Adding upstream version 5.10.209.upstream/5.10.209upstream

Signed-off-by: Daniel Baumann <daniel.baumann@progress-linux.org>

8 files changed, 2140 insertions, 0 deletions

diff --git a/crypto/async_tx/Kconfig b/crypto/async_tx/Kconfig
new file mode 100644
index 000000000..89bafa2ef
--- /dev/null
+++ b/crypto/async_tx/Kconfig
@@ -0,0 +1,28 @@
+# SPDX-License-Identifier: GPL-2.0
+config ASYNC_CORE
+	tristate
+
+config ASYNC_MEMCPY
+	tristate
+	select ASYNC_CORE
+
+config ASYNC_XOR
+	tristate
+	select ASYNC_CORE
+	select XOR_BLOCKS
+
+config ASYNC_PQ
+	tristate
+	select ASYNC_CORE
+
+config ASYNC_RAID6_RECOV
+	tristate
+	select ASYNC_CORE
+	select ASYNC_PQ
+	select ASYNC_XOR
+
+config ASYNC_TX_DISABLE_PQ_VAL_DMA
+	bool
+
+config ASYNC_TX_DISABLE_XOR_VAL_DMA
+	bool
diff --git a/crypto/async_tx/Makefile b/crypto/async_tx/Makefile
new file mode 100644
index 000000000..056e48245
--- /dev/null
+++ b/crypto/async_tx/Makefile
@@ -0,0 +1,7 @@
+# SPDX-License-Identifier: GPL-2.0
+obj-$(CONFIG_ASYNC_CORE) += async_tx.o
+obj-$(CONFIG_ASYNC_MEMCPY) += async_memcpy.o
+obj-$(CONFIG_ASYNC_XOR) += async_xor.o
+obj-$(CONFIG_ASYNC_PQ) += async_pq.o
+obj-$(CONFIG_ASYNC_RAID6_RECOV) += async_raid6_recov.o
+obj-$(CONFIG_ASYNC_RAID6_TEST) += raid6test.o
diff --git a/crypto/async_tx/async_memcpy.c b/crypto/async_tx/async_memcpy.c
new file mode 100644
index 000000000..c538e30e9
--- /dev/null
+++ b/crypto/async_tx/async_memcpy.c
@@ -0,0 +1,97 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * copy offload engine support
+ *
+ * Copyright © 2006, Intel Corporation.
+ *
+ *      Dan Williams <dan.j.williams@intel.com>
+ *
+ *      with architecture considerations by:
+ *      Neil Brown <neilb@suse.de>
+ *      Jeff Garzik <jeff@garzik.org>
+ */
+#include <linux/kernel.h>
+#include <linux/highmem.h>
+#include <linux/module.h>
+#include <linux/mm.h>
+#include <linux/dma-mapping.h>
+#include <linux/async_tx.h>
+
+/**
+ * async_memcpy - attempt to copy memory with a dma engine.
+ * @dest: destination page
+ * @src: src page
+ * @dest_offset: offset into 'dest' to start transaction
+ * @src_offset: offset into 'src' to start transaction
+ * @len: length in bytes
+ * @submit: submission / completion modifiers
+ *
+ * honored flags: ASYNC_TX_ACK
+ */
+struct dma_async_tx_descriptor *
+async_memcpy(struct page *dest, struct page *src, unsigned int dest_offset,
+	     unsigned int src_offset, size_t len,
+	     struct async_submit_ctl *submit)
+{
+	struct dma_chan *chan = async_tx_find_channel(submit, DMA_MEMCPY,
+						      &dest, 1, &src, 1, len);
+	struct dma_device *device = chan ? chan->device : NULL;
+	struct dma_async_tx_descriptor *tx = NULL;
+	struct dmaengine_unmap_data *unmap = NULL;
+
+	if (device)
+		unmap = dmaengine_get_unmap_data(device->dev, 2, GFP_NOWAIT);
+
+	if (unmap && is_dma_copy_aligned(device, src_offset, dest_offset, len)) {
+		unsigned long dma_prep_flags = 0;
+
+		if (submit->cb_fn)
+			dma_prep_flags |= DMA_PREP_INTERRUPT;
+		if (submit->flags & ASYNC_TX_FENCE)
+			dma_prep_flags |= DMA_PREP_FENCE;
+
+		unmap->to_cnt = 1;
+		unmap->addr[0] = dma_map_page(device->dev, src, src_offset, len,
+					      DMA_TO_DEVICE);
+		unmap->from_cnt = 1;
+		unmap->addr[1] = dma_map_page(device->dev, dest, dest_offset, len,
+					      DMA_FROM_DEVICE);
+		unmap->len = len;
+
+		tx = device->device_prep_dma_memcpy(chan, unmap->addr[1],
+						    unmap->addr[0], len,
+						    dma_prep_flags);
+	}
+
+	if (tx) {
+		pr_debug("%s: (async) len: %zu\n", __func__, len);
+
+		dma_set_unmap(tx, unmap);
+		async_tx_submit(chan, tx, submit);
+	} else {
+		void *dest_buf, *src_buf;
+		pr_debug("%s: (sync) len: %zu\n", __func__, len);
+
+		/* wait for any prerequisite operations */
+		async_tx_quiesce(&submit->depend_tx);
+
+		dest_buf = kmap_atomic(dest) + dest_offset;
+		src_buf = kmap_atomic(src) + src_offset;
+
+		memcpy(dest_buf, src_buf, len);
+
+		kunmap_atomic(src_buf);
+		kunmap_atomic(dest_buf);
+
+		async_tx_sync_epilog(submit);
+	}
+
+	dmaengine_unmap_put(unmap);
+
+	return tx;
+}
+EXPORT_SYMBOL_GPL(async_memcpy);
+
+MODULE_AUTHOR("Intel Corporation");
+MODULE_DESCRIPTION("asynchronous memcpy api");
+MODULE_LICENSE("GPL");
diff --git a/crypto/async_tx/async_pq.c b/crypto/async_tx/async_pq.c
new file mode 100644
index 000000000..f9cdc5e91
--- /dev/null
+++ b/crypto/async_tx/async_pq.c
@@ -0,0 +1,463 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Copyright(c) 2007 Yuri Tikhonov <yur@emcraft.com>
+ * Copyright(c) 2009 Intel Corporation
+ */
+#include <linux/kernel.h>
+#include <linux/interrupt.h>
+#include <linux/module.h>
+#include <linux/dma-mapping.h>
+#include <linux/raid/pq.h>
+#include <linux/async_tx.h>
+#include <linux/gfp.h>
+
+/**
+ * pq_scribble_page - space to hold throwaway P or Q buffer for
+ * synchronous gen_syndrome
+ */
+static struct page *pq_scribble_page;
+
+/* the struct page *blocks[] parameter passed to async_gen_syndrome()
+ * and async_syndrome_val() contains the 'P' destination address at
+ * blocks[disks-2] and the 'Q' destination address at blocks[disks-1]
+ *
+ * note: these are macros as they are used as lvalues
+ */
+#define P(b, d) (b[d-2])
+#define Q(b, d) (b[d-1])
+
+#define MAX_DISKS 255
+
+/**
+ * do_async_gen_syndrome - asynchronously calculate P and/or Q
+ */
+static __async_inline struct dma_async_tx_descriptor *
+do_async_gen_syndrome(struct dma_chan *chan,
+		      const unsigned char *scfs, int disks,
+		      struct dmaengine_unmap_data *unmap,
+		      enum dma_ctrl_flags dma_flags,
+		      struct async_submit_ctl *submit)
+{
+	struct dma_async_tx_descriptor *tx = NULL;
+	struct dma_device *dma = chan->device;
+	enum async_tx_flags flags_orig = submit->flags;
+	dma_async_tx_callback cb_fn_orig = submit->cb_fn;
+	dma_async_tx_callback cb_param_orig = submit->cb_param;
+	int src_cnt = disks - 2;
+	unsigned short pq_src_cnt;
+	dma_addr_t dma_dest[2];
+	int src_off = 0;
+
+	while (src_cnt > 0) {
+		submit->flags = flags_orig;
+		pq_src_cnt = min(src_cnt, dma_maxpq(dma, dma_flags));
+		/* if we are submitting additional pqs, leave the chain open,
+		 * clear the callback parameters, and leave the destination
+		 * buffers mapped
+		 */
+		if (src_cnt > pq_src_cnt) {
+			submit->flags &= ~ASYNC_TX_ACK;
+			submit->flags |= ASYNC_TX_FENCE;
+			submit->cb_fn = NULL;
+			submit->cb_param = NULL;
+		} else {
+			submit->cb_fn = cb_fn_orig;
+			submit->cb_param = cb_param_orig;
+			if (cb_fn_orig)
+				dma_flags |= DMA_PREP_INTERRUPT;
+		}
+		if (submit->flags & ASYNC_TX_FENCE)
+			dma_flags |= DMA_PREP_FENCE;
+
+		/* Drivers force forward progress in case they can not provide
+		 * a descriptor
+		 */
+		for (;;) {
+			dma_dest[0] = unmap->addr[disks - 2];
+			dma_dest[1] = unmap->addr[disks - 1];
+			tx = dma->device_prep_dma_pq(chan, dma_dest,
+						     &unmap->addr[src_off],
+						     pq_src_cnt,
+						     &scfs[src_off], unmap->len,
+						     dma_flags);
+			if (likely(tx))
+				break;
+			async_tx_quiesce(&submit->depend_tx);
+			dma_async_issue_pending(chan);
+		}
+
+		dma_set_unmap(tx, unmap);
+		async_tx_submit(chan, tx, submit);
+		submit->depend_tx = tx;
+
+		/* drop completed sources */
+		src_cnt -= pq_src_cnt;
+		src_off += pq_src_cnt;
+
+		dma_flags |= DMA_PREP_CONTINUE;
+	}
+
+	return tx;
+}
+
+/**
+ * do_sync_gen_syndrome - synchronously calculate a raid6 syndrome
+ */
+static void
+do_sync_gen_syndrome(struct page **blocks, unsigned int *offsets, int disks,
+		     size_t len, struct async_submit_ctl *submit)
+{
+	void **srcs;
+	int i;
+	int start = -1, stop = disks - 3;
+
+	if (submit->scribble)
+		srcs = submit->scribble;
+	else
+		srcs = (void **) blocks;
+
+	for (i = 0; i < disks; i++) {
+		if (blocks[i] == NULL) {
+			BUG_ON(i > disks - 3); /* P or Q can't be zero */
+			srcs[i] = (void*)raid6_empty_zero_page;
+		} else {
+			srcs[i] = page_address(blocks[i]) + offsets[i];
+
+			if (i < disks - 2) {
+				stop = i;
+				if (start == -1)
+					start = i;
+			}
+		}
+	}
+	if (submit->flags & ASYNC_TX_PQ_XOR_DST) {
+		BUG_ON(!raid6_call.xor_syndrome);
+		if (start >= 0)
+			raid6_call.xor_syndrome(disks, start, stop, len, srcs);
+	} else
+		raid6_call.gen_syndrome(disks, len, srcs);
+	async_tx_sync_epilog(submit);
+}
+
+static inline bool
+is_dma_pq_aligned_offs(struct dma_device *dev, unsigned int *offs,
+				     int src_cnt, size_t len)
+{
+	int i;
+
+	for (i = 0; i < src_cnt; i++) {
+		if (!is_dma_pq_aligned(dev, offs[i], 0, len))
+			return false;
+	}
+	return true;
+}
+
+/**
+ * async_gen_syndrome - asynchronously calculate a raid6 syndrome
+ * @blocks: source blocks from idx 0..disks-3, P @ disks-2 and Q @ disks-1
+ * @offsets: offset array into each block (src and dest) to start transaction
+ * @disks: number of blocks (including missing P or Q, see below)
+ * @len: length of operation in bytes
+ * @submit: submission/completion modifiers
+ *
+ * General note: This routine assumes a field of GF(2^8) with a
+ * primitive polynomial of 0x11d and a generator of {02}.
