1import re
2import sys
3
4from . import common
5
6if sys.version_info[0] > 2:
7  class string:
8    expandtabs = str.expandtabs
9else:
10  import string
11
12# RegEx: this is where the magic happens.
13
14##### Assembly parser
15
16ASM_FUNCTION_X86_RE = re.compile(
17    r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@(?P=func)\n(?:\s*.Lfunc_begin[^:\n]*:\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)?'  # drop optional cfi noise
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_AMDGPU_RE = re.compile(
38    r'^_?(?P<func>[^:]+):[ \t]*;+[ \t]*@(?P=func)\n[^:]*?'
39    r'(?P<body>.*?)\n' # (body of the function)
40    # This list is incomplete
41    r'.Lfunc_end[0-9]+:\n',
42    flags=(re.M | re.S))
43
44ASM_FUNCTION_MIPS_RE = re.compile(
45    r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@(?P=func)\n[^:]*?' # f: (name of func)
46    r'(?:^[ \t]+\.(frame|f?mask|set).*?\n)+'  # Mips+LLVM standard asm prologue
47    r'(?P<body>.*?)\n'                        # (body of the function)
48    r'(?:^[ \t]+\.(set|end).*?\n)+'           # Mips+LLVM standard asm epilogue
49    r'(\$|\.L)func_end[0-9]+:\n',             # $func_end0: (mips32 - O32) or
50                                              # .Lfunc_end0: (mips64 - NewABI)
51    flags=(re.M | re.S))
52
53ASM_FUNCTION_PPC_RE = re.compile(
54    r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@(?P=func)\n'
55    r'\.Lfunc_begin[0-9]+:\n'
56    r'(?:[ \t]+.cfi_startproc\n)?'
57    r'(?:\.Lfunc_[gl]ep[0-9]+:\n(?:[ \t]+.*?\n)*)*'
58    r'(?P<body>.*?)\n'
59    # This list is incomplete
60    r'(?:^[ \t]*(?:\.long[ \t]+[^\n]+|\.quad[ \t]+[^\n]+)\n)*'
61    r'.Lfunc_end[0-9]+:\n',
62    flags=(re.M | re.S))
63
64ASM_FUNCTION_RISCV_RE = re.compile(
65    r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@(?P=func)\n[^:]*?'
66    r'(?P<body>^##?[ \t]+[^:]+:.*?)\s*'
67    r'.Lfunc_end[0-9]+:\n',
68    flags=(re.M | re.S))
69
70ASM_FUNCTION_SPARC_RE = re.compile(
71    r'^_?(?P<func>[^:]+):[ \t]*!+[ \t]*@(?P=func)\n'
72    r'(?P<body>.*?)\s*'
73    r'.Lfunc_end[0-9]+:\n',
74    flags=(re.M | re.S))
75
76ASM_FUNCTION_SYSTEMZ_RE = re.compile(
77    r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@(?P=func)\n'
78    r'[ \t]+.cfi_startproc\n'
79    r'(?P<body>.*?)\n'
80    r'.Lfunc_end[0-9]+:\n',
81    flags=(re.M | re.S))
82
83
84SCRUB_LOOP_COMMENT_RE = re.compile(
85    r'# =>This Inner Loop Header:.*|# in Loop:.*', flags=re.M)
86
87SCRUB_X86_SHUFFLES_RE = (
88    re.compile(
89        r'^(\s*\w+) [^#\n]+#+ ((?:[xyz]mm\d+|mem)( \{%k\d+\}( \{z\})?)? = .*)$',
90        flags=re.M))
91SCRUB_X86_SPILL_RELOAD_RE = (
92    re.compile(
93        r'-?\d+\(%([er])[sb]p\)(.*(?:Spill|Reload))$',
94        flags=re.M))
95SCRUB_X86_SP_RE = re.compile(r'\d+\(%(esp|rsp)\)')
96SCRUB_X86_RIP_RE = re.compile(r'[.\w]+\(%rip\)')
97SCRUB_X86_LCP_RE = re.compile(r'\.LCPI[0-9]+_[0-9]+')
98SCRUB_X86_RET_RE = re.compile(r'ret[l|q]')
99
100def scrub_asm_x86(asm, args):
101  # Scrub runs of whitespace out of the assembly, but leave the leading
102  # whitespace in place.
103  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
104  # Expand the tabs used for indentation.
105  asm = string.expandtabs(asm, 2)
106  # Detect shuffle asm comments and hide the operands in favor of the comments.
107  asm = SCRUB_X86_SHUFFLES_RE.sub(r'\1 {{.*#+}} \2', asm)
108  # Detect stack spills and reloads and hide their exact offset and whether
109  # they used the stack pointer or frame pointer.
110  asm = SCRUB_X86_SPILL_RELOAD_RE.sub(r'{{[-0-9]+}}(%\1{{[sb]}}p)\2', asm)
111  # Generically match the stack offset of a memory operand.
112  asm = SCRUB_X86_SP_RE.sub(r'{{[0-9]+}}(%\1)', asm)
113  if getattr(args, 'x86_scrub_rip', False):
114    # Generically match a RIP-relative memory operand.
115    asm = SCRUB_X86_RIP_RE.sub(r'{{.*}}(%rip)', asm)
116  # Generically match a LCP symbol.
117  asm = SCRUB_X86_LCP_RE.sub(r'{{\.LCPI.*}}', asm)
118  if getattr(args, 'extra_scrub', False):
119    # Avoid generating different checks for 32- and 64-bit because of 'retl' vs 'retq'.
