Switch to non-logarithmic ALIGN macro.

diff --git a/ChangeLog b/ChangeLog
index 20726f1a2c2f138f9cc6c58874281694c25e2d5f..3ee05f6fad66cc857e07d45562d85872471b7b37 100644 (file)
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,9 @@
 2013-04-16  Niels Möller  <nisse@lysator.liu.se>
 
+       * asm.m4 (m4_log2): New macro, similar to the one in gmp.
+       (ALIGN): Changed to take alignment in bytes. Updated all callers,
+       currently used only in x86 and x86_64 files.
+
        * umac.h (umac32_ctx, umac64_ctx, umac96_ctx, umac128_ctx): Make
        block count an uint64_t. Reorder some elements to put short values
        together.

diff --git a/asm.m4 b/asm.m4
index f24442da27c66b73c3138d107477bad9ade8a298..200b13618ada20853c3f6868aa309286c0d3c483 100644 (file)
--- a/asm.m4
+++ b/asm.m4
@@ -26,16 +26,17 @@ define(<EPILOGUE>,
 <ifelse(ELF_STYLE,yes,
 <.size C_NAME($1), . - C_NAME($1)>,<>)>)
 
-dnl Argument to ALIGN is always logarithmic
-dnl FIXME: the << operator is not supported by Solaris m4,
-dnl and ** is not supported by OpenBSD m4.
-dnl We should switch to non-logarithmic ALIGN instead.
+define(<m4_log2>, <m4_log2_internal($1,1,0)>)
+define(<m4_log2_internal>,
+<ifelse($3, 10, <not-a-power-of-two>,
+$1, $2, $3,
+<m4_log2_internal($1, eval(2*$2), eval(1 + $3))>)>)
+
+dnl Argument to ALIGN is always in bytes, and converted to a
+dnl logarithmic .align if necessary.
 
-dnl Need changequote to be able to use the << operator.
 define(<ALIGN>,
-<changequote([,])dnl
-.align ifelse(ALIGN_LOG,yes,$1,eval(1 << $1))dnl >> balance
-changequote(<,>)dnl
+<.align ifelse(ALIGN_LOG,yes,<m4_log2($1)>,$1)
 >)
 
 dnl Struct defining macros

diff --git a/x86/aes-decrypt-internal.asm b/x86/aes-decrypt-internal.asm
index c19853ff981ec2235b8db76fb10d0526cfbb2af7..64e59283b225ec5c21f99770f1f79f60104828c5 100644 (file)
--- a/x86/aes-decrypt-internal.asm
+++ b/x86/aes-decrypt-internal.asm
@@ -1,4 +1,3 @@
-C -*- mode: asm; asm-comment-char: ?C; -*-  
 C nettle, low-level cryptographics library
 C 
 C Copyright (C) 2001, 2002, 2005 Rafael R. Sevilla, Niels Möller
@@ -61,7 +60,7 @@ C %edi is a temporary, often used as an accumulator.
        C              unsigned length, uint8_t *dst,
        C              uint8_t *src)
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(_nettle_aes_decrypt)
        C save all registers that need to be saved
        pushl   %ebx            C  20(%esp)
@@ -94,7 +93,7 @@ PROLOGUE(_nettle_aes_decrypt)
 
        addl    $16,KEY         C  point to next key
        movl    KEY,FRAME_KEY
-       ALIGN(4)
+       ALIGN(16)
 .Lround_loop:
        AES_ROUND(T, SA,SD,SC,SB, TMP, KEY)
        movl    TMP, TA

