use rijndael instead of blowfish because of faster key setup.

break swap paritions into sections, each section has own
encryption key.  if a section's key becomes unreferenced, erase it.

7 files changed, 762 insertions, 106 deletions

diff --git a/sys/conf/files b/sys/conf/files
index 29f705c950a..86d87d896c5 100644
--- a/sys/conf/files
+++ b/sys/conf/files
@@ -1,4 +1,4 @@
-#	$OpenBSD: files,v 1.162 2000/05/17 18:20:19 mickey Exp $
+#	$OpenBSD: files,v 1.163 2000/05/27 21:06:07 provos Exp $
 #	$NetBSD: files,v 1.87 1996/05/19 17:17:50 jonathan Exp $
 
 #	@(#)files.newconf	7.5 (Berkeley) 5/10/93
@@ -237,6 +237,7 @@ file adosfs/adlookup.c			adosfs
 file adosfs/adutil.c			adosfs
 file adosfs/advfsops.c			adosfs
 file adosfs/advnops.c			adosfs
+file crypto/rijndael.c			uvm_swap_encrypt
 file ddb/db_access.c			ddb
 file ddb/db_aout.c			ddb
 file ddb/db_break.c			ddb
@@ -532,7 +533,7 @@ file netinet/ip_esp.c			inet & ipsec
 file netinet/ip_ah.c			inet & ipsec
 file crypto/rmd160.c			(inet & ipsec) | crypto
 file crypto/sha1.c			(inet & ipsec) | crypto
-file crypto/blf.c			(inet & ipsec) | crypto | uvm_swap_encrypt
+file crypto/blf.c			(inet & ipsec) | crypto
 file crypto/cast.c			(inet & ipsec) | crypto
 file crypto/skipjack.c			(inet & ipsec) | crypto
 file crypto/ecb_enc.c			(inet & ipsec) | crypto
diff --git a/sys/crypto/rijndael.c b/sys/crypto/rijndael.c
new file mode 100644
index 00000000000..242cba31d57
--- /dev/null
+++ b/sys/crypto/rijndael.c
@@ -0,0 +1,494 @@
+
+/* This is an independent implementation of the encryption algorithm:   */
+/*                                                                      */
+/*         RIJNDAEL by Joan Daemen and Vincent Rijmen                   */
+/*                                                                      */
+/* which is a candidate algorithm in the Advanced Encryption Standard   */
+/* programme of the US National Institute of Standards and Technology.  */
+/*                                                                      */
+/* Copyright in this implementation is held by Dr B R Gladman but I     */
+/* hereby give permission for its free direct or derivative use subject */
+/* to acknowledgment of its origin and compliance with any conditions   */
+/* that the originators of the algorithm place on its exploitation.     */
+/*                                                                      */
+/* Dr Brian Gladman (gladman@seven77.demon.co.uk) 14th January 1999     */
+
+/* Timing data for Rijndael (rijndael.c)
+
+Algorithm: rijndael (rijndael.c)
+
+128 bit key:
+Key Setup:    305/1389 cycles (encrypt/decrypt)
+Encrypt:       374 cycles =    68.4 mbits/sec
+Decrypt:       352 cycles =    72.7 mbits/sec
+Mean:          363 cycles =    70.5 mbits/sec
+
+192 bit key:
+Key Setup:    277/1595 cycles (encrypt/decrypt)
+Encrypt:       439 cycles =    58.3 mbits/sec
+Decrypt:       425 cycles =    60.2 mbits/sec
+Mean:          432 cycles =    59.3 mbits/sec
+
+256 bit key:
+Key Setup:    374/1960 cycles (encrypt/decrypt)
+Encrypt:       502 cycles =    51.0 mbits/sec
+Decrypt:       498 cycles =    51.4 mbits/sec
+Mean:          500 cycles =    51.2 mbits/sec
+
+*/
+
+#include <sys/param.h>
+#include <sys/systm.h>
+
+#include <crypto/rijndael.h>
+
+void gen_tabs	__P((void));
+
+/* 3. Basic macros for speeding up generic operations               */
+
+/* Circular rotate of 32 bit values                                 */
+
+#define rotr(x,n)   (((x) >> ((int)(n))) | ((x) << (32 - (int)(n))))
+#define rotl(x,n)   (((x) << ((int)(n))) | ((x) >> (32 - (int)(n))))
+
+/* Invert byte order in a 32 bit variable                           */
+
+#define bswap(x)    (rotl(x, 8) & 0x00ff00ff | rotr(x, 8) & 0xff00ff00)
+
+/* Extract byte from a 32 bit quantity (little endian notation)     */ 
+
+#define byte(x,n)   ((u1byte)((x) >> (8 * n)))
+
+#if BYTE_ORDER != LITTLE_ENDIAN
+#define BLOCK_SWAP
+#endif
+
+/* For inverting byte order in input/output 32 bit words if needed  */
+
+#ifdef  BLOCK_SWAP
+#define BYTE_SWAP
+#define WORD_SWAP
+#endif
+
+#ifdef  BYTE_SWAP
+#define io_swap(x)  bswap(x)
+#else
+#define io_swap(x)  (x)
+#endif
+
+/* For inverting the byte order of input/output blocks if needed    */
+
+#ifdef  WORD_SWAP
+
+#define get_block(x)                            \
+    ((u4byte*)(x))[0] = io_swap(in_blk[3]);     \
+    ((u4byte*)(x))[1] = io_swap(in_blk[2]);     \
+    ((u4byte*)(x))[2] = io_swap(in_blk[1]);     \
+    ((u4byte*)(x))[3] = io_swap(in_blk[0])
+
+#define put_block(x)                            \
+    out_blk[3] = io_swap(((u4byte*)(x))[0]);    \
+    out_blk[2] = io_swap(((u4byte*)(x))[1]);    \
+    out_blk[1] = io_swap(((u4byte*)(x))[2]);    \
+    out_blk[0] = io_swap(((u4byte*)(x))[3])
+
