blob: 0e93b194be9514ff4102a78a62bcd8dca3eb0aa5 [file] [log] [blame]
Fangrui Songee4e2e72018-01-30 00:40:05 +00001import re
Fangrui Song4f0f4262018-02-10 05:01:33 +00002import sys
Fangrui Songee4e2e72018-01-30 00:40:05 +00003
4from . import common
5
Fangrui Song4f0f4262018-02-10 05:01:33 +00006if sys.version_info[0] > 2:
7 class string:
8 expandtabs = str.expandtabs
9else:
10 import string
11
Fangrui Songee4e2e72018-01-30 00:40:05 +000012# RegEx: this is where the magic happens.
13
Fangrui Song4f0f4262018-02-10 05:01:33 +000014##### Assembly parser
15
Fangrui Songee4e2e72018-01-30 00:40:05 +000016ASM_FUNCTION_X86_RE = re.compile(
17 r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@(?P=func)\n[^:]*?'
18 r'(?P<body>^##?[ \t]+[^:]+:.*?)\s*'
19 r'^\s*(?:[^:\n]+?:\s*\n\s*\.size|\.cfi_endproc|\.globl|\.comm|\.(?:sub)?section|#+ -- End function)',
20 flags=(re.M | re.S))
21
22ASM_FUNCTION_ARM_RE = re.compile(
23 r'^(?P<func>[0-9a-zA-Z_]+):\n' # f: (name of function)
24 r'\s+\.fnstart\n' # .fnstart
25 r'(?P<body>.*?)\n' # (body of the function)
26 r'.Lfunc_end[0-9]+:', # .Lfunc_end0: or # -- End function
27 flags=(re.M | re.S))
28
29ASM_FUNCTION_AARCH64_RE = re.compile(
30 r'^_?(?P<func>[^:]+):[ \t]*\/\/[ \t]*@(?P=func)\n'
31 r'[ \t]+.cfi_startproc\n'
32 r'(?P<body>.*?)\n'
33 # This list is incomplete
34 r'.Lfunc_end[0-9]+:\n',
35 flags=(re.M | re.S))
36
37ASM_FUNCTION_MIPS_RE = re.compile(
38 r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@(?P=func)\n[^:]*?' # f: (name of func)
39 r'(?:^[ \t]+\.(frame|f?mask|set).*?\n)+' # Mips+LLVM standard asm prologue
40 r'(?P<body>.*?)\n' # (body of the function)
41 r'(?:^[ \t]+\.(set|end).*?\n)+' # Mips+LLVM standard asm epilogue
42 r'(\$|\.L)func_end[0-9]+:\n', # $func_end0: (mips32 - O32) or
43 # .Lfunc_end0: (mips64 - NewABI)
44 flags=(re.M | re.S))
45
46ASM_FUNCTION_PPC_RE = re.compile(
47 r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@(?P=func)\n'
48 r'\.Lfunc_begin[0-9]+:\n'
49 r'(?:[ \t]+.cfi_startproc\n)?'
50 r'(?:\.Lfunc_[gl]ep[0-9]+:\n(?:[ \t]+.*?\n)*)*'
51 r'(?P<body>.*?)\n'
52 # This list is incomplete
53 r'(?:^[ \t]*(?:\.long[ \t]+[^\n]+|\.quad[ \t]+[^\n]+)\n)*'
54 r'.Lfunc_end[0-9]+:\n',
55 flags=(re.M | re.S))
56
57ASM_FUNCTION_RISCV_RE = re.compile(
58 r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@(?P=func)\n[^:]*?'