diff --git a/x86/aes-encrypt-internal.asm b/x86/aes-encrypt-internal.asm
index fc7d2c403f108d91485159b54f741e7b6bd7ac2e..9fe32fc5e3c7b25add52f86d7d697a54aa4497e2 100644 (file)
--- a/x86/aes-encrypt-internal.asm
+++ b/x86/aes-encrypt-internal.asm
@@ -1,4 +1,3 @@
-C -*- mode: asm; asm-comment-char: ?C; -*-  
 C nettle, low-level cryptographics library
 C 
 C Copyright (C) 2001, 2002, 2005 Rafael R. Sevilla, Niels Möller
@@ -61,7 +60,7 @@ C %edi is a temporary, often used as an accumulator.
        C              unsigned length, uint8_t *dst,
        C              uint8_t *src)
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(_nettle_aes_encrypt)
        C save all registers that need to be saved
        pushl   %ebx            C  20(%esp)
@@ -94,7 +93,7 @@ PROLOGUE(_nettle_aes_encrypt)
 
        addl    $16,KEY         C  point to next key
        movl    KEY,FRAME_KEY
-       ALIGN(4)
+       ALIGN(16)
 .Lround_loop:
        AES_ROUND(T, SA,SB,SC,SD, TMP, KEY)
        movl    TMP, TA

diff --git a/x86/arcfour-crypt.asm b/x86/arcfour-crypt.asm
index 842ae4a56325d23189b6f0363178598870af35e0..89ee7c9e8bb2630a8e2579a3f914cf5dd3c9e3aa 100644 (file)
--- a/x86/arcfour-crypt.asm
+++ b/x86/arcfour-crypt.asm
@@ -23,7 +23,7 @@ C MA 02111-1301, USA.
        C               unsigned length, uint8_t *dst,
        C               const uint8_t *src)
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(nettle_arcfour_crypt)
        C save all registers that need to be saved
        pushl   %ebx            C  12(%esp)
@@ -63,7 +63,7 @@ C Register usage:
        sarl    $1, %edx
        jc      .Lloop_odd
        
-       ALIGN(4)
+       ALIGN(16)
 .Lloop:
        movb    (%ebp, %eax), %cl       C  si.
        addb    %cl, %bl

diff --git a/x86/camellia-crypt-internal.asm b/x86/camellia-crypt-internal.asm
index e8d892d3b54d7d9c50d77e5b5c72c8936211a5ce..7766220e0ce9303930a94b359ee119e61ba8fe36 100644 (file)
--- a/x86/camellia-crypt-internal.asm
+++ b/x86/camellia-crypt-internal.asm
@@ -1,4 +1,3 @@
-C -*- mode: asm; asm-comment-char: ?C; -*-  
 C nettle, low-level cryptographics library
 C 
 C Copyright (C) 2010, Niels Möller
@@ -142,7 +141,7 @@ define(<FLINV>, <
        C                 unsigned length, uint8_t *dst,
        C                 uint8_t *src)
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(_nettle_camellia_crypt)
        C save all registers that need to be saved
        pushl   %ebx            C  32(%esp)

diff --git a/x86/md5-compress.asm b/x86/md5-compress.asm
index 1bdada791b55c3ea9c302ffd4dcb095bb3782fb0..ac0cd900e3146504b17771b60b15c28e07408399 100644 (file)
--- a/x86/md5-compress.asm
+++ b/x86/md5-compress.asm
@@ -68,7 +68,7 @@ define(<ROUND>,<
        C _nettle_md5_compress(uint32_t *state, uint8_t *data)
        
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(_nettle_md5_compress)
        C save all registers that need to be saved
 

diff --git a/x86/sha1-compress.asm b/x86/sha1-compress.asm
index afb8d8c56fae0561a3d2123cc2d36bddec01e04c..777615dc13b1cbd32e80477b1e2c0cdf08cf4dd6 100644 (file)
--- a/x86/sha1-compress.asm
+++ b/x86/sha1-compress.asm
@@ -160,7 +160,7 @@ PROLOGUE(_nettle_sha1_compress)
 