+#define get_key(x,len)                          \
+    ((u4byte*)(x))[4] = ((u4byte*)(x))[5] =     \
+    ((u4byte*)(x))[6] = ((u4byte*)(x))[7] = 0;  \
+    switch((((len) + 63) / 64)) {               \
+    case 2:                                     \
+    ((u4byte*)(x))[0] = io_swap(in_key[3]);     \
+    ((u4byte*)(x))[1] = io_swap(in_key[2]);     \
+    ((u4byte*)(x))[2] = io_swap(in_key[1]);     \
+    ((u4byte*)(x))[3] = io_swap(in_key[0]);     \
+    break;                                      \
+    case 3:                                     \
+    ((u4byte*)(x))[0] = io_swap(in_key[5]);     \
+    ((u4byte*)(x))[1] = io_swap(in_key[4]);     \
+    ((u4byte*)(x))[2] = io_swap(in_key[3]);     \
+    ((u4byte*)(x))[3] = io_swap(in_key[2]);     \
+    ((u4byte*)(x))[4] = io_swap(in_key[1]);     \
+    ((u4byte*)(x))[5] = io_swap(in_key[0]);     \
+    break;                                      \
+    case 4:                                     \
+    ((u4byte*)(x))[0] = io_swap(in_key[7]);     \
+    ((u4byte*)(x))[1] = io_swap(in_key[6]);     \
+    ((u4byte*)(x))[2] = io_swap(in_key[5]);     \
+    ((u4byte*)(x))[3] = io_swap(in_key[4]);     \
+    ((u4byte*)(x))[4] = io_swap(in_key[3]);     \
+    ((u4byte*)(x))[5] = io_swap(in_key[2]);     \
+    ((u4byte*)(x))[6] = io_swap(in_key[1]);     \
+    ((u4byte*)(x))[7] = io_swap(in_key[0]);     \
+    }
+
+#else
+
+#define get_block(x)                            \
+    ((u4byte*)(x))[0] = io_swap(in_blk[0]);     \
+    ((u4byte*)(x))[1] = io_swap(in_blk[1]);     \
+    ((u4byte*)(x))[2] = io_swap(in_blk[2]);     \
+    ((u4byte*)(x))[3] = io_swap(in_blk[3])
+
+#define put_block(x)                            \
+    out_blk[0] = io_swap(((u4byte*)(x))[0]);    \
+    out_blk[1] = io_swap(((u4byte*)(x))[1]);    \
+    out_blk[2] = io_swap(((u4byte*)(x))[2]);    \
+    out_blk[3] = io_swap(((u4byte*)(x))[3])
+
+#define get_key(x,len)                          \
+    ((u4byte*)(x))[4] = ((u4byte*)(x))[5] =     \
+    ((u4byte*)(x))[6] = ((u4byte*)(x))[7] = 0;  \
+    switch((((len) + 63) / 64)) {               \
+    case 4:                                     \
+    ((u4byte*)(x))[6] = io_swap(in_key[6]);     \
+    ((u4byte*)(x))[7] = io_swap(in_key[7]);     \
+    case 3:                                     \
+    ((u4byte*)(x))[4] = io_swap(in_key[4]);     \
+    ((u4byte*)(x))[5] = io_swap(in_key[5]);     \
+    case 2:                                     \
+    ((u4byte*)(x))[0] = io_swap(in_key[0]);     \
+    ((u4byte*)(x))[1] = io_swap(in_key[1]);     \
+    ((u4byte*)(x))[2] = io_swap(in_key[2]);     \
+    ((u4byte*)(x))[3] = io_swap(in_key[3]);     \
+    }
+
+#endif
+
+#define LARGE_TABLES
+
+u1byte  pow_tab[256];
+u1byte  log_tab[256];
+u1byte  sbx_tab[256];
+u1byte  isb_tab[256];
+u4byte  rco_tab[ 10];
+u4byte  ft_tab[4][256];
+u4byte  it_tab[4][256];
+
+#ifdef  LARGE_TABLES
+  u4byte  fl_tab[4][256];
+  u4byte  il_tab[4][256];
+#endif
+
+u4byte  tab_gen = 0;
+
+#define ff_mult(a,b)    (a && b ? pow_tab[(log_tab[a] + log_tab[b]) % 255] : 0)
+
+#define f_rn(bo, bi, n, k)                          \
+    bo[n] =  ft_tab[0][byte(bi[n],0)] ^             \
+             ft_tab[1][byte(bi[(n + 1) & 3],1)] ^   \
+             ft_tab[2][byte(bi[(n + 2) & 3],2)] ^   \
+             ft_tab[3][byte(bi[(n + 3) & 3],3)] ^ *(k + n)
+
+#define i_rn(bo, bi, n, k)                          \
+    bo[n] =  it_tab[0][byte(bi[n],0)] ^             \
+             it_tab[1][byte(bi[(n + 3) & 3],1)] ^   \
+             it_tab[2][byte(bi[(n + 2) & 3],2)] ^   \
+             it_tab[3][byte(bi[(n + 1) & 3],3)] ^ *(k + n)
+
+#ifdef LARGE_TABLES
+
+#define ls_box(x)                \
+    ( fl_tab[0][byte(x, 0)] ^    \
+      fl_tab[1][byte(x, 1)] ^    \
+      fl_tab[2][byte(x, 2)] ^    \
+      fl_tab[3][byte(x, 3)] )
+
+#define f_rl(bo, bi, n, k)                          \
+    bo[n] =  fl_tab[0][byte(bi[n],0)] ^             \
+             fl_tab[1][byte(bi[(n + 1) & 3],1)] ^   \
+             fl_tab[2][byte(bi[(n + 2) & 3],2)] ^   \
+             fl_tab[3][byte(bi[(n + 3) & 3],3)] ^ *(k + n)
+
+#define i_rl(bo, bi, n, k)                          \
+    bo[n] =  il_tab[0][byte(bi[n],0)] ^             \
+             il_tab[1][byte(bi[(n + 3) & 3],1)] ^   \
+             il_tab[2][byte(bi[(n + 2) & 3],2)] ^   \
+             il_tab[3][byte(bi[(n + 1) & 3],3)] ^ *(k + n)
+
+#else
+
+#define ls_box(x)                            \
+    ((u4byte)sbx_tab[byte(x, 0)] <<  0) ^    \
+    ((u4byte)sbx_tab[byte(x, 1)] <<  8) ^    \
+    ((u4byte)sbx_tab[byte(x, 2)] << 16) ^    \
+    ((u4byte)sbx_tab[byte(x, 3)] << 24)
+
+#define f_rl(bo, bi, n, k)                                      \
+    bo[n] = (u4byte)sbx_tab[byte(bi[n],0)] ^                    \
