Merge git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/herbert/crypto-2.6: (46 commits)
  hwrng: via_rng - Fix memory scribbling on some CPUs
  crypto: padlock - Move padlock.h into include/crypto
  hwrng: via_rng - Fix asm constraints
  crypto: n2 - use __devexit not __exit in n2_unregister_algs
  crypto: mark crypto workqueues CPU_INTENSIVE
  crypto: mv_cesa - dont return PTR_ERR() of wrong pointer
  crypto: ripemd - Set module author and update email address
  crypto: omap-sham - backlog handling fix
  crypto: gf128mul - Remove experimental tag
  crypto: af_alg - fix af_alg memory_allocated data type
  crypto: aesni-intel - Fixed build with binutils 2.16
  crypto: af_alg - Make sure sk_security is initialized on accept()ed sockets
  net: Add missing lockdep class names for af_alg
  include: Install linux/if_alg.h for user-space crypto API
  crypto: omap-aes - checkpatch --file warning fixes
  crypto: omap-aes - initialize aes module once per request
  crypto: omap-aes - unnecessary code removed
  crypto: omap-aes - error handling implementation improved
  crypto: omap-aes - redundant locking is removed
  crypto: omap-aes - DMA initialization fixes for OMAP off mode
  ...

8 files changed, 361 insertions, 320 deletions

diff --git a/drivers/char/hw_random/via-rng.c b/drivers/char/hw_random/via-rng.c
index 794aacb..d0387a8 100644
--- a/drivers/char/hw_random/via-rng.c
+++ b/drivers/char/hw_random/via-rng.c
@@ -24,6 +24,7 @@
  * warranty of any kind, whether express or implied.
  */
 
+#include <crypto/padlock.h>
 #include <linux/module.h>
 #include <linux/kernel.h>
 #include <linux/hw_random.h>
@@ -34,7 +35,6 @@
 #include <asm/i387.h>
 
 
-#define PFX	KBUILD_MODNAME ": "
 
 
 enum {
@@ -81,8 +81,7 @@ static inline u32 xstore(u32 *addr, u32 edx_in)
 	ts_state = irq_ts_save();
 
 	asm(".byte 0x0F,0xA7,0xC0 /* xstore %%edi (addr=%0) */"
-		:"=m"(*addr), "=a"(eax_out)
-		:"D"(addr), "d"(edx_in));
+		: "=m" (*addr), "=a" (eax_out), "+d" (edx_in), "+D" (addr));
 
 	irq_ts_restore(ts_state);
 	return eax_out;
@@ -90,8 +89,10 @@ static inline u32 xstore(u32 *addr, u32 edx_in)
 
 static int via_rng_data_present(struct hwrng *rng, int wait)
 {
+	char buf[16 + PADLOCK_ALIGNMENT - STACK_ALIGN] __attribute__
+		((aligned(STACK_ALIGN)));
+	u32 *via_rng_datum = (u32 *)PTR_ALIGN(&buf[0], PADLOCK_ALIGNMENT);
 	u32 bytes_out;
-	u32 *via_rng_datum = (u32 *)(&rng->priv);
 	int i;
 
 	/* We choose the recommended 1-byte-per-instruction RNG rate,
@@ -115,6 +116,7 @@ static int via_rng_data_present(struct hwrng *rng, int wait)
 			break;
 		udelay(10);
 	}
+	rng->priv = *via_rng_datum;
 	return bytes_out ? 1 : 0;
 }
 
diff --git a/drivers/crypto/mv_cesa.c b/drivers/crypto/mv_cesa.c
index 7d279e5..c99305a 100644
--- a/drivers/crypto/mv_cesa.c
+++ b/drivers/crypto/mv_cesa.c
@@ -857,7 +857,7 @@ static int mv_cra_hash_init(struct crypto_tfm *tfm, const char *base_hash_name,
 			printk(KERN_WARNING MV_CESA
 			       "Base driver '%s' could not be loaded!\n",
 			       base_hash_name);
-			err = PTR_ERR(fallback_tfm);
+			err = PTR_ERR(base_hash);
 			goto err_bad_base;
 		}
 	}
diff --git a/drivers/crypto/n2_core.c b/drivers/crypto/n2_core.c
index 76141262..80dc094 100644
--- a/drivers/crypto/n2_core.c
+++ b/drivers/crypto/n2_core.c
@@ -1542,7 +1542,7 @@ out:
 	return err;
 }
 
-static void __exit n2_unregister_algs(void)
+static void __devexit n2_unregister_algs(void)
 {
 	mutex_lock(&spu_lock);
 	if (!--algs_registered)
diff --git a/drivers/crypto/omap-aes.c b/drivers/crypto/omap-aes.c
index 799ca51..add2a1a 100644
--- a/drivers/crypto/omap-aes.c
+++ b/drivers/crypto/omap-aes.c
@@ -74,11 +74,9 @@
 #define FLAGS_CBC		BIT(1)
 #define FLAGS_GIV		BIT(2)
 
-#define FLAGS_NEW_KEY		BIT(4)
-#define FLAGS_NEW_IV		BIT(5)
-#define FLAGS_INIT		BIT(6)
-#define FLAGS_FAST		BIT(7)
-#define FLAGS_BUSY		8
+#define FLAGS_INIT		BIT(4)
+#define FLAGS_FAST		BIT(5)
+#define FLAGS_BUSY		BIT(6)
 
 struct omap_aes_ctx {
 	struct omap_aes_dev *dd;
@@ -98,19 +96,18 @@ struct omap_aes_reqctx {
 struct omap_aes_dev {
 	struct list_head	list;
 	unsigned long		phys_base;
-	void __iomem 		*io_base;
+	void __iomem		*io_base;
 	struct clk		*iclk;
 	struct omap_aes_ctx	*ctx;
 	struct device		*dev;
 	unsigned long		flags;
+	int			err;
 
-	u32			*iv;
-	u32			ctrl;
+	spinlock_t		lock;
+	struct crypto_queue	queue;
 
-	spinlock_t			lock;
-	struct crypto_queue		queue;
-
-	struct tasklet_struct		task;
+	struct tasklet_struct	done_task;
+	struct tasklet_struct	queue_task;
 
 	struct ablkcipher_request	*req;
 	size_t				total;
@@ -179,9 +176,13 @@ static int omap_aes_wait(struct omap_aes_dev *dd, u32 offset, u32 bit)
 
 static int omap_aes_hw_init(struct omap_aes_dev *dd)
 {
-	int err = 0;
-
+	/*
+	 * clocks are enabled when request starts and disabled when finished.
+	 * It may be long delays between requests.
+	 * Device might go to off mode to save power.
+	 */
 	clk_enable(dd->iclk);
+
 	if (!(dd->flags & FLAGS_INIT)) {
 		/* is it necessary to reset before every operation? */
 		omap_aes_write_mask(dd, AES_REG_MASK, AES_REG_MASK_SOFTRESET,
@@ -193,39 +194,26 @@ static int omap_aes_hw_init(struct omap_aes_dev *dd)
 		__asm__ __volatile__("nop");
 		__asm__ __volatile__("nop");
 
-		err = omap_aes_wait(dd, AES_REG_SYSSTATUS,
-				AES_REG_SYSSTATUS_RESETDONE);
-		if (!err)
-			dd->flags |= FLAGS_INIT;
-	}
+		if (omap_aes_wait(dd, AES_REG_SYSSTATUS,
+				AES_REG_SYSSTATUS_RESETDONE))
+			return -ETIMEDOUT;
 
