs/exofs/ore.c b/fs/exofs/ore.c
index 25305af..fcfa86a 100644
--- a/fs/exofs/ore.c
+++ b/fs/exofs/ore.c
@@ -24,76 +24,287 @@
 
 #include <linux/slab.h>
 #include <asm/div64.h>
+#include <linux/lcm.h>
 
-#include <scsi/osd_ore.h>
+#include "ore_raid.h"
 
-#define ORE_ERR(fmt, a...) printk(KERN_ERR "ore: " fmt, ##a)
+MODULE_AUTHOR("Boaz Harrosh <bharrosh@panasas.com>");
+MODULE_DESCRIPTION("Objects Raid Engine ore.ko");
+MODULE_LICENSE("GPL");
+
+/* ore_verify_layout does a couple of things:
+ * 1. Given a minimum number of needed parameters fixes up the rest of the
+ *    members to be operatonals for the ore. The needed parameters are those
+ *    that are defined by the pnfs-objects layout STD.
+ * 2. Check to see if the current ore code actually supports these parameters
+ *    for example stripe_unit must be a multple of the system PAGE_SIZE,
+ *    and etc...
+ * 3. Cache some havily used calculations that will be needed by users.
+ */
+
+enum { BIO_MAX_PAGES_KMALLOC =
+		(PAGE_SIZE - sizeof(struct bio)) / sizeof(struct bio_vec),};
 
-#ifdef CONFIG_EXOFS_DEBUG
-#define ORE_DBGMSG(fmt, a...) \
-	printk(KERN_NOTICE "ore @%s:%d: " fmt, __func__, __LINE__, ##a)
-#else
-#define ORE_DBGMSG(fmt, a...) \
-	do { if (0) printk(fmt, ##a); } while (0)
-#endif
+int ore_verify_layout(unsigned total_comps, struct ore_layout *layout)
+{
+	u64 stripe_length;
+
+	switch (layout->raid_algorithm) {
+	case PNFS_OSD_RAID_0:
+		layout->parity = 0;
+		break;
+	case PNFS_OSD_RAID_5:
+		layout->parity = 1;
+		break;
+	case PNFS_OSD_RAID_PQ:
+	case PNFS_OSD_RAID_4:
+	default:
+		ORE_ERR("Only RAID_0/5 for now\n");
+		return -EINVAL;
+	}
+	if (0 != (layout->stripe_unit & ~PAGE_MASK)) {
+		ORE_ERR("Stripe Unit(0x%llx)"
+			  " must be Multples of PAGE_SIZE(0x%lx)\n",
+			  _LLU(layout->stripe_unit), PAGE_SIZE);
+		return -EINVAL;
+	}
+	if (layout->group_width) {
+		if (!layout->group_depth) {
+			ORE_ERR("group_depth == 0 && group_width != 0\n");
+			return -EINVAL;
+		}
+		if (total_comps < (layout->group_width * layout->mirrors_p1)) {
+			ORE_ERR("Data Map wrong, "
+				"numdevs=%d < group_width=%d * mirrors=%d\n",
+				total_comps, layout->group_width,
+				layout->mirrors_p1);
+			return -EINVAL;
+		}
+		layout->group_count = total_comps / layout->mirrors_p1 /
+						layout->group_width;
+	} else {
+		if (layout->group_depth) {
+			printk(KERN_NOTICE "Warning: group_depth ignored "
+				"group_width == 0 && group_depth == %lld\n",
+				_LLU(layout->group_depth));
+		}
+		layout->group_width = total_comps / layout->mirrors_p1;
+		layout->group_depth = -1;
+		layout->group_count = 1;
+	}
 
-/* u64 has problems with printk this will cast it to unsigned long long */
-#define _LLU(x) (unsigned long long)(x)
+	stripe_length = (u64)layout->group_width * layout->stripe_unit;
+	if (stripe_length >= (1ULL << 32)) {
+		ORE_ERR("Stripe_length(0x%llx) >= 32bit is not supported\n",
+			_LLU(stripe_length));
+		return -EINVAL;
+	}
 
-#define ORE_DBGMSG2(M...) do {} while (0)
-/* #define ORE_DBGMSG2 ORE_DBGMSG */
+	layout->max_io_length =
+		(BIO_MAX_PAGES_KMALLOC * PAGE_SIZE - layout->stripe_unit) *
+							layout->group_width;
+	if (layout->parity) {
+		unsigned stripe_length =
+				(layout->group_width - layout->parity) *
+				layout->stripe_unit;
 