+        rotl(((u4byte)sbx_tab[byte(bi[(n + 1) & 3],1)]),  8) ^  \
+        rotl(((u4byte)sbx_tab[byte(bi[(n + 2) & 3],2)]), 16) ^  \
+        rotl(((u4byte)sbx_tab[byte(bi[(n + 3) & 3],3)]), 24) ^ *(k + n)
+
+#define i_rl(bo, bi, n, k)                                      \
+    bo[n] = (u4byte)isb_tab[byte(bi[n],0)] ^                    \
+        rotl(((u4byte)isb_tab[byte(bi[(n + 3) & 3],1)]),  8) ^  \
+        rotl(((u4byte)isb_tab[byte(bi[(n + 2) & 3],2)]), 16) ^  \
+        rotl(((u4byte)isb_tab[byte(bi[(n + 1) & 3],3)]), 24) ^ *(k + n)
+
+#endif
+
+void
+gen_tabs(void)
+{
+	u4byte  i, t;
+	u1byte  p, q;
+
+	/* log and power tables for GF(2**8) finite field with  */
+	/* 0x11b as modular polynomial - the simplest prmitive  */
+	/* root is 0x11, used here to generate the tables       */
+
+	for(i = 0,p = 1; i < 256; ++i) {
+		pow_tab[i] = (u1byte)p; log_tab[p] = (u1byte)i;
+
+		p = p ^ (p << 1) ^ (p & 0x80 ? 0x01b : 0);
+	}
+
+	log_tab[1] = 0; p = 1;
+
+	for(i = 0; i < 10; ++i) {
+		rco_tab[i] = p; 
+
+		p = (p << 1) ^ (p & 0x80 ? 0x1b : 0);
+	}
+
+	/* note that the affine byte transformation matrix in   */
+	/* rijndael specification is in big endian format with  */
+	/* bit 0 as the most significant bit. In the remainder  */
+	/* of the specification the bits are numbered from the  */
+	/* least significant end of a byte.                     */
+
+	for(i = 0; i < 256; ++i) {
+		p = (i ? pow_tab[255 - log_tab[i]] : 0); q = p; 
+		q = (q >> 7) | (q << 1); p ^= q; 
+		q = (q >> 7) | (q << 1); p ^= q; 
+		q = (q >> 7) | (q << 1); p ^= q; 
+		q = (q >> 7) | (q << 1); p ^= q ^ 0x63; 
+		sbx_tab[i] = (u1byte)p; isb_tab[p] = (u1byte)i;
+	}
+
+	for(i = 0; i < 256; ++i) {
+		p = sbx_tab[i]; 
+
+#ifdef  LARGE_TABLES        
+        
+		t = p; fl_tab[0][i] = t;
+		fl_tab[1][i] = rotl(t,  8);
+		fl_tab[2][i] = rotl(t, 16);
+		fl_tab[3][i] = rotl(t, 24);
+#endif
+		t = ((u4byte)ff_mult(2, p)) |
+			((u4byte)p <<  8) |
+			((u4byte)p << 16) |
+			((u4byte)ff_mult(3, p) << 24);
+        
+		ft_tab[0][i] = t;
+		ft_tab[1][i] = rotl(t,  8);
+		ft_tab[2][i] = rotl(t, 16);
+		ft_tab[3][i] = rotl(t, 24);
+
+		p = isb_tab[i]; 
+
+#ifdef  LARGE_TABLES        
+        
+		t = p; il_tab[0][i] = t; 
+		il_tab[1][i] = rotl(t,  8); 
+		il_tab[2][i] = rotl(t, 16); 
+		il_tab[3][i] = rotl(t, 24);
+#endif 
+		t = ((u4byte)ff_mult(14, p)) |
+			((u4byte)ff_mult( 9, p) <<  8) |
+			((u4byte)ff_mult(13, p) << 16) |
+			((u4byte)ff_mult(11, p) << 24);
+        
+		it_tab[0][i] = t; 
+		it_tab[1][i] = rotl(t,  8); 
+		it_tab[2][i] = rotl(t, 16); 
+		it_tab[3][i] = rotl(t, 24); 
+	}
+
+	tab_gen = 1;
+};
+
+#define star_x(x) (((x) & 0x7f7f7f7f) << 1) ^ ((((x) & 0x80808080) >> 7) * 0x1b)
+
+#define imix_col(y,x)       \
+    u   = star_x(x);        \
+    v   = star_x(u);        \
+    w   = star_x(v);        \
+    t   = w ^ (x);          \
+   (y)  = u ^ v ^ w;        \
+   (y) ^= rotr(u ^ t,  8) ^ \
+          rotr(v ^ t, 16) ^ \
+          rotr(t,24)
+
+/* initialise the key schedule from the user supplied key   */
+
+#define loop4(i)                                    \
+{   t = ls_box(rotr(t,  8)) ^ rco_tab[i];           \
+    t ^= e_key[4 * i];     e_key[4 * i + 4] = t;    \
+    t ^= e_key[4 * i + 1]; e_key[4 * i + 5] = t;    \
+    t ^= e_key[4 * i + 2]; e_key[4 * i + 6] = t;    \
+    t ^= e_key[4 * i + 3]; e_key[4 * i + 7] = t;    \
+}
+
+#define loop6(i)                                    \
+{   t = ls_box(rotr(t,  8)) ^ rco_tab[i];           \
+    t ^= e_key[6 * i];     e_key[6 * i + 6] = t;    \
+    t ^= e_key[6 * i + 1]; e_key[6 * i + 7] = t;    \
+    t ^= e_key[6 * i + 2]; e_key[6 * i + 8] = t;    \
+    t ^= e_key[6 * i + 3]; e_key[6 * i + 9] = t;    \
+    t ^= e_key[6 * i + 4]; e_key[6 * i + 10] = t;   \
+    t ^= e_key[6 * i + 5]; e_key[6 * i + 11] = t;   \
+}
+
+#define loop8(i)                                    \
+{   t = ls_box(rotr(t,  8)) ^ rco_tab[i];           \
+    t ^= e_key[8 * i];     e_key[8 * i + 8] = t;    \
+    t ^= e_key[8 * i + 1]; e_key[8 * i + 9] = t;    \
+    t ^= e_key[8 * i + 2]; e_key[8 * i + 10] = t;   \
+    t ^= e_key[8 * i + 3]; e_key[8 * i + 11] = t;   \
+    t  = e_key[8 * i + 4] ^ ls_box(t);              \
+    e_key[8 * i + 12] = t;                          \
+    t ^= e_key[8 * i + 5]; e_key[8 * i + 13] = t;   \
+    t ^= e_key[8 * i + 6]; e_key[8 * i + 14] = t;   \
+    t ^= e_key[8 * i + 7]; e_key[8 * i + 15] = t;   \
+}
+
+rijndael_ctx *
+rijndael_set_key(rijndael_ctx *ctx, const u4byte *in_key, const u4byte key_len,
+		 int encrypt)
+{  
+	u4byte  i, t, u, v, w;
+	u4byte *e_key = ctx->e_key;
+	u4byte *d_key = ctx->d_key;
+
+	ctx->decrypt = !encrypt;
+
+	if(!tab_gen)
+		gen_tabs();
+
+	ctx->k_len = (key_len + 31) / 32;
+