+ *
+ * 'disks' note: callers can optionally omit either P or Q (but not
+ * both) from the calculation by setting blocks[disks-2] or
+ * blocks[disks-1] to NULL.  When P or Q is omitted 'len' must be <=
+ * PAGE_SIZE as a temporary buffer of this size is used in the
+ * synchronous path.  'disks' always accounts for both destination
+ * buffers.  If any source buffers (blocks[i] where i < disks - 2) are
+ * set to NULL those buffers will be replaced with the raid6_zero_page
+ * in the synchronous path and omitted in the hardware-asynchronous
+ * path.
+ */
+struct dma_async_tx_descriptor *
+async_gen_syndrome(struct page **blocks, unsigned int *offsets, int disks,
+		   size_t len, struct async_submit_ctl *submit)
+{
+	int src_cnt = disks - 2;
+	struct dma_chan *chan = async_tx_find_channel(submit, DMA_PQ,
+						      &P(blocks, disks), 2,
+						      blocks, src_cnt, len);
+	struct dma_device *device = chan ? chan->device : NULL;
+	struct dmaengine_unmap_data *unmap = NULL;
+
+	BUG_ON(disks > MAX_DISKS || !(P(blocks, disks) || Q(blocks, disks)));
+
+	if (device)
+		unmap = dmaengine_get_unmap_data(device->dev, disks, GFP_NOWAIT);
+
+	/* XORing P/Q is only implemented in software */
+	if (unmap && !(submit->flags & ASYNC_TX_PQ_XOR_DST) &&
+	    (src_cnt <= dma_maxpq(device, 0) ||
+	     dma_maxpq(device, DMA_PREP_CONTINUE) > 0) &&
+	    is_dma_pq_aligned_offs(device, offsets, disks, len)) {
+		struct dma_async_tx_descriptor *tx;
+		enum dma_ctrl_flags dma_flags = 0;
+		unsigned char coefs[MAX_DISKS];
+		int i, j;
+
+		/* run the p+q asynchronously */
+		pr_debug("%s: (async) disks: %d len: %zu\n",
+			 __func__, disks, len);
+
+		/* convert source addresses being careful to collapse 'empty'
+		 * sources and update the coefficients accordingly
+		 */
+		unmap->len = len;
+		for (i = 0, j = 0; i < src_cnt; i++) {
+			if (blocks[i] == NULL)
+				continue;
+			unmap->addr[j] = dma_map_page(device->dev, blocks[i],
+						offsets[i], len, DMA_TO_DEVICE);
+			coefs[j] = raid6_gfexp[i];
+			unmap->to_cnt++;
+			j++;
+		}
+
+		/*
+		 * DMAs use destinations as sources,
+		 * so use BIDIRECTIONAL mapping
+		 */
+		unmap->bidi_cnt++;
+		if (P(blocks, disks))
+			unmap->addr[j++] = dma_map_page(device->dev, P(blocks, disks),
+							P(offsets, disks),
+							len, DMA_BIDIRECTIONAL);
+		else {
+			unmap->addr[j++] = 0;
+			dma_flags |= DMA_PREP_PQ_DISABLE_P;
+		}
+
+		unmap->bidi_cnt++;
+		if (Q(blocks, disks))
+			unmap->addr[j++] = dma_map_page(device->dev, Q(blocks, disks),
+							Q(offsets, disks),
+							len, DMA_BIDIRECTIONAL);
+		else {
+			unmap->addr[j++] = 0;
+			dma_flags |= DMA_PREP_PQ_DISABLE_Q;
+		}
+
+		tx = do_async_gen_syndrome(chan, coefs, j, unmap, dma_flags, submit);
+		dmaengine_unmap_put(unmap);
+		return tx;
+	}
+
+	dmaengine_unmap_put(unmap);
+
+	/* run the pq synchronously */
+	pr_debug("%s: (sync) disks: %d len: %zu\n", __func__, disks, len);
+
+	/* wait for any prerequisite operations */
+	async_tx_quiesce(&submit->depend_tx);
+
+	if (!P(blocks, disks)) {
+		P(blocks, disks) = pq_scribble_page;
+		P(offsets, disks) = 0;
+	}
+	if (!Q(blocks, disks)) {
+		Q(blocks, disks) = pq_scribble_page;
+		Q(offsets, disks) = 0;
+	}
+	do_sync_gen_syndrome(blocks, offsets, disks, len, submit);
+
+	return NULL;
+}
+EXPORT_SYMBOL_GPL(async_gen_syndrome);
+
+static inline struct dma_chan *
+pq_val_chan(struct async_submit_ctl *submit, struct page **blocks, int disks, size_t len)
+{
+	#ifdef CONFIG_ASYNC_TX_DISABLE_PQ_VAL_DMA
+	return NULL;
+	#endif
+	return async_tx_find_channel(submit, DMA_PQ_VAL, NULL, 0,  blocks,
+				     disks, len);
+}
+
+/**
+ * async_syndrome_val - asynchronously validate a raid6 syndrome
+ * @blocks: source blocks from idx 0..disks-3, P @ disks-2 and Q @ disks-1
+ * @offset: common offset into each block (src and dest) to start transaction
+ * @disks: number of blocks (including missing P or Q, see below)
+ * @len: length of operation in bytes
+ * @pqres: on val failure SUM_CHECK_P_RESULT and/or SUM_CHECK_Q_RESULT are set
+ * @spare: temporary result buffer for the synchronous case
+ * @s_off: spare buffer page offset
+ * @submit: submission / completion modifiers
+ *
+ * The same notes from async_gen_syndrome apply to the 'blocks',
+ * and 'disks' parameters of this routine.  The synchronous path
+ * requires a temporary result buffer and submit->scribble to be
+ * specified.