        C Loop-mixed to 520 cycles (for the complete function call) on
        C AMD K7.
-ALIGN(5)
+ALIGN(32)
        mov     88(%esp), T2
        mov     OFFSET(2)(T2), %ecx
        mov     OFFSET(0)(T2), %eax

diff --git a/x86_64/aes-decrypt-internal.asm b/x86_64/aes-decrypt-internal.asm
index de97de32b0282100f4fa89c9b09b868a459da0c2..0d4f2f92cff7da863634d5aedfe4f868cb9baca0 100644 (file)
--- a/x86_64/aes-decrypt-internal.asm
+++ b/x86_64/aes-decrypt-internal.asm
@@ -53,7 +53,7 @@ define(<TMP>,<%rbp>)
        C              unsigned length, uint8_t *dst,
        C              uint8_t *src)
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(_nettle_aes_decrypt)
        W64_ENTRY(5, 0)
        test    PARAM_LENGTH, PARAM_LENGTH
@@ -81,7 +81,7 @@ PROLOGUE(_nettle_aes_decrypt)
        subl    $1, COUNT
 
        add     $16,KEY         C  point to next key
-       ALIGN(4)
+       ALIGN(16)
 .Lround_loop:
        AES_ROUND(TABLE, SA,SD,SC,SB, TA, TMP)
        AES_ROUND(TABLE, SB,SA,SD,SC, TB, TMP)

diff --git a/x86_64/aes-encrypt-internal.asm b/x86_64/aes-encrypt-internal.asm
index fbfcdb660259e15ef8e3c45e99beb7dd5c9e3c4f..4ae0ec858b07851106a50f023a61e31d3987c852 100644 (file)
--- a/x86_64/aes-encrypt-internal.asm
+++ b/x86_64/aes-encrypt-internal.asm
@@ -53,7 +53,7 @@ define(<TMP>,<%rbp>)
        C              unsigned length, uint8_t *dst,
        C              uint8_t *src)
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(_nettle_aes_encrypt)
        W64_ENTRY(5, 0)
        test    PARAM_LENGTH, PARAM_LENGTH
@@ -81,7 +81,7 @@ PROLOGUE(_nettle_aes_encrypt)
        subl    $1, COUNT
 
        add     $16,KEY         C  point to next key
-       ALIGN(4)
+       ALIGN(16)
 .Lround_loop:
        AES_ROUND(TABLE, SA,SB,SC,SD, TA, TMP)
        AES_ROUND(TABLE, SB,SC,SD,SA, TB, TMP)

diff --git a/x86_64/camellia-crypt-internal.asm b/x86_64/camellia-crypt-internal.asm
index 8fc2fc1b47bd07ac2b7d74dffdbc0261abe7b7b4..e44a3dec676bc2403fa7307f3b61f0f447b98d24 100644 (file)
--- a/x86_64/camellia-crypt-internal.asm
+++ b/x86_64/camellia-crypt-internal.asm
@@ -1,4 +1,3 @@
-C -*- mode: asm; asm-comment-char: ?C; -*-  
 C nettle, low-level cryptographics library
 C 
 C Copyright (C) 2010, Niels Möller
@@ -122,7 +121,7 @@ C   xorl    XREG(TMP), XREG($1)
        C                 unsigned length, uint8_t *dst,
        C                 uint8_t *src)
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(_nettle_camellia_crypt)
 
        W64_ENTRY(5, 0)

diff --git a/x86_64/ecc-192-modp.asm b/x86_64/ecc-192-modp.asm
index 5812070b88908f0a3c48513d25fa5af48a07d80b..f3fe4958e79995bcd5106fb18c36e50958597a85 100644 (file)
--- a/x86_64/ecc-192-modp.asm
+++ b/x86_64/ecc-192-modp.asm
@@ -30,7 +30,7 @@ define(<C2>, <%r11>)
 