+	e_key[0] = in_key[0]; e_key[1] = in_key[1];
+	e_key[2] = in_key[2]; e_key[3] = in_key[3];
+	
+	switch(ctx->k_len) {
+        case 4: t = e_key[3];
+                for(i = 0; i < 10; ++i) 
+			loop4(i);
+                break;
+
+        case 6: e_key[4] = in_key[4]; t = e_key[5] = in_key[5];
+                for(i = 0; i < 8; ++i) 
+			loop6(i);
+                break;
+
+        case 8: e_key[4] = in_key[4]; e_key[5] = in_key[5];
+                e_key[6] = in_key[6]; t = e_key[7] = in_key[7];
+                for(i = 0; i < 7; ++i) 
+			loop8(i);
+                break;
+	}
+
+	if (!encrypt) {
+		d_key[0] = e_key[0]; d_key[1] = e_key[1];
+		d_key[2] = e_key[2]; d_key[3] = e_key[3];
+
+		for(i = 4; i < 4 * ctx->k_len + 24; ++i) {
+			imix_col(d_key[i], e_key[i]);
+		}
+	}
+
+	return ctx;
+};
+
+/* encrypt a block of text  */
+
+#define f_nround(bo, bi, k) \
+    f_rn(bo, bi, 0, k);     \
+    f_rn(bo, bi, 1, k);     \
+    f_rn(bo, bi, 2, k);     \
+    f_rn(bo, bi, 3, k);     \
+    k += 4
+
+#define f_lround(bo, bi, k) \
+    f_rl(bo, bi, 0, k);     \
+    f_rl(bo, bi, 1, k);     \
+    f_rl(bo, bi, 2, k);     \
+    f_rl(bo, bi, 3, k)
+
+void
+rijndael_encrypt(rijndael_ctx *ctx, const u4byte *in_blk, u4byte *out_blk)
+{   
+	u4byte k_len = ctx->k_len;
+	u4byte *e_key = ctx->e_key;
+	u4byte  b0[4], b1[4], *kp;
+
+	b0[0] = in_blk[0] ^ e_key[0]; b0[1] = in_blk[1] ^ e_key[1];
+	b0[2] = in_blk[2] ^ e_key[2]; b0[3] = in_blk[3] ^ e_key[3];
+
+	kp = e_key + 4;
+
+	if(k_len > 6) {
+		f_nround(b1, b0, kp); f_nround(b0, b1, kp);
+	}
+
+	if(k_len > 4) {
+		f_nround(b1, b0, kp); f_nround(b0, b1, kp);
+	}
+
+	f_nround(b1, b0, kp); f_nround(b0, b1, kp);
+	f_nround(b1, b0, kp); f_nround(b0, b1, kp);
+	f_nround(b1, b0, kp); f_nround(b0, b1, kp);
+	f_nround(b1, b0, kp); f_nround(b0, b1, kp);
+	f_nround(b1, b0, kp); f_lround(b0, b1, kp);
+
+	out_blk[0] = b0[0]; out_blk[1] = b0[1];
+	out_blk[2] = b0[2]; out_blk[3] = b0[3];
+};
+
+/* decrypt a block of text  */
+
+#define i_nround(bo, bi, k) \
+    i_rn(bo, bi, 0, k);     \
+    i_rn(bo, bi, 1, k);     \
+    i_rn(bo, bi, 2, k);     \
+    i_rn(bo, bi, 3, k);     \
+    k -= 4
+
+#define i_lround(bo, bi, k) \
+    i_rl(bo, bi, 0, k);     \
+    i_rl(bo, bi, 1, k);     \
+    i_rl(bo, bi, 2, k);     \
+    i_rl(bo, bi, 3, k)
+
+void
+rijndael_decrypt(rijndael_ctx *ctx, const u4byte *in_blk, u4byte *out_blk)
+{  
+	u4byte  b0[4], b1[4], *kp;
+	u4byte k_len = ctx->k_len;
+	u4byte *e_key = ctx->e_key;
+	u4byte *d_key = ctx->d_key;
+
+	b0[0] = in_blk[0] ^ e_key[4 * k_len + 24]; b0[1] = in_blk[1] ^ e_key[4 * k_len + 25];
+	b0[2] = in_blk[2] ^ e_key[4 * k_len + 26]; b0[3] = in_blk[3] ^ e_key[4 * k_len + 27];
+
+	kp = d_key + 4 * (k_len + 5);
+
+	if(k_len > 6) {
+		i_nround(b1, b0, kp); i_nround(b0, b1, kp);
+	}
+
+	if(k_len > 4) {
+		i_nround(b1, b0, kp); i_nround(b0, b1, kp);
+	}
+
+	i_nround(b1, b0, kp); i_nround(b0, b1, kp);
+	i_nround(b1, b0, kp); i_nround(b0, b1, kp);
+	i_nround(b1, b0, kp); i_nround(b0, b1, kp);
+	i_nround(b1, b0, kp); i_nround(b0, b1, kp);
+	i_nround(b1, b0, kp); i_lround(b0, b1, kp);
+
+	out_blk[0] = b0[0]; out_blk[1] = b0[1];
+	out_blk[2] = b0[2]; out_blk[3] = b0[3];
+};
diff --git a/sys/crypto/rijndael.h b/sys/crypto/rijndael.h
new file mode 100644
index 00000000000..c13f18c9590
--- /dev/null
+++ b/sys/crypto/rijndael.h
@@ -0,0 +1,31 @@
+#ifndef _RIJNDAEL_H_
+#define _RIJNDAEL_H_
+
+/* 1. Standard types for AES cryptography source code               */
+
+typedef u_int8_t   u1byte; /* an 8 bit unsigned character type */
+typedef u_int16_t  u2byte; /* a 16 bit unsigned integer type   */
+typedef u_int32_t  u4byte; /* a 32 bit unsigned integer type   */
+
+typedef int8_t     s1byte; /* an 8 bit signed character type   */
+typedef int16_t    s2byte; /* a 16 bit signed integer type     */
+typedef int32_t    s4byte; /* a 32 bit signed integer type     */
+
+typedef struct _rijndael_ctx {
+	u4byte  k_len;
+	int decrypt;
+	u4byte  e_key[64];
+	u4byte  d_key[64];
+} rijndael_ctx;
+
+
+/* 2. Standard interface for AES cryptographic routines             */
+
+/* These are all based on 32 bit unsigned values and will therefore */
+/* require endian conversions for big-endian architectures          */
+
+rijndael_ctx *rijndael_set_key  __P((rijndael_ctx *, const u4byte *, u4byte, int));
+void rijndael_encrypt __P((rijndael_ctx *, const u4byte *, u4byte *));
+void rijndael_decrypt __P((rijndael_ctx *, const u4byte *, u4byte *));
+
+#endif /* _RIJNDAEL_H_ */
diff --git a/sys/uvm/uvm.h b/sys/uvm/uvm.h
index 6cdb784715f..dde898ab7c2 100644
--- a/sys/uvm/uvm.h
+++ b/sys/uvm/uvm.h
@@ -57,6 +57,9 @@
 #include <uvm/uvm_pager.h>
 #include <uvm/uvm_pdaemon.h>
 #include <uvm/uvm_swap.h>
+#ifdef UVM_SWAP_ENCRYPT
+#include <uvm/uvm_swap_encrypt.h>
+#endif
 
