Merge branch 'mainline' into upstream-linus

6 files changed, 466 insertions, 26 deletions

diff --git a/arch/x86_64/crypto/Makefile b/arch/x86_64/crypto/Makefile
index 426d20f4b72e..15b538a8b7f7 100644
--- a/arch/x86_64/crypto/Makefile
+++ b/arch/x86_64/crypto/Makefile
@@ -5,5 +5,8 @@
 # 
 
 obj-$(CONFIG_CRYPTO_AES_X86_64) += aes-x86_64.o
+obj-$(CONFIG_CRYPTO_TWOFISH_X86_64) += twofish-x86_64.o
 
 aes-x86_64-y := aes-x86_64-asm.o aes.o
+twofish-x86_64-y := twofish-x86_64-asm.o twofish.o
+
diff --git a/arch/x86_64/crypto/aes.c b/arch/x86_64/crypto/aes.c
index 68866fab37aa..5cdb13ea5cc2 100644
--- a/arch/x86_64/crypto/aes.c
+++ b/arch/x86_64/crypto/aes.c
@@ -228,13 +228,14 @@ static void __init gen_tabs(void)
 }
 
 static int aes_set_key(struct crypto_tfm *tfm, const u8 *in_key,
-                       unsigned int key_len, u32 *flags)
+                       unsigned int key_len)
 {
         struct aes_ctx *ctx = crypto_tfm_ctx(tfm);
         const __le32 *key = (const __le32 *)in_key;
+        u32 *flags = &tfm->crt_flags;
         u32 i, j, t, u, v, w;
 