-MODULE_AUTHOR("Boaz Harrosh <bharrosh@panasas.com>");
-MODULE_DESCRIPTION("Objects Raid Engine ore.ko");
-MODULE_LICENSE("GPL");
+		layout->max_io_length /= stripe_length;
+		layout->max_io_length *= stripe_length;
+	}
+	return 0;
+}
+EXPORT_SYMBOL(ore_verify_layout);
 
 static u8 *_ios_cred(struct ore_io_state *ios, unsigned index)
 {
-	return ios->comps->comps[index & ios->comps->single_comp].cred;
+	return ios->oc->comps[index & ios->oc->single_comp].cred;
 }
 
 static struct osd_obj_id *_ios_obj(struct ore_io_state *ios, unsigned index)
 {
-	return &ios->comps->comps[index & ios->comps->single_comp].obj;
+	return &ios->oc->comps[index & ios->oc->single_comp].obj;
 }
 
 static struct osd_dev *_ios_od(struct ore_io_state *ios, unsigned index)
 {
-	return ios->comps->ods[index];
+	ORE_DBGMSG2("oc->first_dev=%d oc->numdevs=%d i=%d oc->ods=%p\n",
+		    ios->oc->first_dev, ios->oc->numdevs, index,
+		    ios->oc->ods);
+
+	return ore_comp_dev(ios->oc, index);
 }
 
-int  ore_get_rw_state(struct ore_layout *layout, struct ore_components *comps,
+int  _ore_get_io_state(struct ore_layout *layout,
+			struct ore_components *oc, unsigned numdevs,
+			unsigned sgs_per_dev, unsigned num_par_pages,
+			struct ore_io_state **pios)
+{
+	struct ore_io_state *ios;
+	struct page **pages;
+	struct osd_sg_entry *sgilist;
+	struct __alloc_all_io_state {
+		struct ore_io_state ios;
+		struct ore_per_dev_state per_dev[numdevs];
+		union {
+			struct osd_sg_entry sglist[sgs_per_dev * numdevs];
+			struct page *pages[num_par_pages];
+		};
+	} *_aios;
+
+	if (likely(sizeof(*_aios) <= PAGE_SIZE)) {
+		_aios = kzalloc(sizeof(*_aios), GFP_KERNEL);
+		if (unlikely(!_aios)) {
+			ORE_DBGMSG("Failed kzalloc bytes=%zd\n",
+				   sizeof(*_aios));
+			*pios = NULL;
+			return -ENOMEM;
+		}
+		pages = num_par_pages ? _aios->pages : NULL;
+		sgilist = sgs_per_dev ? _aios->sglist : NULL;
+		ios = &_aios->ios;
+	} else {
+		struct __alloc_small_io_state {
+			struct ore_io_state ios;
+			struct ore_per_dev_state per_dev[numdevs];
+		} *_aio_small;
+		union __extra_part {
+			struct osd_sg_entry sglist[sgs_per_dev * numdevs];
+			struct page *pages[num_par_pages];
+		} *extra_part;
+
+		_aio_small = kzalloc(sizeof(*_aio_small), GFP_KERNEL);
+		if (unlikely(!_aio_small)) {
+			ORE_DBGMSG("Failed alloc first part bytes=%zd\n",
+				   sizeof(*_aio_small));
+			*pios = NULL;
+			return -ENOMEM;
+		}
+		extra_part = kzalloc(sizeof(*extra_part), GFP_KERNEL);
+		if (unlikely(!extra_part)) {
+			ORE_DBGMSG("Failed alloc second part bytes=%zd\n",
+				   sizeof(*extra_part));
+			kfree(_aio_small);
+			*pios = NULL;
+			return -ENOMEM;
+		}
+
+		pages = num_par_pages ? extra_part->pages : NULL;
+		sgilist = sgs_per_dev ? extra_part->sglist : NULL;
+		/* In this case the per_dev[0].sgilist holds the pointer to
+		 * be freed
+		 */
+		ios = &_aio_small->ios;
+		ios->extra_part_alloc = true;
+	}
+
+	if (pages) {
+		ios->parity_pages = pages;
+		ios->max_par_pages = num_par_pages;
+	}
+	if (sgilist) {
+		unsigned d;
+
+		for (d = 0; d < numdevs; ++d) {
+			ios->per_dev[d].sglist = sgilist;
+			sgilist += sgs_per_dev;
+		}
+		ios->sgs_per_dev = sgs_per_dev;
+	}
+
+	ios->layout = layout;
+	ios->oc = oc;
+	*pios = ios;
+	return 0;
+}
+
+/* Allocate an io_state for only a single group of devices
+ *
+ * If a user needs to call ore_read/write() this version must be used becase it
+ * allocates extra stuff for striping and raid.
+ * The ore might decide to only IO less then @length bytes do to alignmets
+ * and constrains as follows:
+ * - The IO cannot cross group boundary.
+ * - In raid5/6 The end of the IO must align at end of a stripe eg.
+ *   (@offset + @length) % strip_size == 0. Or the complete range is within a
+ *   single stripe.
+ * - Memory condition only permitted a shorter IO. (A user can use @length=~0
+ *   And check the returned ios->length for max_io_size.)
+ *
+ * The caller must check returned ios->length (and/or ios->nr_pages) and
+ * re-issue these pages that fall outside of ios->length
+ */
+int  ore_get_rw_state(struct ore_layout *layout, struct ore_components *oc,
 		      bool is_reading, u64 offset, u64 length,
 		      struct ore_io_state **pios)
 {
 	struct ore_io_state *ios;
+	unsigned numdevs = layout->group_width * layout->mirrors_p1;
+	unsigned sgs_per_dev = 0, max_par_pages = 0;
+	int ret;
 