-	return err;
-}
+		dd->flags |= FLAGS_INIT;
+		dd->err = 0;
+	}
 
-static void omap_aes_hw_cleanup(struct omap_aes_dev *dd)
-{
-	clk_disable(dd->iclk);
+	return 0;
 }
 
-static void omap_aes_write_ctrl(struct omap_aes_dev *dd)
+static int omap_aes_write_ctrl(struct omap_aes_dev *dd)
 {
 	unsigned int key32;
-	int i;
+	int i, err;
 	u32 val, mask;
 
-	val = FLD_VAL(((dd->ctx->keylen >> 3) - 1), 4, 3);
-	if (dd->flags & FLAGS_CBC)
-		val |= AES_REG_CTRL_CBC;
-	if (dd->flags & FLAGS_ENCRYPT)
-		val |= AES_REG_CTRL_DIRECTION;
-
-	if (dd->ctrl == val && !(dd->flags & FLAGS_NEW_IV) &&
-		   !(dd->ctx->flags & FLAGS_NEW_KEY))
-		goto out;
-
-	/* only need to write control registers for new settings */
-
-	dd->ctrl = val;
+	err = omap_aes_hw_init(dd);
+	if (err)
+		return err;
 
 	val = 0;
 	if (dd->dma_lch_out >= 0)
@@ -237,30 +225,43 @@ static void omap_aes_write_ctrl(struct omap_aes_dev *dd)
 
 	omap_aes_write_mask(dd, AES_REG_MASK, val, mask);
 
-	pr_debug("Set key\n");
 	key32 = dd->ctx->keylen / sizeof(u32);
-	/* set a key */
+
+	/* it seems a key should always be set even if it has not changed */
 	for (i = 0; i < key32; i++) {
 		omap_aes_write(dd, AES_REG_KEY(i),
 			__le32_to_cpu(dd->ctx->key[i]));
 	}
-	dd->ctx->flags &= ~FLAGS_NEW_KEY;
 
-	if (dd->flags & FLAGS_NEW_IV) {
-		pr_debug("Set IV\n");
-		omap_aes_write_n(dd, AES_REG_IV(0), dd->iv, 4);
-		dd->flags &= ~FLAGS_NEW_IV;
-	}
+	if ((dd->flags & FLAGS_CBC) && dd->req->info)
+		omap_aes_write_n(dd, AES_REG_IV(0), dd->req->info, 4);
+
+	val = FLD_VAL(((dd->ctx->keylen >> 3) - 1), 4, 3);
+	if (dd->flags & FLAGS_CBC)
+		val |= AES_REG_CTRL_CBC;
+	if (dd->flags & FLAGS_ENCRYPT)
+		val |= AES_REG_CTRL_DIRECTION;
 
 	mask = AES_REG_CTRL_CBC | AES_REG_CTRL_DIRECTION |
 			AES_REG_CTRL_KEY_SIZE;
 
-	omap_aes_write_mask(dd, AES_REG_CTRL, dd->ctrl, mask);
+	omap_aes_write_mask(dd, AES_REG_CTRL, val, mask);
 
-out:
-	/* start DMA or disable idle mode */
-	omap_aes_write_mask(dd, AES_REG_MASK, AES_REG_MASK_START,
-			    AES_REG_MASK_START);
+	/* IN */
+	omap_set_dma_dest_params(dd->dma_lch_in, 0, OMAP_DMA_AMODE_CONSTANT,
+				 dd->phys_base + AES_REG_DATA, 0, 4);
+
+	omap_set_dma_dest_burst_mode(dd->dma_lch_in, OMAP_DMA_DATA_BURST_4);
+	omap_set_dma_src_burst_mode(dd->dma_lch_in, OMAP_DMA_DATA_BURST_4);
+
+	/* OUT */
+	omap_set_dma_src_params(dd->dma_lch_out, 0, OMAP_DMA_AMODE_CONSTANT,
+				dd->phys_base + AES_REG_DATA, 0, 4);
+
+	omap_set_dma_src_burst_mode(dd->dma_lch_out, OMAP_DMA_DATA_BURST_4);
+	omap_set_dma_dest_burst_mode(dd->dma_lch_out, OMAP_DMA_DATA_BURST_4);
+
+	return 0;
 }
 
 static struct omap_aes_dev *omap_aes_find_dev(struct omap_aes_ctx *ctx)
@@ -288,8 +289,16 @@ static void omap_aes_dma_callback(int lch, u16 ch_status, void *data)
 {
 	struct omap_aes_dev *dd = data;
 
-	if (lch == dd->dma_lch_out)
-		tasklet_schedule(&dd->task);
+	if (ch_status != OMAP_DMA_BLOCK_IRQ) {
+		pr_err("omap-aes DMA error status: 0x%hx\n", ch_status);
+		dd->err = -EIO;
+		dd->flags &= ~FLAGS_INIT; /* request to re-initialize */
+	} else if (lch == dd->dma_lch_in) {
+		return;
+	}
+
+	/* dma_lch_out - completed */
+	tasklet_schedule(&dd->done_task);
 }
 
 static int omap_aes_dma_init(struct omap_aes_dev *dd)
@@ -339,18 +348,6 @@ static int omap_aes_dma_init(struct omap_aes_dev *dd)
 		goto err_dma_out;
 	}
 
-	omap_set_dma_dest_params(dd->dma_lch_in, 0, OMAP_DMA_AMODE_CONSTANT,
-				 dd->phys_base + AES_REG_DATA, 0, 4);
-
-	omap_set_dma_dest_burst_mode(dd->dma_lch_in, OMAP_DMA_DATA_BURST_4);
-	omap_set_dma_src_burst_mode(dd->dma_lch_in, OMAP_DMA_DATA_BURST_4);
-
-	omap_set_dma_src_params(dd->dma_lch_out, 0, OMAP_DMA_AMODE_CONSTANT,
-				dd->phys_base + AES_REG_DATA, 0, 4);
-
-	omap_set_dma_src_burst_mode(dd->dma_lch_out, OMAP_DMA_DATA_BURST_4);
-	omap_set_dma_dest_burst_mode(dd->dma_lch_out, OMAP_DMA_DATA_BURST_4);
-
 	return 0;
 
 err_dma_out:
@@ -406,6 +403,11 @@ static int sg_copy(struct scatterlist **sg, size_t *offset, void *buf,
 		if (!count)
 			return off;
 
+		/*
+		 * buflen and total are AES_BLOCK_SIZE size aligned,
+		 * so count should be also aligned
+		 */
+
 		sg_copy_buf(buf + off, *sg, *offset, count, out);
 
 		off += count;
@@ -461,7 +463,9 @@ static int omap_aes_crypt_dma(struct crypto_tfm *tfm, dma_addr_t dma_addr_in,
 	omap_start_dma(dd->dma_lch_in);
 	omap_start_dma(dd->dma_lch_out);
 
-	omap_aes_write_ctrl(dd);
+	/* start DMA or disable idle mode */
+	omap_aes_write_mask(dd, AES_REG_MASK, AES_REG_MASK_START,
+			    AES_REG_MASK_START);
 
 	return 0;
 }
@@ -488,8 +492,10 @@ static int omap_aes_crypt_dma_start(struct omap_aes_dev *dd)
 		count = min(dd->total, sg_dma_len(dd->in_sg));
 		count = min(count, sg_dma_len(dd->out_sg));
 
-		if (count != dd->total)
+		if (count != dd->total) {
+			pr_err("request length != buffer length\n");
 			return -EINVAL;
+		}
 
 		pr_debug("fast\n");
 