        C ecc_192_modp (const struct ecc_curve *ecc, mp_limb_t *rp)
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(nettle_ecc_192_modp)
        W64_ENTRY(2, 0)
        mov     16(RP), T2

diff --git a/x86_64/memxor.asm b/x86_64/memxor.asm
index c2ccc920604827cd19bf9d53db5de4e9471f8efc..b22a4721bc74d91e95a60078bcb77d795e4ae5f1 100644 (file)
--- a/x86_64/memxor.asm
+++ b/x86_64/memxor.asm
@@ -1,4 +1,3 @@
-C -*- mode: asm; asm-comment-char: ?C; -*-  
 C nettle, low-level cryptographics library
 C 
 C Copyright (C) 2010, Niels Möller
@@ -37,7 +36,7 @@ define(<USE_SSE2>, <no>)
 
        C memxor(uint8_t *dst, const uint8_t *src, size_t n)
        C                 %rdi               %rsi      %rdx
-       ALIGN(4)
+       ALIGN(16)
 
 PROLOGUE(memxor)
        W64_ENTRY(3, 0)
@@ -48,7 +47,7 @@ EPILOGUE(memxor)
 
        C memxor3(uint8_t *dst, const uint8_t *a, const uint8_t *b, size_t n)
        C                 %rdi              %rsi              %rdx      %rcx
-       ALIGN(4)
+       ALIGN(16)
        
 PROLOGUE(memxor3)
        W64_ENTRY(4, 0)
@@ -124,7 +123,7 @@ ifelse(USE_SSE2, yes, <
        jz      .Ldone
        jmp     .Lshift_next
 
-       ALIGN(4)
+       ALIGN(16)
 
 .Lshift_loop:
        mov     8(AP, N), S0
@@ -177,7 +176,7 @@ C   jz      .Ldone
        
        jmp     .Lword_next
 
-       ALIGN(4)
+       ALIGN(16)
 
 .Lword_loop:
        mov     8(AP, N), TMP
@@ -234,7 +233,7 @@ ifelse(USE_SSE2, yes, <
        mov     TMP, (DST, N)
        jmp     .Lsse2_next
 
-       ALIGN(4)
+       ALIGN(16)
 .Lsse2_loop:
        movdqu  (AP, N), %xmm0
        movdqu  (BP, N), %xmm1

diff --git a/x86_64/salsa20-core-internal.asm b/x86_64/salsa20-core-internal.asm
index 81ca2cc83d14712654e1d9148f440e3e3685cb28..0e0cdf6ac0eeccf32cddd5d5a30ef3bf770e8082 100644 (file)
--- a/x86_64/salsa20-core-internal.asm
+++ b/x86_64/salsa20-core-internal.asm
@@ -34,7 +34,7 @@ include_src(<x86_64/salsa20.m4>)
 
        C _salsa20_core(uint32_t *dst, const uint32_t *src, unsigned rounds)
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(_nettle_salsa20_core)
        W64_ENTRY(3, 9) 
 
@@ -58,7 +58,7 @@ PROLOGUE(_nettle_salsa20_core)
 
        shrl    $1, XREG(COUNT)
 
-       ALIGN(4)
+       ALIGN(16)
 .Loop:
        QROUND(X0, X1, X2, X3)
        pshufd  $0x93, X1, X1   C       11 00 01 10 (least sign. left)

diff --git a/x86_64/salsa20-crypt.asm b/x86_64/salsa20-crypt.asm
index 5d11980475d72f0350f591235300d471221be1bb..25b7e4973b298a100fd6d5e75fc6cc2573da962e 100644 (file)
--- a/x86_64/salsa20-crypt.asm
+++ b/x86_64/salsa20-crypt.asm
@@ -50,7 +50,7 @@ C registers.
        C salsa20_crypt(struct salsa20_ctx *ctx, unsigned length,
        C               uint8_t *dst, const uint8_t *src)
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(nettle_salsa20_crypt)
        W64_ENTRY(4, 9) 
 