-	/*TODO: Maybe use kmem_cach per sbi of size
-	 * exofs_io_state_size(layout->s_numdevs)
-	 */
-	ios = kzalloc(ore_io_state_size(comps->numdevs), GFP_KERNEL);
-	if (unlikely(!ios)) {
-		ORE_DBGMSG("Failed kzalloc bytes=%d\n",
-			     ore_io_state_size(comps->numdevs));
-		*pios = NULL;
-		return -ENOMEM;
+	if (layout->parity && length) {
+		unsigned data_devs = layout->group_width - layout->parity;
+		unsigned stripe_size = layout->stripe_unit * data_devs;
+		unsigned pages_in_unit = layout->stripe_unit / PAGE_SIZE;
+		u32 remainder;
+		u64 num_stripes;
+		u64 num_raid_units;
+
+		num_stripes = div_u64_rem(length, stripe_size, &remainder);
+		if (remainder)
+			++num_stripes;
+
+		num_raid_units =  num_stripes * layout->parity;
+
+		if (is_reading) {
+			/* For reads add per_dev sglist array */
+			/* TODO: Raid 6 we need twice more. Actually:
+			*         num_stripes / LCMdP(W,P);
+			*         if (W%P != 0) num_stripes *= parity;
+			*/
+
+			/* first/last seg is split */
+			num_raid_units += layout->group_width;
+			sgs_per_dev = div_u64(num_raid_units, data_devs);
+		} else {
+			/* For Writes add parity pages array. */
+			max_par_pages = num_raid_units * pages_in_unit *
+						sizeof(struct page *);
+		}
 	}
 
-	ios->layout = layout;
-	ios->comps = comps;
-	ios->offset = offset;
-	ios->length = length;
+	ret = _ore_get_io_state(layout, oc, numdevs, sgs_per_dev, max_par_pages,
+				pios);
+	if (unlikely(ret))
+		return ret;
+
+	ios = *pios;
 	ios->reading = is_reading;
+	ios->offset = offset;
+
+	if (length) {
+		ore_calc_stripe_info(layout, offset, length, &ios->si);
+		ios->length = ios->si.length;
+		ios->nr_pages = (ios->length + PAGE_SIZE - 1) / PAGE_SIZE;
+		if (layout->parity)
+			_ore_post_alloc_raid_stuff(ios);
+	}
 
-	*pios = ios;
 	return 0;
 }
 EXPORT_SYMBOL(ore_get_rw_state);
 
-int  ore_get_io_state(struct ore_layout *layout, struct ore_components *comps,
-		      struct ore_io_state **ios)
+/* Allocate an io_state for all the devices in the comps array
+ *
+ * This version of io_state allocation is used mostly by create/remove
+ * and trunc where we currently need all the devices. The only wastful
+ * bit is the read/write_attributes with no IO. Those sites should
+ * be converted to use ore_get_rw_state() with length=0
+ */
+int  ore_get_io_state(struct ore_layout *layout, struct ore_components *oc,
+		      struct ore_io_state **pios)
 {
-	return ore_get_rw_state(layout, comps, true, 0, 0, ios);
+	return _ore_get_io_state(layout, oc, oc->numdevs, 0, 0, pios);
 }
 EXPORT_SYMBOL(ore_get_io_state);
 
@@ -111,6 +322,7 @@ void ore_put_io_state(struct ore_io_state *ios)
 				bio_put(per_dev->bio);
 		}
 
+		_ore_free_raid_stuff(ios);
 		kfree(ios);
 	}
 }
@@ -138,7 +350,7 @@ static void _done_io(struct osd_request *or, void *p)
 	kref_put(&ios->kref, _last_io);
 }
 
-static int ore_io_execute(struct ore_io_state *ios)
+int ore_io_execute(struct ore_io_state *ios)
 {
 	DECLARE_COMPLETION_ONSTACK(wait);
 	bool sync = (ios->done == NULL);
@@ -198,7 +410,7 @@ static void _clear_bio(struct bio *bio)
 	}
 }
 
-int ore_check_io(struct ore_io_state *ios, u64 *resid)
+int ore_check_io(struct ore_io_state *ios, ore_on_dev_error on_dev_error)
 {
 	enum osd_err_priority acumulated_osd_err = 0;
 	int acumulated_lin_err = 0;
@@ -206,7 +418,8 @@ int ore_check_io(struct ore_io_state *ios, u64 *resid)
 
 	for (i = 0; i < ios->numdevs; i++) {
 		struct osd_sense_info osi;
-		struct osd_request *or = ios->per_dev[i].or;
+		struct ore_per_dev_state *per_dev = &ios->per_dev[i];
+		struct osd_request *or = per_dev->or;
 		int ret;
 
 		if (unlikely(!or))
@@ -218,29 +431,31 @@ int ore_check_io(struct ore_io_state *ios, u64 *resid)
 