 /*
  * pull in VM_NFREELIST
diff --git a/sys/uvm/uvm_swap.c b/sys/uvm/uvm_swap.c
index ae519285c46..00f8789b829 100644
--- a/sys/uvm/uvm_swap.c
+++ b/sys/uvm/uvm_swap.c
@@ -53,7 +53,7 @@
 
 #include <uvm/uvm.h>
 #ifdef UVM_SWAP_ENCRYPT
-#include <uvm/uvm_swap_encrypt.h>
+#include <sys/syslog.h>
 #endif
 
 #include <miscfs/specfs/specdev.h>
@@ -154,6 +154,9 @@ struct swapdev {
 	struct ucred		*swd_cred;	/* cred for file access */
 #endif
 #ifdef UVM_SWAP_ENCRYPT
+#define SWD_KEY_SHIFT		7		/* One key per 0.5 MByte */
+#define SWD_KEY(x,y)		&((x)->swd_keys[((y) - (x)->swd_drumoffset) >> SWD_KEY_SHIFT])
+
 #define SWD_DCRYPT_SHIFT	5
 #define SWD_DCRYPT_BITS		32
 #define SWD_DCRYPT_MASK		(SWD_DCRYPT_BITS - 1)
@@ -161,6 +164,8 @@ struct swapdev {
 #define SWD_DCRYPT_BIT(x)	((x) & SWD_DCRYPT_MASK)
 #define SWD_DCRYPT_SIZE(x)	(SWD_DCRYPT_OFF((x) + SWD_DCRYPT_MASK) * sizeof(u_int32_t))
 	u_int32_t		*swd_decrypt;	/* bitmap for decryption */
+	struct swap_key		*swd_keys;	/* keys for different parts */
+	int			swd_nkeys;	/* active keys */
 #endif
 };
 
@@ -390,6 +395,10 @@ uvm_swap_initcrypt(struct swapdev *sdp, int npages)
 	 */
 	sdp->swd_decrypt = malloc(SWD_DCRYPT_SIZE(npages), M_VMSWAP, M_WAITOK);
 	bzero(sdp->swd_decrypt, SWD_DCRYPT_SIZE(npages));
+	sdp->swd_keys = malloc((npages >> SWD_KEY_SHIFT) * sizeof(struct swap_key),
+			       M_VMSWAP, M_WAITOK);
+	bzero(sdp->swd_keys, (npages >> SWD_KEY_SHIFT) * sizeof(struct swap_key));
+	sdp->swd_nkeys = 0;
 }
 
 boolean_t
@@ -1247,8 +1256,11 @@ swap_off(p, sdp)
 	return ENODEV;
 
