xref: /llvm-project/llvm/utils/UpdateTestChecks/asm.py (revision 15be15073ef6eb2c4fd734917322909d405f16e2)
1from __future__ import print_function
2import re
3import sys
4
5from . import common
6
7if sys.version_info[0] > 2:
8  class string:
9    expandtabs = str.expandtabs
10else:
11  import string
12
13# RegEx: this is where the magic happens.
14
15##### Assembly parser
16
17ASM_FUNCTION_X86_RE = re.compile(
18    r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*(@"?(?P=func)"?| -- Begin function (?P=func))\n(?:\s*\.?Lfunc_begin[^:\n]*:\n)?'
19    r'(?:\.L[^$]+\$local:\n)?'      # drop .L<func>$local:
20    r'(?:[ \t]+.cfi_startproc\n|.seh_proc[^\n]+\n)?'  # drop optional cfi
21    r'(?P<body>^##?[ \t]+[^:]+:.*?)\s*'
22    r'^\s*(?:[^:\n]+?:\s*\n\s*\.size|\.cfi_endproc|\.globl|\.comm|\.(?:sub)?section|#+ -- End function)',
23    flags=(re.M | re.S))
24
25ASM_FUNCTION_ARM_RE = re.compile(
26        r'^(?P<func>[0-9a-zA-Z_]+):\n' # f: (name of function)
27        r'\s+\.fnstart\n' # .fnstart
28        r'(?P<body>.*?)\n' # (body of the function)
29        r'.Lfunc_end[0-9]+:', # .Lfunc_end0: or # -- End function
30        flags=(re.M | re.S))
31
32ASM_FUNCTION_AARCH64_RE = re.compile(
33     r'^_?(?P<func>[^:]+):[ \t]*\/\/[ \t]*@"?(?P=func)"?( (Function|Tail Call))?\n'
34     r'(?:[ \t]+.cfi_startproc\n)?'  # drop optional cfi noise
35     r'(?P<body>.*?)\n'
36     # This list is incomplete
37     r'.Lfunc_end[0-9]+:\n',
38     flags=(re.M | re.S))
39
40ASM_FUNCTION_AMDGPU_RE = re.compile(
41    r'^_?(?P<func>[^:]+):[ \t]*;+[ \t]*@"?(?P=func)"?\n[^:]*?'
42    r'(?P<body>.*?)\n' # (body of the function)
43    # This list is incomplete
44    r'^\s*(\.Lfunc_end[0-9]+:\n|\.section)',
45    flags=(re.M | re.S))
46
47ASM_FUNCTION_HEXAGON_RE = re.compile(
48    r'^_?(?P<func>[^:]+):[ \t]*//[ \t]*@"?(?P=func)"?\n[^:]*?'
49    r'(?P<body>.*?)\n' # (body of the function)
50    # This list is incomplete
51    r'.Lfunc_end[0-9]+:\n',
52    flags=(re.M | re.S))
53
54ASM_FUNCTION_M68K_RE = re.compile(
55    r'^_?(?P<func>[^:]+):[ \t]*;[ \t]*@"?(?P=func)"?\n'
56    r'(?P<body>.*?)\s*' # (body of the function)
57    r'.Lfunc_end[0-9]+:\n',
58    flags=(re.M | re.S))
59
60ASM_FUNCTION_MIPS_RE = re.compile(
61    r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@"?(?P=func)"?\n[^:]*?' # f: (name of func)
62    r'(?:^[ \t]+\.(frame|f?mask|set).*?\n)+'  # Mips+LLVM standard asm prologue
63    r'(?P<body>.*?)\n'                        # (body of the function)
64    # Mips+LLVM standard asm epilogue
65    r'(?:(^[ \t]+\.set[^\n]*?\n)*^[ \t]+\.end.*?\n)'
66    r'(\$|\.L)func_end[0-9]+:\n',             # $func_end0: (mips32 - O32) or
67                                              # .Lfunc_end0: (mips64 - NewABI)
68    flags=(re.M | re.S))
69
70ASM_FUNCTION_MSP430_RE = re.compile(
71    r'^_?(?P<func>[^:]+):[ \t]*;+[ \t]*@"?(?P=func)"?\n[^:]*?'