@@ -525,23 +531,25 @@ static int omap_aes_crypt_dma_start(struct omap_aes_dev *dd)
 
 	dd->total -= count;
 
-	err = omap_aes_hw_init(dd);
-
 	err = omap_aes_crypt_dma(tfm, addr_in, addr_out, count);
+	if (err) {
+		dma_unmap_sg(dd->dev, dd->in_sg, 1, DMA_TO_DEVICE);
+		dma_unmap_sg(dd->dev, dd->out_sg, 1, DMA_TO_DEVICE);
+	}
 
 	return err;
 }
 
 static void omap_aes_finish_req(struct omap_aes_dev *dd, int err)
 {
-	struct omap_aes_ctx *ctx;
+	struct ablkcipher_request *req = dd->req;
 
 	pr_debug("err: %d\n", err);
 
-	ctx = crypto_ablkcipher_ctx(crypto_ablkcipher_reqtfm(dd->req));
+	clk_disable(dd->iclk);
+	dd->flags &= ~FLAGS_BUSY;
 
-	if (!dd->total)
-		dd->req->base.complete(&dd->req->base, err);
+	req->base.complete(&req->base, err);
 }
 
 static int omap_aes_crypt_dma_stop(struct omap_aes_dev *dd)
@@ -553,8 +561,6 @@ static int omap_aes_crypt_dma_stop(struct omap_aes_dev *dd)
 
 	omap_aes_write_mask(dd, AES_REG_MASK, 0, AES_REG_MASK_START);
 
-	omap_aes_hw_cleanup(dd);
-
 	omap_stop_dma(dd->dma_lch_in);
 	omap_stop_dma(dd->dma_lch_out);
 
@@ -574,40 +580,39 @@ static int omap_aes_crypt_dma_stop(struct omap_aes_dev *dd)
 		}
 	}
 
-	if (err || !dd->total)
-		omap_aes_finish_req(dd, err);
-
 	return err;
 }
 
-static int omap_aes_handle_req(struct omap_aes_dev *dd)
+static int omap_aes_handle_queue(struct omap_aes_dev *dd,
+			       struct ablkcipher_request *req)
 {
 	struct crypto_async_request *async_req, *backlog;
 	struct omap_aes_ctx *ctx;
 	struct omap_aes_reqctx *rctx;
-	struct ablkcipher_request *req;
 	unsigned long flags;
-
-	if (dd->total)
-		goto start;
+	int err, ret = 0;
 
 	spin_lock_irqsave(&dd->lock, flags);
+	if (req)
+		ret = ablkcipher_enqueue_request(&dd->queue, req);
+	if (dd->flags & FLAGS_BUSY) {
+		spin_unlock_irqrestore(&dd->lock, flags);
+		return ret;
+	}
 	backlog = crypto_get_backlog(&dd->queue);
 	async_req = crypto_dequeue_request(&dd->queue);
-	if (!async_req)
-		clear_bit(FLAGS_BUSY, &dd->flags);
+	if (async_req)
+		dd->flags |= FLAGS_BUSY;
 	spin_unlock_irqrestore(&dd->lock, flags);
 
 	if (!async_req)
-		return 0;
+		return ret;
 
 	if (backlog)
 		backlog->complete(backlog, -EINPROGRESS);
 
 	req = ablkcipher_request_cast(async_req);
 
-	pr_debug("get new req\n");
-
 	/* assign new request to device */
 	dd->req = req;
 	dd->total = req->nbytes;
@@ -621,27 +626,22 @@ static int omap_aes_handle_req(struct omap_aes_dev *dd)
 	rctx->mode &= FLAGS_MODE_MASK;
 	dd->flags = (dd->flags & ~FLAGS_MODE_MASK) | rctx->mode;
 
-	dd->iv = req->info;
-	if ((dd->flags & FLAGS_CBC) && dd->iv)
-		dd->flags |= FLAGS_NEW_IV;
-	else
-		dd->flags &= ~FLAGS_NEW_IV;
-
+	dd->ctx = ctx;
 	ctx->dd = dd;
-	if (dd->ctx != ctx) {
-		/* assign new context to device */
-		dd->ctx = ctx;
-		ctx->flags |= FLAGS_NEW_KEY;
-	}
 
-	if (!IS_ALIGNED(req->nbytes, AES_BLOCK_SIZE))
-		pr_err("request size is not exact amount of AES blocks\n");
+	err = omap_aes_write_ctrl(dd);
+	if (!err)
+		err = omap_aes_crypt_dma_start(dd);
+	if (err) {
+		/* aes_task will not finish it, so do it here */
+		omap_aes_finish_req(dd, err);
+		tasklet_schedule(&dd->queue_task);
+	}
 
-start:
-	return omap_aes_crypt_dma_start(dd);
+	return ret; /* return ret, which is enqueue return value */
 }
 
-static void omap_aes_task(unsigned long data)
+static void omap_aes_done_task(unsigned long data)
 {
 	struct omap_aes_dev *dd = (struct omap_aes_dev *)data;
 	int err;
@@ -650,40 +650,50 @@ static void omap_aes_task(unsigned long data)
 
 	err = omap_aes_crypt_dma_stop(dd);
 
-	err = omap_aes_handle_req(dd);
+	err = dd->err ? : err;
+
+	if (dd->total && !err) {
+		err = omap_aes_crypt_dma_start(dd);
+		if (!err)
+			return; /* DMA started. Not fininishing. */
+	}
+
+	omap_aes_finish_req(dd, err);
+	omap_aes_handle_queue(dd, NULL);
 
 	pr_debug("exit\n");
 }
 
+static void omap_aes_queue_task(unsigned long data)
+{
+	struct omap_aes_dev *dd = (struct omap_aes_dev *)data;
+
+	omap_aes_handle_queue(dd, NULL);
+}
+
 static int omap_aes_crypt(struct ablkcipher_request *req, unsigned long mode)
 {
 	struct omap_aes_ctx *ctx = crypto_ablkcipher_ctx(
 			crypto_ablkcipher_reqtfm(req));
 	struct omap_aes_reqctx *rctx = ablkcipher_request_ctx(req);
 	struct omap_aes_dev *dd;
-	unsigned long flags;
-	int err;
 
 	pr_debug("nbytes: %d, enc: %d, cbc: %d\n", req->nbytes,
 		  !!(mode & FLAGS_ENCRYPT),
 		  !!(mode & FLAGS_CBC));
 
+	if (!IS_ALIGNED(req->nbytes, AES_BLOCK_SIZE)) {
+		pr_err("request size is not exact amount of AES blocks\n");
+		return -EINVAL;
+	}
+
 	dd = omap_aes_find_dev(ctx);
 	if (!dd)
 		return -ENODEV;
 
 	rctx->mode = mode;
 
-	spin_lock_irqsave(&dd->lock, flags);
-	err = ablkcipher_enqueue_request(&dd->queue, req);
-	spin_unlock_irqrestore(&dd->lock, flags);
-
-	if (!test_and_set_bit(FLAGS_BUSY, &dd->flags))
-		omap_aes_handle_req(dd);
-
-	pr_debug("exit\n");
-
-	return err;
+	return omap_aes_handle_queue(dd, req);
 }
 