 		if (OSD_ERR_PRI_CLEAR_PAGES == osi.osd_err_pri) {
 			/* start read offset passed endof file */
-			_clear_bio(ios->per_dev[i].bio);
+			_clear_bio(per_dev->bio);
 			ORE_DBGMSG("start read offset passed end of file "
 				"offset=0x%llx, length=0x%llx\n",
-				_LLU(ios->per_dev[i].offset),
-				_LLU(ios->per_dev[i].length));
+				_LLU(per_dev->offset),
+				_LLU(per_dev->length));
 
 			continue; /* we recovered */
 		}
 
+		if (on_dev_error) {
+			u64 residual = ios->reading ?
+					or->in.residual : or->out.residual;
+			u64 offset = (ios->offset + ios->length) - residual;
+			struct ore_dev *od = ios->oc->ods[
+					per_dev->dev - ios->oc->first_dev];
+
+			on_dev_error(ios, od, per_dev->dev, osi.osd_err_pri,
+				     offset, residual);
+		}
 		if (osi.osd_err_pri >= acumulated_osd_err) {
 			acumulated_osd_err = osi.osd_err_pri;
 			acumulated_lin_err = ret;
 		}
 	}
 
-	/* TODO: raid specific residual calculations */
-	if (resid) {
-		if (likely(!acumulated_lin_err))
-			*resid = 0;
-		else
-			*resid = ios->length;
-	}
-
 	return acumulated_lin_err;
 }
 EXPORT_SYMBOL(ore_check_io);
@@ -248,61 +463,65 @@ EXPORT_SYMBOL(ore_check_io);
 /*
  * L - logical offset into the file
  *
- * U - The number of bytes in a stripe within a group
+ * D - number of Data devices
+ *	D = group_width - parity
  *
- *	U = stripe_unit * group_width
+ * U - The number of bytes in a stripe within a group
+ *	U =  stripe_unit * D
  *
  * T - The number of bytes striped within a group of component objects
  *     (before advancing to the next group)
- *
- *	T = stripe_unit * group_width * group_depth
+ *	T = U * group_depth
  *
  * S - The number of bytes striped across all component objects
  *     before the pattern repeats
+ *	S = T * group_count
  *
- *	S = stripe_unit * group_width * group_depth * group_count
- *
- * M - The "major" (i.e., across all components) stripe number
- *
+ * M - The "major" (i.e., across all components) cycle number
  *	M = L / S
  *
- * G - Counts the groups from the beginning of the major stripe
- *
+ * G - Counts the groups from the beginning of the major cycle
  *	G = (L - (M * S)) / T	[or (L % S) / T]
  *
  * H - The byte offset within the group
- *
  *	H = (L - (M * S)) % T	[or (L % S) % T]
  *
  * N - The "minor" (i.e., across the group) stripe number
- *
  *	N = H / U
  *
  * C - The component index coresponding to L
  *
- *	C = (H - (N * U)) / stripe_unit + G * group_width
- *	[or (L % U) / stripe_unit + G * group_width]
+ *	C = (H - (N * U)) / stripe_unit + G * D
+ *	[or (L % U) / stripe_unit + G * D]
  *
  * O - The component offset coresponding to L
- *
  *	O = L % stripe_unit + N * stripe_unit + M * group_depth * stripe_unit
+ *
+ * LCMdP – Parity cycle: Lowest Common Multiple of group_width, parity
+ *          divide by parity
+ *	LCMdP = lcm(group_width, parity) / parity
+ *
+ * R - The parity Rotation stripe
+ *     (Note parity cycle always starts at a group's boundary)
+ *	R = N % LCMdP
+ *
+ * I = the first parity device index
+ *	I = (group_width + group_width - R*parity - parity) % group_width
+ *
+ * Craid - The component index Rotated
+ *	Craid = (group_width + C - R*parity) % group_width
+ *      (We add the group_width to avoid negative numbers modulo math)
  */
-struct _striping_info {
-	u64 obj_offset;
-	u64 group_length;
-	u64 M; /* for truncate */
-	unsigned dev;
-	unsigned unit_off;
-};
-
-static void _calc_stripe_info(struct ore_layout *layout, u64 file_offset,
-			      struct _striping_info *si)
+void ore_calc_stripe_info(struct ore_layout *layout, u64 file_offset,
+			  u64 length, struct ore_striping_info *si)
 {
 	u32	stripe_unit = layout->stripe_unit;
 	u32	group_width = layout->group_width;
 	u64	group_depth = layout->group_depth;
+	u32	parity      = layout->parity;
 
-	u32	U = stripe_unit * group_width;
+	u32	D = group_width - parity;
+	u32	U = D * stripe_unit;
 	u64	T = U * group_depth;
 	u64	S = T * layout->group_count;
 	u64	M = div64_u64(file_offset, S);
@@ -318,39 +537,65 @@ static void _calc_stripe_info(struct ore_layout *layout, u64 file_offset,
 	u32	N = div_u64(H, U);
 