+ */
+struct dma_async_tx_descriptor *
+async_syndrome_val(struct page **blocks, unsigned int *offsets, int disks,
+		   size_t len, enum sum_check_flags *pqres, struct page *spare,
+		   unsigned int s_off, struct async_submit_ctl *submit)
+{
+	struct dma_chan *chan = pq_val_chan(submit, blocks, disks, len);
+	struct dma_device *device = chan ? chan->device : NULL;
+	struct dma_async_tx_descriptor *tx;
+	unsigned char coefs[MAX_DISKS];
+	enum dma_ctrl_flags dma_flags = submit->cb_fn ? DMA_PREP_INTERRUPT : 0;
+	struct dmaengine_unmap_data *unmap = NULL;
+
+	BUG_ON(disks < 4 || disks > MAX_DISKS);
+
+	if (device)
+		unmap = dmaengine_get_unmap_data(device->dev, disks, GFP_NOWAIT);
+
+	if (unmap && disks <= dma_maxpq(device, 0) &&
+	    is_dma_pq_aligned_offs(device, offsets, disks, len)) {
+		struct device *dev = device->dev;
+		dma_addr_t pq[2];
+		int i, j = 0, src_cnt = 0;
+
+		pr_debug("%s: (async) disks: %d len: %zu\n",
+			 __func__, disks, len);
+
+		unmap->len = len;
+		for (i = 0; i < disks-2; i++)
+			if (likely(blocks[i])) {
+				unmap->addr[j] = dma_map_page(dev, blocks[i],
+							      offsets[i], len,
+							      DMA_TO_DEVICE);
+				coefs[j] = raid6_gfexp[i];
+				unmap->to_cnt++;
+				src_cnt++;
+				j++;
+			}
+
+		if (!P(blocks, disks)) {
+			pq[0] = 0;
+			dma_flags |= DMA_PREP_PQ_DISABLE_P;
+		} else {
+			pq[0] = dma_map_page(dev, P(blocks, disks),
+					     P(offsets, disks), len,
+					     DMA_TO_DEVICE);
+			unmap->addr[j++] = pq[0];
+			unmap->to_cnt++;
+		}
+		if (!Q(blocks, disks)) {
+			pq[1] = 0;
+			dma_flags |= DMA_PREP_PQ_DISABLE_Q;
+		} else {
+			pq[1] = dma_map_page(dev, Q(blocks, disks),
+					     Q(offsets, disks), len,
+					     DMA_TO_DEVICE);
+			unmap->addr[j++] = pq[1];
+			unmap->to_cnt++;
+		}
+
+		if (submit->flags & ASYNC_TX_FENCE)
+			dma_flags |= DMA_PREP_FENCE;
+		for (;;) {
+			tx = device->device_prep_dma_pq_val(chan, pq,
+							    unmap->addr,
+							    src_cnt,
+							    coefs,
+							    len, pqres,
+							    dma_flags);
+			if (likely(tx))
+				break;
+			async_tx_quiesce(&submit->depend_tx);
+			dma_async_issue_pending(chan);
+		}
+
+		dma_set_unmap(tx, unmap);
+		async_tx_submit(chan, tx, submit);
+	} else {
+		struct page *p_src = P(blocks, disks);
+		unsigned int p_off = P(offsets, disks);
+		struct page *q_src = Q(blocks, disks);
+		unsigned int q_off = Q(offsets, disks);
+		enum async_tx_flags flags_orig = submit->flags;
+		dma_async_tx_callback cb_fn_orig = submit->cb_fn;
+		void *scribble = submit->scribble;
+		void *cb_param_orig = submit->cb_param;
+		void *p, *q, *s;
+
+		pr_debug("%s: (sync) disks: %d len: %zu\n",
+			 __func__, disks, len);
+
+		/* caller must provide a temporary result buffer and
+		 * allow the input parameters to be preserved
+		 */
+		BUG_ON(!spare || !scribble);
+
+		/* wait for any prerequisite operations */
+		async_tx_quiesce(&submit->depend_tx);
+
+		/* recompute p and/or q into the temporary buffer and then
+		 * check to see the result matches the current value
+		 */
+		tx = NULL;
+		*pqres = 0;
+		if (p_src) {
+			init_async_submit(submit, ASYNC_TX_XOR_ZERO_DST, NULL,
+					  NULL, NULL, scribble);
+			tx = async_xor_offs(spare, s_off,
+					blocks, offsets, disks-2, len, submit);
+			async_tx_quiesce(&tx);
+			p = page_address(p_src) + p_off;
+			s = page_address(spare) + s_off;
+			*pqres |= !!memcmp(p, s, len) << SUM_CHECK_P;
+		}
+
+		if (q_src) {
+			P(blocks, disks) = NULL;
+			Q(blocks, disks) = spare;
+			Q(offsets, disks) = s_off;
+			init_async_submit(submit, 0, NULL, NULL, NULL, scribble);
+			tx = async_gen_syndrome(blocks, offsets, disks,
+					len, submit);
+			async_tx_quiesce(&tx);
+			q = page_address(q_src) + q_off;
+			s = page_address(spare) + s_off;
+			*pqres |= !!memcmp(q, s, len) << SUM_CHECK_Q;
+		}
+
+		/* restore P, Q and submit */
+		P(blocks, disks) = p_src;
+		P(offsets, disks) = p_off;
+		Q(blocks, disks) = q_src;
+		Q(offsets, disks) = q_off;
+
+		submit->cb_fn = cb_fn_orig;
+		submit->cb_param = cb_param_orig;
+		submit->flags = flags_orig;
+		async_tx_sync_epilog(submit);
+		tx = NULL;
+	}
+	dmaengine_unmap_put(unmap);
+
+	return tx;
+}
+EXPORT_SYMBOL_GPL(async_syndrome_val);
+
+static int __init async_pq_init(void)
+{
+	pq_scribble_page = alloc_page(GFP_KERNEL);
+
+	if (pq_scribble_page)
+		return 0;
+
+	pr_err("%s: failed to allocate required spare page\n", __func__);
+
+	return -ENOMEM;
+}
+
+static void __exit async_pq_exit(void)
+{
+	__free_page(pq_scribble_page);
+}
+
+module_init(async_pq_init);
+module_exit(async_pq_exit);
+
+MODULE_DESCRIPTION("asynchronous raid6 syndrome generation/validation");
+MODULE_LICENSE("GPL");
diff --git a/crypto/async_tx/async_raid6_recov.c b/crypto/async_tx/async_raid6_recov.c
new file mode 100644
index 000000000..354b8cd55
--- /dev/null
+++ b/crypto/async_tx/async_raid6_recov.c
@@ -0,0 +1,594 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Asynchronous RAID-6 recovery calculations ASYNC_TX API.
+ * Copyright(c) 2009 Intel Corporation
+ *
+ * based on raid6recov.c:
+ *   Copyright 2002 H. Peter Anvin
+ */
+#include <linux/kernel.h>
+#include <linux/interrupt.h>
+#include <linux/module.h>
+#include <linux/dma-mapping.h>
+#include <linux/raid/pq.h>
+#include <linux/async_tx.h>
+#include <linux/dmaengine.h>
+
+static struct dma_async_tx_descriptor *
+async_sum_product(struct page *dest, unsigned int d_off,
+		struct page **srcs, unsigned int *src_offs, unsigned char *coef,
+		size_t len, struct async_submit_ctl *submit)
+{
+	struct dma_chan *chan = async_tx_find_channel(submit, DMA_PQ,
+						      &dest, 1, srcs, 2, len);
+	struct dma_device *dma = chan ? chan->device : NULL;
+	struct dmaengine_unmap_data *unmap = NULL;
+	const u8 *amul, *bmul;
+	u8 ax, bx;
+	u8 *a, *b, *c;
+
+	if (dma)
+		unmap = dmaengine_get_unmap_data(dma->dev, 3, GFP_NOWAIT);
+
+	if (unmap) {
+		struct device *dev = dma->dev;
+		dma_addr_t pq[2];
+		struct dma_async_tx_descriptor *tx;
+		enum dma_ctrl_flags dma_flags = DMA_PREP_PQ_DISABLE_P;
+
+		if (submit->flags & ASYNC_TX_FENCE)
+			dma_flags |= DMA_PREP_FENCE;
+		unmap->addr[0] = dma_map_page(dev, srcs[0], src_offs[0],
+						len, DMA_TO_DEVICE);
+		unmap->addr[1] = dma_map_page(dev, srcs[1], src_offs[1],
+						len, DMA_TO_DEVICE);
+		unmap->to_cnt = 2;
+
+		unmap->addr[2] = dma_map_page(dev, dest, d_off,
+						len, DMA_BIDIRECTIONAL);
+		unmap->bidi_cnt = 1;
+		/* engine only looks at Q, but expects it to follow P */
+		pq[1] = unmap->addr[2];
+
+		unmap->len = len;
+		tx = dma->device_prep_dma_pq(chan, pq, unmap->addr, 2, coef,
+					     len, dma_flags);
+		if (tx) {
+			dma_set_unmap(tx, unmap);
+			async_tx_submit(chan, tx, submit);
+			dmaengine_unmap_put(unmap);
+			return tx;
+		}
+
+		/* could not get a descriptor, unmap and fall through to
+		 * the synchronous path
+		 */
+		dmaengine_unmap_put(unmap);
+	}
+
+	/* run the operation synchronously */
+	async_tx_quiesce(&submit->depend_tx);
+	amul = raid6_gfmul[coef[0]];
+	bmul = raid6_gfmul[coef[1]];
+	a = page_address(srcs[0]) + src_offs[0];
+	b = page_address(srcs[1]) + src_offs[1];
+	c = page_address(dest) + d_off;
+
+	while (len--) {
+		ax    = amul[*a++];
+		bx    = bmul[*b++];
+		*c++ = ax ^ bx;
+	}
+
+	return NULL;
+}
+
+static struct dma_async_tx_descriptor *
+async_mult(struct page *dest, unsigned int d_off, struct page *src,
+		unsigned int s_off, u8 coef, size_t len,
+		struct async_submit_ctl *submit)
+{
+	struct dma_chan *chan = async_tx_find_channel(submit, DMA_PQ,
+						      &dest, 1, &src, 1, len);
+	struct dma_device *dma = chan ? chan->device : NULL;
+	struct dmaengine_unmap_data *unmap = NULL;
+	const u8 *qmul; /* Q multiplier table */
+	u8 *d, *s;
+
+	if (dma)
+		unmap = dmaengine_get_unmap_data(dma->dev, 3, GFP_NOWAIT);
+
+	if (unmap) {
+		dma_addr_t dma_dest[2];
+		struct device *dev = dma->dev;
+		struct dma_async_tx_descriptor *tx;
+		enum dma_ctrl_flags dma_flags = DMA_PREP_PQ_DISABLE_P;
+
+		if (submit->flags & ASYNC_TX_FENCE)
+			dma_flags |= DMA_PREP_FENCE;
+		unmap->addr[0] = dma_map_page(dev, src, s_off,
+						len, DMA_TO_DEVICE);
+		unmap->to_cnt++;
+		unmap->addr[1] = dma_map_page(dev, dest, d_off,
+						len, DMA_BIDIRECTIONAL);
+		dma_dest[1] = unmap->addr[1];
+		unmap->bidi_cnt++;
+		unmap->len = len;
+
+		/* this looks funny, but the engine looks for Q at
+		 * dma_dest[1] and ignores dma_dest[0] as a dest
+		 * due to DMA_PREP_PQ_DISABLE_P
+		 */