 /* ********************** ALG API ************************************ */
@@ -701,7 +711,6 @@ static int omap_aes_setkey(struct crypto_ablkcipher *tfm, const u8 *key,
 
 	memcpy(ctx->key, key, keylen);
 	ctx->keylen = keylen;
-	ctx->flags |= FLAGS_NEW_KEY;
 
 	return 0;
 }
@@ -750,7 +759,7 @@ static struct crypto_alg algs[] = {
 	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
 	.cra_blocksize		= AES_BLOCK_SIZE,
 	.cra_ctxsize		= sizeof(struct omap_aes_ctx),
-	.cra_alignmask	 	= 0,
+	.cra_alignmask		= 0,
 	.cra_type		= &crypto_ablkcipher_type,
 	.cra_module		= THIS_MODULE,
 	.cra_init		= omap_aes_cra_init,
@@ -770,7 +779,7 @@ static struct crypto_alg algs[] = {
 	.cra_flags		= CRYPTO_ALG_TYPE_ABLKCIPHER | CRYPTO_ALG_ASYNC,
 	.cra_blocksize		= AES_BLOCK_SIZE,
 	.cra_ctxsize		= sizeof(struct omap_aes_ctx),
-	.cra_alignmask	 	= 0,
+	.cra_alignmask		= 0,
 	.cra_type		= &crypto_ablkcipher_type,
 	.cra_module		= THIS_MODULE,
 	.cra_init		= omap_aes_cra_init,
@@ -849,7 +858,8 @@ static int omap_aes_probe(struct platform_device *pdev)
 		 (reg & AES_REG_REV_MAJOR) >> 4, reg & AES_REG_REV_MINOR);
 	clk_disable(dd->iclk);
 
-	tasklet_init(&dd->task, omap_aes_task, (unsigned long)dd);
+	tasklet_init(&dd->done_task, omap_aes_done_task, (unsigned long)dd);
+	tasklet_init(&dd->queue_task, omap_aes_queue_task, (unsigned long)dd);
 
 	err = omap_aes_dma_init(dd);
 	if (err)
@@ -876,7 +886,8 @@ err_algs:
 		crypto_unregister_alg(&algs[j]);
 	omap_aes_dma_cleanup(dd);
 err_dma:
-	tasklet_kill(&dd->task);
+	tasklet_kill(&dd->done_task);
+	tasklet_kill(&dd->queue_task);
 	iounmap(dd->io_base);
 err_io:
 	clk_put(dd->iclk);
@@ -903,7 +914,8 @@ static int omap_aes_remove(struct platform_device *pdev)
 	for (i = 0; i < ARRAY_SIZE(algs); i++)
 		crypto_unregister_alg(&algs[i]);
 
-	tasklet_kill(&dd->task);
+	tasklet_kill(&dd->done_task);
+	tasklet_kill(&dd->queue_task);
 	omap_aes_dma_cleanup(dd);
 	iounmap(dd->io_base);
 	clk_put(dd->iclk);
diff --git a/drivers/crypto/omap-sham.c b/drivers/crypto/omap-sham.c
index a081c7c..2e71123 100644
--- a/drivers/crypto/omap-sham.c
+++ b/drivers/crypto/omap-sham.c
@@ -72,10 +72,9 @@
 
 #define DEFAULT_TIMEOUT_INTERVAL	HZ
 
-#define FLAGS_FIRST		0x0001
 #define FLAGS_FINUP		0x0002
 #define FLAGS_FINAL		0x0004
-#define FLAGS_FAST		0x0008
+#define FLAGS_SG		0x0008
 #define FLAGS_SHA1		0x0010
 #define FLAGS_DMA_ACTIVE	0x0020
 #define FLAGS_OUTPUT_READY	0x0040
@@ -83,13 +82,17 @@
 #define FLAGS_INIT		0x0100
 #define FLAGS_CPU		0x0200
 #define FLAGS_HMAC		0x0400
-
-/* 3rd byte */
-#define FLAGS_BUSY		16
+#define FLAGS_ERROR		0x0800
+#define FLAGS_BUSY		0x1000
 
 #define OP_UPDATE	1
 #define OP_FINAL	2
 
+#define OMAP_ALIGN_MASK		(sizeof(u32)-1)
+#define OMAP_ALIGNED		__attribute__((aligned(sizeof(u32))))
+
+#define BUFLEN		PAGE_SIZE
+
 struct omap_sham_dev;
 
 struct omap_sham_reqctx {
@@ -97,8 +100,8 @@ struct omap_sham_reqctx {
 	unsigned long		flags;
 	unsigned long		op;
 
+	u8			digest[SHA1_DIGEST_SIZE] OMAP_ALIGNED;
 	size_t			digcnt;
-	u8			*buffer;
 	size_t			bufcnt;
 	size_t			buflen;
 	dma_addr_t		dma_addr;
@@ -107,6 +110,8 @@ struct omap_sham_reqctx {
 	struct scatterlist	*sg;
 	unsigned int		offset;	/* offset in current sg */
 	unsigned int		total;	/* total request */
+
+	u8			buffer[0] OMAP_ALIGNED;
 };
 
 struct omap_sham_hmac_ctx {
@@ -136,6 +141,7 @@ struct omap_sham_dev {
 	int			irq;
 	struct clk		*iclk;
 	spinlock_t		lock;
+	int			err;
 	int			dma;
 	int			dma_lch;
 	struct tasklet_struct	done_task;
@@ -194,53 +200,68 @@ static inline int omap_sham_wait(struct omap_sham_dev *dd, u32 offset, u32 bit)
 static void omap_sham_copy_hash(struct ahash_request *req, int out)
 {
 	struct omap_sham_reqctx *ctx = ahash_request_ctx(req);
+	u32 *hash = (u32 *)ctx->digest;
+	int i;
+
+	/* MD5 is almost unused. So copy sha1 size to reduce code */
+	for (i = 0; i < SHA1_DIGEST_SIZE / sizeof(u32); i++) {
+		if (out)
+			hash[i] = omap_sham_read(ctx->dd,
+						SHA_REG_DIGEST(i));
+		else
+			omap_sham_write(ctx->dd,
+					SHA_REG_DIGEST(i), hash[i]);
+	}
+}
+
+static void omap_sham_copy_ready_hash(struct ahash_request *req)
+{
+	struct omap_sham_reqctx *ctx = ahash_request_ctx(req);
+	u32 *in = (u32 *)ctx->digest;
 	u32 *hash = (u32 *)req->result;
 	int i;
 
+	if (!hash)
+		return;
+
 	if (likely(ctx->flags & FLAGS_SHA1)) {
 		/* SHA1 results are in big endian */
 		for (i = 0; i < SHA1_DIGEST_SIZE / sizeof(u32); i++)
-			if (out)
-				hash[i] = be32_to_cpu(omap_sham_read(ctx->dd,
-							SHA_REG_DIGEST(i)));
-			else
-				omap_sham_write(ctx->dd, SHA_REG_DIGEST(i),
-							cpu_to_be32(hash[i]));
+			hash[i] = be32_to_cpu(in[i]);
 	} else {
 		/* MD5 results are in little endian */
 		for (i = 0; i < MD5_DIGEST_SIZE / sizeof(u32); i++)
-			if (out)
-				hash[i] = le32_to_cpu(omap_sham_read(ctx->dd,
-							SHA_REG_DIGEST(i)));
-			else
-				omap_sham_write(ctx->dd, SHA_REG_DIGEST(i),
-							cpu_to_le32(hash[i]));
+			hash[i] = le32_to_cpu(in[i]);
 	}
 }
 
-static int omap_sham_write_ctrl(struct omap_sham_dev *dd, size_t length,
-				 int final, int dma)
+static int omap_sham_hw_init(struct omap_sham_dev *dd)
 {
-	struct omap_sham_reqctx *ctx = ahash_request_ctx(dd->req);
-	u32 val = length << 5, mask;
+	clk_enable(dd->iclk);
 