 	/* "H - (N * U)" is just "H % U" so it's bound to u32 */
-	si->dev = (u32)(H - (N * U)) / stripe_unit + G * group_width;
-	si->dev *= layout->mirrors_p1;
+	u32	C = (u32)(H - (N * U)) / stripe_unit + G * group_width;
 
 	div_u64_rem(file_offset, stripe_unit, &si->unit_off);
 
 	si->obj_offset = si->unit_off + (N * stripe_unit) +
 				  (M * group_depth * stripe_unit);
 
-	si->group_length = T - H;
+	if (parity) {
+		u32 LCMdP = lcm(group_width, parity) / parity;
+		/* R     = N % LCMdP; */
+		u32 RxP   = (N % LCMdP) * parity;
+		u32 first_dev = C - C % group_width;
+
+		si->par_dev = (group_width + group_width - parity - RxP) %
+			      group_width + first_dev;
+		si->dev = (group_width + C - RxP) % group_width + first_dev;
+		si->bytes_in_stripe = U;
+		si->first_stripe_start = M * S + G * T + N * U;
+	} else {
+		/* Make the math correct see _prepare_one_group */
+		si->par_dev = group_width;
+		si->dev = C;
+	}
+
+	si->dev *= layout->mirrors_p1;
+	si->par_dev *= layout->mirrors_p1;
+	si->offset = file_offset;
+	si->length = T - H;
+	if (si->length > length)
+		si->length = length;
 	si->M = M;
 }
+EXPORT_SYMBOL(ore_calc_stripe_info);
 
-static int _add_stripe_unit(struct ore_io_state *ios,  unsigned *cur_pg,
-		unsigned pgbase, struct ore_per_dev_state *per_dev,
-		int cur_len)
+int _ore_add_stripe_unit(struct ore_io_state *ios,  unsigned *cur_pg,
+			 unsigned pgbase, struct page **pages,
+			 struct ore_per_dev_state *per_dev, int cur_len)
 {
 	unsigned pg = *cur_pg;
 	struct request_queue *q =
 			osd_request_queue(_ios_od(ios, per_dev->dev));
-
-	per_dev->length += cur_len;
+	unsigned len = cur_len;
+	int ret;
 
 	if (per_dev->bio == NULL) {
 		unsigned pages_in_stripe = ios->layout->group_width *
 					(ios->layout->stripe_unit / PAGE_SIZE);
-		unsigned bio_size = (ios->nr_pages + pages_in_stripe) /
-						ios->layout->group_width;
+		unsigned nr_pages = ios->nr_pages * ios->layout->group_width /
+					(ios->layout->group_width -
+					 ios->layout->parity);
+		unsigned bio_size = (nr_pages + pages_in_stripe) /
+					ios->layout->group_width;
 
 		per_dev->bio = bio_kmalloc(GFP_KERNEL, bio_size);
 		if (unlikely(!per_dev->bio)) {
 			ORE_DBGMSG("Failed to allocate BIO size=%u\n",
 				     bio_size);
-			return -ENOMEM;
+			ret = -ENOMEM;
+			goto out;
 		}
 	}
 
@@ -358,64 +603,90 @@ static int _add_stripe_unit(struct ore_io_state *ios,  unsigned *cur_pg,
 		unsigned pglen = min_t(unsigned, PAGE_SIZE - pgbase, cur_len);
 		unsigned added_len;
 
-		BUG_ON(ios->nr_pages <= pg);
 		cur_len -= pglen;
 
-		added_len = bio_add_pc_page(q, per_dev->bio, ios->pages[pg],
+		added_len = bio_add_pc_page(q, per_dev->bio, pages[pg],
 					    pglen, pgbase);
-		if (unlikely(pglen != added_len))
-			return -ENOMEM;
+		if (unlikely(pglen != added_len)) {
+			ORE_DBGMSG("Failed bio_add_pc_page bi_vcnt=%u\n",
+				   per_dev->bio->bi_vcnt);
+			ret = -ENOMEM;
+			goto out;
+		}
+		_add_stripe_page(ios->sp2d, &ios->si, pages[pg]);
+
 		pgbase = 0;
 		++pg;
 	}
 	BUG_ON(cur_len);
 
+	per_dev->length += len;
 	*cur_pg = pg;
-	return 0;
+	ret = 0;
+out:	/* we fail the complete unit on an error eg don't advance
+	 * per_dev->length and cur_pg. This means that we might have a bigger
+	 * bio than the CDB requested length (per_dev->length). That's fine
+	 * only the oposite is fatal.
+	 */
+	return ret;
 }
 
-static int _prepare_one_group(struct ore_io_state *ios, u64 length,
-			      struct _striping_info *si)
+static int _prepare_for_striping(struct ore_io_state *ios)
 {
+	struct ore_striping_info *si = &ios->si;
 	unsigned stripe_unit = ios->layout->stripe_unit;
 	unsigned mirrors_p1 = ios->layout->mirrors_p1;
-	unsigned devs_in_group = ios->layout->group_width * mirrors_p1;
+	unsigned group_width = ios->layout->group_width;
+	unsigned devs_in_group = group_width * mirrors_p1;
 	unsigned dev = si->dev;
 	unsigned first_dev = dev - (dev % devs_in_group);
-	unsigned max_comp = ios->numdevs ? ios->numdevs - mirrors_p1 : 0;
+	unsigned dev_order;
 	unsigned cur_pg = ios->pages_consumed;
+	u64 length = ios->length;
 	int ret = 0;
 