@@ -92,7 +92,7 @@ PROLOGUE(nettle_salsa20_crypt)
        SWAP(X0, X2, M0011)     
 
        movl    $10, XREG(COUNT)
-       ALIGN(4)
+       ALIGN(16)
 .Loop:
        QROUND(X0, X1, X2, X3)
        C For the row operations, we first rotate the rows, to get

diff --git a/x86_64/serpent-decrypt.asm b/x86_64/serpent-decrypt.asm
index 02a857cecb272dfe9397c84a7c26fcd38296275b..d6bacb5d8e4f3977578c7304a29acc5046314a77 100644 (file)
--- a/x86_64/serpent-decrypt.asm
+++ b/x86_64/serpent-decrypt.asm
@@ -522,7 +522,7 @@ define(<WLTI>, <
        C                 unsigned length, uint8_t *dst,
        C                 const uint8_t *src)
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(nettle_serpent_decrypt)
         C save all registers that need to be saved
        W64_ENTRY(4, 13)
@@ -557,7 +557,7 @@ PROLOGUE(nettle_serpent_decrypt)
 
        jmp     .Lwround_start
 
-       ALIGN(4)
+       ALIGN(16)
 
 .Lwround_loop:
        WLTI(X0,X1,X2,X3)
@@ -624,7 +624,7 @@ PROLOGUE(nettle_serpent_decrypt)
        mov     $384, CNT
        jmp     .Lround_start
 
-       ALIGN(4)
+       ALIGN(16)
 .Lround_loop:
        LTI(x0,x1,x2,x3)
 .Lround_start:

diff --git a/x86_64/serpent-encrypt.asm b/x86_64/serpent-encrypt.asm
index 5362bad1f06ab8ab68407bd20e40ec4a61110307..613ef41e003b1612a215b824859a7d533c7b354d 100644 (file)
--- a/x86_64/serpent-encrypt.asm
+++ b/x86_64/serpent-encrypt.asm
@@ -549,7 +549,7 @@ define(<WLT>, <
        C                 unsigned length, uint8_t *dst,
        C                 const uint8_t *src)
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(nettle_serpent_encrypt)
         C save all registers that need to be saved
        W64_ENTRY(4, 13)
@@ -583,7 +583,7 @@ PROLOGUE(nettle_serpent_encrypt)
        mov     $-512, CNT
        jmp     .Lwround_start
 
-       ALIGN(4)
+       ALIGN(16)
 .Lwround_loop:
        WLT(X0,X1,X2,X3)
 .Lwround_start:
@@ -653,7 +653,7 @@ C parallell.
        mov     $-512, CNT
        jmp     .Lround_start
        
-       ALIGN(4)
+       ALIGN(16)
 .Lround_loop:
        LT(x0,x1,x2,x3)
 .Lround_start:

diff --git a/x86_64/sha1-compress.asm b/x86_64/sha1-compress.asm
index ffa28d0b9fa76151c4b38f5648e7e43426c05c9d..5155683ca332f1016138cb88237f0e337b76c5f7 100644 (file)
--- a/x86_64/sha1-compress.asm
+++ b/x86_64/sha1-compress.asm
@@ -123,7 +123,7 @@ C adding, and then rotating back.
        C _nettle_sha1_compress(uint32_t *state, uint8_t *input)
        
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(_nettle_sha1_compress)
        C save all registers that need to be saved
        W64_ENTRY(2, 0)

diff --git a/x86_64/sha256-compress.asm b/x86_64/sha256-compress.asm
index 59f922e759676feed068989dd0eb83c27d128862..6bfb7a78c9f4087a88f70c29b18c897d1eabb1ae 100644 (file)
--- a/x86_64/sha256-compress.asm
+++ b/x86_64/sha256-compress.asm
@@ -114,7 +114,7 @@ define(<NOEXPN>, <
        C _nettle_sha256_compress(uint32_t *state, const uint8_t *input, const uint32_t *k)
 
        .text
-       ALIGN(4)
+       ALIGN(16)
 
 PROLOGUE(_nettle_sha256_compress)
        W64_ENTRY(3, 0)
@@ -137,7 +137,7 @@ PROLOGUE(_nettle_sha256_compress)
        movl    24(STATE), SG
        movl    28(STATE), SH
        xor     COUNT, COUNT
-       ALIGN(4)
+       ALIGN(16)
 