59 r'(?P<body>^##?[ \t]+[^:]+:.*?)\s*'
60 r'.Lfunc_end[0-9]+:\n',
61 flags=(re.M | re.S))
62
63ASM_FUNCTION_SYSTEMZ_RE = re.compile(
64 r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@(?P=func)\n'
65 r'[ \t]+.cfi_startproc\n'
66 r'(?P<body>.*?)\n'
67 r'.Lfunc_end[0-9]+:\n',
68 flags=(re.M | re.S))
69
70
71SCRUB_LOOP_COMMENT_RE = re.compile(
72 r'# =>This Inner Loop Header:.*|# in Loop:.*', flags=re.M)
73
74SCRUB_X86_SHUFFLES_RE = (
75 re.compile(
76 r'^(\s*\w+) [^#\n]+#+ ((?:[xyz]mm\d+|mem)( \{%k\d+\}( \{z\})?)? = .*)$',
77 flags=re.M))
Chandler Carruth6646bec2018-04-03 09:57:05 +000078SCRUB_X86_SPILL_RELOAD_RE = (
79 re.compile(
80 r'-?\d+\(%([er])[sb]p\)(.*(?:Spill|Reload))$',
81 flags=re.M))
Fangrui Songee4e2e72018-01-30 00:40:05 +000082SCRUB_X86_SP_RE = re.compile(r'\d+\(%(esp|rsp)\)')
83SCRUB_X86_RIP_RE = re.compile(r'[.\w]+\(%rip\)')
84SCRUB_X86_LCP_RE = re.compile(r'\.LCPI[0-9]+_[0-9]+')
85SCRUB_X86_RET_RE = re.compile(r'ret[l|q]')
86
87def scrub_asm_x86(asm, args):
88 # Scrub runs of whitespace out of the assembly, but leave the leading
89 # whitespace in place.
90 asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
91 # Expand the tabs used for indentation.
92 asm = string.expandtabs(asm, 2)
93 # Detect shuffle asm comments and hide the operands in favor of the comments.
94 asm = SCRUB_X86_SHUFFLES_RE.sub(r'\1 {{.*#+}} \2', asm)
Chandler Carruth6646bec2018-04-03 09:57:05 +000095 # Detect stack spills and reloads and hide their exact offset and whether
96 # they used the stack pointer or frame pointer.
Chandler Carruthff2f4fc2018-04-03 10:28:56 +000097 asm = SCRUB_X86_SPILL_RELOAD_RE.sub(r'{{[-0-9]+}}(%\1{{[sb]}}p)\2', asm)
Fangrui Songee4e2e72018-01-30 00:40:05 +000098 # Generically match the stack offset of a memory operand.
99 asm = SCRUB_X86_SP_RE.sub(r'{{[0-9]+}}(%\1)', asm)
100 # Generically match a RIP-relative memory operand.
101 asm = SCRUB_X86_RIP_RE.sub(r'{{.*}}(%rip)', asm)
102 # Generically match a LCP symbol.
103 asm = SCRUB_X86_LCP_RE.sub(r'{{\.LCPI.*}}', asm)
Fangrui Song0a301a12018-03-02 17:37:04 +0000104 if getattr(args, 'x86_extra_scrub', False):
Fangrui Songee4e2e72018-01-30 00:40:05 +0000105 # Avoid generating different checks for 32- and 64-bit because of 'retl' vs 'retq'.
106 asm = SCRUB_X86_RET_RE.sub(r'ret{{[l|q]}}', asm)
107 # Strip kill operands inserted into the asm.
108 asm = common.SCRUB_KILL_COMMENT_RE.sub('', asm)
109 # Strip trailing whitespace.
110 asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
111 return asm
112
113def scrub_asm_arm_eabi(asm, args):
114 # Scrub runs of whitespace out of the assembly, but leave the leading
115 # whitespace in place.
116 asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
117 # Expand the tabs used for indentation.
118 asm = string.expandtabs(asm, 2)
119 # Strip kill operands inserted into the asm.
120 asm = common.SCRUB_KILL_COMMENT_RE.sub('', asm)
121 # Strip trailing whitespace.
122 asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
123 return asm
124
125def scrub_asm_powerpc64(asm, args):
126 # Scrub runs of whitespace out of the assembly, but leave the leading
127 # whitespace in place.