+	if (!ios->pages) {
+		ios->numdevs = ios->layout->mirrors_p1;
+		return 0;
+	}
+
+	BUG_ON(length > si->length);
+
+	dev_order = _dev_order(devs_in_group, mirrors_p1, si->par_dev, dev);
+	si->cur_comp = dev_order;
+	si->cur_pg = si->unit_off / PAGE_SIZE;
+
 	while (length) {
-		struct ore_per_dev_state *per_dev = &ios->per_dev[dev];
+		unsigned comp = dev - first_dev;
+		struct ore_per_dev_state *per_dev = &ios->per_dev[comp];
 		unsigned cur_len, page_off = 0;
 
 		if (!per_dev->length) {
 			per_dev->dev = dev;
-			if (dev < si->dev) {
-				per_dev->offset = si->obj_offset + stripe_unit -
-								   si->unit_off;
-				cur_len = stripe_unit;
-			} else if (dev == si->dev) {
+			if (dev == si->dev) {
+				WARN_ON(dev == si->par_dev);
 				per_dev->offset = si->obj_offset;
 				cur_len = stripe_unit - si->unit_off;
 				page_off = si->unit_off & ~PAGE_MASK;
 				BUG_ON(page_off && (page_off != ios->pgbase));
-			} else { /* dev > si->dev */
-				per_dev->offset = si->obj_offset - si->unit_off;
+			} else {
+				if (si->cur_comp > dev_order)
+					per_dev->offset =
+						si->obj_offset - si->unit_off;
+				else /* si->cur_comp < dev_order */
+					per_dev->offset =
+						si->obj_offset + stripe_unit -
+								   si->unit_off;
 				cur_len = stripe_unit;
 			}
-
-			if (max_comp < dev)
-				max_comp = dev;
 		} else {
 			cur_len = stripe_unit;
 		}
 		if (cur_len >= length)
 			cur_len = length;
 
-		ret = _add_stripe_unit(ios, &cur_pg, page_off , per_dev,
-				       cur_len);
+		ret = _ore_add_stripe_unit(ios, &cur_pg, page_off, ios->pages,
+					   per_dev, cur_len);
 		if (unlikely(ret))
 			goto out;
 
@@ -423,60 +694,60 @@ static int _prepare_one_group(struct ore_io_state *ios, u64 length,
 		dev = (dev % devs_in_group) + first_dev;
 
 		length -= cur_len;
-	}
-out:
-	ios->numdevs = max_comp + mirrors_p1;
-	ios->pages_consumed = cur_pg;
-	return ret;
-}
-
-static int _prepare_for_striping(struct ore_io_state *ios)
-{
-	u64 length = ios->length;
-	u64 offset = ios->offset;
-	struct _striping_info si;
-	int ret = 0;
 
-	if (!ios->pages) {
-		if (ios->kern_buff) {
-			struct ore_per_dev_state *per_dev = &ios->per_dev[0];
+		si->cur_comp = (si->cur_comp + 1) % group_width;
+		if (unlikely((dev == si->par_dev) || (!length && ios->sp2d))) {
+			if (!length && ios->sp2d) {
+				/* If we are writing and this is the very last
+				 * stripe. then operate on parity dev.
+				 */
+				dev = si->par_dev;
+			}
+			if (ios->sp2d)
+				/* In writes cur_len just means if it's the
+				 * last one. See _ore_add_parity_unit.
+				 */
+				cur_len = length;
+			per_dev = &ios->per_dev[dev - first_dev];
+			if (!per_dev->length) {
+				/* Only/always the parity unit of the first
+				 * stripe will be empty. So this is a chance to
+				 * initialize the per_dev info.
+				 */
+				per_dev->dev = dev;
+				per_dev->offset = si->obj_offset - si->unit_off;
+			}
 
-			_calc_stripe_info(ios->layout, ios->offset, &si);
-			per_dev->offset = si.obj_offset;
-			per_dev->dev = si.dev;
+			ret = _ore_add_parity_unit(ios, si, per_dev, cur_len);
+			if (unlikely(ret))
+					goto out;
 