-        if (key_len != 16 && key_len != 24 && key_len != 32) {
+        if (key_len % 8) {
                 *flags |= CRYPTO_TFM_RES_BAD_KEY_LEN;
                 return -EINVAL;
         }
diff --git a/arch/x86_64/crypto/twofish-x86_64-asm.S b/arch/x86_64/crypto/twofish-x86_64-asm.S
new file mode 100644
index 000000000000..35974a586615
--- /dev/null
+++ b/arch/x86_64/crypto/twofish-x86_64-asm.S
@@ -0,0 +1,324 @@
+/***************************************************************************
+*   Copyright (C) 2006 by Joachim Fritschi, <jfritschi@freenet.de>        *
+*                                                                         *
+*   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.                                   *
+*                                                                         *
+*   This program is distributed in the hope that it will be useful,       *
+*   but WITHOUT ANY WARRANTY; without even the implied warranty of        *
+*   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the         *
+*   GNU General Public License for more details.                          *
+*                                                                         *
+*   You should have received a copy of the GNU General Public License     *
+*   along with this program; if not, write to the                         *
+*   Free Software Foundation, Inc.,                                       *
+*   59 Temple Place - Suite 330, Boston, MA  02111-1307, USA.             *
+***************************************************************************/
+
+.file "twofish-x86_64-asm.S"
+.text
+
+#include <asm/asm-offsets.h>
+
+#define a_offset        0
+#define b_offset        4
+#define c_offset        8
+#define d_offset        12
+
+/* Structure of the crypto context struct*/
+
+#define s0      0       /* S0 Array 256 Words each */
+#define s1      1024    /* S1 Array */
+#define s2      2048    /* S2 Array */
+#define s3      3072    /* S3 Array */
+#define w       4096    /* 8 whitening keys (word) */
+#define k       4128    /* key 1-32 ( word ) */
+
+/* define a few register aliases to allow macro substitution */
+
+#define R0     %rax
+#define R0D    %eax
+#define R0B    %al
+#define R0H    %ah
+
+#define R1     %rbx
+#define R1D    %ebx
+#define R1B    %bl
+#define R1H    %bh
+
+#define R2     %rcx
+#define R2D    %ecx
+#define R2B    %cl
+#define R2H    %ch
+
+#define R3     %rdx
+#define R3D    %edx
+#define R3B    %dl
+#define R3H    %dh
+
+
+/* performs input whitening */
+#define input_whitening(src,context,offset)\
+        xor     w+offset(context),      src;
+
+/* performs input whitening */
+#define output_whitening(src,context,offset)\
+        xor     w+16+offset(context),   src;
+
+
+/*
+ * a input register containing a (rotated 16)
+ * b input register containing b
+ * c input register containing c
+ * d input register containing d (already rol $1)
+ * operations on a and b are interleaved to increase performance
+ */
+#define encrypt_round(a,b,c,d,round)\
+        movzx   b ## B,         %edi;\
+        mov     s1(%r11,%rdi,4),%r8d;\
+        movzx   a ## B,         %edi;\
+        mov     s2(%r11,%rdi,4),%r9d;\
+        movzx   b ## H,         %edi;\
+        ror     $16,            b ## D;\
+        xor     s2(%r11,%rdi,4),%r8d;\
+        movzx   a ## H,         %edi;\
+        ror     $16,            a ## D;\
+        xor     s3(%r11,%rdi,4),%r9d;\
+        movzx   b ## B,         %edi;\
+        xor     s3(%r11,%rdi,4),%r8d;\
+        movzx   a ## B,         %edi;\
+        xor     (%r11,%rdi,4),  %r9d;\
+        movzx   b ## H,         %edi;\
+        ror     $15,            b ## D;\
+        xor     (%r11,%rdi,4),  %r8d;\
+        movzx   a ## H,         %edi;\
+        xor     s1(%r11,%rdi,4),%r9d;\
+        add     %r8d,           %r9d;\
+        add     %r9d,           %r8d;\
+        add     k+round(%r11),  %r9d;\
+        xor     %r9d,           c ## D;\
+        rol     $15,            c ## D;\
+        add     k+4+round(%r11),%r8d;\
+        xor     %r8d,           d ## D;
+
+/*
+ * a input register containing a(rotated 16)
+ * b input register containing b
+ * c input register containing c
+ * d input register containing d (already rol $1)
+ * operations on a and b are interleaved to increase performance
+ * during the round a and b are prepared for the output whitening
+ */
+#define encrypt_last_round(a,b,c,d,round)\
+        mov     b ## D,         %r10d;\
+        shl     $32,            %r10;\
+        movzx   b ## B,         %edi;\
+        mov     s1(%r11,%rdi,4),%r8d;\
+        movzx   a ## B,         %edi;\
+        mov     s2(%r11,%rdi,4),%r9d;\
+        movzx   b ## H,         %edi;\
+        ror     $16,            b ## D;\
+        xor     s2(%r11,%rdi,4),%r8d;\
+        movzx   a ## H,         %edi;\
+        ror     $16,            a ## D;\
+        xor     s3(%r11,%rdi,4),%r9d;\
+        movzx   b ## B,         %edi;\
+        xor     s3(%r11,%rdi,4),%r8d;\
+        movzx   a ## B,         %edi;\
+        xor     (%r11,%rdi,4),  %r9d;\
+        xor     a,              %r10;\
+        movzx   b ## H,         %edi;\
+        xor     (%r11,%rdi,4),  %r8d;\
+        movzx   a ## H,         %edi;\
+        xor     s1(%r11,%rdi,4),%r9d;\
+        add     %r8d,           %r9d;\
+        add     %r9d,           %r8d;\
+        add     k+round(%r11),  %r9d;\
+        xor     %r9d,           c ## D;\
+        ror     $1,             c ## D;\
+        add     k+4+round(%r11),%r8d;\
+        xor     %r8d,           d ## D
+
+/*
+ * a input register containing a
+ * b input register containing b (rotated 16)
+ * c input register containing c (already rol $1)
+ * d input register containing d
+ * operations on a and b are interleaved to increase performance
+ */
+#define decrypt_round(a,b,c,d,round)\
+        movzx   a ## B,         %edi;\
+        mov     (%r11,%rdi,4),  %r9d;\
+        movzx   b ## B,         %edi;\
+        mov     s3(%r11,%rdi,4),%r8d;\
+        movzx   a ## H,         %edi;\
+        ror     $16,            a ## D;\
+        xor     s1(%r11,%rdi,4),%r9d;\
+        movzx   b ## H,         %edi;\
+        ror     $16,            b ## D;\
+        xor     (%r11,%rdi,4),  %r8d;\
+        movzx   a ## B,         %edi;\
+        xor     s2(%r11,%rdi,4),%r9d;\
+        movzx   b ## B,         %edi;\
+        xor     s1(%r11,%rdi,4),%r8d;\
+        movzx   a ## H,         %edi;\
+        ror     $15,            a ## D;\
+        xor     s3(%r11,%rdi,4),%r9d;\
+        movzx   b ## H,         %edi;\
+        xor     s2(%r11,%rdi,4),%r8d;\
+        add     %r8d,           %r9d;\
+        add     %r9d,           %r8d;\
+        add     k+round(%r11),  %r9d;\
+        xor     %r9d,           c ## D;\
+        add     k+4+round(%r11),%r8d;\
+        xor     %r8d,           d ## D;\
+        rol     $15,            d ## D;
+
+/*
+ * a input register containing a
+ * b input register containing b
+ * c input register containing c (already rol $1)
+ * d input register containing d
+ * operations on a and b are interleaved to increase performance
+ * during the round a and b are prepared for the output whitening
+ */
+#define decrypt_last_round(a,b,c,d,round)\
+        movzx   a ## B,         %edi;\
+        mov     (%r11,%rdi,4),  %r9d;\
+        movzx   b ## B,         %edi;\
+        mov     s3(%r11,%rdi,4),%r8d;\
+        movzx   b ## H,         %edi;\
+        ror     $16,            b ## D;\
+        xor     (%r11,%rdi,4),  %r8d;\
+        movzx   a ## H,         %edi;\
+        mov     b ## D,         %r10d;\
+        shl     $32,            %r10;\
+        xor     a,              %r10;\
