summaryrefslogtreecommitdiff
blob: a70eb32e6f085fb93c6cc60c1b9a3d62422413df (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
--- mysql/strings/longlong2str-x86.s	2005-07-15 14:06:49.000000000 +0200
+++ mysql-fixed/strings/longlong2str-x86.s	2005-07-15 14:07:42.000000000 +0200
@@ -13,8 +13,9 @@
 # along with this program; if not, write to the Free Software
 # Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA
 
-# Optimized longlong2str function for Intel 80x86  (gcc/gas syntax) 
-# Some set sequences are optimized for pentuimpro II 
+# longlong2str function for Intel 80x86  (gcc/gas syntax) 
+
+# See longlong2str(dst,radix,val) description in longlong2str.c.
 
 	.file	"longlong2str.s"
 	.version "1.01"
@@ -24,99 +25,88 @@
 
 .globl	longlong2str
 	.type	 longlong2str,@function
-	
+
 longlong2str:
-	subl $80,%esp
+	subl  $80,%esp          # Temporary buffer for up to 64 radix-2 digits
 	pushl %ebp
 	pushl %esi
 	pushl %edi
 	pushl %ebx
-	movl 100(%esp),%esi	# Lower part of val 
-	movl 104(%esp),%ebp	# Higher part of val 
-	movl 108(%esp),%edi	# get dst 
-	movl 112(%esp),%ebx	# Radix 
-	movl %ebx,%eax
-	testl %eax,%eax
-	jge .L144
-
-	addl $36,%eax
-	cmpl $34,%eax
-	ja .Lerror		# Wrong radix 
+
+	movl  100(%esp),%esi	# esi = Lower part of val 
+	movl  112(%esp),%ebx	# ebx = Radix
+	movl  104(%esp),%ebp	# ebp = Higher part of val 
+	movl  108(%esp),%edi    # edi -> dst
+		 
+	testl %ebx,%ebx
+	jge   .L144             # (Radix >= 0)
+
 	testl %ebp,%ebp
-	jge .L146
-	movb $45,(%edi)		# Add sign 
-	incl %edi		# Change sign of val 
-	negl %esi
-	adcl $0,%ebp
-	negl %ebp
+	jge   .L146             # (Higher part of val >= 0)
+	movb  $45,(%edi)	# Add '-' sign 
+	incl  %edi
+	negl  %esi              # Change val to positive
+	adcl  $0,%ebp
+	negl  %ebp
 .L146:
-	negl %ebx		# Change radix to positive 
-	jmp .L148
-	.align 4
+	negl  %ebx		# Change radix to positive 
 .L144:
-	addl $-2,%eax
-	cmpl $34,%eax
-	ja .Lerror		# Radix in range 
-
-.L148:
-	movl %esi,%eax		# Test if zero (for easy loop) 
-	orl %ebp,%eax
-	jne .L150
-	movb $48,(%edi)
-	incl %edi
-	jmp .L10_end
-	.align 4
-
-.L150:
-	leal 92(%esp),%ecx	# End of buffer 
-	jmp  .L155
-	.align 4
+	cmpl  $36,%ebx          # Radix must be between 2 and 36
+	ja    .Lerror		# (Radix not in range)
+	cmpl  $2,%ebx
+	jb    .Lerror		# (Radix not in range)
+
+	movl  %esi,%eax         # eax = lower part of val ...
+	orl   %ebp,%eax         # and it stays thus if ebp=0
+	je    Lzero             # (Treat zero as special case)
 