-			/* no cross device without page array */
-			BUG_ON((ios->layout->group_width > 1) &&
-			       (si.unit_off + ios->length >
-				ios->layout->stripe_unit));
+			/* Rotate next par_dev backwards with wraping */
+			si->par_dev = (devs_in_group + si->par_dev -
+				       ios->layout->parity * mirrors_p1) %
+				      devs_in_group + first_dev;
+			/* Next stripe, start fresh */
+			si->cur_comp = 0;
+			si->cur_pg = 0;
 		}
-		ios->numdevs = ios->layout->mirrors_p1;
-		return 0;
-	}
-
-	while (length) {
-		_calc_stripe_info(ios->layout, offset, &si);
-
-		if (length < si.group_length)
-			si.group_length = length;
-
-		ret = _prepare_one_group(ios, si.group_length, &si);
-		if (unlikely(ret))
-			goto out;
-
-		offset += si.group_length;
-		length -= si.group_length;
 	}
-
 out:
-	return ret;
+	ios->numdevs = devs_in_group;
+	ios->pages_consumed = cur_pg;
+	if (unlikely(ret)) {
+		if (length == ios->length)
+			return ret;
+		else
+			ios->length -= length;
+	}
+	return 0;
 }
 
 int ore_create(struct ore_io_state *ios)
 {
 	int i, ret;
 
-	for (i = 0; i < ios->comps->numdevs; i++) {
+	for (i = 0; i < ios->oc->numdevs; i++) {
 		struct osd_request *or;
 
 		or = osd_start_request(_ios_od(ios, i), GFP_KERNEL);
@@ -501,7 +772,7 @@ int ore_remove(struct ore_io_state *ios)
 {
 	int i, ret;
 
-	for (i = 0; i < ios->comps->numdevs; i++) {
+	for (i = 0; i < ios->oc->numdevs; i++) {
 		struct osd_request *or;
 
 		or = osd_start_request(_ios_od(ios, i), GFP_KERNEL);
@@ -543,7 +814,6 @@ static int _write_mirror(struct ore_io_state *ios, int cur_comp)
 			goto out;
 		}
 		per_dev->or = or;
-		per_dev->offset = master_dev->offset;
 
 		if (ios->pages) {
 			struct bio *bio;
@@ -562,6 +832,7 @@ static int _write_mirror(struct ore_io_state *ios, int cur_comp)
 				__bio_clone(bio, master_dev->bio);
 				bio->bi_bdev = NULL;
 				bio->bi_next = NULL;
+				per_dev->offset = master_dev->offset;
 				per_dev->length = master_dev->length;
 				per_dev->bio =  bio;
 				per_dev->dev = dev;
@@ -579,7 +850,15 @@ static int _write_mirror(struct ore_io_state *ios, int cur_comp)
 				     _LLU(per_dev->offset),
 				     _LLU(per_dev->length), dev);
 		} else if (ios->kern_buff) {
-			ret = osd_req_write_kern(or, _ios_obj(ios, dev),
+			per_dev->offset = ios->si.obj_offset;
+			per_dev->dev = ios->si.dev + dev;
+
+			/* no cross device without page array */
+			BUG_ON((ios->layout->group_width > 1) &&
+			       (ios->si.unit_off + ios->length >
+				ios->layout->stripe_unit));
+
+			ret = osd_req_write_kern(or, _ios_obj(ios, per_dev->dev),
 						 per_dev->offset,
 						 ios->kern_buff, ios->length);
 			if (unlikely(ret))
@@ -588,7 +867,7 @@ static int _write_mirror(struct ore_io_state *ios, int cur_comp)
 				      "length=0x%llx dev=%d\n",
 				     _LLU(_ios_obj(ios, dev)->id),
 				     _LLU(per_dev->offset),
-				     _LLU(ios->length), dev);
+				     _LLU(ios->length), per_dev->dev);
 		} else {
 			osd_req_set_attributes(or, _ios_obj(ios, dev));
 			ORE_DBGMSG2("obj(0x%llx) set_attributes=%d dev=%d\n",
@@ -614,6 +893,14 @@ int ore_write(struct ore_io_state *ios)
 	int i;
 	int ret;
 
+	if (unlikely(ios->sp2d && !ios->r4w)) {
+		/* A library is attempting a RAID-write without providing
+		 * a pages lock interface.
+		 */
+		WARN_ON_ONCE(1);
+		return -ENOTSUPP;
+	}
+
 	ret = _prepare_for_striping(ios);
 	if (unlikely(ret))
 		return ret;
@@ -629,7 +916,7 @@ int ore_write(struct ore_io_state *ios)
 }
 EXPORT_SYMBOL(ore_write);
 
-static int _read_mirror(struct ore_io_state *ios, unsigned cur_comp)
+int _ore_read_mirror(struct ore_io_state *ios, unsigned cur_comp)
 {
 	struct osd_request *or;
 	struct ore_per_dev_state *per_dev = &ios->per_dev[cur_comp];
@@ -648,22 +935,27 @@ static int _read_mirror(struct ore_io_state *ios, unsigned cur_comp)
 	per_dev->or = or;
 