+		tx = dma->device_prep_dma_pq(chan, dma_dest, unmap->addr,
+					     1, &coef, len, dma_flags);
+
+		if (tx) {
+			dma_set_unmap(tx, unmap);
+			dmaengine_unmap_put(unmap);
+			async_tx_submit(chan, tx, submit);
+			return tx;
+		}
+
+		/* could not get a descriptor, unmap and fall through to
+		 * the synchronous path
+		 */
+		dmaengine_unmap_put(unmap);
+	}
+
+	/* no channel available, or failed to allocate a descriptor, so
+	 * perform the operation synchronously
+	 */
+	async_tx_quiesce(&submit->depend_tx);
+	qmul  = raid6_gfmul[coef];
+	d = page_address(dest) + d_off;
+	s = page_address(src) + s_off;
+
+	while (len--)
+		*d++ = qmul[*s++];
+
+	return NULL;
+}
+
+static struct dma_async_tx_descriptor *
+__2data_recov_4(int disks, size_t bytes, int faila, int failb,
+		struct page **blocks, unsigned int *offs,
+		struct async_submit_ctl *submit)
+{
+	struct dma_async_tx_descriptor *tx = NULL;
+	struct page *p, *q, *a, *b;
+	unsigned int p_off, q_off, a_off, b_off;
+	struct page *srcs[2];
+	unsigned int src_offs[2];
+	unsigned char coef[2];
+	enum async_tx_flags flags = submit->flags;
+	dma_async_tx_callback cb_fn = submit->cb_fn;
+	void *cb_param = submit->cb_param;
+	void *scribble = submit->scribble;
+
+	p = blocks[disks-2];
+	p_off = offs[disks-2];
+	q = blocks[disks-1];
+	q_off = offs[disks-1];
+
+	a = blocks[faila];
+	a_off = offs[faila];
+	b = blocks[failb];
+	b_off = offs[failb];
+
+	/* in the 4 disk case P + Pxy == P and Q + Qxy == Q */
+	/* Dx = A*(P+Pxy) + B*(Q+Qxy) */
+	srcs[0] = p;
+	src_offs[0] = p_off;
+	srcs[1] = q;
+	src_offs[1] = q_off;
+	coef[0] = raid6_gfexi[failb-faila];
+	coef[1] = raid6_gfinv[raid6_gfexp[faila]^raid6_gfexp[failb]];
+	init_async_submit(submit, ASYNC_TX_FENCE, tx, NULL, NULL, scribble);
+	tx = async_sum_product(b, b_off, srcs, src_offs, coef, bytes, submit);
+
+	/* Dy = P+Pxy+Dx */
+	srcs[0] = p;
+	src_offs[0] = p_off;
+	srcs[1] = b;
+	src_offs[1] = b_off;
+	init_async_submit(submit, flags | ASYNC_TX_XOR_ZERO_DST, tx, cb_fn,
+			  cb_param, scribble);
+	tx = async_xor_offs(a, a_off, srcs, src_offs, 2, bytes, submit);
+
+	return tx;
+
+}
+
+static struct dma_async_tx_descriptor *
+__2data_recov_5(int disks, size_t bytes, int faila, int failb,
+		struct page **blocks, unsigned int *offs,
+		struct async_submit_ctl *submit)
+{
+	struct dma_async_tx_descriptor *tx = NULL;
+	struct page *p, *q, *g, *dp, *dq;
+	unsigned int p_off, q_off, g_off, dp_off, dq_off;
+	struct page *srcs[2];
+	unsigned int src_offs[2];
+	unsigned char coef[2];
+	enum async_tx_flags flags = submit->flags;
+	dma_async_tx_callback cb_fn = submit->cb_fn;
+	void *cb_param = submit->cb_param;
+	void *scribble = submit->scribble;
+	int good_srcs, good, i;
+
+	good_srcs = 0;
+	good = -1;
+	for (i = 0; i < disks-2; i++) {
+		if (blocks[i] == NULL)
+			continue;
+		if (i == faila || i == failb)
+			continue;
+		good = i;
+		good_srcs++;
+	}
+	BUG_ON(good_srcs > 1);
+
+	p = blocks[disks-2];
+	p_off = offs[disks-2];
+	q = blocks[disks-1];
+	q_off = offs[disks-1];
+	g = blocks[good];
+	g_off = offs[good];
+
+	/* Compute syndrome with zero for the missing data pages
+	 * Use the dead data pages as temporary storage for delta p and
+	 * delta q
+	 */
+	dp = blocks[faila];
+	dp_off = offs[faila];
+	dq = blocks[failb];
+	dq_off = offs[failb];
+
+	init_async_submit(submit, ASYNC_TX_FENCE, tx, NULL, NULL, scribble);
+	tx = async_memcpy(dp, g, dp_off, g_off, bytes, submit);
+	init_async_submit(submit, ASYNC_TX_FENCE, tx, NULL, NULL, scribble);
+	tx = async_mult(dq, dq_off, g, g_off,
+			raid6_gfexp[good], bytes, submit);
+
+	/* compute P + Pxy */
+	srcs[0] = dp;
+	src_offs[0] = dp_off;
+	srcs[1] = p;
+	src_offs[1] = p_off;
+	init_async_submit(submit, ASYNC_TX_FENCE|ASYNC_TX_XOR_DROP_DST, tx,
+			  NULL, NULL, scribble);
+	tx = async_xor_offs(dp, dp_off, srcs, src_offs, 2, bytes, submit);
+
+	/* compute Q + Qxy */
+	srcs[0] = dq;
+	src_offs[0] = dq_off;
+	srcs[1] = q;
+	src_offs[1] = q_off;
+	init_async_submit(submit, ASYNC_TX_FENCE|ASYNC_TX_XOR_DROP_DST, tx,
+			  NULL, NULL, scribble);
+	tx = async_xor_offs(dq, dq_off, srcs, src_offs, 2, bytes, submit);
+
+	/* Dx = A*(P+Pxy) + B*(Q+Qxy) */
+	srcs[0] = dp;
+	src_offs[0] = dp_off;
+	srcs[1] = dq;
+	src_offs[1] = dq_off;
+	coef[0] = raid6_gfexi[failb-faila];
+	coef[1] = raid6_gfinv[raid6_gfexp[faila]^raid6_gfexp[failb]];
+	init_async_submit(submit, ASYNC_TX_FENCE, tx, NULL, NULL, scribble);
+	tx = async_sum_product(dq, dq_off, srcs, src_offs, coef, bytes, submit);
+
+	/* Dy = P+Pxy+Dx */
+	srcs[0] = dp;
+	src_offs[0] = dp_off;
+	srcs[1] = dq;
+	src_offs[1] = dq_off;
+	init_async_submit(submit, flags | ASYNC_TX_XOR_DROP_DST, tx, cb_fn,
+			  cb_param, scribble);
+	tx = async_xor_offs(dp, dp_off, srcs, src_offs, 2, bytes, submit);
+
+	return tx;
+}
+
+static struct dma_async_tx_descriptor *
+__2data_recov_n(int disks, size_t bytes, int faila, int failb,
+	      struct page **blocks, unsigned int *offs,
+		  struct async_submit_ctl *submit)
+{
+	struct dma_async_tx_descriptor *tx = NULL;
+	struct page *p, *q, *dp, *dq;
+	unsigned int p_off, q_off, dp_off, dq_off;
+	struct page *srcs[2];
+	unsigned int src_offs[2];
+	unsigned char coef[2];
+	enum async_tx_flags flags = submit->flags;
+	dma_async_tx_callback cb_fn = submit->cb_fn;
+	void *cb_param = submit->cb_param;
+	void *scribble = submit->scribble;
+
+	p = blocks[disks-2];
+	p_off = offs[disks-2];
+	q = blocks[disks-1];
+	q_off = offs[disks-1];
+
+	/* Compute syndrome with zero for the missing data pages
+	 * Use the dead data pages as temporary storage for
+	 * delta p and delta q
+	 */
+	dp = blocks[faila];
+	dp_off = offs[faila];
+	blocks[faila] = NULL;
+	blocks[disks-2] = dp;
+	offs[disks-2] = dp_off;
+	dq = blocks[failb];
+	dq_off = offs[failb];
+	blocks[failb] = NULL;
+	blocks[disks-1] = dq;
+	offs[disks-1] = dq_off;
+
+	init_async_submit(submit, ASYNC_TX_FENCE, tx, NULL, NULL, scribble);
+	tx = async_gen_syndrome(blocks, offs, disks, bytes, submit);
+
+	/* Restore pointer table */
+	blocks[faila]   = dp;
+	offs[faila] = dp_off;
+	blocks[failb]   = dq;
+	offs[failb] = dq_off;
+	blocks[disks-2] = p;
+	offs[disks-2] = p_off;
+	blocks[disks-1] = q;
+	offs[disks-1] = q_off;
+
+	/* compute P + Pxy */
+	srcs[0] = dp;
+	src_offs[0] = dp_off;
+	srcs[1] = p;
+	src_offs[1] = p_off;
+	init_async_submit(submit, ASYNC_TX_FENCE|ASYNC_TX_XOR_DROP_DST, tx,
+			  NULL, NULL, scribble);
+	tx = async_xor_offs(dp, dp_off, srcs, src_offs, 2, bytes, submit);
+
+	/* compute Q + Qxy */
+	srcs[0] = dq;
+	src_offs[0] = dq_off;
+	srcs[1] = q;
+	src_offs[1] = q_off;
+	init_async_submit(submit, ASYNC_TX_FENCE|ASYNC_TX_XOR_DROP_DST, tx,
+			  NULL, NULL, scribble);
+	tx = async_xor_offs(dq, dq_off, srcs, src_offs, 2, bytes, submit);
+
+	/* Dx = A*(P+Pxy) + B*(Q+Qxy) */
+	srcs[0] = dp;
+	src_offs[0] = dp_off;
+	srcs[1] = dq;
+	src_offs[1] = dq_off;
+	coef[0] = raid6_gfexi[failb-faila];
+	coef[1] = raid6_gfinv[raid6_gfexp[faila]^raid6_gfexp[failb]];
+	init_async_submit(submit, ASYNC_TX_FENCE, tx, NULL, NULL, scribble);
+	tx = async_sum_product(dq, dq_off, srcs, src_offs, coef, bytes, submit);
+
+	/* Dy = P+Pxy+Dx */
+	srcs[0] = dp;
+	src_offs[0] = dp_off;
+	srcs[1] = dq;
+	src_offs[1] = dq_off;
+	init_async_submit(submit, flags | ASYNC_TX_XOR_DROP_DST, tx, cb_fn,
+			  cb_param, scribble);
+	tx = async_xor_offs(dp, dp_off, srcs, src_offs, 2, bytes, submit);
+
+	return tx;
+}
+
+/**
+ * async_raid6_2data_recov - asynchronously calculate two missing data blocks
+ * @disks: number of disks in the RAID-6 array
+ * @bytes: block size
+ * @faila: first failed drive index
+ * @failb: second failed drive index
+ * @blocks: array of source pointers where the last two entries are p and q
+ * @offs: array of offset for pages in blocks
+ * @submit: submission/completion modifiers
+ */
+struct dma_async_tx_descriptor *
+async_raid6_2data_recov(int disks, size_t bytes, int faila, int failb,
+			struct page **blocks, unsigned int *offs,
+			struct async_submit_ctl *submit)
+{
+	void *scribble = submit->scribble;
+	int non_zero_srcs, i;
+
+	BUG_ON(faila == failb);
+	if (failb < faila)
+		swap(faila, failb);
+
+	pr_debug("%s: disks: %d len: %zu\n", __func__, disks, bytes);
+
+	/* if a dma resource is not available or a scribble buffer is not
+	 * available punt to the synchronous path.  In the 'dma not
+	 * available' case be sure to use the scribble buffer to
+	 * preserve the content of 'blocks' as the caller intended.