 #ifdef UVM_SWAP_ENCRYPT
-	if (sdp->swd_decrypt)
+	if (sdp->swd_decrypt) {
 		free(sdp->swd_decrypt);
+		bzero(sdp->swd_keys, (sdp->swd_npages >> SWD_KEY_SHIFT) * sizeof(struct swap_key));
+		free(sdp->swd_keys);
+	}
 #endif
 	extent_free(swapmap, sdp->swd_mapoffset, sdp->swd_mapsize, EX_WAITOK);
 	name = sdp->swd_ex->ex_name;
@@ -1826,6 +1838,20 @@ uvm_swap_free(startslot, nslots)
 	if (sdp->swd_npginuse < 0)
 		panic("uvm_swap_free: inuse < 0");
 #endif
+#ifdef UVM_SWAP_ENCRYPT
+	{
+		int i;
+		if (swap_encrypt_initalized) {
+			/* Dereference keys */
+			for (i = 0; i < nslots; i++)
+				if (uvm_swap_needdecrypt(sdp, startslot + i))
+					SWAP_KEY_PUT(sdp, SWD_KEY(sdp, startslot + i));
+
+			/* Mark range as not decrypt */
+			uvm_swap_markdecrypt(sdp, startslot, nslots, 0);
+		}
+	}
+#endif UVM_SWAP_ENCRYPT
 	simple_unlock(&uvm.swap_data_lock);
 }
 
@@ -1937,14 +1963,7 @@ uvm_swap_io(pps, startslot, npages, flags)
 		return (VM_PAGER_AGAIN);
 
 #ifdef UVM_SWAP_ENCRYPT
-	/* 
-	 * encrypt to swap
-	 */
 	if ((flags & B_READ) == 0) {
-		int i, opages;
-		caddr_t src, dst;
-		u_int64_t block;
-
 		/*
 		 * Check if we need to do swap encryption on old pages.
 		 * Later we need a different scheme, that swap encrypts
@@ -1953,8 +1972,31 @@ uvm_swap_io(pps, startslot, npages, flags)
 		 * in the cluster, and avoid the memory overheard in 
 		 * swapping.
 		 */
-		if (!uvm_doswapencrypt)
-				goto noswapencrypt;
+		if (uvm_doswapencrypt)
+			encrypt = 1;
+	}
+
+	if (swap_encrypt_initalized  || encrypt) { 
+		/*
+		 * we need to know the swap device that we are swapping to/from
+		 * to see if the pages need to be marked for decryption or
+		 * actually need to be decrypted.
+		 * XXX - does this information stay the same over the whole 
+		 * execution of this function?
+		 */
+		simple_lock(&uvm.swap_data_lock);
+		sdp = swapdrum_getsdp(startslot);
+		simple_unlock(&uvm.swap_data_lock);
+	}
+
+	/* 
+	 * encrypt to swap
+	 */
+	if ((flags & B_READ) == 0 && encrypt) {
+		int i, opages;
+		caddr_t src, dst;
+		struct swap_key *key;
+		u_int64_t block;
 
 		if (!uvm_swap_allocpages(tpps, npages)) {
 			uvm_pagermapout(kva, npages);
@@ -1972,9 +2014,12 @@ uvm_swap_io(pps, startslot, npages, flags)
 		dst = (caddr_t) dstkva;
 		block = startblk;
 		for (i = 0; i < npages; i++) {
+			key = SWD_KEY(sdp, startslot + i);
+			SWAP_KEY_GET(sdp, key);	/* add reference */
+
 			/* mark for async writes */
 			tpps[i]->pqflags |= PQ_ENCRYPT;
-			swap_encrypt(src, dst, block, 1 << PAGE_SHIFT);
+			swap_encrypt(key, src, dst, block, 1 << PAGE_SHIFT);
 			src += 1 << PAGE_SHIFT;
 			dst += 1 << PAGE_SHIFT;
 			block += btodb(1 << PAGE_SHIFT);
@@ -1988,9 +2033,6 @@ uvm_swap_io(pps, startslot, npages, flags)
 				      PGO_PDFREECLUST, 0);
 
 		kva = dstkva;
-
-		encrypt = 1;
-	noswapencrypt:
 	}
 #endif /* UVM_SWAP_ENCRYPT */
 
@@ -2011,7 +2053,12 @@ uvm_swap_io(pps, startslot, npages, flags)
 	if (sbp == NULL) {
 #ifdef UVM_SWAP_ENCRYPT
 		if ((flags & B_READ) == 0 && encrypt) {
+			int i;
+
 			/* swap encrypt needs cleanup */
+			for (i = 0; i < npages; i++)
+				SWAP_KEY_PUT(sdp, SWD_KEY(sdp, startslot + i));
+
 			uvm_pagermapout(kva, npages);
 			uvm_swap_freepages(tpps, npages);
 		}
@@ -2050,21 +2097,6 @@ uvm_swap_io(pps, startslot, npages, flags)
 		bp->b_dev = swapdev_vp->v_rdev;
 	bp->b_bcount = npages << PAGE_SHIFT;
 