72    r'(?P<body>.*?)\n'
73    r'(\$|\.L)func_end[0-9]+:\n',             # $func_end0:
74    flags=(re.M | re.S))
75
76ASM_FUNCTION_AVR_RE = re.compile(
77    r'^_?(?P<func>[^:]+):[ \t]*;+[ \t]*@"?(?P=func)"?\n[^:]*?'
78    r'(?P<body>.*?)\n'
79    r'.Lfunc_end[0-9]+:\n',
80    flags=(re.M | re.S))
81
82ASM_FUNCTION_PPC_RE = re.compile(
83    r'#[ \-\t]*Begin function (?P<func>[^.:]+)\n'
84    r'.*?'
85    r'^[_.]?(?P=func):(?:[ \t]*#+[ \t]*@"?(?P=func)"?)?\n'
86    r'(?:^[^#]*\n)*'
87    r'(?P<body>.*?)\n'
88    # This list is incomplete
89    r'(?:^[ \t]*(?:\.(?:long|quad|v?byte)[ \t]+[^\n]+)\n)*'
90    r'(?:\.Lfunc_end|L\.\.(?P=func))[0-9]+:\n',
91    flags=(re.M | re.S))
92
93ASM_FUNCTION_RISCV_RE = re.compile(
94    r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@"?(?P=func)"?\n'
95    r'(?:\s*\.?L(?P=func)\$local:\n)?'  # optional .L<func>$local: due to -fno-semantic-interposition
96    r'(?:\s*\.?Lfunc_begin[^:\n]*:\n)?[^:]*?'
97    r'(?P<body>^##?[ \t]+[^:]+:.*?)\s*'
98    r'.Lfunc_end[0-9]+:\n',
99    flags=(re.M | re.S))
100
101ASM_FUNCTION_LANAI_RE = re.compile(
102    r'^_?(?P<func>[^:]+):[ \t]*!+[ \t]*@"?(?P=func)"?\n'
103    r'(?:[ \t]+.cfi_startproc\n)?'  # drop optional cfi noise
104    r'(?P<body>.*?)\s*'
105    r'.Lfunc_end[0-9]+:\n',
106    flags=(re.M | re.S))
107
108ASM_FUNCTION_SPARC_RE = re.compile(
109    r'^_?(?P<func>[^:]+):[ \t]*!+[ \t]*@"?(?P=func)"?\n'
110    r'(?P<body>.*?)\s*'
111    r'.Lfunc_end[0-9]+:\n',
112    flags=(re.M | re.S))
113
114ASM_FUNCTION_SYSTEMZ_RE = re.compile(
115    r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@"?(?P=func)"?\n'
116    r'[ \t]+.cfi_startproc\n'
117    r'(?P<body>.*?)\n'
118    r'.Lfunc_end[0-9]+:\n',
119    flags=(re.M | re.S))
120
121ASM_FUNCTION_AARCH64_DARWIN_RE = re.compile(
122     r'^_(?P<func>[^:]+):[ \t]*;[ \t]@"?(?P=func)"?\n'
123     r'([ \t]*.cfi_startproc\n[\s]*)?'
124     r'(?P<body>.*?)'
125     r'([ \t]*.cfi_endproc\n[\s]*)?'
126     r'^[ \t]*;[ \t]--[ \t]End[ \t]function',
127     flags=(re.M | re.S))
128
129ASM_FUNCTION_ARM_DARWIN_RE = re.compile(
130     r'^[ \t]*\.globl[ \t]*_(?P<func>[^ \t])[ \t]*@[ \t]--[ \t]Begin[ \t]function[ \t]"?(?P=func)"?'
131     r'(?P<directives>.*?)'
132     r'^_(?P=func):\n[ \t]*'
133     r'(?P<body>.*?)'
134     r'^[ \t]*@[ \t]--[ \t]End[ \t]function',
135     flags=(re.M | re.S ))
136
137ASM_FUNCTION_ARM_MACHO_RE = re.compile(
138     r'^_(?P<func>[^:]+):[ \t]*\n'
139     r'([ \t]*.cfi_startproc\n[ \t]*)?'
140     r'(?P<body>.*?)\n'
141     r'[ \t]*\.cfi_endproc\n',
142     flags=(re.M | re.S))
143
144ASM_FUNCTION_ARM_IOS_RE = re.compile(
145     r'^_(?P<func>[^:]+):\n'
146     r'(?P<body>.*?)'