+	 */
+	if (!async_dma_find_channel(DMA_PQ) || !scribble) {
+		void **ptrs = scribble ? scribble : (void **) blocks;
+
+		async_tx_quiesce(&submit->depend_tx);
+		for (i = 0; i < disks; i++)
+			if (blocks[i] == NULL)
+				ptrs[i] = (void *) raid6_empty_zero_page;
+			else
+				ptrs[i] = page_address(blocks[i]) + offs[i];
+
+		raid6_2data_recov(disks, bytes, faila, failb, ptrs);
+
+		async_tx_sync_epilog(submit);
+
+		return NULL;
+	}
+
+	non_zero_srcs = 0;
+	for (i = 0; i < disks-2 && non_zero_srcs < 4; i++)
+		if (blocks[i])
+			non_zero_srcs++;
+	switch (non_zero_srcs) {
+	case 0:
+	case 1:
+		/* There must be at least 2 sources - the failed devices. */
+		BUG();
+
+	case 2:
+		/* dma devices do not uniformly understand a zero source pq
+		 * operation (in contrast to the synchronous case), so
+		 * explicitly handle the special case of a 4 disk array with
+		 * both data disks missing.
+		 */
+		return __2data_recov_4(disks, bytes, faila, failb,
+				blocks, offs, submit);
+	case 3:
+		/* dma devices do not uniformly understand a single
+		 * source pq operation (in contrast to the synchronous
+		 * case), so explicitly handle the special case of a 5 disk
+		 * array with 2 of 3 data disks missing.
+		 */
+		return __2data_recov_5(disks, bytes, faila, failb,
+				blocks, offs, submit);
+	default:
+		return __2data_recov_n(disks, bytes, faila, failb,
+				blocks, offs, submit);
+	}
+}
+EXPORT_SYMBOL_GPL(async_raid6_2data_recov);
+
+/**
+ * async_raid6_datap_recov - asynchronously calculate a data and the 'p' block
+ * @disks: number of disks in the RAID-6 array
+ * @bytes: block size
+ * @faila: failed drive index
+ * @blocks: array of source pointers where the last two entries are p and q
+ * @offs: array of offset for pages in blocks
+ * @submit: submission/completion modifiers
+ */
+struct dma_async_tx_descriptor *
+async_raid6_datap_recov(int disks, size_t bytes, int faila,
+			struct page **blocks, unsigned int *offs,
+			struct async_submit_ctl *submit)
+{
+	struct dma_async_tx_descriptor *tx = NULL;
+	struct page *p, *q, *dq;
+	unsigned int p_off, q_off, dq_off;
+	u8 coef;
+	enum async_tx_flags flags = submit->flags;
+	dma_async_tx_callback cb_fn = submit->cb_fn;
+	void *cb_param = submit->cb_param;
+	void *scribble = submit->scribble;
+	int good_srcs, good, i;
+	struct page *srcs[2];
+	unsigned int src_offs[2];
+
+	pr_debug("%s: disks: %d len: %zu\n", __func__, disks, bytes);
+
+	/* if a dma resource is not available or a scribble buffer is not
+	 * available punt to the synchronous path.  In the 'dma not
+	 * available' case be sure to use the scribble buffer to
+	 * preserve the content of 'blocks' as the caller intended.
+	 */
+	if (!async_dma_find_channel(DMA_PQ) || !scribble) {
+		void **ptrs = scribble ? scribble : (void **) blocks;
+
+		async_tx_quiesce(&submit->depend_tx);
+		for (i = 0; i < disks; i++)
+			if (blocks[i] == NULL)
+				ptrs[i] = (void*)raid6_empty_zero_page;
+			else
+				ptrs[i] = page_address(blocks[i]) + offs[i];
+
+		raid6_datap_recov(disks, bytes, faila, ptrs);
+
+		async_tx_sync_epilog(submit);
+
+		return NULL;
+	}
+
+	good_srcs = 0;
+	good = -1;
+	for (i = 0; i < disks-2; i++) {
+		if (i == faila)
+			continue;
+		if (blocks[i]) {
+			good = i;
+			good_srcs++;
+			if (good_srcs > 1)
+				break;
+		}
+	}
+	BUG_ON(good_srcs == 0);
+
+	p = blocks[disks-2];
+	p_off = offs[disks-2];
+	q = blocks[disks-1];
+	q_off = offs[disks-1];
+
+	/* Compute syndrome with zero for the missing data page
+	 * Use the dead data page as temporary storage for delta q
+	 */
+	dq = blocks[faila];
+	dq_off = offs[faila];
+	blocks[faila] = NULL;
+	blocks[disks-1] = dq;
+	offs[disks-1] = dq_off;
+
+	/* in the 4-disk case we only need to perform a single source
+	 * multiplication with the one good data block.
+	 */
+	if (good_srcs == 1) {
+		struct page *g = blocks[good];
+		unsigned int g_off = offs[good];
+
+		init_async_submit(submit, ASYNC_TX_FENCE, tx, NULL, NULL,
+				  scribble);
+		tx = async_memcpy(p, g, p_off, g_off, bytes, submit);
+
+		init_async_submit(submit, ASYNC_TX_FENCE, tx, NULL, NULL,
+				  scribble);
+		tx = async_mult(dq, dq_off, g, g_off,
+				raid6_gfexp[good], bytes, submit);
+	} else {
+		init_async_submit(submit, ASYNC_TX_FENCE, tx, NULL, NULL,
+				  scribble);
+		tx = async_gen_syndrome(blocks, offs, disks, bytes, submit);
+	}
+
+	/* Restore pointer table */
+	blocks[faila]   = dq;
+	offs[faila] = dq_off;
+	blocks[disks-1] = q;
+	offs[disks-1] = q_off;
+
+	/* calculate g^{-faila} */
+	coef = raid6_gfinv[raid6_gfexp[faila]];
+
+	srcs[0] = dq;
+	src_offs[0] = dq_off;
+	srcs[1] = q;
+	src_offs[1] = q_off;
+	init_async_submit(submit, ASYNC_TX_FENCE|ASYNC_TX_XOR_DROP_DST, tx,
+			  NULL, NULL, scribble);
+	tx = async_xor_offs(dq, dq_off, srcs, src_offs, 2, bytes, submit);
+
+	init_async_submit(submit, ASYNC_TX_FENCE, tx, NULL, NULL, scribble);
+	tx = async_mult(dq, dq_off, dq, dq_off, coef, bytes, submit);
+
+	srcs[0] = p;
+	src_offs[0] = p_off;
+	srcs[1] = dq;
+	src_offs[1] = dq_off;
+	init_async_submit(submit, flags | ASYNC_TX_XOR_DROP_DST, tx, cb_fn,
+			  cb_param, scribble);
+	tx = async_xor_offs(p, p_off, srcs, src_offs, 2, bytes, submit);
+
+	return tx;
+}
+EXPORT_SYMBOL_GPL(async_raid6_datap_recov);
+
+MODULE_AUTHOR("Dan Williams <dan.j.williams@intel.com>");
+MODULE_DESCRIPTION("asynchronous RAID-6 recovery api");
+MODULE_LICENSE("GPL");
diff --git a/crypto/async_tx/async_tx.c b/crypto/async_tx/async_tx.c
new file mode 100644
index 000000000..925693431
--- /dev/null
+++ b/crypto/async_tx/async_tx.c
@@ -0,0 +1,281 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * core routines for the asynchronous memory transfer/transform api
+ *
+ * Copyright © 2006, Intel Corporation.
+ *
+ *	Dan Williams <dan.j.williams@intel.com>
+ *
+ *	with architecture considerations by:
+ *	Neil Brown <neilb@suse.de>
+ *	Jeff Garzik <jeff@garzik.org>
+ */
+#include <linux/rculist.h>
+#include <linux/module.h>
+#include <linux/kernel.h>
+#include <linux/async_tx.h>
+
+#ifdef CONFIG_DMA_ENGINE
+static int __init async_tx_init(void)
+{
+	async_dmaengine_get();
+
+	printk(KERN_INFO "async_tx: api initialized (async)\n");
+
+	return 0;
+}
+
+static void __exit async_tx_exit(void)
+{
+	async_dmaengine_put();
+}
+
+module_init(async_tx_init);
+module_exit(async_tx_exit);
+
+/**
+ * __async_tx_find_channel - find a channel to carry out the operation or let
+ *	the transaction execute synchronously
+ * @submit: transaction dependency and submission modifiers
+ * @tx_type: transaction type
+ */
+struct dma_chan *
+__async_tx_find_channel(struct async_submit_ctl *submit,
+			enum dma_transaction_type tx_type)
+{
+	struct dma_async_tx_descriptor *depend_tx = submit->depend_tx;
+
+	/* see if we can keep the chain on one channel */
+	if (depend_tx &&
+	    dma_has_cap(tx_type, depend_tx->chan->device->cap_mask))
+		return depend_tx->chan;
+	return async_dma_find_channel(tx_type);
+}
+EXPORT_SYMBOL_GPL(__async_tx_find_channel);
+#endif
+
+
+/**
+ * async_tx_channel_switch - queue an interrupt descriptor with a dependency
+ * 	pre-attached.
+ * @depend_tx: the operation that must finish before the new operation runs
+ * @tx: the new operation
+ */
+static void
+async_tx_channel_switch(struct dma_async_tx_descriptor *depend_tx,
+			struct dma_async_tx_descriptor *tx)
+{
+	struct dma_chan *chan = depend_tx->chan;
+	struct dma_device *device = chan->device;
+	struct dma_async_tx_descriptor *intr_tx = (void *) ~0;
+
+	/* first check to see if we can still append to depend_tx */
+	txd_lock(depend_tx);
+	if (txd_parent(depend_tx) && depend_tx->chan == tx->chan) {
+		txd_chain(depend_tx, tx);
+		intr_tx = NULL;
+	}
+	txd_unlock(depend_tx);
+
+	/* attached dependency, flush the parent channel */
+	if (!intr_tx) {
+		device->device_issue_pending(chan);
+		return;
+	}
+
+	/* see if we can schedule an interrupt
+	 * otherwise poll for completion
+	 */
+	if (dma_has_cap(DMA_INTERRUPT, device->cap_mask))
+		intr_tx = device->device_prep_dma_interrupt(chan, 0);
+	else
+		intr_tx = NULL;
+
+	if (intr_tx) {
+		intr_tx->callback = NULL;
+		intr_tx->callback_param = NULL;
+		/* safe to chain outside the lock since we know we are
+		 * not submitted yet
+		 */
+		txd_chain(intr_tx, tx);
+
+		/* check if we need to append */
+		txd_lock(depend_tx);
+		if (txd_parent(depend_tx)) {
+			txd_chain(depend_tx, intr_tx);
+			async_tx_ack(intr_tx);
+			intr_tx = NULL;
+		}
+		txd_unlock(depend_tx);
+
+		if (intr_tx) {
+			txd_clear_parent(intr_tx);
+			intr_tx->tx_submit(intr_tx);
+			async_tx_ack(intr_tx);
+		}
+		device->device_issue_pending(chan);
+	} else {
+		if (dma_wait_for_async_tx(depend_tx) != DMA_COMPLETE)
+			panic("%s: DMA error waiting for depend_tx\n",
+			      __func__);
+		tx->tx_submit(tx);
+	}
+}
+
+
+/**
+ * submit_disposition - flags for routing an incoming operation
+ * @ASYNC_TX_SUBMITTED: we were able to append the new operation under the lock
+ * @ASYNC_TX_CHANNEL_SWITCH: when the lock is dropped schedule a channel switch
+ * @ASYNC_TX_DIRECT_SUBMIT: when the lock is dropped submit directly
+ *
+ * while holding depend_tx->lock we must avoid submitting new operations
+ * to prevent a circular locking dependency with drivers that already
+ * hold a channel lock when calling async_tx_run_dependencies.