+        ror     $16,            a ## D;\
+        xor     s1(%r11,%rdi,4),%r9d;\
+        movzx   b ## B,         %edi;\
+        xor     s1(%r11,%rdi,4),%r8d;\
+        movzx   a ## B,         %edi;\
+        xor     s2(%r11,%rdi,4),%r9d;\
+        movzx   b ## H,         %edi;\
+        xor     s2(%r11,%rdi,4),%r8d;\
+        movzx   a ## H,         %edi;\
+        xor     s3(%r11,%rdi,4),%r9d;\
+        add     %r8d,           %r9d;\
+        add     %r9d,           %r8d;\
+        add     k+round(%r11),  %r9d;\
+        xor     %r9d,           c ## D;\
+        add     k+4+round(%r11),%r8d;\
+        xor     %r8d,           d ## D;\
+        ror     $1,             d ## D;
+
+.align 8
+.global twofish_enc_blk
+.global twofish_dec_blk
+
+twofish_enc_blk:
+        pushq    R1
+
+        /* %rdi contains the crypto tfm adress */
+        /* %rsi contains the output adress */
+        /* %rdx contains the input adress */
+        add     $crypto_tfm_ctx_offset, %rdi    /* set ctx adress */
+        /* ctx adress is moved to free one non-rex register
+        as target for the 8bit high operations */
+        mov     %rdi,           %r11
+
+        movq    (R3),   R1
+        movq    8(R3),  R3
+        input_whitening(R1,%r11,a_offset)
+        input_whitening(R3,%r11,c_offset)
+        mov     R1D,    R0D
+        rol     $16,    R0D
+        shr     $32,    R1
+        mov     R3D,    R2D
+        shr     $32,    R3
+        rol     $1,     R3D
+
+        encrypt_round(R0,R1,R2,R3,0);
+        encrypt_round(R2,R3,R0,R1,8);
+        encrypt_round(R0,R1,R2,R3,2*8);
+        encrypt_round(R2,R3,R0,R1,3*8);
+        encrypt_round(R0,R1,R2,R3,4*8);
+        encrypt_round(R2,R3,R0,R1,5*8);
+        encrypt_round(R0,R1,R2,R3,6*8);
+        encrypt_round(R2,R3,R0,R1,7*8);
+        encrypt_round(R0,R1,R2,R3,8*8);
+        encrypt_round(R2,R3,R0,R1,9*8);
+        encrypt_round(R0,R1,R2,R3,10*8);
+        encrypt_round(R2,R3,R0,R1,11*8);
+        encrypt_round(R0,R1,R2,R3,12*8);
+        encrypt_round(R2,R3,R0,R1,13*8);
+        encrypt_round(R0,R1,R2,R3,14*8);
+        encrypt_last_round(R2,R3,R0,R1,15*8);
+
+
+        output_whitening(%r10,%r11,a_offset)
+        movq    %r10,   (%rsi)
+
+        shl     $32,    R1
+        xor     R0,     R1
+
+        output_whitening(R1,%r11,c_offset)
+        movq    R1,     8(%rsi)
+
+        popq    R1
+        movq    $1,%rax
+        ret
+
+twofish_dec_blk:
+        pushq    R1
+
+        /* %rdi contains the crypto tfm adress */
+        /* %rsi contains the output adress */
+        /* %rdx contains the input adress */
+        add     $crypto_tfm_ctx_offset, %rdi    /* set ctx adress */
+        /* ctx adress is moved to free one non-rex register
+        as target for the 8bit high operations */
+        mov     %rdi,           %r11
+
+        movq    (R3),   R1
+        movq    8(R3),  R3
+        output_whitening(R1,%r11,a_offset)
+        output_whitening(R3,%r11,c_offset)
+        mov     R1D,    R0D
+        shr     $32,    R1
+        rol     $16,    R1D
+        mov     R3D,    R2D
+        shr     $32,    R3
+        rol     $1,     R2D
+
+        decrypt_round(R0,R1,R2,R3,15*8);
+        decrypt_round(R2,R3,R0,R1,14*8);
+        decrypt_round(R0,R1,R2,R3,13*8);
+        decrypt_round(R2,R3,R0,R1,12*8);
+        decrypt_round(R0,R1,R2,R3,11*8);
+        decrypt_round(R2,R3,R0,R1,10*8);
+        decrypt_round(R0,R1,R2,R3,9*8);
+        decrypt_round(R2,R3,R0,R1,8*8);
+        decrypt_round(R0,R1,R2,R3,7*8);
+        decrypt_round(R2,R3,R0,R1,6*8);
+        decrypt_round(R0,R1,R2,R3,5*8);
+        decrypt_round(R2,R3,R0,R1,4*8);
+        decrypt_round(R0,R1,R2,R3,3*8);
+        decrypt_round(R2,R3,R0,R1,2*8);
+        decrypt_round(R0,R1,R2,R3,1*8);
+        decrypt_last_round(R2,R3,R0,R1,0);
+
+        input_whitening(%r10,%r11,a_offset)
+        movq    %r10,   (%rsi)
+
+        shl     $32,    R1
+        xor     R0,     R1
+
+        input_whitening(R1,%r11,c_offset)
+        movq    R1,     8(%rsi)
+
+        popq    R1
+        movq    $1,%rax
+        ret
diff --git a/arch/x86_64/crypto/twofish.c b/arch/x86_64/crypto/twofish.c
new file mode 100644
index 000000000000..182d91d5cfb9
--- /dev/null
+++ b/arch/x86_64/crypto/twofish.c
@@ -0,0 +1,97 @@
+/*
+ * Glue Code for optimized x86_64 assembler version of TWOFISH
+ *
+ * Originally Twofish for GPG
+ * By Matthew Skala <mskala@ansuz.sooke.bc.ca>, July 26, 1998
+ * 256-bit key length added March 20, 1999
+ * Some modifications to reduce the text size by Werner Koch, April, 1998
+ * Ported to the kerneli patch by Marc Mutz <Marc@Mutz.com>
+ * Ported to CryptoAPI by Colin Slater <hoho@tacomeat.net>
+ *
+ * The original author has disclaimed all copyright interest in this
+ * code and thus put it in the public domain. The subsequent authors
+ * have put this under the GNU General Public License.
+ *
+ * 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.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307
+ * USA
+ *
+ * This code is a "clean room" implementation, written from the paper
+ * _Twofish: A 128-Bit Block Cipher_ by Bruce Schneier, John Kelsey,
+ * Doug Whiting, David Wagner, Chris Hall, and Niels Ferguson, available
+ * through http://www.counterpane.com/twofish.html
+ *
+ * For background information on multiplication in finite fields, used for
+ * the matrix operations in the key schedule, see the book _Contemporary
+ * Abstract Algebra_ by Joseph A. Gallian, especially chapter 22 in the
+ * Third Edition.
+ */
+
+#include <crypto/twofish.h>
+#include <linux/crypto.h>
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/types.h>
+
+asmlinkage void twofish_enc_blk(struct crypto_tfm *tfm, u8 *dst, const u8 *src);
+asmlinkage void twofish_dec_blk(struct crypto_tfm *tfm, u8 *dst, const u8 *src);
+
+static void twofish_encrypt(struct crypto_tfm *tfm, u8 *dst, const u8 *src)
+{
+        twofish_enc_blk(tfm, dst, src);
+}
+
+static void twofish_decrypt(struct crypto_tfm *tfm, u8 *dst, const u8 *src)
+{
+        twofish_dec_blk(tfm, dst, src);
+}
+
+static struct crypto_alg alg = {
+        .cra_name               =       "twofish",
+        .cra_driver_name        =       "twofish-x86_64",
+        .cra_priority           =       200,
+        .cra_flags              =       CRYPTO_ALG_TYPE_CIPHER,
+        .cra_blocksize          =       TF_BLOCK_SIZE,
+        .cra_ctxsize            =       sizeof(struct twofish_ctx),
+        .cra_alignmask          =       3,
+        .cra_module             =       THIS_MODULE,
+        .cra_list               =       LIST_HEAD_INIT(alg.cra_list),
+        .cra_u                  =       {
+                .cipher = {
+                        .cia_min_keysize        =       TF_MIN_KEY_SIZE,
+                        .cia_max_keysize        =       TF_MAX_KEY_SIZE,
+                        .cia_setkey             =       twofish_setkey,
+                        .cia_encrypt            =       twofish_encrypt,
+                        .cia_decrypt            =       twofish_decrypt
+                }
+        }
+};
+
+static int __init init(void)
+{
+        return crypto_register_alg(&alg);
+}
+
+static void __exit fini(void)
+{
+        crypto_unregister_alg(&alg);
+}
+
+module_init(init);
+module_exit(fini);
+
+MODULE_LICENSE("GPL");
+MODULE_DESCRIPTION ("Twofish Cipher Algorithm, x86_64 asm optimized");
+MODULE_ALIAS("twofish");
diff --git a/arch/x86_64/kernel/e820.c b/arch/x86_64/kernel/e820.c
index 764bf23c7103..d6d7f731f6f0 100644
--- a/arch/x86_64/kernel/e820.c
+++ b/arch/x86_64/kernel/e820.c
@@ -108,6 +108,35 @@ e820_any_mapped(unsigned long start, unsigned long end, unsigned type)
         return 0;
 }
 