 	if (ios->pages) {
-		osd_req_read(or, obj, per_dev->offset,
-				per_dev->bio, per_dev->length);
+		if (per_dev->cur_sg) {
+			/* finalize the last sg_entry */
+			_ore_add_sg_seg(per_dev, 0, false);
+			if (unlikely(!per_dev->cur_sg))
+				return 0; /* Skip parity only device */
+
+			osd_req_read_sg(or, obj, per_dev->bio,
+					per_dev->sglist, per_dev->cur_sg);
+		} else {
+			/* The no raid case */
+			osd_req_read(or, obj, per_dev->offset,
+				     per_dev->bio, per_dev->length);
+		}
+
 		ORE_DBGMSG("read(0x%llx) offset=0x%llx length=0x%llx"
-			     " dev=%d\n", _LLU(obj->id),
+			     " dev=%d sg_len=%d\n", _LLU(obj->id),
 			     _LLU(per_dev->offset), _LLU(per_dev->length),
-			     first_dev);
-	} else if (ios->kern_buff) {
-		int ret = osd_req_read_kern(or, obj, per_dev->offset,
-					    ios->kern_buff, ios->length);
-		ORE_DBGMSG2("read_kern(0x%llx) offset=0x%llx "
-			      "length=0x%llx dev=%d ret=>%d\n",
-			      _LLU(obj->id), _LLU(per_dev->offset),
-			      _LLU(ios->length), first_dev, ret);
-		if (unlikely(ret))
-			return ret;
+			     first_dev, per_dev->cur_sg);
 	} else {
+		BUG_ON(ios->kern_buff);
+
 		osd_req_get_attributes(or, obj);
 		ORE_DBGMSG2("obj(0x%llx) get_attributes=%d dev=%d\n",
 			      _LLU(obj->id),
@@ -688,7 +980,7 @@ int ore_read(struct ore_io_state *ios)
 		return ret;
 
 	for (i = 0; i < ios->numdevs; i += ios->layout->mirrors_p1) {
-		ret = _read_mirror(ios, i);
+		ret = _ore_read_mirror(ios, i);
 		if (unlikely(ret))
 			return ret;
 	}
@@ -744,31 +1036,29 @@ static int _truncate_mirrors(struct ore_io_state *ios, unsigned cur_comp,
 }
 
 struct _trunc_info {
-	struct _striping_info si;
+	struct ore_striping_info si;
 	u64 prev_group_obj_off;
 	u64 next_group_obj_off;
 
 	unsigned first_group_dev;
 	unsigned nex_group_dev;
-	unsigned max_devs;
 };
 
-void _calc_trunk_info(struct ore_layout *layout, u64 file_offset,
-		       struct _trunc_info *ti)
+static void _calc_trunk_info(struct ore_layout *layout, u64 file_offset,
+			     struct _trunc_info *ti)
 {
 	unsigned stripe_unit = layout->stripe_unit;
 
-	_calc_stripe_info(layout, file_offset, &ti->si);
+	ore_calc_stripe_info(layout, file_offset, 0, &ti->si);
 
 	ti->prev_group_obj_off = ti->si.M * stripe_unit;
 	ti->next_group_obj_off = ti->si.M ? (ti->si.M - 1) * stripe_unit : 0;
 
 	ti->first_group_dev = ti->si.dev - (ti->si.dev % layout->group_width);
 	ti->nex_group_dev = ti->first_group_dev + layout->group_width;
-	ti->max_devs = layout->group_width * layout->group_count;
 }
 
-int ore_truncate(struct ore_layout *layout, struct ore_components *comps,
+int ore_truncate(struct ore_layout *layout, struct ore_components *oc,
 		   u64 size)
 {
 	struct ore_io_state *ios;
@@ -779,22 +1069,22 @@ int ore_truncate(struct ore_layout *layout, struct ore_components *comps,
 	struct _trunc_info ti;
 	int i, ret;
 
-	ret = ore_get_io_state(layout, comps, &ios);
+	ret = ore_get_io_state(layout, oc, &ios);
 	if (unlikely(ret))
 		return ret;
 
 	_calc_trunk_info(ios->layout, size, &ti);
 
-	size_attrs = kcalloc(ti.max_devs, sizeof(*size_attrs),
+	size_attrs = kcalloc(ios->oc->numdevs, sizeof(*size_attrs),
 			     GFP_KERNEL);
 	if (unlikely(!size_attrs)) {
 		ret = -ENOMEM;
 		goto out;
 	}
 
-	ios->numdevs = ios->comps->numdevs;
+	ios->numdevs = ios->oc->numdevs;
 
-	for (i = 0; i < ti.max_devs; ++i) {
+	for (i = 0; i < ios->numdevs; ++i) {
 		struct exofs_trunc_attr *size_attr = &size_attrs[i];
 		u64 obj_size;
 
@@ -815,7 +1105,7 @@ int ore_truncate(struct ore_layout *layout, struct ore_components *comps,
 		size_attr->attr.val_ptr = &size_attr->newsize;
 
 		ORE_DBGMSG("trunc(0x%llx) obj_offset=0x%llx dev=%d\n",
-			     _LLU(comps->comps->obj.id), _LLU(obj_size), i);
+			     _LLU(oc->comps->obj.id), _LLU(obj_size), i);
 		ret = _truncate_mirrors(ios, i * ios->layout->mirrors_p1,
 					&size_attr->attr);
 		if (unlikely(ret))