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:
crypto: padlock - fix VIA PadLock instruction usage with irq_ts_save/restore()
crypto: hash - Add missing top-level functions
crypto: hash - Fix digest size check for digest type
crypto: tcrypt - Fix AEAD chunk testing
crypto: talitos - Add handling for SEC 3.x treatment of link table

crypto/digest.c

···
 	struct ahash_tfm  *crt  = &tfm->crt_ahash;
 	struct digest_alg *dalg = &tfm->__crt_alg->cra_digest;
 
-	if (dalg->dia_digestsize > crypto_tfm_alg_blocksize(tfm))
+	if (dalg->dia_digestsize > PAGE_SIZE / 8)
 		return -EINVAL;
 
 	crt->init       = digest_async_init;

crypto/tcrypt.c

···
 
 			for (k = 0, temp = 0; k < template[i].np; k++) {
 				printk(KERN_INFO "page %u\n", k);
-				q = &axbuf[IDX[k]];
-				hexdump(q, template[i].tap[k]);
+				q = &xbuf[IDX[k]];
+
+				n = template[i].tap[k];
+				if (k == template[i].np - 1)
+					n += enc ? authsize : -authsize;
+				hexdump(q, n);
 				printk(KERN_INFO "%s\n",
-				       memcmp(q, template[i].result + temp,
-					      template[i].tap[k] -
-					      (k < template[i].np - 1 || enc ?
-					       0 : authsize)) ?
+				       memcmp(q, template[i].result + temp, n) ?
 				       "fail" : "pass");
 
-				for (n = 0; q[template[i].tap[k] + n]; n++)
-					;
+				q += n;
+				if (k == template[i].np - 1 && !enc) {
+					if (memcmp(q, template[i].input +
+						      temp + n, authsize))
+						n = authsize;
+					else
+						n = 0;
+				} else {
+					for (n = 0; q[n]; n++)
+						;
+				}
 				if (n) {
 					printk("Result buffer corruption %u "
 					       "bytes:\n", n);
-					hexdump(&q[template[i].tap[k]], n);
+					hexdump(q, n);
 				}
 
 				temp += template[i].tap[k];

drivers/char/hw_random/via-rng.c

···
 #include <asm/io.h>
 #include <asm/msr.h>
 #include <asm/cpufeature.h>
+#include <asm/i387.h>
 
 
 #define PFX	KBUILD_MODNAME ": "
···
  * Another possible performance boost may come from simply buffering
  * until we have 4 bytes, thus returning a u32 at a time,
  * instead of the current u8-at-a-time.
+ *
+ * Padlock instructions can generate a spurious DNA fault, so
+ * we have to call them in the context of irq_ts_save/restore()
  */
 
 static inline u32 xstore(u32 *addr, u32 edx_in)
 {
 	u32 eax_out;
+	int ts_state;
+
+	ts_state = irq_ts_save();
 
 	asm(".byte 0x0F,0xA7,0xC0 /* xstore %%edi (addr=%0) */"
 		:"=m"(*addr), "=a"(eax_out)
 		:"D"(addr), "d"(edx_in));
 
+	irq_ts_restore(ts_state);
 	return eax_out;
 }
 

drivers/crypto/padlock-aes.c

···
 #include <linux/interrupt.h>
 #include <linux/kernel.h>
 #include <asm/byteorder.h>
+#include <asm/i387.h>
 #include "padlock.h"
 
 /* Control word. */
···
 	asm volatile ("pushfl; popfl");
 }
 
+/*
+ * While the padlock instructions don't use FP/SSE registers, they
+ * generate a spurious DNA fault when cr0.ts is '1'. These instructions
+ * should be used only inside the irq_ts_save/restore() context
+ */
+
 static inline void padlock_xcrypt(const u8 *input, u8 *output, void *key,
 				  void *control_word)
 {
···
 static void aes_encrypt(struct crypto_tfm *tfm, u8 *out, const u8 *in)
 {
 	struct aes_ctx *ctx = aes_ctx(tfm);
+	int ts_state;
 	padlock_reset_key();
+
+	ts_state = irq_ts_save();
 	aes_crypt(in, out, ctx->E, &ctx->cword.encrypt);
+	irq_ts_restore(ts_state);
 }
 
 static void aes_decrypt(struct crypto_tfm *tfm, u8 *out, const u8 *in)
 {
 	struct aes_ctx *ctx = aes_ctx(tfm);
+	int ts_state;
 	padlock_reset_key();
+
+	ts_state = irq_ts_save();
 	aes_crypt(in, out, ctx->D, &ctx->cword.decrypt);
+	irq_ts_restore(ts_state);
 }
 