-	if (unlikely(!ctx->digcnt)) {
+	if (!(dd->flags & FLAGS_INIT)) {
+		omap_sham_write_mask(dd, SHA_REG_MASK,
+			SHA_REG_MASK_SOFTRESET, SHA_REG_MASK_SOFTRESET);
 
-		clk_enable(dd->iclk);
+		if (omap_sham_wait(dd, SHA_REG_SYSSTATUS,
+					SHA_REG_SYSSTATUS_RESETDONE))
+			return -ETIMEDOUT;
 
-		if (!(dd->flags & FLAGS_INIT)) {
-			omap_sham_write_mask(dd, SHA_REG_MASK,
-				SHA_REG_MASK_SOFTRESET, SHA_REG_MASK_SOFTRESET);
+		dd->flags |= FLAGS_INIT;
+		dd->err = 0;
+	}
 
-			if (omap_sham_wait(dd, SHA_REG_SYSSTATUS,
-						SHA_REG_SYSSTATUS_RESETDONE))
-				return -ETIMEDOUT;
+	return 0;
+}
 
-			dd->flags |= FLAGS_INIT;
-		}
-	} else {
+static void omap_sham_write_ctrl(struct omap_sham_dev *dd, size_t length,
+				 int final, int dma)
+{
+	struct omap_sham_reqctx *ctx = ahash_request_ctx(dd->req);
+	u32 val = length << 5, mask;
+
+	if (likely(ctx->digcnt))
 		omap_sham_write(dd, SHA_REG_DIGCNT, ctx->digcnt);
-	}
 
 	omap_sham_write_mask(dd, SHA_REG_MASK,
 		SHA_REG_MASK_IT_EN | (dma ? SHA_REG_MASK_DMA_EN : 0),
@@ -260,29 +281,26 @@ static int omap_sham_write_ctrl(struct omap_sham_dev *dd, size_t length,
 			SHA_REG_CTRL_ALGO | SHA_REG_CTRL_LENGTH;
 
 	omap_sham_write_mask(dd, SHA_REG_CTRL, val, mask);
-
-	return 0;
 }
 
 static int omap_sham_xmit_cpu(struct omap_sham_dev *dd, const u8 *buf,
 			      size_t length, int final)
 {
 	struct omap_sham_reqctx *ctx = ahash_request_ctx(dd->req);
-	int err, count, len32;
+	int count, len32;
 	const u32 *buffer = (const u32 *)buf;
 
 	dev_dbg(dd->dev, "xmit_cpu: digcnt: %d, length: %d, final: %d\n",
 						ctx->digcnt, length, final);
 
-	err = omap_sham_write_ctrl(dd, length, final, 0);
-	if (err)
-		return err;
+	omap_sham_write_ctrl(dd, length, final, 0);
+
+	/* should be non-zero before next lines to disable clocks later */
+	ctx->digcnt += length;
 
 	if (omap_sham_wait(dd, SHA_REG_CTRL, SHA_REG_CTRL_INPUT_READY))
 		return -ETIMEDOUT;
 
-	ctx->digcnt += length;
-
 	if (final)
 		ctx->flags |= FLAGS_FINAL; /* catch last interrupt */
 
@@ -298,16 +316,11 @@ static int omap_sham_xmit_dma(struct omap_sham_dev *dd, dma_addr_t dma_addr,
 			      size_t length, int final)
 {
 	struct omap_sham_reqctx *ctx = ahash_request_ctx(dd->req);
-	int err, len32;
+	int len32;
 
 	dev_dbg(dd->dev, "xmit_dma: digcnt: %d, length: %d, final: %d\n",
 						ctx->digcnt, length, final);
 
-	/* flush cache entries related to our page */
-	if (dma_addr == ctx->dma_addr)
-		dma_sync_single_for_device(dd->dev, dma_addr, length,
-					   DMA_TO_DEVICE);
-
 	len32 = DIV_ROUND_UP(length, sizeof(u32));
 
 	omap_set_dma_transfer_params(dd->dma_lch, OMAP_DMA_DATA_TYPE_S32, len32,
@@ -317,9 +330,7 @@ static int omap_sham_xmit_dma(struct omap_sham_dev *dd, dma_addr_t dma_addr,
 	omap_set_dma_src_params(dd->dma_lch, 0, OMAP_DMA_AMODE_POST_INC,
 				dma_addr, 0, 0);
 
-	err = omap_sham_write_ctrl(dd, length, final, 1);
-	if (err)
-		return err;
+	omap_sham_write_ctrl(dd, length, final, 1);
 
 	ctx->digcnt += length;
 
@@ -371,15 +382,29 @@ static size_t omap_sham_append_sg(struct omap_sham_reqctx *ctx)
 	return 0;
 }
 
+static int omap_sham_xmit_dma_map(struct omap_sham_dev *dd,
+					struct omap_sham_reqctx *ctx,
+					size_t length, int final)
+{
+	ctx->dma_addr = dma_map_single(dd->dev, ctx->buffer, ctx->buflen,
+				       DMA_TO_DEVICE);
+	if (dma_mapping_error(dd->dev, ctx->dma_addr)) {
+		dev_err(dd->dev, "dma %u bytes error\n", ctx->buflen);
+		return -EINVAL;
+	}
+
+	ctx->flags &= ~FLAGS_SG;
+
+	/* next call does not fail... so no unmap in the case of error */
+	return omap_sham_xmit_dma(dd, ctx->dma_addr, length, final);
+}
+
 static int omap_sham_update_dma_slow(struct omap_sham_dev *dd)
 {
 	struct omap_sham_reqctx *ctx = ahash_request_ctx(dd->req);
 	unsigned int final;
 	size_t count;
 
-	if (!ctx->total)
-		return 0;
-
 	omap_sham_append_sg(ctx);
 
 	final = (ctx->flags & FLAGS_FINUP) && !ctx->total;
@@ -390,30 +415,68 @@ static int omap_sham_update_dma_slow(struct omap_sham_dev *dd)
 	if (final || (ctx->bufcnt == ctx->buflen && ctx->total)) {
 		count = ctx->bufcnt;
 		ctx->bufcnt = 0;
-		return omap_sham_xmit_dma(dd, ctx->dma_addr, count, final);
+		return omap_sham_xmit_dma_map(dd, ctx, count, final);
 	}
 
 	return 0;
 }
 
-static int omap_sham_update_dma_fast(struct omap_sham_dev *dd)
+/* Start address alignment */
+#define SG_AA(sg)	(IS_ALIGNED(sg->offset, sizeof(u32)))
+/* SHA1 block size alignment */
+#define SG_SA(sg)	(IS_ALIGNED(sg->length, SHA1_MD5_BLOCK_SIZE))
+
+static int omap_sham_update_dma_start(struct omap_sham_dev *dd)
 {
 	struct omap_sham_reqctx *ctx = ahash_request_ctx(dd->req);
-	unsigned int length;
+	unsigned int length, final, tail;
+	struct scatterlist *sg;
 
-	ctx->flags |= FLAGS_FAST;
+	if (!ctx->total)
+		return 0;
+
+	if (ctx->bufcnt || ctx->offset)
+		return omap_sham_update_dma_slow(dd);
+
+	dev_dbg(dd->dev, "fast: digcnt: %d, bufcnt: %u, total: %u\n",
+			ctx->digcnt, ctx->bufcnt, ctx->total);
+
+	sg = ctx->sg;
 