+/*
+ * This function checks if the entire range <start,end> is mapped with type.
+ *
+ * Note: this function only works correct if the e820 table is sorted and
+ * not-overlapping, which is the case
+ */
+int __init e820_all_mapped(unsigned long start, unsigned long end, unsigned type)
+{
+        int i;
+        for (i = 0; i < e820.nr_map; i++) {
+                struct e820entry *ei = &e820.map[i];
+                if (type && ei->type != type)
+                        continue;
+                /* is the region (part) in overlap with the current region ?*/
+                if (ei->addr >= end || ei->addr + ei->size <= start)
+                        continue;
+
+                /* if the region is at the beginning of <start,end> we move
+                 * start to the end of the region since it's ok until there
+                 */
+                if (ei->addr <= start)
+                        start = ei->addr + ei->size;
+                /* if start is now at or beyond end, we're done, full coverage */
+                if (start >= end)
+                        return 1; /* we're done */
+        }
+        return 0;
+}
+
 /* 
  * Find a free area in a specific range. 
  */ 
diff --git a/arch/x86_64/pci/mmconfig.c b/arch/x86_64/pci/mmconfig.c
index 2d48a7941d48..3c55c76c6fd5 100644
--- a/arch/x86_64/pci/mmconfig.c
+++ b/arch/x86_64/pci/mmconfig.c
@@ -9,7 +9,6 @@
 #include <linux/init.h>
 #include <linux/acpi.h>
 #include <linux/bitmap.h>
-#include <linux/dmi.h>
 #include <asm/e820.h>
 