147     r'^[ \t]*@[ \t]--[ \t]End[ \t]function',
148     flags=(re.M | re.S))
149
150ASM_FUNCTION_WASM32_RE = re.compile(
151    r'^_?(?P<func>[^:]+):[ \t]*#+[ \t]*@"?(?P=func)"?\n'
152    r'(?P<body>.*?)\n'
153    r'^\s*(\.Lfunc_end[0-9]+:\n|end_function)',
154    flags=(re.M | re.S))
155
156SCRUB_X86_SHUFFLES_RE = (
157    re.compile(
158        r'^(\s*\w+) [^#\n]+#+ ((?:[xyz]mm\d+|mem)( \{%k\d+\}( \{z\})?)? = .*)$',
159        flags=re.M))
160
161SCRUB_X86_SHUFFLES_NO_MEM_RE = (
162    re.compile(
163        r'^(\s*\w+) [^#\n]+#+ ((?:[xyz]mm\d+|mem)( \{%k\d+\}( \{z\})?)? = (?!.*(?:mem)).*)$',
164        flags=re.M))
165
166SCRUB_X86_SPILL_RELOAD_RE = (
167    re.compile(
168        r'-?\d+\(%([er])[sb]p\)(.*(?:Spill|Reload))$',
169        flags=re.M))
170SCRUB_X86_SP_RE = re.compile(r'\d+\(%(esp|rsp)\)')
171SCRUB_X86_RIP_RE = re.compile(r'[.\w]+\(%rip\)')
172SCRUB_X86_LCP_RE = re.compile(r'\.?LCPI[0-9]+_[0-9]+')
173SCRUB_X86_RET_RE = re.compile(r'ret[l|q]')
174
175def scrub_asm_x86(asm, args):
176  # Scrub runs of whitespace out of the assembly, but leave the leading
177  # whitespace in place.
178  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
179  # Expand the tabs used for indentation.
180  asm = string.expandtabs(asm, 2)
181
182  # Detect shuffle asm comments and hide the operands in favor of the comments.
183  if getattr(args, 'no_x86_scrub_mem_shuffle', True):
184    asm = SCRUB_X86_SHUFFLES_NO_MEM_RE.sub(r'\1 {{.*#+}} \2', asm)
185  else:
186    asm = SCRUB_X86_SHUFFLES_RE.sub(r'\1 {{.*#+}} \2', asm)
187
188  # Detect stack spills and reloads and hide their exact offset and whether
189  # they used the stack pointer or frame pointer.
190  asm = SCRUB_X86_SPILL_RELOAD_RE.sub(r'{{[-0-9]+}}(%\1{{[sb]}}p)\2', asm)
191  if getattr(args, 'x86_scrub_sp', True):
192    # Generically match the stack offset of a memory operand.
193    asm = SCRUB_X86_SP_RE.sub(r'{{[0-9]+}}(%\1)', asm)
194  if getattr(args, 'x86_scrub_rip', False):
195    # Generically match a RIP-relative memory operand.
196    asm = SCRUB_X86_RIP_RE.sub(r'{{.*}}(%rip)', asm)
197  # Generically match a LCP symbol.
198  asm = SCRUB_X86_LCP_RE.sub(r'{{\.?LCPI[0-9]+_[0-9]+}}', asm)
199  if getattr(args, 'extra_scrub', False):
200    # Avoid generating different checks for 32- and 64-bit because of 'retl' vs 'retq'.
201    asm = SCRUB_X86_RET_RE.sub(r'ret{{[l|q]}}', asm)
202  # Strip kill operands inserted into the asm.
203  asm = common.SCRUB_KILL_COMMENT_RE.sub('', asm)
204  # Strip trailing whitespace.
205  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
206  return asm
207
208def scrub_asm_amdgpu(asm, args):
209  # Scrub runs of whitespace out of the assembly, but leave the leading
210  # whitespace in place.
211  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
212  # Expand the tabs used for indentation.
213  asm = string.expandtabs(asm, 2)
214  # Strip trailing whitespace.
215  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
216  return asm
217
218def scrub_asm_arm_eabi(asm, args):
219  # Scrub runs of whitespace out of the assembly, but leave the leading
220  # whitespace in place.