-	length = min(ctx->total, sg_dma_len(ctx->sg));
-	ctx->total = length;
+	if (!SG_AA(sg))
+		return omap_sham_update_dma_slow(dd);
+
+	if (!sg_is_last(sg) && !SG_SA(sg))
+		/* size is not SHA1_BLOCK_SIZE aligned */
+		return omap_sham_update_dma_slow(dd);
+
+	length = min(ctx->total, sg->length);
+
+	if (sg_is_last(sg)) {
+		if (!(ctx->flags & FLAGS_FINUP)) {
+			/* not last sg must be SHA1_MD5_BLOCK_SIZE aligned */
+			tail = length & (SHA1_MD5_BLOCK_SIZE - 1);
+			/* without finup() we need one block to close hash */
+			if (!tail)
+				tail = SHA1_MD5_BLOCK_SIZE;
+			length -= tail;
+		}
+	}
 
 	if (!dma_map_sg(dd->dev, ctx->sg, 1, DMA_TO_DEVICE)) {
 		dev_err(dd->dev, "dma_map_sg  error\n");
 		return -EINVAL;
 	}
 
+	ctx->flags |= FLAGS_SG;
+
 	ctx->total -= length;
+	ctx->offset = length; /* offset where to start slow */
 
-	return omap_sham_xmit_dma(dd, sg_dma_address(ctx->sg), length, 1);
+	final = (ctx->flags & FLAGS_FINUP) && !ctx->total;
+
+	/* next call does not fail... so no unmap in the case of error */
+	return omap_sham_xmit_dma(dd, sg_dma_address(ctx->sg), length, final);
 }
 
 static int omap_sham_update_cpu(struct omap_sham_dev *dd)
@@ -433,8 +496,17 @@ static int omap_sham_update_dma_stop(struct omap_sham_dev *dd)
 	struct omap_sham_reqctx *ctx = ahash_request_ctx(dd->req);
 
 	omap_stop_dma(dd->dma_lch);
-	if (ctx->flags & FLAGS_FAST)
+	if (ctx->flags & FLAGS_SG) {
 		dma_unmap_sg(dd->dev, ctx->sg, 1, DMA_TO_DEVICE);
+		if (ctx->sg->length == ctx->offset) {
+			ctx->sg = sg_next(ctx->sg);
+			if (ctx->sg)
+				ctx->offset = 0;
+		}
+	} else {
+		dma_unmap_single(dd->dev, ctx->dma_addr, ctx->buflen,
+				 DMA_TO_DEVICE);
+	}
 
 	return 0;
 }
@@ -454,14 +526,7 @@ static void omap_sham_cleanup(struct ahash_request *req)
 	spin_unlock_irqrestore(&dd->lock, flags);
 
 	if (ctx->digcnt)
-		clk_disable(dd->iclk);
-
-	if (ctx->dma_addr)
-		dma_unmap_single(dd->dev, ctx->dma_addr, ctx->buflen,
-				 DMA_TO_DEVICE);
-
-	if (ctx->buffer)
-		free_page((unsigned long)ctx->buffer);
+		omap_sham_copy_ready_hash(req);
 
 	dev_dbg(dd->dev, "digcnt: %d, bufcnt: %d\n", ctx->digcnt, ctx->bufcnt);
 }
@@ -489,8 +554,6 @@ static int omap_sham_init(struct ahash_request *req)
 
 	ctx->flags = 0;
 
-	ctx->flags |= FLAGS_FIRST;
-
 	dev_dbg(dd->dev, "init: digest size: %d\n",
 		crypto_ahash_digestsize(tfm));
 
@@ -499,21 +562,7 @@ static int omap_sham_init(struct ahash_request *req)
 
 	ctx->bufcnt = 0;
 	ctx->digcnt = 0;
-
-	ctx->buflen = PAGE_SIZE;
-	ctx->buffer = (void *)__get_free_page(
-				(req->base.flags & CRYPTO_TFM_REQ_MAY_SLEEP) ?
-				GFP_KERNEL : GFP_ATOMIC);
-	if (!ctx->buffer)
-		return -ENOMEM;
-
-	ctx->dma_addr = dma_map_single(dd->dev, ctx->buffer, ctx->buflen,
-					DMA_TO_DEVICE);
-	if (dma_mapping_error(dd->dev, ctx->dma_addr)) {
-		dev_err(dd->dev, "dma %u bytes error\n", ctx->buflen);
-		free_page((unsigned long)ctx->buffer);
-		return -EINVAL;
-	}
+	ctx->buflen = BUFLEN;
 
 	if (tctx->flags & FLAGS_HMAC) {
 		struct omap_sham_hmac_ctx *bctx = tctx->base;
@@ -538,10 +587,8 @@ static int omap_sham_update_req(struct omap_sham_dev *dd)
 
 	if (ctx->flags & FLAGS_CPU)
 		err = omap_sham_update_cpu(dd);
-	else if (ctx->flags & FLAGS_FAST)
-		err = omap_sham_update_dma_fast(dd);
 	else
-		err = omap_sham_update_dma_slow(dd);
+		err = omap_sham_update_dma_start(dd);
 
 	/* wait for dma completion before can take more data */
 	dev_dbg(dd->dev, "update: err: %d, digcnt: %d\n", err, ctx->digcnt);
@@ -560,15 +607,12 @@ static int omap_sham_final_req(struct omap_sham_dev *dd)
 		use_dma = 0;
 
 	if (use_dma)
-		err = omap_sham_xmit_dma(dd, ctx->dma_addr, ctx->bufcnt, 1);
+		err = omap_sham_xmit_dma_map(dd, ctx, ctx->bufcnt, 1);
 	else
 		err = omap_sham_xmit_cpu(dd, ctx->buffer, ctx->bufcnt, 1);
 
 	ctx->bufcnt = 0;
 
-	if (err != -EINPROGRESS)
-		omap_sham_cleanup(req);
-
 	dev_dbg(dd->dev, "final_req: err: %d\n", err);
 
 	return err;
@@ -576,6 +620,7 @@ static int omap_sham_final_req(struct omap_sham_dev *dd)
 
 static int omap_sham_finish_req_hmac(struct ahash_request *req)
 {
+	struct omap_sham_reqctx *ctx = ahash_request_ctx(req);
 	struct omap_sham_ctx *tctx = crypto_tfm_ctx(req->base.tfm);
 	struct omap_sham_hmac_ctx *bctx = tctx->base;
 	int bs = crypto_shash_blocksize(bctx->shash);
@@ -590,48 +635,56 @@ static int omap_sham_finish_req_hmac(struct ahash_request *req)
 
 	return crypto_shash_init(&desc.shash) ?:
 	       crypto_shash_update(&desc.shash, bctx->opad, bs) ?:
-	       crypto_shash_finup(&desc.shash, req->result, ds, req->result);
+	       crypto_shash_finup(&desc.shash, ctx->digest, ds, ctx->digest);
 }
 
 static void omap_sham_finish_req(struct ahash_request *req, int err)
 {
 	struct omap_sham_reqctx *ctx = ahash_request_ctx(req);
+	struct omap_sham_dev *dd = ctx->dd;
 
 	if (!err) {
 		omap_sham_copy_hash(ctx->dd->req, 1);
 		if (ctx->flags & FLAGS_HMAC)
 			err = omap_sham_finish_req_hmac(req);
+	} else {
+		ctx->flags |= FLAGS_ERROR;
 	}
 
-	if (ctx->flags & FLAGS_FINAL)
+	if ((ctx->flags & FLAGS_FINAL) || err)
 		omap_sham_cleanup(req);
 