+ */
+enum submit_disposition {
+	ASYNC_TX_SUBMITTED,
+	ASYNC_TX_CHANNEL_SWITCH,
+	ASYNC_TX_DIRECT_SUBMIT,
+};
+
+void
+async_tx_submit(struct dma_chan *chan, struct dma_async_tx_descriptor *tx,
+		struct async_submit_ctl *submit)
+{
+	struct dma_async_tx_descriptor *depend_tx = submit->depend_tx;
+
+	tx->callback = submit->cb_fn;
+	tx->callback_param = submit->cb_param;
+
+	if (depend_tx) {
+		enum submit_disposition s;
+
+		/* sanity check the dependency chain:
+		 * 1/ if ack is already set then we cannot be sure
+		 * we are referring to the correct operation
+		 * 2/ dependencies are 1:1 i.e. two transactions can
+		 * not depend on the same parent
+		 */
+		BUG_ON(async_tx_test_ack(depend_tx) || txd_next(depend_tx) ||
+		       txd_parent(tx));
+
+		/* the lock prevents async_tx_run_dependencies from missing
+		 * the setting of ->next when ->parent != NULL
+		 */
+		txd_lock(depend_tx);
+		if (txd_parent(depend_tx)) {
+			/* we have a parent so we can not submit directly
+			 * if we are staying on the same channel: append
+			 * else: channel switch
+			 */
+			if (depend_tx->chan == chan) {
+				txd_chain(depend_tx, tx);
+				s = ASYNC_TX_SUBMITTED;
+			} else
+				s = ASYNC_TX_CHANNEL_SWITCH;
+		} else {
+			/* we do not have a parent so we may be able to submit
+			 * directly if we are staying on the same channel
+			 */
+			if (depend_tx->chan == chan)
+				s = ASYNC_TX_DIRECT_SUBMIT;
+			else
+				s = ASYNC_TX_CHANNEL_SWITCH;
+		}
+		txd_unlock(depend_tx);
+
+		switch (s) {
+		case ASYNC_TX_SUBMITTED:
+			break;
+		case ASYNC_TX_CHANNEL_SWITCH:
+			async_tx_channel_switch(depend_tx, tx);
+			break;
+		case ASYNC_TX_DIRECT_SUBMIT:
+			txd_clear_parent(tx);
+			tx->tx_submit(tx);
+			break;
+		}
+	} else {
+		txd_clear_parent(tx);
+		tx->tx_submit(tx);
+	}
+
+	if (submit->flags & ASYNC_TX_ACK)
+		async_tx_ack(tx);
+
+	if (depend_tx)
+		async_tx_ack(depend_tx);
+}
+EXPORT_SYMBOL_GPL(async_tx_submit);
+
+/**
+ * async_trigger_callback - schedules the callback function to be run
+ * @submit: submission and completion parameters
+ *
+ * honored flags: ASYNC_TX_ACK
+ *
+ * The callback is run after any dependent operations have completed.
+ */
+struct dma_async_tx_descriptor *
+async_trigger_callback(struct async_submit_ctl *submit)
+{
+	struct dma_chan *chan;
+	struct dma_device *device;
+	struct dma_async_tx_descriptor *tx;
+	struct dma_async_tx_descriptor *depend_tx = submit->depend_tx;
+
+	if (depend_tx) {
+		chan = depend_tx->chan;
+		device = chan->device;
+
+		/* see if we can schedule an interrupt
+		 * otherwise poll for completion
+		 */
+		if (device && !dma_has_cap(DMA_INTERRUPT, device->cap_mask))
+			device = NULL;
+
+		tx = device ? device->device_prep_dma_interrupt(chan, 0) : NULL;
+	} else
+		tx = NULL;
+
+	if (tx) {
+		pr_debug("%s: (async)\n", __func__);
+
+		async_tx_submit(chan, tx, submit);
+	} else {
+		pr_debug("%s: (sync)\n", __func__);
+
+		/* wait for any prerequisite operations */
+		async_tx_quiesce(&submit->depend_tx);
+
+		async_tx_sync_epilog(submit);
+	}
+
+	return tx;
+}
+EXPORT_SYMBOL_GPL(async_trigger_callback);
+
+/**
+ * async_tx_quiesce - ensure tx is complete and freeable upon return
+ * @tx - transaction to quiesce
+ */
+void async_tx_quiesce(struct dma_async_tx_descriptor **tx)
+{
+	if (*tx) {
+		/* if ack is already set then we cannot be sure
+		 * we are referring to the correct operation
+		 */
+		BUG_ON(async_tx_test_ack(*tx));
+		if (dma_wait_for_async_tx(*tx) != DMA_COMPLETE)
+			panic("%s: DMA error waiting for transaction\n",
+			      __func__);
+		async_tx_ack(*tx);
+		*tx = NULL;
+	}
+}
+EXPORT_SYMBOL_GPL(async_tx_quiesce);
+
+MODULE_AUTHOR("Intel Corporation");
+MODULE_DESCRIPTION("Asynchronous Bulk Memory Transactions API");
+MODULE_LICENSE("GPL");
diff --git a/crypto/async_tx/async_xor.c b/crypto/async_tx/async_xor.c
new file mode 100644
index 000000000..d8a915211
--- /dev/null
+++ b/crypto/async_tx/async_xor.c
@@ -0,0 +1,420 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * xor offload engine api
+ *
+ * Copyright © 2006, Intel Corporation.
+ *
+ *      Dan Williams <dan.j.williams@intel.com>
+ *
+ *      with architecture considerations by:
+ *      Neil Brown <neilb@suse.de>
+ *      Jeff Garzik <jeff@garzik.org>
+ */
+#include <linux/kernel.h>
+#include <linux/interrupt.h>
+#include <linux/module.h>
+#include <linux/mm.h>
+#include <linux/dma-mapping.h>
+#include <linux/raid/xor.h>
+#include <linux/async_tx.h>
+
+/* do_async_xor - dma map the pages and perform the xor with an engine */
+static __async_inline struct dma_async_tx_descriptor *
+do_async_xor(struct dma_chan *chan, struct dmaengine_unmap_data *unmap,
+	     struct async_submit_ctl *submit)
+{
+	struct dma_device *dma = chan->device;
+	struct dma_async_tx_descriptor *tx = NULL;
+	dma_async_tx_callback cb_fn_orig = submit->cb_fn;
+	void *cb_param_orig = submit->cb_param;
+	enum async_tx_flags flags_orig = submit->flags;
+	enum dma_ctrl_flags dma_flags = 0;
+	int src_cnt = unmap->to_cnt;
+	int xor_src_cnt;
+	dma_addr_t dma_dest = unmap->addr[unmap->to_cnt];
+	dma_addr_t *src_list = unmap->addr;
+
+	while (src_cnt) {
+		dma_addr_t tmp;
+
+		submit->flags = flags_orig;
+		xor_src_cnt = min(src_cnt, (int)dma->max_xor);
+		/* if we are submitting additional xors, leave the chain open
+		 * and clear the callback parameters
+		 */
+		if (src_cnt > xor_src_cnt) {
+			submit->flags &= ~ASYNC_TX_ACK;
+			submit->flags |= ASYNC_TX_FENCE;
+			submit->cb_fn = NULL;
+			submit->cb_param = NULL;
+		} else {
+			submit->cb_fn = cb_fn_orig;
+			submit->cb_param = cb_param_orig;
+		}
+		if (submit->cb_fn)
+			dma_flags |= DMA_PREP_INTERRUPT;
+		if (submit->flags & ASYNC_TX_FENCE)
+			dma_flags |= DMA_PREP_FENCE;
+
+		/* Drivers force forward progress in case they can not provide a
+		 * descriptor
+		 */
+		tmp = src_list[0];
+		if (src_list > unmap->addr)
+			src_list[0] = dma_dest;
+		tx = dma->device_prep_dma_xor(chan, dma_dest, src_list,
+					      xor_src_cnt, unmap->len,
+					      dma_flags);
+
+		if (unlikely(!tx))
+			async_tx_quiesce(&submit->depend_tx);
+
+		/* spin wait for the preceding transactions to complete */
+		while (unlikely(!tx)) {
+			dma_async_issue_pending(chan);
+			tx = dma->device_prep_dma_xor(chan, dma_dest,
+						      src_list,
+						      xor_src_cnt, unmap->len,
+						      dma_flags);
+		}
+		src_list[0] = tmp;
+
+		dma_set_unmap(tx, unmap);
+		async_tx_submit(chan, tx, submit);
+		submit->depend_tx = tx;
+
+		if (src_cnt > xor_src_cnt) {
+			/* drop completed sources */
+			src_cnt -= xor_src_cnt;
+			/* use the intermediate result a source */
+			src_cnt++;
+			src_list += xor_src_cnt - 1;
+		} else
+			break;
+	}
+
+	return tx;
+}
+
+static void
+do_sync_xor_offs(struct page *dest, unsigned int offset,
+		struct page **src_list, unsigned int *src_offs,
+	    int src_cnt, size_t len, struct async_submit_ctl *submit)
+{
+	int i;
+	int xor_src_cnt = 0;
+	int src_off = 0;
+	void *dest_buf;
+	void **srcs;
+
+	if (submit->scribble)
+		srcs = submit->scribble;
+	else
+		srcs = (void **) src_list;
+
+	/* convert to buffer pointers */
+	for (i = 0; i < src_cnt; i++)
+		if (src_list[i])
+			srcs[xor_src_cnt++] = page_address(src_list[i]) +
+				(src_offs ? src_offs[i] : offset);
+	src_cnt = xor_src_cnt;
+	/* set destination address */
+	dest_buf = page_address(dest) + offset;
+
+	if (submit->flags & ASYNC_TX_XOR_ZERO_DST)
+		memset(dest_buf, 0, len);
+
+	while (src_cnt > 0) {
+		/* process up to 'MAX_XOR_BLOCKS' sources */
+		xor_src_cnt = min(src_cnt, MAX_XOR_BLOCKS);
+		xor_blocks(xor_src_cnt, len, dest_buf, &srcs[src_off]);
+
+		/* drop completed sources */
+		src_cnt -= xor_src_cnt;
+		src_off += xor_src_cnt;
+	}
+
+	async_tx_sync_epilog(submit);
+}
+
+static inline bool
+dma_xor_aligned_offsets(struct dma_device *device, unsigned int offset,
+		unsigned int *src_offs, int src_cnt, int len)
+{
+	int i;
+
+	if (!is_dma_xor_aligned(device, offset, 0, len))
+		return false;
+
+	if (!src_offs)
+		return true;
+
+	for (i = 0; i < src_cnt; i++) {
+		if (!is_dma_xor_aligned(device, src_offs[i], 0, len))
+			return false;
+	}
+	return true;
+}
+
+/**
+ * async_xor_offs - attempt to xor a set of blocks with a dma engine.