 static struct crypto_alg aes_alg = {
···
 	struct aes_ctx *ctx = blk_aes_ctx(desc->tfm);
 	struct blkcipher_walk walk;
 	int err;
+	int ts_state;
 
 	padlock_reset_key();
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
 
+	ts_state = irq_ts_save();
 	while ((nbytes = walk.nbytes)) {
 		padlock_xcrypt_ecb(walk.src.virt.addr, walk.dst.virt.addr,
 				   ctx->E, &ctx->cword.encrypt,
···
 		nbytes &= AES_BLOCK_SIZE - 1;
 		err = blkcipher_walk_done(desc, &walk, nbytes);
 	}
+	irq_ts_restore(ts_state);
 
 	return err;
 }
···
 	struct aes_ctx *ctx = blk_aes_ctx(desc->tfm);
 	struct blkcipher_walk walk;
 	int err;
+	int ts_state;
 
 	padlock_reset_key();
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
 
+	ts_state = irq_ts_save();
 	while ((nbytes = walk.nbytes)) {
 		padlock_xcrypt_ecb(walk.src.virt.addr, walk.dst.virt.addr,
 				   ctx->D, &ctx->cword.decrypt,
···
 		nbytes &= AES_BLOCK_SIZE - 1;
 		err = blkcipher_walk_done(desc, &walk, nbytes);
 	}
-
+	irq_ts_restore(ts_state);
 	return err;
 }
 
···
 	struct aes_ctx *ctx = blk_aes_ctx(desc->tfm);
 	struct blkcipher_walk walk;
 	int err;
+	int ts_state;
 
 	padlock_reset_key();
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
 
+	ts_state = irq_ts_save();
 	while ((nbytes = walk.nbytes)) {
 		u8 *iv = padlock_xcrypt_cbc(walk.src.virt.addr,
 					    walk.dst.virt.addr, ctx->E,
···
 		nbytes &= AES_BLOCK_SIZE - 1;
 		err = blkcipher_walk_done(desc, &walk, nbytes);
 	}
+	irq_ts_restore(ts_state);
 
 	return err;
 }
···
 	struct aes_ctx *ctx = blk_aes_ctx(desc->tfm);
 	struct blkcipher_walk walk;
 	int err;
+	int ts_state;
 
 	padlock_reset_key();
 
 	blkcipher_walk_init(&walk, dst, src, nbytes);
 	err = blkcipher_walk_virt(desc, &walk);
 
+	ts_state = irq_ts_save();
 	while ((nbytes = walk.nbytes)) {
 		padlock_xcrypt_cbc(walk.src.virt.addr, walk.dst.virt.addr,
 				   ctx->D, walk.iv, &ctx->cword.decrypt,
···
 		err = blkcipher_walk_done(desc, &walk, nbytes);
 	}
 
+	irq_ts_restore(ts_state);
 	return err;
 }
 

drivers/crypto/padlock-sha.c

···
 #include <linux/interrupt.h>
 #include <linux/kernel.h>
 #include <linux/scatterlist.h>
+#include <asm/i387.h>
 #include "padlock.h"
 
 #define SHA1_DEFAULT_FALLBACK	"sha1-generic"
···
 	 *     PadLock microcode needs it that big. */
 	char buf[128+16];
 	char *result = NEAREST_ALIGNED(buf);
+	int ts_state;
 
 	((uint32_t *)result)[0] = SHA1_H0;
 	((uint32_t *)result)[1] = SHA1_H1;
···
 	((uint32_t *)result)[3] = SHA1_H3;
 	((uint32_t *)result)[4] = SHA1_H4;
  
+	/* prevent taking the spurious DNA fault with padlock. */
+	ts_state = irq_ts_save();
 	asm volatile (".byte 0xf3,0x0f,0xa6,0xc8" /* rep xsha1 */
 		      : "+S"(in), "+D"(result)
 		      : "c"(count), "a"(0));
+	irq_ts_restore(ts_state);
 
 	padlock_output_block((uint32_t *)result, (uint32_t *)out, 5);
 }
···
 	 *     PadLock microcode needs it that big. */
 	char buf[128+16];
 	char *result = NEAREST_ALIGNED(buf);
+	int ts_state;
 
 	((uint32_t *)result)[0] = SHA256_H0;
 	((uint32_t *)result)[1] = SHA256_H1;
···
 	((uint32_t *)result)[6] = SHA256_H6;
 	((uint32_t *)result)[7] = SHA256_H7;
 