 #include "pci.h"
@@ -165,33 +164,11 @@ static __init void unreachable_devices(void)
         }
 }
 
-static int disable_mcfg(struct dmi_system_id *d)
-{
-        printk("PCI: %s detected. Disabling MCFG.\n", d->ident);
-        pci_probe &= ~PCI_PROBE_MMCONF;
-        return 0;
-}
-
-static struct dmi_system_id __initdata dmi_bad_mcfg[] = {
-        /* Has broken MCFG table that makes the system hang when used */
-        {
-         .callback = disable_mcfg,
-         .ident = "Intel D3C5105 SDV",
-         .matches = {
-                     DMI_MATCH(DMI_BIOS_VENDOR, "Intel"),
-                     DMI_MATCH(DMI_BOARD_NAME, "D26928"),
-                     },
-         },
-         {}
-};
-
 void __init pci_mmcfg_init(void)
 {
         int i;
 
-        dmi_check_system(dmi_bad_mcfg);
-
-        if ((pci_probe & (PCI_PROBE_MMCONF|PCI_PROBE_MMCONF_FORCE)) == 0)
+        if ((pci_probe & PCI_PROBE_MMCONF) == 0)
                 return;
 
         acpi_table_parse(ACPI_MCFG, acpi_parse_mcfg);
@@ -200,6 +177,15 @@ void __init pci_mmcfg_init(void)
             (pci_mmcfg_config[0].base_address == 0))
                 return;
 
+        if (!e820_all_mapped(pci_mmcfg_config[0].base_address,
+                        pci_mmcfg_config[0].base_address + MMCONFIG_APER_MIN,
+                        E820_RESERVED)) {
+                printk(KERN_ERR "PCI: BIOS Bug: MCFG area at %x is not E820-reserved\n",
+                                pci_mmcfg_config[0].base_address);
+                printk(KERN_ERR "PCI: Not using MMCONFIG.\n");
+                return;
+        }
+
         /* RED-PEN i386 doesn't do _nocache right now */
         pci_mmcfg_virt = kmalloc(sizeof(*pci_mmcfg_virt) * pci_mmcfg_config_num, GFP_KERNEL);
         if (pci_mmcfg_virt == NULL) {