+ * @dest: destination page
+ * @offset: dst offset to start transaction
+ * @src_list: array of source pages
+ * @src_offs: array of source pages offset, NULL means common src/dst offset
+ * @src_cnt: number of source pages
+ * @len: length in bytes
+ * @submit: submission / completion modifiers
+ *
+ * honored flags: ASYNC_TX_ACK, ASYNC_TX_XOR_ZERO_DST, ASYNC_TX_XOR_DROP_DST
+ *
+ * xor_blocks always uses the dest as a source so the
+ * ASYNC_TX_XOR_ZERO_DST flag must be set to not include dest data in
+ * the calculation.  The assumption with dma eninges is that they only
+ * use the destination buffer as a source when it is explicity specified
+ * in the source list.
+ *
+ * src_list note: if the dest is also a source it must be at index zero.
+ * The contents of this array will be overwritten if a scribble region
+ * is not specified.
+ */
+struct dma_async_tx_descriptor *
+async_xor_offs(struct page *dest, unsigned int offset,
+		struct page **src_list, unsigned int *src_offs,
+		int src_cnt, size_t len, struct async_submit_ctl *submit)
+{
+	struct dma_chan *chan = async_tx_find_channel(submit, DMA_XOR,
+						      &dest, 1, src_list,
+						      src_cnt, len);
+	struct dma_device *device = chan ? chan->device : NULL;
+	struct dmaengine_unmap_data *unmap = NULL;
+
+	BUG_ON(src_cnt <= 1);
+
+	if (device)
+		unmap = dmaengine_get_unmap_data(device->dev, src_cnt+1, GFP_NOWAIT);
+
+	if (unmap && dma_xor_aligned_offsets(device, offset,
+				src_offs, src_cnt, len)) {
+		struct dma_async_tx_descriptor *tx;
+		int i, j;
+
+		/* run the xor asynchronously */
+		pr_debug("%s (async): len: %zu\n", __func__, len);
+
+		unmap->len = len;
+		for (i = 0, j = 0; i < src_cnt; i++) {
+			if (!src_list[i])
+				continue;
+			unmap->to_cnt++;
+			unmap->addr[j++] = dma_map_page(device->dev, src_list[i],
+					src_offs ? src_offs[i] : offset,
+					len, DMA_TO_DEVICE);
+		}
+
+		/* map it bidirectional as it may be re-used as a source */
+		unmap->addr[j] = dma_map_page(device->dev, dest, offset, len,
+					      DMA_BIDIRECTIONAL);
+		unmap->bidi_cnt = 1;
+
+		tx = do_async_xor(chan, unmap, submit);
+		dmaengine_unmap_put(unmap);
+		return tx;
+	} else {
+		dmaengine_unmap_put(unmap);
+		/* run the xor synchronously */
+		pr_debug("%s (sync): len: %zu\n", __func__, len);
+		WARN_ONCE(chan, "%s: no space for dma address conversion\n",
+			  __func__);
+
+		/* in the sync case the dest is an implied source
+		 * (assumes the dest is the first source)
+		 */
+		if (submit->flags & ASYNC_TX_XOR_DROP_DST) {
+			src_cnt--;
+			src_list++;
+			if (src_offs)
+				src_offs++;
+		}
+
+		/* wait for any prerequisite operations */
+		async_tx_quiesce(&submit->depend_tx);
+
+		do_sync_xor_offs(dest, offset, src_list, src_offs,
+				src_cnt, len, submit);
+
+		return NULL;
+	}
+}
+EXPORT_SYMBOL_GPL(async_xor_offs);
+
+/**
+ * async_xor - attempt to xor a set of blocks with a dma engine.
+ * @dest: destination page
+ * @src_list: array of source pages
+ * @offset: common src/dst offset to start transaction
+ * @src_cnt: number of source pages
+ * @len: length in bytes
+ * @submit: submission / completion modifiers
+ *
+ * honored flags: ASYNC_TX_ACK, ASYNC_TX_XOR_ZERO_DST, ASYNC_TX_XOR_DROP_DST
+ *
+ * xor_blocks always uses the dest as a source so the
+ * ASYNC_TX_XOR_ZERO_DST flag must be set to not include dest data in
+ * the calculation.  The assumption with dma eninges is that they only
+ * use the destination buffer as a source when it is explicity specified
+ * in the source list.
+ *
+ * src_list note: if the dest is also a source it must be at index zero.
+ * The contents of this array will be overwritten if a scribble region
+ * is not specified.
+ */
+struct dma_async_tx_descriptor *
+async_xor(struct page *dest, struct page **src_list, unsigned int offset,
+	  int src_cnt, size_t len, struct async_submit_ctl *submit)
+{
+	return async_xor_offs(dest, offset, src_list, NULL,
+			src_cnt, len, submit);
+}
+EXPORT_SYMBOL_GPL(async_xor);
+
+static int page_is_zero(struct page *p, unsigned int offset, size_t len)
+{
+	return !memchr_inv(page_address(p) + offset, 0, len);
+}
+
+static inline struct dma_chan *
+xor_val_chan(struct async_submit_ctl *submit, struct page *dest,
+		 struct page **src_list, int src_cnt, size_t len)
+{
+	#ifdef CONFIG_ASYNC_TX_DISABLE_XOR_VAL_DMA
+	return NULL;
+	#endif
+	return async_tx_find_channel(submit, DMA_XOR_VAL, &dest, 1, src_list,
+				     src_cnt, len);
+}
+
+/**
+ * async_xor_val_offs - attempt a xor parity check with a dma engine.
+ * @dest: destination page used if the xor is performed synchronously
+ * @offset: des offset in pages to start transaction
+ * @src_list: array of source pages
+ * @src_offs: array of source pages offset, NULL means common src/det offset
+ * @src_cnt: number of source pages
+ * @len: length in bytes
+ * @result: 0 if sum == 0 else non-zero
+ * @submit: submission / completion modifiers
+ *
+ * honored flags: ASYNC_TX_ACK
+ *
+ * src_list note: if the dest is also a source it must be at index zero.
+ * The contents of this array will be overwritten if a scribble region
+ * is not specified.
+ */
+struct dma_async_tx_descriptor *
+async_xor_val_offs(struct page *dest, unsigned int offset,
+		struct page **src_list, unsigned int *src_offs,
+		int src_cnt, size_t len, enum sum_check_flags *result,
+		struct async_submit_ctl *submit)
+{
+	struct dma_chan *chan = xor_val_chan(submit, dest, src_list, src_cnt, len);
+	struct dma_device *device = chan ? chan->device : NULL;
+	struct dma_async_tx_descriptor *tx = NULL;
+	struct dmaengine_unmap_data *unmap = NULL;
+
+	BUG_ON(src_cnt <= 1);
+
+	if (device)
+		unmap = dmaengine_get_unmap_data(device->dev, src_cnt, GFP_NOWAIT);
+
+	if (unmap && src_cnt <= device->max_xor &&
+	    dma_xor_aligned_offsets(device, offset, src_offs, src_cnt, len)) {
+		unsigned long dma_prep_flags = 0;
+		int i;
+
+		pr_debug("%s: (async) len: %zu\n", __func__, len);
+
+		if (submit->cb_fn)
+			dma_prep_flags |= DMA_PREP_INTERRUPT;
+		if (submit->flags & ASYNC_TX_FENCE)
+			dma_prep_flags |= DMA_PREP_FENCE;
+
+		for (i = 0; i < src_cnt; i++) {
+			unmap->addr[i] = dma_map_page(device->dev, src_list[i],
+					src_offs ? src_offs[i] : offset,
+					len, DMA_TO_DEVICE);
+			unmap->to_cnt++;
+		}
+		unmap->len = len;
+
+		tx = device->device_prep_dma_xor_val(chan, unmap->addr, src_cnt,
+						     len, result,
+						     dma_prep_flags);
+		if (unlikely(!tx)) {
+			async_tx_quiesce(&submit->depend_tx);
+
+			while (!tx) {
+				dma_async_issue_pending(chan);
+				tx = device->device_prep_dma_xor_val(chan,
+					unmap->addr, src_cnt, len, result,
+					dma_prep_flags);
+			}
+		}
+		dma_set_unmap(tx, unmap);
+		async_tx_submit(chan, tx, submit);
+	} else {
+		enum async_tx_flags flags_orig = submit->flags;
+
+		pr_debug("%s: (sync) len: %zu\n", __func__, len);
+		WARN_ONCE(device && src_cnt <= device->max_xor,
+			  "%s: no space for dma address conversion\n",
+			  __func__);
+
+		submit->flags |= ASYNC_TX_XOR_DROP_DST;
+		submit->flags &= ~ASYNC_TX_ACK;
+
+		tx = async_xor_offs(dest, offset, src_list, src_offs,
+				src_cnt, len, submit);
+
+		async_tx_quiesce(&tx);
+
+		*result = !page_is_zero(dest, offset, len) << SUM_CHECK_P;
+
+		async_tx_sync_epilog(submit);
+		submit->flags = flags_orig;
+	}
+	dmaengine_unmap_put(unmap);
+
+	return tx;
+}
+EXPORT_SYMBOL_GPL(async_xor_val_offs);
+
+/**
+ * async_xor_val - attempt a xor parity check with a dma engine.