-	clear_bit(FLAGS_BUSY, &ctx->dd->flags);
+	clk_disable(dd->iclk);
+	dd->flags &= ~FLAGS_BUSY;
 
 	if (req->base.complete)
 		req->base.complete(&req->base, err);
 }
 
-static int omap_sham_handle_queue(struct omap_sham_dev *dd)
+static int omap_sham_handle_queue(struct omap_sham_dev *dd,
+				  struct ahash_request *req)
 {
 	struct crypto_async_request *async_req, *backlog;
 	struct omap_sham_reqctx *ctx;
-	struct ahash_request *req, *prev_req;
+	struct ahash_request *prev_req;
 	unsigned long flags;
-	int err = 0;
-
-	if (test_and_set_bit(FLAGS_BUSY, &dd->flags))
-		return 0;
+	int err = 0, ret = 0;
 
 	spin_lock_irqsave(&dd->lock, flags);
+	if (req)
+		ret = ahash_enqueue_request(&dd->queue, req);
+	if (dd->flags & FLAGS_BUSY) {
+		spin_unlock_irqrestore(&dd->lock, flags);
+		return ret;
+	}
 	backlog = crypto_get_backlog(&dd->queue);
 	async_req = crypto_dequeue_request(&dd->queue);
-	if (!async_req)
-		clear_bit(FLAGS_BUSY, &dd->flags);
+	if (async_req)
+		dd->flags |= FLAGS_BUSY;
 	spin_unlock_irqrestore(&dd->lock, flags);
 
 	if (!async_req)
-		return 0;
+		return ret;
 
 	if (backlog)
 		backlog->complete(backlog, -EINPROGRESS);
@@ -646,7 +699,22 @@ static int omap_sham_handle_queue(struct omap_sham_dev *dd)
 	dev_dbg(dd->dev, "handling new req, op: %lu, nbytes: %d\n",
 						ctx->op, req->nbytes);
 
-	if (req != prev_req && ctx->digcnt)
+
+	err = omap_sham_hw_init(dd);
+	if (err)
+		goto err1;
+
+	omap_set_dma_dest_params(dd->dma_lch, 0,
+			OMAP_DMA_AMODE_CONSTANT,
+			dd->phys_base + SHA_REG_DIN(0), 0, 16);
+
+	omap_set_dma_dest_burst_mode(dd->dma_lch,
+			OMAP_DMA_DATA_BURST_16);
+
+	omap_set_dma_src_burst_mode(dd->dma_lch,
+			OMAP_DMA_DATA_BURST_4);
+
+	if (ctx->digcnt)
 		/* request has changed - restore hash */
 		omap_sham_copy_hash(req, 0);
 
@@ -658,7 +726,7 @@ static int omap_sham_handle_queue(struct omap_sham_dev *dd)
 	} else if (ctx->op == OP_FINAL) {
 		err = omap_sham_final_req(dd);
 	}
-
+err1:
 	if (err != -EINPROGRESS) {
 		/* done_task will not finish it, so do it here */
 		omap_sham_finish_req(req, err);
@@ -667,7 +735,7 @@ static int omap_sham_handle_queue(struct omap_sham_dev *dd)
 
 	dev_dbg(dd->dev, "exit, err: %d\n", err);
 
-	return err;
+	return ret;
 }
 
 static int omap_sham_enqueue(struct ahash_request *req, unsigned int op)
@@ -675,18 +743,10 @@ static int omap_sham_enqueue(struct ahash_request *req, unsigned int op)
 	struct omap_sham_reqctx *ctx = ahash_request_ctx(req);
 	struct omap_sham_ctx *tctx = crypto_tfm_ctx(req->base.tfm);
 	struct omap_sham_dev *dd = tctx->dd;
-	unsigned long flags;
-	int err;
 
 	ctx->op = op;
 
-	spin_lock_irqsave(&dd->lock, flags);
-	err = ahash_enqueue_request(&dd->queue, req);
-	spin_unlock_irqrestore(&dd->lock, flags);
-
-	omap_sham_handle_queue(dd);
-
-	return err;
+	return omap_sham_handle_queue(dd, req);
 }
 
 static int omap_sham_update(struct ahash_request *req)
@@ -709,21 +769,13 @@ static int omap_sham_update(struct ahash_request *req)
 			*/
 			omap_sham_append_sg(ctx);
 			return 0;
-		} else if (ctx->bufcnt + ctx->total <= 64) {
+		} else if (ctx->bufcnt + ctx->total <= SHA1_MD5_BLOCK_SIZE) {
+			/*
+			* faster to use CPU for short transfers
+			*/
 			ctx->flags |= FLAGS_CPU;
-		} else if (!ctx->bufcnt && sg_is_last(ctx->sg)) {
-			/* may be can use faster functions */
-			int aligned = IS_ALIGNED((u32)ctx->sg->offset,
-								sizeof(u32));
-
-			if (aligned && (ctx->flags & FLAGS_FIRST))
-				/* digest: first and final */
-				ctx->flags |= FLAGS_FAST;
-
-			ctx->flags &= ~FLAGS_FIRST;
 		}
-	} else if (ctx->bufcnt + ctx->total <= ctx->buflen) {
-		/* if not finaup -> not fast */
+	} else if (ctx->bufcnt + ctx->total < ctx->buflen) {
 		omap_sham_append_sg(ctx);
 		return 0;
 	}
@@ -761,12 +813,14 @@ static int omap_sham_final(struct ahash_request *req)
 
 	ctx->flags |= FLAGS_FINUP;
 
-	/* OMAP HW accel works only with buffers >= 9 */
-	/* HMAC is always >= 9 because of ipad */
-	if ((ctx->digcnt + ctx->bufcnt) < 9)
-		err = omap_sham_final_shash(req);
-	else if (ctx->bufcnt)
-		return omap_sham_enqueue(req, OP_FINAL);
+	if (!(ctx->flags & FLAGS_ERROR)) {
+		/* OMAP HW accel works only with buffers >= 9 */
+		/* HMAC is always >= 9 because of ipad */
+		if ((ctx->digcnt + ctx->bufcnt) < 9)
+			err = omap_sham_final_shash(req);
+		else if (ctx->bufcnt)
+			return omap_sham_enqueue(req, OP_FINAL);
+	}
 
 	omap_sham_cleanup(req);
 
@@ -836,6 +890,8 @@ static int omap_sham_cra_init_alg(struct crypto_tfm *tfm, const char *alg_base)
 	struct omap_sham_ctx *tctx = crypto_tfm_ctx(tfm);
 	const char *alg_name = crypto_tfm_alg_name(tfm);
 