+	/* prevent taking the spurious DNA fault with padlock. */
+	ts_state = irq_ts_save();
 	asm volatile (".byte 0xf3,0x0f,0xa6,0xd0" /* rep xsha256 */
 		      : "+S"(in), "+D"(result)
 		      : "c"(count), "a"(0));
+	irq_ts_restore(ts_state);
 
 	padlock_output_block((uint32_t *)result, (uint32_t *)out, 8);
 }

drivers/crypto/talitos.c

···
 	unsigned int exec_units;
 	unsigned int desc_types;
 
+	/* SEC Compatibility info */
+	unsigned long features;
+
 	/* next channel to be assigned next incoming descriptor */
 	atomic_t last_chan;
 
···
 	/* hwrng device */
 	struct hwrng rng;
 };
+
+/* .features flag */
+#define TALITOS_FTR_SRC_LINK_TBL_LEN_INCLUDES_EXTENT 0x00000001
 
 /*
  * map virtual single (contiguous) pointer to h/w descriptor pointer
···
 	/* copy the generated ICV to dst */
 	if (edesc->dma_len) {
 		icvdata = &edesc->link_tbl[edesc->src_nents +
-					   edesc->dst_nents + 1];
+					   edesc->dst_nents + 2];
 		sg = sg_last(areq->dst, edesc->dst_nents);
 		memcpy((char *)sg_virt(sg) + sg->length - ctx->authsize,
 		       icvdata, ctx->authsize);
···
 		/* auth check */
 		if (edesc->dma_len)
 			icvdata = &edesc->link_tbl[edesc->src_nents +
-						   edesc->dst_nents + 1];
+						   edesc->dst_nents + 2];
 		else
 			icvdata = &edesc->link_tbl[0];
 
···
 		sg_count = sg_to_link_tbl(areq->src, sg_count, cryptlen,
 					  &edesc->link_tbl[0]);
 		if (sg_count > 1) {
+			struct talitos_ptr *link_tbl_ptr =
+				&edesc->link_tbl[sg_count-1];
+			struct scatterlist *sg;
+			struct talitos_private *priv = dev_get_drvdata(dev);
+
 			desc->ptr[4].j_extent |= DESC_PTR_LNKTBL_JUMP;
 			desc->ptr[4].ptr = cpu_to_be32(edesc->dma_link_tbl);
 			dma_sync_single_for_device(ctx->dev, edesc->dma_link_tbl,
 						   edesc->dma_len, DMA_BIDIRECTIONAL);
+			/* If necessary for this SEC revision,
+			 * add a link table entry for ICV.
+			 */
+			if ((priv->features &
+			     TALITOS_FTR_SRC_LINK_TBL_LEN_INCLUDES_EXTENT) &&
+			    (edesc->desc.hdr & DESC_HDR_MODE0_ENCRYPT) == 0) {
+				link_tbl_ptr->j_extent = 0;
+				link_tbl_ptr++;
+				link_tbl_ptr->j_extent = DESC_PTR_LNKTBL_RETURN;
+				link_tbl_ptr->len = cpu_to_be16(authsize);
+				sg = sg_last(areq->src, edesc->src_nents ? : 1);
+				link_tbl_ptr->ptr = cpu_to_be32(
+						(char *)sg_dma_address(sg)
+						+ sg->length - authsize);
+			}
 		} else {
 			/* Only one segment now, so no link tbl needed */
 			desc->ptr[4].ptr = cpu_to_be32(sg_dma_address(areq->src));
···
 		desc->ptr[5].ptr = cpu_to_be32(sg_dma_address(areq->dst));
 	} else {
 		struct talitos_ptr *link_tbl_ptr =
-			&edesc->link_tbl[edesc->src_nents];
-		struct scatterlist *sg;
+			&edesc->link_tbl[edesc->src_nents + 1];
 
