|
Line 0
Link Here
|
|
|
1 |
--- src/checksum.c.orig Mon Oct 20 16:57:38 2003 |
| 2 |
+++ src/checksum.c Mon Oct 20 17:01:34 2003 |
| 3 |
@@ -23,68 +23,62 @@ |
| 4 |
csum_partial(const u_char * buff, int len, u_int sum) |
| 5 |
{ |
| 6 |
__asm__ ( |
| 7 |
-/* "pushl %esi |
| 8 |
- pushl %ebx |
| 9 |
- movl 20(%esp),%eax # Function arg: u_int sum |
| 10 |
- movl 16(%esp),%ecx # Function arg: int len |
| 11 |
- movl 12(%esp),%esi # Function arg: u_char *buff*/ |
| 12 |
- |
| 13 |
- "testl $2, %%esi |
| 14 |
- jz 2f |
| 15 |
- subl $2, %%ecx |
| 16 |
- jae 1f |
| 17 |
- addl $2, %%ecx |
| 18 |
- jmp 4f |
| 19 |
-1: movw (%%esi), %%bx |
| 20 |
- addl $2, %%esi |
| 21 |
- addw %%bx, %%ax |
| 22 |
- adcl $0, %%eax |
| 23 |
-2: |
| 24 |
- movl %%ecx, %%edx |
| 25 |
- shrl $5, %%ecx |
| 26 |
- jz 2f |
| 27 |
- testl %%esi, %%esi |
| 28 |
-1: movl (%%esi), %%ebx |
| 29 |
- adcl %%ebx, %%eax |
| 30 |
- movl 4(%%esi), %%ebx |
| 31 |
- adcl %%ebx, %%eax |
| 32 |
- movl 8(%%esi), %%ebx |
| 33 |
- adcl %%ebx, %%eax |
| 34 |
- movl 12(%%esi), %%ebx |
| 35 |
- adcl %%ebx, %%eax |
| 36 |
- movl 16(%%esi), %%ebx |
| 37 |
- adcl %%ebx, %%eax |
| 38 |
- movl 20(%%esi), %%ebx |
| 39 |
- adcl %%ebx, %%eax |
| 40 |
- movl 24(%%esi), %%ebx |
| 41 |
- adcl %%ebx, %%eax |
| 42 |
- movl 28(%%esi), %%ebx |
| 43 |
- adcl %%ebx, %%eax |
| 44 |
- lea 32(%%esi), %%esi |
| 45 |
- dec %%ecx |
| 46 |
- jne 1b |
| 47 |
- adcl $0, %%eax |
| 48 |
-2: movl %%edx, %%ecx |
| 49 |
- andl $0x1c, %%edx |
| 50 |
- je 4f |
| 51 |
- shrl $2, %%edx |
| 52 |
-3: adcl (%%esi), %%eax |
| 53 |
- lea 4(%%esi), %%esi |
| 54 |
- dec %%edx |
| 55 |
- jne 3b |
| 56 |
- adcl $0, %%eax |
| 57 |
-4: andl $3, %%ecx |
| 58 |
- jz 7f |
| 59 |
- cmpl $2, %%ecx |
| 60 |
- jb 5f |
| 61 |
- movw (%%esi),%%cx |
| 62 |
- leal 2(%%esi),%%esi |
| 63 |
- je 6f |
| 64 |
- shll $16,%%ecx |
| 65 |
-5: movb (%%esi),%%cl |
| 66 |
-6: addl %%ecx,%%eax |
| 67 |
- adcl $0, %%eax |
| 68 |
-7: " |
| 69 |
+ "testl $2, %%esi\n"\ |
| 70 |
+ "jz 2f\n"\ |
| 71 |
+ "subl $2, %%ecx\n"\ |
| 72 |
+ "jae 1f\n"\ |
| 73 |
+ "addl $2, %%ecx\n"\ |
| 74 |
+ "jmp 4f\n"\ |
| 75 |
+"1: movw (%%esi), %%bx\n"\ |
| 76 |
+" addl $2, %%esi\n"\ |
| 77 |
+" addw %%bx, %%ax\n"\ |
| 78 |
+" adcl $0, %%eax\n"\ |
| 79 |
+"2:\n"\ |
| 80 |
+" movl %%ecx, %%edx\n"\ |
| 81 |
+" shrl $5, %%ecx\n"\ |
| 82 |
+" jz 2f\n"\ |
| 83 |
+" testl %%esi, %%esi\n"\ |
| 84 |
+"1: movl (%%esi), %%ebx\n"\ |
| 85 |
+" adcl %%ebx, %%eax\n"\ |
| 86 |
+" movl 4(%%esi), %%ebx\n"\ |
| 87 |
+" adcl %%ebx, %%eax\n"\ |
| 88 |
+" movl 8(%%esi), %%ebx\n"\ |
| 89 |
+" adcl %%ebx, %%eax\n"\ |
| 90 |
+" movl 12(%%esi), %%ebx\n"\ |
| 91 |
+" adcl %%ebx, %%eax\n"\ |
| 92 |
+" movl 16(%%esi), %%ebx\n"\ |
| 93 |
+" adcl %%ebx, %%eax\n"\ |
| 94 |
+" movl 20(%%esi), %%ebx\n"\ |
| 95 |
+" adcl %%ebx, %%eax\n"\ |
| 96 |
+" movl 24(%%esi), %%ebx\n"\ |
| 97 |
+" adcl %%ebx, %%eax\n"\ |
| 98 |
+" movl 28(%%esi), %%ebx\n"\ |
| 99 |
+" adcl %%ebx, %%eax\n"\ |
| 100 |