 .Loop1:
        NOEXPN(0) ROUND(SA,SB,SC,SD,SE,SF,SG,SH,0)

diff --git a/x86_64/sha3-permute.asm b/x86_64/sha3-permute.asm
index 64c4cfc4907a473d8598c557c8a193c109b9ea1b..7f9a6b79108d570278a4cbff0a801412ffd5e20d 100644 (file)
--- a/x86_64/sha3-permute.asm
+++ b/x86_64/sha3-permute.asm
@@ -94,7 +94,7 @@ define(<ROTL64>, <
        
        C sha3_permute(struct sha3_state *ctx)
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(nettle_sha3_permute)
        W64_ENTRY(1, 16)
        push    %rbp
@@ -139,7 +139,7 @@ PROLOGUE(nettle_sha3_permute)
        pxor    A2122, C12
        pxor    A2324, C34
        
-       ALIGN(4)
+       ALIGN(16)
 .Loop:
        C The theta step. Combine parity bits, then xor to state.
        C D0 = C4 ^ (C1 <<< 1)
@@ -483,7 +483,7 @@ PROLOGUE(nettle_sha3_permute)
 
 EPILOGUE(nettle_sha3_permute)
 
-ALIGN(4)
+ALIGN(16)
 .rc:   C In reverse order
        .quad   0x8000000080008008
        .quad   0x0000000080000001

diff --git a/x86_64/sha512-compress.asm b/x86_64/sha512-compress.asm
index d54ebda33373b477de093287d284ab9f3c539828..21df82a2310e0f33f94f3fa973ad3c2985870489 100644 (file)
--- a/x86_64/sha512-compress.asm
+++ b/x86_64/sha512-compress.asm
@@ -114,7 +114,7 @@ define(<NOEXPN>, <
        C _nettle_sha512_compress(uint64_t *state, const uint8_t *input, const uint64_t *k)
 
        .text
-       ALIGN(4)
+       ALIGN(16)
 
 PROLOGUE(_nettle_sha512_compress)
        W64_ENTRY(3, 0)
@@ -137,7 +137,7 @@ PROLOGUE(_nettle_sha512_compress)
        mov     48(STATE), SG
        mov     56(STATE), SH
        xor     COUNT, COUNT
-       ALIGN(4)
+       ALIGN(16)
 
 .Loop1:
        NOEXPN(0) ROUND(SA,SB,SC,SD,SE,SF,SG,SH,0)

diff --git a/x86_64/umac-nh-n.asm b/x86_64/umac-nh-n.asm
index 06e74269f81850ab837c22d1260230476060e9c8..bcb994874ea87b16bbbd1fea7e8b4bebbff29b9c 100644 (file)
--- a/x86_64/umac-nh-n.asm
+++ b/x86_64/umac-nh-n.asm
@@ -49,7 +49,7 @@ C aligned.
        C umac_nh_n(uint64_t *out, unsigned n, const uint32_t *key,
        C           unsigned length, const uint8_t *msg)
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(_nettle_umac_nh_n)
        W64_ENTRY(5, 14)
        pxor    XY0, XY0

diff --git a/x86_64/umac-nh.asm b/x86_64/umac-nh.asm
index f9230cd8d68fa497e5f5ada8ca915a245fca19f9..8e88df6a4fb4e8728b0d0d74e2ebf4fd6bf5acc4 100644 (file)
--- a/x86_64/umac-nh.asm
+++ b/x86_64/umac-nh.asm
@@ -36,7 +36,7 @@ C aligned.
        
        C umac_nh(const uint32_t *key, unsigned length, const uint8_t *msg)
        .text
-       ALIGN(4)
+       ALIGN(16)
 PROLOGUE(_nettle_umac_nh)
        W64_ENTRY(3, 7)
        pxor    XY, XY