+ * @dest: destination page used if the xor is performed synchronously
+ * @src_list: array of source pages
+ * @offset: offset in pages to start transaction
+ * @src_cnt: number of source pages
+ * @len: length in bytes
+ * @result: 0 if sum == 0 else non-zero
+ * @submit: submission / completion modifiers
+ *
+ * honored flags: ASYNC_TX_ACK
+ *
+ * src_list note: if the dest is also a source it must be at index zero.
+ * The contents of this array will be overwritten if a scribble region
+ * is not specified.
+ */
+struct dma_async_tx_descriptor *
+async_xor_val(struct page *dest, struct page **src_list, unsigned int offset,
+	      int src_cnt, size_t len, enum sum_check_flags *result,
+	      struct async_submit_ctl *submit)
+{
+	return async_xor_val_offs(dest, offset, src_list, NULL, src_cnt,
+			len, result, submit);
+}
+EXPORT_SYMBOL_GPL(async_xor_val);
+
+MODULE_AUTHOR("Intel Corporation");
+MODULE_DESCRIPTION("asynchronous xor/xor-zero-sum api");
+MODULE_LICENSE("GPL");
diff --git a/crypto/async_tx/raid6test.c b/crypto/async_tx/raid6test.c
new file mode 100644
index 000000000..66db82e5a
--- /dev/null
+++ b/crypto/async_tx/raid6test.c
@@ -0,0 +1,250 @@
+// SPDX-License-Identifier: GPL-2.0-only
+/*
+ * asynchronous raid6 recovery self test
+ * Copyright (c) 2009, Intel Corporation.
+ *
+ * based on drivers/md/raid6test/test.c:
+ * 	Copyright 2002-2007 H. Peter Anvin
+ */
+#include <linux/async_tx.h>
+#include <linux/gfp.h>
+#include <linux/mm.h>
+#include <linux/random.h>
+#include <linux/module.h>
+
+#undef pr
+#define pr(fmt, args...) pr_info("raid6test: " fmt, ##args)
+
+#define NDISKS 64 /* Including P and Q */
+
+static struct page *dataptrs[NDISKS];
+unsigned int dataoffs[NDISKS];
+static addr_conv_t addr_conv[NDISKS];
+static struct page *data[NDISKS+3];
+static struct page *spare;
+static struct page *recovi;
+static struct page *recovj;
+
+static void callback(void *param)
+{
+	struct completion *cmp = param;
+
+	complete(cmp);
+}
+
+static void makedata(int disks)
+{
+	int i;
+
+	for (i = 0; i < disks; i++) {
+		prandom_bytes(page_address(data[i]), PAGE_SIZE);
+		dataptrs[i] = data[i];
+		dataoffs[i] = 0;
+	}
+}
+
+static char disk_type(int d, int disks)
+{
+	if (d == disks - 2)
+		return 'P';
+	else if (d == disks - 1)
+		return 'Q';
+	else
+		return 'D';
+}
+
+/* Recover two failed blocks. */
+static void raid6_dual_recov(int disks, size_t bytes, int faila, int failb,
+		struct page **ptrs, unsigned int *offs)
+{
+	struct async_submit_ctl submit;
+	struct completion cmp;
+	struct dma_async_tx_descriptor *tx = NULL;
+	enum sum_check_flags result = ~0;
+
+	if (faila > failb)
+		swap(faila, failb);
+
+	if (failb == disks-1) {
+		if (faila == disks-2) {
+			/* P+Q failure.  Just rebuild the syndrome. */
+			init_async_submit(&submit, 0, NULL, NULL, NULL, addr_conv);
+			tx = async_gen_syndrome(ptrs, offs,
+					disks, bytes, &submit);
+		} else {
+			struct page *blocks[NDISKS];
+			struct page *dest;
+			int count = 0;
+			int i;
+
+			BUG_ON(disks > NDISKS);
+
+			/* data+Q failure.  Reconstruct data from P,
+			 * then rebuild syndrome
+			 */
+			for (i = disks; i-- ; ) {
+				if (i == faila || i == failb)
+					continue;
+				blocks[count++] = ptrs[i];
+			}
+			dest = ptrs[faila];
+			init_async_submit(&submit, ASYNC_TX_XOR_ZERO_DST, NULL,
+					  NULL, NULL, addr_conv);
+			tx = async_xor(dest, blocks, 0, count, bytes, &submit);
+
+			init_async_submit(&submit, 0, tx, NULL, NULL, addr_conv);
+			tx = async_gen_syndrome(ptrs, offs,
+					disks, bytes, &submit);
+		}
+	} else {
+		if (failb == disks-2) {
+			/* data+P failure. */
+			init_async_submit(&submit, 0, NULL, NULL, NULL, addr_conv);
+			tx = async_raid6_datap_recov(disks, bytes,
+					faila, ptrs, offs, &submit);
+		} else {
+			/* data+data failure. */
+			init_async_submit(&submit, 0, NULL, NULL, NULL, addr_conv);
+			tx = async_raid6_2data_recov(disks, bytes,
+					faila, failb, ptrs, offs, &submit);
+		}
+	}
+	init_completion(&cmp);
+	init_async_submit(&submit, ASYNC_TX_ACK, tx, callback, &cmp, addr_conv);
+	tx = async_syndrome_val(ptrs, offs,
+			disks, bytes, &result, spare, 0, &submit);
+	async_tx_issue_pending(tx);
+
+	if (wait_for_completion_timeout(&cmp, msecs_to_jiffies(3000)) == 0)
+		pr("%s: timeout! (faila: %d failb: %d disks: %d)\n",
+		   __func__, faila, failb, disks);
+
+	if (result != 0)
+		pr("%s: validation failure! faila: %d failb: %d sum_check_flags: %x\n",
+		   __func__, faila, failb, result);
+}
+
+static int test_disks(int i, int j, int disks)
+{
+	int erra, errb;
+
+	memset(page_address(recovi), 0xf0, PAGE_SIZE);
+	memset(page_address(recovj), 0xba, PAGE_SIZE);
+
+	dataptrs[i] = recovi;
+	dataptrs[j] = recovj;
+
+	raid6_dual_recov(disks, PAGE_SIZE, i, j, dataptrs, dataoffs);
+
+	erra = memcmp(page_address(data[i]), page_address(recovi), PAGE_SIZE);
+	errb = memcmp(page_address(data[j]), page_address(recovj), PAGE_SIZE);
+
+	pr("%s(%d, %d): faila=%3d(%c)  failb=%3d(%c)  %s\n",
+	   __func__, i, j, i, disk_type(i, disks), j, disk_type(j, disks),
+	   (!erra && !errb) ? "OK" : !erra ? "ERRB" : !errb ? "ERRA" : "ERRAB");
+
+	dataptrs[i] = data[i];
+	dataptrs[j] = data[j];
+
+	return erra || errb;
+}
+
+static int test(int disks, int *tests)
+{
+	struct dma_async_tx_descriptor *tx;
+	struct async_submit_ctl submit;
+	struct completion cmp;
+	int err = 0;
+	int i, j;
+
+	recovi = data[disks];
+	recovj = data[disks+1];
+	spare  = data[disks+2];
+
+	makedata(disks);
+
+	/* Nuke syndromes */
+	memset(page_address(data[disks-2]), 0xee, PAGE_SIZE);
+	memset(page_address(data[disks-1]), 0xee, PAGE_SIZE);
+
+	/* Generate assumed good syndrome */
+	init_completion(&cmp);
+	init_async_submit(&submit, ASYNC_TX_ACK, NULL, callback, &cmp, addr_conv);
+	tx = async_gen_syndrome(dataptrs, dataoffs, disks, PAGE_SIZE, &submit);
+	async_tx_issue_pending(tx);
+
+	if (wait_for_completion_timeout(&cmp, msecs_to_jiffies(3000)) == 0) {
+		pr("error: initial gen_syndrome(%d) timed out\n", disks);
+		return 1;
+	}
+
+	pr("testing the %d-disk case...\n", disks);
+	for (i = 0; i < disks-1; i++)
+		for (j = i+1; j < disks; j++) {
+			(*tests)++;
+			err += test_disks(i, j, disks);
+		}
+
+	return err;
+}
+
+
+static int raid6_test(void)
+{
+	int err = 0;
+	int tests = 0;
+	int i;
+
+	for (i = 0; i < NDISKS+3; i++) {
+		data[i] = alloc_page(GFP_KERNEL);
+		if (!data[i]) {
+			while (i--)
+				put_page(data[i]);
+			return -ENOMEM;
+		}
+	}
+
+	/* the 4-disk and 5-disk cases are special for the recovery code */
+	if (NDISKS > 4)
+		err += test(4, &tests);
+	if (NDISKS > 5)
+		err += test(5, &tests);
+	/* the 11 and 12 disk cases are special for ioatdma (p-disabled
+	 * q-continuation without extended descriptor)
+	 */
+	if (NDISKS > 12) {
+		err += test(11, &tests);
+		err += test(12, &tests);
+	}
+
+	/* the 24 disk case is special for ioatdma as it is the boudary point
+	 * at which it needs to switch from 8-source ops to 16-source
+	 * ops for continuation (assumes DMA_HAS_PQ_CONTINUE is not set)
+	 */
+	if (NDISKS > 24)
+		err += test(24, &tests);
+
+	err += test(NDISKS, &tests);
+
+	pr("\n");
+	pr("complete (%d tests, %d failure%s)\n",
+	   tests, err, err == 1 ? "" : "s");
+
+	for (i = 0; i < NDISKS+3; i++)
+		put_page(data[i]);
+
+	return 0;
+}
+
+static void raid6_test_exit(void)
+{
+}
+
+/* when compiled-in wait for drivers to load first (assumes dma drivers
+ * are also compliled-in)
+ */
+late_initcall(raid6_test);
+module_exit(raid6_test_exit);
+MODULE_AUTHOR("Dan Williams <dan.j.williams@intel.com>");
+MODULE_DESCRIPTION("asynchronous RAID-6 recovery self tests");
+MODULE_LICENSE("GPL");