 		desc->ptr[5].ptr = cpu_to_be32((struct talitos_ptr *)
 					       edesc->dma_link_tbl +
-					       edesc->src_nents);
+					       edesc->src_nents + 1);
 		if (areq->src == areq->dst) {
 			memcpy(link_tbl_ptr, &edesc->link_tbl[0],
 			       edesc->src_nents * sizeof(struct talitos_ptr));
···
 			sg_count = sg_to_link_tbl(areq->dst, sg_count, cryptlen,
 						  link_tbl_ptr);
 		}
+		/* Add an entry to the link table for ICV data */
 		link_tbl_ptr += sg_count - 1;
-
-		/* handle case where sg_last contains the ICV exclusively */
-		sg = sg_last(areq->dst, edesc->dst_nents);
-		if (sg->length == ctx->authsize)
-			link_tbl_ptr--;
-
 		link_tbl_ptr->j_extent = 0;
+		sg_count++;
 		link_tbl_ptr++;
 		link_tbl_ptr->j_extent = DESC_PTR_LNKTBL_RETURN;
 		link_tbl_ptr->len = cpu_to_be16(authsize);
···
 		link_tbl_ptr->ptr = cpu_to_be32((struct talitos_ptr *)
 						edesc->dma_link_tbl +
 					        edesc->src_nents +
-						edesc->dst_nents + 1);
+						edesc->dst_nents + 2);
 
 		desc->ptr[5].j_extent |= DESC_PTR_LNKTBL_JUMP;
 		dma_sync_single_for_device(ctx->dev, edesc->dma_link_tbl,
···
 
 	/*
 	 * allocate space for base edesc plus the link tables,
-	 * allowing for a separate entry for the generated ICV (+ 1),
+	 * allowing for two separate entries for ICV and generated ICV (+ 2),
 	 * and the ICV data itself
 	 */
 	alloc_len = sizeof(struct ipsec_esp_edesc);
 	if (src_nents || dst_nents) {
-		dma_len = (src_nents + dst_nents + 1) *
+		dma_len = (src_nents + dst_nents + 2) *
 				 sizeof(struct talitos_ptr) + ctx->authsize;
 		alloc_len += dma_len;
 	} else {
···
 	/* stash incoming ICV for later cmp with ICV generated by the h/w */
 	if (edesc->dma_len)
 		icvdata = &edesc->link_tbl[edesc->src_nents +
-					   edesc->dst_nents + 1];
+					   edesc->dst_nents + 2];
 	else
 		icvdata = &edesc->link_tbl[0];
 
···
 		err = -EINVAL;
 		goto err_out;
 	}
+
+	if (of_device_is_compatible(np, "fsl,sec3.0"))
+		priv->features |= TALITOS_FTR_SRC_LINK_TBL_LEN_INCLUDES_EXTENT;
 
 	priv->head_lock = kmalloc(sizeof(spinlock_t) * priv->num_channels,
 				  GFP_KERNEL);

include/asm-x86/i387.h

···
 #include <linux/sched.h>
 #include <linux/kernel_stat.h>
 #include <linux/regset.h>
+#include <linux/hardirq.h>
 #include <asm/asm.h>
 #include <asm/processor.h>
 #include <asm/sigcontext.h>
···
 {
 	stts();
 	preempt_enable();
+}
+
+/*
+ * Some instructions like VIA's padlock instructions generate a spurious
+ * DNA fault but don't modify SSE registers. And these instructions
+ * get used from interrupt context aswell. To prevent these kernel instructions
+ * in interrupt context interact wrongly with other user/kernel fpu usage, we
+ * should use them only in the context of irq_ts_save/restore()
+ */
+static inline int irq_ts_save(void)
+{
+	/*
+	 * If we are in process context, we are ok to take a spurious DNA fault.
+	 * Otherwise, doing clts() in process context require pre-emption to
+	 * be disabled or some heavy lifting like kernel_fpu_begin()
+	 */
+	if (!in_interrupt())
+		return 0;
+
+	if (read_cr0() & X86_CR0_TS) {
+		clts();
+		return 1;
+	}
+
+	return 0;
+}
+
+static inline void irq_ts_restore(int TS_state)
+{
+	if (TS_state)
+		stts();
 }
 
 #ifdef CONFIG_X86_64

include/crypto/hash.h

···
 	return crt->digest(req);
 }
 
+static inline int crypto_ahash_init(struct ahash_request *req)
+{
+	struct ahash_tfm *crt = crypto_ahash_crt(crypto_ahash_reqtfm(req));
+	return crt->init(req);
+}
+
+static inline int crypto_ahash_update(struct ahash_request *req)
+{
+	struct ahash_tfm *crt = crypto_ahash_crt(crypto_ahash_reqtfm(req));
+	return crt->update(req);
+}
+
+static inline int crypto_ahash_final(struct ahash_request *req)
+{
+	struct ahash_tfm *crt = crypto_ahash_crt(crypto_ahash_reqtfm(req));
+	return crt->final(req);
+}
+
 static inline void ahash_request_set_tfm(struct ahash_request *req,
 					 struct crypto_ahash *tfm)
 {