-#ifdef UVM_SWAP_ENCRYPT
-	if (swap_encrypt_initalized) { 
-		/*
-		 * we need to know the swap device that we are swapping to/from
-		 * to see if the pages need to be marked for decryption or
-		 * actually need to be decrypted.
-		 * XXX - does this information stay the same over the whole 
-		 * execution of this function?
-		 */
-		simple_lock(&uvm.swap_data_lock);
-		sdp = swapdrum_getsdp(startslot);
-		simple_unlock(&uvm.swap_data_lock);
-	}
-#endif
-
 	/* 
 	 * for pageouts we must set "dirtyoff" [NFS client code needs it].
 	 * and we bump v_numoutput (counter of number of active outputs).
@@ -2121,11 +2153,15 @@ uvm_swap_io(pps, startslot, npages, flags)
 		int i;
 		caddr_t data = bp->b_data;
 		u_int64_t block = startblk;
+		struct swap_key *key = NULL;
+
 		for (i = 0; i < npages; i++) {
 			/* Check if we need to decrypt */
-			if (uvm_swap_needdecrypt(sdp, startslot + i))
-				swap_decrypt(data, data, block,
+			if (uvm_swap_needdecrypt(sdp, startslot + i)) {
+				key = SWD_KEY(sdp, startslot + i);
+				swap_decrypt(key, data, data, block,
 					     1 << PAGE_SHIFT);
+			}
 			data += 1 << PAGE_SHIFT;
 			block += btodb(1 << PAGE_SHIFT);
 		}
diff --git a/sys/uvm/uvm_swap_encrypt.c b/sys/uvm/uvm_swap_encrypt.c
index c995bef8240..97fefd244f2 100644
--- a/sys/uvm/uvm_swap_encrypt.c
+++ b/sys/uvm/uvm_swap_encrypt.c
@@ -30,34 +30,47 @@
 
 #include <sys/param.h>
 #include <sys/systm.h>
+#include <sys/kernel.h>
+#include <sys/malloc.h>
+#include <sys/time.h>
 #include <dev/rndvar.h>
-#include <crypto/blf.h>
+#include <crypto/rijndael.h>
 
-#include <uvm/uvm_swap_encrypt.h>
+#include <vm/vm.h>
+#include <vm/vm_conf.h>
 
-blf_ctx swap_key;
+#include <uvm/uvm.h>
+
+struct swap_key *kcur = NULL;
+rijndael_ctx swap_key;
 
 int uvm_doswapencrypt = 0;
-int swap_encrypt_initalized = 0;
+u_int uvm_swpkeyscreated = 0;
+u_int uvm_swpkeysdeleted = 0;
 
-/*
- * Initalize the key from the kernel random number generator.  This is
- * done once on startup.
- */
+int swap_encrypt_initalized = 0;
 
 void
-swap_encrypt_init(caddr_t data, size_t len)
+swap_key_create(struct swap_key *key)
 {
 	int i;
-	u_int32_t *key = (u_int32_t *)data;
+	u_int32_t *p = key->key;
 
-	if (swap_encrypt_initalized)
-		return;
+	key->refcount = 0;
+	for (i = 0; i < sizeof(key->key) / sizeof(u_int32_t); i++)
+		*p++ = arc4random();
+
+	uvm_swpkeyscreated++;
+}
 
-	for (i = 0; i < len / sizeof(u_int32_t); i++)
-		*key++ = arc4random();
+void
+swap_key_delete(struct swap_key *key)
+{
+	/* Make sure that this key gets removed if we just used it */
+	swap_key_cleanup(key);
 
-	swap_encrypt_initalized = 1;
+	bzero(key, sizeof(*key));
+	uvm_swpkeysdeleted++;
 }
 
 /*
@@ -66,34 +79,43 @@ swap_encrypt_init(caddr_t data, size_t len)
  */
 
 void
-swap_encrypt(caddr_t src, caddr_t dst, u_int64_t block, size_t count)
+swap_encrypt(struct swap_key *key, caddr_t src, caddr_t dst,
+	     u_int64_t block, size_t count)
 {
-  u_int32_t *dsrc = (u_int32_t *)src;
-  u_int32_t *ddst = (u_int32_t *)dst;
-  u_int32_t iv[2];
-  u_int32_t iv1, iv2;
-
-  if (!swap_encrypt_initalized)
-	swap_encrypt_init((caddr_t)&swap_key, sizeof(swap_key));
-
-  count /= sizeof(u_int32_t);
-
-  iv[0] = block >> 32; iv[1] = block;
-  Blowfish_encipher(&swap_key, iv);
-  iv1 = iv[0]; iv2 = iv[1];
-  for (; count > 0; count -= 2) {
-    ddst[0] = dsrc[0] ^ iv1;
-    ddst[1] = dsrc[1] ^ iv2;
-    /*
-     * Do not worry about endianess, it only needs to decrypt on this machine
-     */
-    Blowfish_encipher(&swap_key, ddst);
-    iv1 = ddst[0];
-    iv2 = ddst[1];
-
-    dsrc += 2;
-    ddst += 2;
-  }
+	u_int32_t *dsrc = (u_int32_t *)src;
+	u_int32_t *ddst = (u_int32_t *)dst;
+	u_int32_t iv[4];
+	u_int32_t iv1, iv2, iv3, iv4;
+
+	if (!swap_encrypt_initalized)
+		swap_encrypt_initalized = 1;
+
+	swap_key_prepare(key, 1);
+
+	count /= sizeof(u_int32_t);
+
+	iv[0] = block >> 32; iv[1] = block; iv[2] = ~iv[0]; iv[3] = ~iv[1];
+	rijndael_encrypt(&swap_key, iv, iv); 
+	iv1 = iv[0]; iv2 = iv[1]; iv3 = iv[2]; iv4 = iv[3];
+
+	for (; count > 0; count -= 4) {
+		ddst[0] = dsrc[0] ^ iv1;
+		ddst[1] = dsrc[1] ^ iv2;
+		ddst[2] = dsrc[2] ^ iv3;
+		ddst[3] = dsrc[3] ^ iv4;
+		/*
+		 * Do not worry about endianess, it only needs to decrypt
+		 * on this machine
+		 */
+		rijndael_encrypt(&swap_key, ddst, ddst);
+		iv1 = ddst[0];
+		iv2 = ddst[1];
+		iv3 = ddst[2];
+		iv4 = ddst[3];
+
+		dsrc += 4;
+		ddst += 4;
+	}
 }
 