+" lea 32(%%esi), %%esi\n"\ |
| 101 |
+" dec %%ecx\n"\ |
| 102 |
+" jne 1b\n"\ |
| 103 |
+" adcl $0, %%eax\n"\ |
| 104 |
+"2: movl %%edx, %%ecx\n"\ |
| 105 |
+" andl $0x1c, %%edx\n"\ |
| 106 |
+" je 4f\n"\ |
| 107 |
+" shrl $2, %%edx\n"\ |
| 108 |
+"3: adcl (%%esi), %%eax\n"\ |
| 109 |
+" lea 4(%%esi), %%esi\n"\ |
| 110 |
+" dec %%edx\n"\ |
| 111 |
+" jne 3b\n"\ |
| 112 |
+" adcl $0, %%eax\n"\ |
| 113 |
+"4: andl $3, %%ecx\n"\ |
| 114 |
+" jz 7f\n"\ |
| 115 |
+" cmpl $2, %%ecx\n"\ |
| 116 |
+" jb 5f\n"\ |
| 117 |
+" movw (%%esi),%%cx\n"\ |
| 118 |
+" leal 2(%%esi),%%esi\n"\ |
| 119 |
+" je 6f\n"\ |
| 120 |
+" shll $16,%%ecx\n"\ |
| 121 |
+"5: movb (%%esi),%%cl\n"\ |
| 122 |
+"6: addl %%ecx,%%eax\n"\ |
| 123 |
+" adcl $0, %%eax\n"\ |
| 124 |
+"7: " |
| 125 |
: "=a"(sum) |
| 126 |
: "0"(sum), "c"(len), "S"(buff) |
| 127 |
: "bx", "dx"); |
| 128 |
@@ -103,25 +97,24 @@ |
| 129 |
{ |
| 130 |
u_int sum; |
| 131 |
|
| 132 |
- __asm__ __volatile__(" |
| 133 |
- movl (%1), %0 |
| 134 |
- subl $4, %2 |
| 135 |
- jbe 2f |
| 136 |
- addl 4(%1), %0 |
| 137 |
- adcl 8(%1), %0 |
| 138 |
- adcl 12(%1), %0 |
| 139 |
-1: adcl 16(%1), %0 |
| 140 |
- lea 4(%1), %1 |
| 141 |
- decl %2 |
| 142 |
- jne 1b |
| 143 |
- adcl $0, %0 |
| 144 |
- movl %0, %2 |
| 145 |
- shrl $16, %0 |
| 146 |
- addw %w2, %w0 |
| 147 |
- adcl $0, %0 |
| 148 |
- notl %0 |
| 149 |
-2: |
| 150 |
- " |
| 151 |
+ __asm__ __volatile__( |
| 152 |
+" movl (%1), %0\n"\ |
| 153 |
+" subl $4, %2\n"\ |
| 154 |
+" jbe 2f\n"\ |
| 155 |
+" addl 4(%1), %0\n"\ |
| 156 |
+" adcl 8(%1), %0\n"\ |
| 157 |
+" adcl 12(%1), %0\n"\ |
| 158 |
+"1: adcl 16(%1), %0\n"\ |
| 159 |
+" lea 4(%1), %1\n"\ |
| 160 |
+" decl %2\n"\ |
| 161 |
+" jne 1b\n"\ |
| 162 |
+" adcl $0, %0\n"\ |
| 163 |
+" movl %0, %2\n"\ |
| 164 |
+" shrl $16, %0\n"\ |
| 165 |
+" addw %w2, %w0\n"\ |
| 166 |
+" adcl $0, %0\n"\ |
| 167 |
+" notl %0\n"\ |
| 168 |
+"2:\n" |
| 169 |
/* |
| 170 |
Since the input registers which are loaded with iph and ipl |
| 171 |
are modified, we must also specify them as outputs, or gcc |
| 172 |
@@ -137,10 +130,9 @@ |
| 173 |
static inline u_int |
| 174 |
csum_fold(u_int sum) |
| 175 |
{ |
| 176 |
- __asm__(" |
| 177 |
- addl %1, %0 |
| 178 |
- adcl $0xffff, %0 |
| 179 |
- " |
| 180 |
+ __asm__( |
| 181 |
+" addl %1, %0\n"\ |
| 182 |
+" adcl $0xffff, %0\n" |
| 183 |
: "=r" (sum) |
| 184 |
: "r" (sum << 16), "0" (sum & 0xffff0000) |
| 185 |
); |
| 186 |
@@ -155,12 +147,11 @@ |
| 187 |
csum_tcpudp_magic(u_int saddr, u_int daddr, u_short len, |
| 188 |
u_short proto, u_int sum) |
| 189 |
{ |
| 190 |
- __asm__(" |
| 191 |
- addl %1, %0 |
| 192 |
- adcl %2, %0 |
| 193 |
- adcl %3, %0 |
| 194 |
- adcl $0, %0 |
| 195 |
- " |
| 196 |
+ __asm__( |
| 197 |
+" addl %1, %0\n"\ |
| 198 |
+" adcl %2, %0\n"\ |
| 199 |
+" adcl %3, %0\n"\ |
| 200 |
+" adcl $0, %0\n" |
| 201 |
: "=r" (sum) |
| 202 |
: "g" (daddr), "g"(saddr), "g"((ntohs(len) << 16) + proto * 256), "0"(sum)); |
| 203 |
return (csum_fold(sum)); |