-.L153:
-	# val is stored in in ebp:esi 
-
-	movl %ebp,%eax		# High part of value 
-	xorl %edx,%edx
-	divl %ebx
-	movl %eax,%ebp
-	movl %esi,%eax
-	divl %ebx
-	decl %ecx
-	movl %eax,%esi		# quotent in ebp:esi 
-	movb _dig_vec_upper(%edx),%al   # al is faster than dl 
-	movb %al,(%ecx)		# store value in buff 
-	.align 4
-.L155:
+	leal  92(%esp),%ecx     # ecx -> End of temporary buffer
+	
 	testl %ebp,%ebp
-	ja .L153
-	testl %esi,%esi		# rest value 
-	jl .L153
-	je .L10_mov		# Ready 
-	movl %esi,%eax
-	movl $_dig_vec_upper,%ebp
-	.align 4
-
-.L154:				# Do rest with integer precision 
-	cltd
-	divl %ebx
-	decl %ecx
-	movb (%edx,%ebp),%dl	# bh is always zero as ebx=radix < 36 
-	testl %eax,%eax
-	movb %dl,(%ecx)
-	jne .L154
-
-.L10_mov:
-	movl %ecx,%esi
-	leal 92(%esp),%ecx	# End of buffer 
-	subl %esi,%ecx
-	rep
-	movsb
-
-.L10_end:
-	movl %edi,%eax		# Pointer to end null 
-	movb $0,(%edi)		# Store the end null 
-
-.L165:
+	je    Llow              # (Higher part of val = 0)
+	
+Lhigh:
+	#val in ebp:esi. div the high part by the radix,
+        #then div remainder + low part by the radix.
+	movl  %ebp,%eax	  	# edx=0,eax=high(from ebp) 
+	xorl  %edx,%edx
+        decl  %ecx
+        divl  %ebx         
+	movl  %eax,%ebp         # edx=result of last, eax=low(from esi)
+	movl  %esi,%eax    
+	divl  %ebx
+	movl  %eax,%esi	  	# ebp:esi = quotient 
+	movb  %dl,(%ecx)        # store byte in temporary buffer
+	testl %ebp,%ebp 
+	jne   Lhigh             # (Higher part of val still > 0)
+
+Llow:
+	#val in 0:eax. div 0 + low part by the radix.
+	xorl  %edx,%edx
+        decl  %ecx
+	divl  %ebx
+ 	movb  %dl,(%ecx)        # store byte in temporary buffer
+        testl %eax,%eax
+	jne   Llow              # (Lower part of val still <> 0)
+
+	leal  92(%esp),%esi     # esi -> End of temporary buffer
+
+Lmov:
+	movb  (%ecx),%dl        # dl = byte from temporary buffer
+	movb  $-1,%bl
+	cmpb  $10,%dl           # add 7 if dl > '9'
+	adcb  $0,%bl
+	addb  $48,%dl           # add '0'  
+	andb  $7,%bl
+	addb  %bl,%dl
+        incl  %ecx
+	movb  %dl,(%edi)        # put dl in dst
+	incl  %edi
+	cmpl  %ecx,%esi
+	ja    Lmov              # (more bytes exist in temporary buffer)
+	movb  $0,(%edi)         # trailing '\0' in dst
+	movl  %edi,%eax		# eax = return value = pointer to '\0'
+.Lret:
 	popl %ebx
 	popl %edi
 	popl %esi
@@ -126,20 +116,28 @@
 
 .Lerror:
 	xorl %eax,%eax		# Wrong radix 
-	jmp .L165
+	jmp .Lret
 
-.Lfe3:
-	.size	 longlong2str,.Lfe3-longlong2str
+Lzero:
+        # Treat 0 as a special case. Unnecessary but we
+	# expect 0 will be frequent.
+	movl  108(%esp),%eax    # eax = dst
+	popl  %ebx
+        movb  $48,(%eax)        # '0'
+	popl  %edi
+        incl  %eax
+	popl  %esi
+	addl  $80,%esp
+        movb  $0,(%eax)         # '\0'
+	popl  %ebp
+	ret
 
 #
 # This is almost equal to the above, except that we can do the final
 # loop much more efficient	
-#	
+#
 
 	.align 4
-.Ltmp:
-        .long 0xcccccccd
-	.align 4
 	
 .globl	longlong10_to_str
 	.type	 longlong10_to_str,@function
@@ -202,7 +200,7 @@
 
 	# The following code uses some tricks to change division by 10 to
 	# multiplication and shifts
-	movl .Ltmp,%esi		# set %esi to 0xcccccccd
+	movl $0xcccccccd,%esi   # set %esi to 0xcccccccd
 	
 .L10_40:
         movl %ebx,%eax
@@ -217,7 +215,27 @@
         movl %edx,%ebx
         testl %ebx,%ebx
 	jne .L10_40
-	jmp .L10_mov		# Shared end with longlong10_to_str
+#	jmp .L10_mov		# Shared end with longlong10_to_str
+
 
+.L10_mov:
+	movl %ecx,%esi
+	leal 92(%esp),%ecx	# End of buffer 
+	subl %esi,%ecx
+	rep
+	movsb
+
+.L10_end:
+	movl %edi,%eax		# Pointer to end null 
+	movb $0,(%edi)		# Store the end null 
+
+.L165:
+	popl %ebx
+	popl %edi
+	popl %esi
+	popl %ebp
+	addl $80,%esp
+	ret
+	
 .L10end:
 	.size	 longlong10_to_str,.L10end-longlong10_to_str