 /*
@@ -102,32 +124,76 @@ swap_encrypt(caddr_t src, caddr_t dst, u_int64_t block, size_t count)
  */
 
 void
-swap_decrypt(caddr_t src, caddr_t dst, u_int64_t block, size_t count)
+swap_decrypt(struct swap_key *key, caddr_t src, caddr_t dst,
+	     u_int64_t block, size_t count)
+{
+	u_int32_t *dsrc = (u_int32_t *)src;
+	u_int32_t *ddst = (u_int32_t *)dst;
+	u_int32_t iv[4];
+	u_int32_t iv1, iv2, iv3, iv4, niv1, niv2, niv3, niv4;
+
+	if (!swap_encrypt_initalized)
+		panic("swap_decrypt: key not initalized");
+
+	swap_key_prepare(key, 0);
+
+	count /= sizeof(u_int32_t);
+
+	iv[0] = block >> 32; iv[1] = block; iv[2] = ~iv[0]; iv[3] = ~iv[1];
+	rijndael_encrypt(&swap_key, iv, iv); 
+	iv1 = iv[0]; iv2 = iv[1]; iv3 = iv[2]; iv4 = iv[3];
+
+	for (; count > 0; count -= 4) {
+		ddst[0] = niv1 = dsrc[0];
+		ddst[1] = niv2 = dsrc[1];
+		ddst[2] = niv3 = dsrc[2];
+		ddst[3] = niv4 = dsrc[3];
+		rijndael_decrypt(&swap_key, ddst, ddst);
+		ddst[0] ^= iv1;
+		ddst[1] ^= iv2;
+		ddst[2] ^= iv3;
+		ddst[3] ^= iv4;
+
+		iv1 = niv1;
+		iv2 = niv2;
+		iv3 = niv3;
+		iv4 = niv4;
+
+		dsrc += 4;
+		ddst += 4;
+	}
+}
+
+void
+swap_key_prepare(struct swap_key *key, int encrypt)
 {
-  u_int32_t *dsrc = (u_int32_t *)src;
-  u_int32_t *ddst = (u_int32_t *)dst;
-  u_int32_t iv[2];
-  u_int32_t iv1, iv2, niv1, niv2;
-
-  if (!swap_encrypt_initalized)
-    panic("swap_decrypt: key not initalized");
-
-  count /= sizeof(u_int32_t);
-
-  iv[0] = block >> 32; iv[1] = block;
-  Blowfish_encipher(&swap_key, iv);
-  iv1 = iv[0]; iv2 = iv[1];
-  for (; count > 0; count -= 2) {
-    ddst[0] = niv1 = dsrc[0];
-    ddst[1] = niv2 = dsrc[1];
-    Blowfish_decipher(&swap_key, ddst);
-    ddst[0] ^= iv1;
-    ddst[1] ^= iv2;
-
-    iv1 = niv1;
-    iv2 = niv2;
-
-    dsrc += 2;
-    ddst += 2;
-  }
+	/* Check if we have prepared for this key already,
+	 * if we only have the encryption schedule, we have
+	 * to recompute ang get the decryption schedule also
+	 */
+	if (kcur == key && (encrypt || swap_key.decrypt))
+		return;
+
+	rijndael_set_key(&swap_key, key->key,
+			 sizeof(key->key) * 8,
+			 encrypt);
+
+	kcur = key;
+}
+
+/*
+ * Make sure that a specific key is no longer available.
+ */
+
+void
+swap_key_cleanup(struct swap_key *key)
+{
+	/* Check if we have a key */
+	if (kcur == NULL || kcur != key)
+		return;
+
+	/* Zero out the subkeys */
+	bzero(&swap_key, sizeof(swap_key));
+
+	kcur = NULL;
 }
diff --git a/sys/uvm/uvm_swap_encrypt.h b/sys/uvm/uvm_swap_encrypt.h
index 1eb03550158..842cfa5b381 100644
--- a/sys/uvm/uvm_swap_encrypt.h
+++ b/sys/uvm/uvm_swap_encrypt.h
@@ -31,11 +31,36 @@
 #ifndef _UVM_SWAP_ENCRYPT_H
 #define _UVM_SWAP_ENCRYPT_H
 
-void swap_encrypt_init __P((caddr_t, size_t));
-void swap_encrypt __P((caddr_t, caddr_t, u_int64_t, size_t));
-void swap_decrypt __P((caddr_t, caddr_t, u_int64_t, size_t));
+#define SWAP_KEY_EXPIRE (120 /*60 * 60*/)	/* time after that keys expire */
+#define SWAP_KEY_SIZE	4		/* 128-bit keys */
+
+struct swap_key {
+	u_int32_t key[SWAP_KEY_SIZE];	/* secret key for swap range */
+	u_int16_t refcount;		/* pages that still need it */
+};
+
+void swap_encrypt __P((struct swap_key *,caddr_t, caddr_t, u_int64_t, size_t));
+void swap_decrypt __P((struct swap_key *,caddr_t, caddr_t, u_int64_t, size_t));
+
+void swap_key_cleanup __P((struct swap_key *));
+void swap_key_prepare __P((struct swap_key *, int));
+
+#define SWAP_KEY_GET(s,x) do { if ((x)->refcount == 0) {\
+					swap_key_create(x); \
+			       } \
+			       (x)->refcount++; } while(0);
+#define SWAP_KEY_PUT(s,x) do { (x)->refcount--; \
+			       if ((x)->refcount == 0) { \
+					swap_key_delete(x); \
+			       } \
+			     } while(0);
+
+void swap_key_create __P((struct swap_key *));
+void swap_key_delete __P((struct swap_key *));
 
 extern int uvm_doswapencrypt;		/* swapencrypt enabled/disabled */
+extern int uvm_swprekeyprint;
+extern u_int uvm_swpkeyexpire;		/* expiry time for keys (tR) */
 extern int swap_encrypt_initalized;
 
 #endif /* _UVM_SWAP_ENCRYPT_H */