120    asm = SCRUB_X86_RET_RE.sub(r'ret{{[l|q]}}', asm)
121  # Strip kill operands inserted into the asm.
122  asm = common.SCRUB_KILL_COMMENT_RE.sub('', asm)
123  # Strip trailing whitespace.
124  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
125  return asm
126
127def scrub_asm_amdgpu(asm, args):
128  # Scrub runs of whitespace out of the assembly, but leave the leading
129  # whitespace in place.
130  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
131  # Expand the tabs used for indentation.
132  asm = string.expandtabs(asm, 2)
133  # Strip trailing whitespace.
134  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
135  return asm
136
137def scrub_asm_arm_eabi(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 kill operands inserted into the asm.
144  asm = common.SCRUB_KILL_COMMENT_RE.sub('', asm)
145  # Strip trailing whitespace.
146  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
147  return asm
148
149def scrub_asm_powerpc64(asm, args):
150  # Scrub runs of whitespace out of the assembly, but leave the leading
151  # whitespace in place.
152  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
153  # Expand the tabs used for indentation.
154  asm = string.expandtabs(asm, 2)
155  # Stripe unimportant comments
156  asm = SCRUB_LOOP_COMMENT_RE.sub(r'', asm)
157  # Strip trailing whitespace.
158  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
159  return asm
160
161def scrub_asm_mips(asm, args):
162  # Scrub runs of whitespace out of the assembly, but leave the leading
163  # whitespace in place.
164  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
165  # Expand the tabs used for indentation.
166  asm = string.expandtabs(asm, 2)
167  # Strip trailing whitespace.
168  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
169  return asm
170
171def scrub_asm_riscv(asm, args):
172  # Scrub runs of whitespace out of the assembly, but leave the leading
173  # whitespace in place.
174  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
175  # Expand the tabs used for indentation.
176  asm = string.expandtabs(asm, 2)
177  # Strip trailing whitespace.
178  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
179  return asm
180
181def scrub_asm_sparc(asm, args):
182  # Scrub runs of whitespace out of the assembly, but leave the leading
183  # whitespace in place.
184  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
185  # Expand the tabs used for indentation.
186  asm = string.expandtabs(asm, 2)
187  # Strip trailing whitespace.
188  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
189  return asm
190
191def scrub_asm_systemz(asm, args):
192  # Scrub runs of whitespace out of the assembly, but leave the leading
193  # whitespace in place.
194  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
195  # Expand the tabs used for indentation.
196  asm = string.expandtabs(asm, 2)
197  # Strip trailing whitespace.
198  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
199  return asm
200
201
202def build_function_body_dictionary_for_triple(args, raw_tool_output, triple, prefixes, func_dict):
203  target_handlers = {
204      'x86_64': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
205      'i686': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
206      'x86': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
207      'i386': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
208      'aarch64': (scrub_asm_arm_eabi, ASM_FUNCTION_AARCH64_RE),
209      'r600': (scrub_asm_amdgpu, ASM_FUNCTION_AMDGPU_RE),
210      'amdgcn': (scrub_asm_amdgpu, ASM_FUNCTION_AMDGPU_RE),
211      'arm-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
212      'thumb-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
213      'thumbv6': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
214      'thumbv6-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
215      'thumbv6t2': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
216      'thumbv6t2-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
217      'thumbv6m': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
218      'thumbv6m-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
219      'thumbv7': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
220      'thumbv7-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
221      'thumbv7m': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
222      'thumbv7m-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
223      'thumbv8-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
224      'thumbv8m.base': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
225      'thumbv8m.main': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
226      'armv6': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
227      'armv7': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
228      'armv7-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
229      'armeb-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
230      'armv7eb-eabi': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
231      'armv7eb': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
232      'mips': (scrub_asm_mips, ASM_FUNCTION_MIPS_RE),
233      'powerpc64': (scrub_asm_powerpc64, ASM_FUNCTION_PPC_RE),
234      'powerpc64le': (scrub_asm_powerpc64, ASM_FUNCTION_PPC_RE),
235      'riscv32': (scrub_asm_riscv, ASM_FUNCTION_RISCV_RE),
236      'riscv64': (scrub_asm_riscv, ASM_FUNCTION_RISCV_RE),
237      'sparc': (scrub_asm_sparc, ASM_FUNCTION_SPARC_RE),
238      'sparcv9': (scrub_asm_sparc, ASM_FUNCTION_SPARC_RE),
239      's390x': (scrub_asm_systemz, ASM_FUNCTION_SYSTEMZ_RE),
240  }
241  handlers = None
242  for prefix, s in target_handlers.items():
243    if triple.startswith(prefix):
244      handlers = s
245      break
246  else:
247    raise KeyError('Triple %r is not supported' % (triple))
248
249  scrubber, function_re = handlers
250  common.build_function_body_dictionary(
251          function_re, scrubber, [args], raw_tool_output, prefixes,
252          func_dict, args.verbose)
253
254##### Generator of assembly CHECK lines
255
256def add_asm_checks(output_lines, comment_marker, prefix_list, func_dict, func_name):
257  # Label format is based on ASM string.
258  check_label_format = '{} %s-LABEL: %s:'.format(comment_marker)
259  common.add_checks(output_lines, comment_marker, prefix_list, func_dict, func_name, check_label_format, True, False)
260