128 asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
129 # Expand the tabs used for indentation.
130 asm = string.expandtabs(asm, 2)
131 # Stripe unimportant comments
132 asm = SCRUB_LOOP_COMMENT_RE.sub(r'', asm)
133 # Strip trailing whitespace.
134 asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
135 return asm
136
137def scrub_asm_mips(asm, args):
138 # Scrub runs of whitespace out of the assembly, but leave the leading
139 # whitespace in place.
140 asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
141 # Expand the tabs used for indentation.
142 asm = string.expandtabs(asm, 2)
143 # Strip trailing whitespace.
144 asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
145 return asm
146
147def scrub_asm_riscv(asm, args):
148 # Scrub runs of whitespace out of the assembly, but leave the leading
149 # whitespace in place.
150 asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
151 # Expand the tabs used for indentation.
152 asm = string.expandtabs(asm, 2)
153 # Strip trailing whitespace.
154 asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
155 return asm
156
157def scrub_asm_systemz(asm, args):
158 # Scrub runs of whitespace out of the assembly, but leave the leading
159 # whitespace in place.
160 asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
161 # Expand the tabs used for indentation.
162 asm = string.expandtabs(asm, 2)
163 # Strip trailing whitespace.
164 asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
165 return asm
166
167
168def build_function_body_dictionary_for_triple(args, raw_tool_output, triple, prefixes, func_dict):
169 target_handlers = {
170 'x86_64': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
171 'i686': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
172 'x86': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
173 'i386': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
174 'aarch64': (scrub_asm_arm_eabi, ASM_FUNCTION_AARCH64_RE),
175 'arm-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
176 'thumb-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
177 'thumbv6': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
178 'thumbv6-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
179 'thumbv6t2': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
180 'thumbv6t2-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
181 'thumbv6m': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
182 'thumbv6m-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
183 'thumbv7': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
184 'thumbv7-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
185 'thumbv7m': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
186 'thumbv7m-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
187 'thumbv8-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
188 'thumbv8m.base': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
189 'thumbv8m.main': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
190 'armv6': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
191 'armv7': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
192 'armv7-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
193 'armeb-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
194 'armv7eb-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
195 'armv7eb': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
196 'mips': (scrub_asm_mips, ASM_FUNCTION_MIPS_RE),
197 'powerpc64': (scrub_asm_powerpc64, ASM_FUNCTION_PPC_RE),
198 'powerpc64le': (scrub_asm_powerpc64, ASM_FUNCTION_PPC_RE),
199 'riscv32': (scrub_asm_riscv, ASM_FUNCTION_RISCV_RE),
200 'riscv64': (scrub_asm_riscv, ASM_FUNCTION_RISCV_RE),
201 's390x': (scrub_asm_systemz, ASM_FUNCTION_SYSTEMZ_RE),
202 }
203 handlers = None
204 for prefix, s in target_handlers.items():
205 if triple.startswith(prefix):
206 handlers = s
207 break
208 else:
209 raise KeyError('Triple %r is not supported' % (triple))
210
211 scrubber, function_re = handlers
212 common.build_function_body_dictionary(
213 function_re, scrubber, [args], raw_tool_output, prefixes,
214 func_dict, args.verbose)
Fangrui Song4f0f4262018-02-10 05:01:33 +0000215
216##### Generator of assembly CHECK lines
217
Simon Pilgrim702ec042018-04-05 09:50:58 +0000218def add_asm_checks(output_lines, comment_marker, prefix_list, func_dict, func_name):
219 # Label format is based on ASM string.
220 check_label_format = '{} %s-LABEL: %s:'.format(comment_marker)
Simon Pilgrim82962292018-04-05 10:48:38 +0000221 common.add_checks(output_lines, comment_marker, prefix_list, func_dict, func_name, check_label_format, True)