+	pr_info("enter\n");
+
 	/* Allocate a fallback and abort if it failed. */
 	tctx->fallback = crypto_alloc_shash(alg_name, 0,
 					    CRYPTO_ALG_NEED_FALLBACK);
@@ -846,7 +902,7 @@ static int omap_sham_cra_init_alg(struct crypto_tfm *tfm, const char *alg_base)
 	}
 
 	crypto_ahash_set_reqsize(__crypto_ahash_cast(tfm),
-				 sizeof(struct omap_sham_reqctx));
+				 sizeof(struct omap_sham_reqctx) + BUFLEN);
 
 	if (alg_base) {
 		struct omap_sham_hmac_ctx *bctx = tctx->base;
@@ -932,7 +988,7 @@ static struct ahash_alg algs[] = {
 						CRYPTO_ALG_NEED_FALLBACK,
 		.cra_blocksize		= SHA1_BLOCK_SIZE,
 		.cra_ctxsize		= sizeof(struct omap_sham_ctx),
-		.cra_alignmask		= 0,
+		.cra_alignmask		= OMAP_ALIGN_MASK,
 		.cra_module		= THIS_MODULE,
 		.cra_init		= omap_sham_cra_init,
 		.cra_exit		= omap_sham_cra_exit,
@@ -956,7 +1012,7 @@ static struct ahash_alg algs[] = {
 		.cra_blocksize		= SHA1_BLOCK_SIZE,
 		.cra_ctxsize		= sizeof(struct omap_sham_ctx) +
 					sizeof(struct omap_sham_hmac_ctx),
-		.cra_alignmask		= 0,
+		.cra_alignmask		= OMAP_ALIGN_MASK,
 		.cra_module		= THIS_MODULE,
 		.cra_init		= omap_sham_cra_sha1_init,
 		.cra_exit		= omap_sham_cra_exit,
@@ -980,7 +1036,7 @@ static struct ahash_alg algs[] = {
 		.cra_blocksize		= SHA1_BLOCK_SIZE,
 		.cra_ctxsize		= sizeof(struct omap_sham_ctx) +
 					sizeof(struct omap_sham_hmac_ctx),
-		.cra_alignmask		= 0,
+		.cra_alignmask		= OMAP_ALIGN_MASK,
 		.cra_module		= THIS_MODULE,
 		.cra_init		= omap_sham_cra_md5_init,
 		.cra_exit		= omap_sham_cra_exit,
@@ -993,7 +1049,7 @@ static void omap_sham_done_task(unsigned long data)
 	struct omap_sham_dev *dd = (struct omap_sham_dev *)data;
 	struct ahash_request *req = dd->req;
 	struct omap_sham_reqctx *ctx = ahash_request_ctx(req);
-	int ready = 1;
+	int ready = 0, err = 0;
 
 	if (ctx->flags & FLAGS_OUTPUT_READY) {
 		ctx->flags &= ~FLAGS_OUTPUT_READY;
@@ -1003,15 +1059,18 @@ static void omap_sham_done_task(unsigned long data)
 	if (dd->flags & FLAGS_DMA_ACTIVE) {
 		dd->flags &= ~FLAGS_DMA_ACTIVE;
 		omap_sham_update_dma_stop(dd);
-		omap_sham_update_dma_slow(dd);
+		if (!dd->err)
+			err = omap_sham_update_dma_start(dd);
 	}
 
-	if (ready && !(dd->flags & FLAGS_DMA_ACTIVE)) {
-		dev_dbg(dd->dev, "update done\n");
+	err = dd->err ? : err;
+
+	if (err != -EINPROGRESS && (ready || err)) {
+		dev_dbg(dd->dev, "update done: err: %d\n", err);
 		/* finish curent request */
-		omap_sham_finish_req(req, 0);
+		omap_sham_finish_req(req, err);
 		/* start new request */
-		omap_sham_handle_queue(dd);
+		omap_sham_handle_queue(dd, NULL);
 	}
 }
 
@@ -1019,7 +1078,7 @@ static void omap_sham_queue_task(unsigned long data)
 {
 	struct omap_sham_dev *dd = (struct omap_sham_dev *)data;
 
-	omap_sham_handle_queue(dd);
+	omap_sham_handle_queue(dd, NULL);
 }
 
 static irqreturn_t omap_sham_irq(int irq, void *dev_id)
@@ -1041,6 +1100,7 @@ static irqreturn_t omap_sham_irq(int irq, void *dev_id)
 	omap_sham_read(dd, SHA_REG_CTRL);
 
 	ctx->flags |= FLAGS_OUTPUT_READY;
+	dd->err = 0;
 	tasklet_schedule(&dd->done_task);
 
 	return IRQ_HANDLED;
@@ -1050,8 +1110,13 @@ static void omap_sham_dma_callback(int lch, u16 ch_status, void *data)
 {
 	struct omap_sham_dev *dd = data;
 
-	if (likely(lch == dd->dma_lch))
-		tasklet_schedule(&dd->done_task);
+	if (ch_status != OMAP_DMA_BLOCK_IRQ) {
+		pr_err("omap-sham DMA error status: 0x%hx\n", ch_status);
+		dd->err = -EIO;
+		dd->flags &= ~FLAGS_INIT; /* request to re-initialize */
+	}
+
+	tasklet_schedule(&dd->done_task);
 }
 
 static int omap_sham_dma_init(struct omap_sham_dev *dd)
@@ -1066,15 +1131,6 @@ static int omap_sham_dma_init(struct omap_sham_dev *dd)
 		dev_err(dd->dev, "Unable to request DMA channel\n");
 		return err;
 	}
-	omap_set_dma_dest_params(dd->dma_lch, 0,
-			OMAP_DMA_AMODE_CONSTANT,
-			dd->phys_base + SHA_REG_DIN(0), 0, 16);
-
-	omap_set_dma_dest_burst_mode(dd->dma_lch,
-			OMAP_DMA_DATA_BURST_16);
-
-	omap_set_dma_src_burst_mode(dd->dma_lch,
-			OMAP_DMA_DATA_BURST_4);
 
 	return 0;
 }
diff --git a/drivers/crypto/padlock-aes.c b/drivers/crypto/padlock-aes.c
index 8a515ba..db33d30 100644
--- a/drivers/crypto/padlock-aes.c
+++ b/drivers/crypto/padlock-aes.c
@@ -9,6 +9,7 @@
 
 #include <crypto/algapi.h>
 #include <crypto/aes.h>
+#include <crypto/padlock.h>
 #include <linux/module.h>
 #include <linux/init.h>
 #include <linux/types.h>
@@ -21,7 +22,6 @@
 #include <asm/byteorder.h>
 #include <asm/processor.h>
 #include <asm/i387.h>
-#include "padlock.h"
 
 /*
  * Number of data blocks actually fetched for each xcrypt insn.
diff --git a/drivers/crypto/padlock-sha.c b/drivers/crypto/padlock-sha.c
index d3a27e0..adf075b 100644
--- a/drivers/crypto/padlock-sha.c
+++ b/drivers/crypto/padlock-sha.c
@@ -13,6 +13,7 @@
  */
 
 #include <crypto/internal/hash.h>
+#include <crypto/padlock.h>
 #include <crypto/sha.h>
 #include <linux/err.h>
 #include <linux/module.h>
@@ -22,13 +23,6 @@
 #include <linux/kernel.h>
 #include <linux/scatterlist.h>
 #include <asm/i387.h>
-#include "padlock.h"
-
-#ifdef CONFIG_64BIT
-#define STACK_ALIGN 16
-#else
-#define STACK_ALIGN 4
-#endif
 
 struct padlock_sha_desc {
 	struct shash_desc fallback;
diff --git a/drivers/crypto/padlock.h b/drivers/crypto/padlock.h
deleted file mode 100644
index b728e45..0000000
--- a/drivers/crypto/padlock.h
+++ /dev/null
@@ -1,23 +0,0 @@
-/*
- * Driver for VIA PadLock
- *
- * Copyright (c) 2004 Michal Ludvig <michal@logix.cz>
- *
- * This program is free software; you can redistribute it and/or modify it
- * under the terms of the GNU General Public License as published by the Free
- * Software Foundation; either version 2 of the License, or (at your option) 
- * any later version.
- *
- */
-
-#ifndef _CRYPTO_PADLOCK_H
-#define _CRYPTO_PADLOCK_H
-
-#define PADLOCK_ALIGNMENT 16
-
-#define PFX	"padlock: "
-
-#define PADLOCK_CRA_PRIORITY	300
-#define PADLOCK_COMPOSITE_PRIORITY 400
-
-#endif	/* _CRYPTO_PADLOCK_H */