221  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
222  # Expand the tabs used for indentation.
223  asm = string.expandtabs(asm, 2)
224  # Strip kill operands inserted into the asm.
225  asm = common.SCRUB_KILL_COMMENT_RE.sub('', asm)
226  # Strip trailing whitespace.
227  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
228  return asm
229
230def scrub_asm_hexagon(asm, args):
231  # Scrub runs of whitespace out of the assembly, but leave the leading
232  # whitespace in place.
233  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
234  # Expand the tabs used for indentation.
235  asm = string.expandtabs(asm, 2)
236  # Strip trailing whitespace.
237  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
238  return asm
239
240def scrub_asm_powerpc(asm, args):
241  # Scrub runs of whitespace out of the assembly, but leave the leading
242  # whitespace in place.
243  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
244  # Expand the tabs used for indentation.
245  asm = string.expandtabs(asm, 2)
246  # Strip unimportant comments, but leave the token '#' in place.
247  asm = common.SCRUB_LOOP_COMMENT_RE.sub(r'#', asm)
248  # Strip trailing whitespace.
249  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
250  # Strip the tailing token '#', except the line only has token '#'.
251  asm = common.SCRUB_TAILING_COMMENT_TOKEN_RE.sub(r'', asm)
252  return asm
253
254def scrub_asm_m68k(asm, args):
255  # Scrub runs of whitespace out of the assembly, but leave the leading
256  # whitespace in place.
257  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
258  # Expand the tabs used for indentation.
259  asm = string.expandtabs(asm, 2)
260  # Strip trailing whitespace.
261  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
262  return asm
263
264def scrub_asm_mips(asm, args):
265  # Scrub runs of whitespace out of the assembly, but leave the leading
266  # whitespace in place.
267  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
268  # Expand the tabs used for indentation.
269  asm = string.expandtabs(asm, 2)
270  # Strip trailing whitespace.
271  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
272  return asm
273
274def scrub_asm_msp430(asm, args):
275  # Scrub runs of whitespace out of the assembly, but leave the leading
276  # whitespace in place.
277  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
278  # Expand the tabs used for indentation.
279  asm = string.expandtabs(asm, 2)
280  # Strip trailing whitespace.
281  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
282  return asm
283
284def scrub_asm_avr(asm, args):
285  # Scrub runs of whitespace out of the assembly, but leave the leading
286  # whitespace in place.
287  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
288  # Expand the tabs used for indentation.
289  asm = string.expandtabs(asm, 2)
290  # Strip trailing whitespace.
291  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
292  return asm
293
294def scrub_asm_riscv(asm, args):
295  # Scrub runs of whitespace out of the assembly, but leave the leading
296  # whitespace in place.
297  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
298  # Expand the tabs used for indentation.
299  asm = string.expandtabs(asm, 2)
300  # Strip trailing whitespace.
301  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
302  return asm
303
304def scrub_asm_lanai(asm, args):
305  # Scrub runs of whitespace out of the assembly, but leave the leading
306  # whitespace in place.
307  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
308  # Expand the tabs used for indentation.
309  asm = string.expandtabs(asm, 2)
310  # Strip trailing whitespace.
311  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
312  return asm
313
314def scrub_asm_sparc(asm, args):
315  # Scrub runs of whitespace out of the assembly, but leave the leading
316  # whitespace in place.
317  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
318  # Expand the tabs used for indentation.
319  asm = string.expandtabs(asm, 2)
320  # Strip trailing whitespace.
321  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
322  return asm
323
324def scrub_asm_systemz(asm, args):
325  # Scrub runs of whitespace out of the assembly, but leave the leading
326  # whitespace in place.
327  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
328  # Expand the tabs used for indentation.
329  asm = string.expandtabs(asm, 2)
330  # Strip trailing whitespace.
331  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
332  return asm
333
334def scrub_asm_wasm32(asm, args):
335  # Scrub runs of whitespace out of the assembly, but leave the leading
336  # whitespace in place.
337  asm = common.SCRUB_WHITESPACE_RE.sub(r' ', asm)
338  # Expand the tabs used for indentation.
339  asm = string.expandtabs(asm, 2)
340  # Strip trailing whitespace.
341  asm = common.SCRUB_TRAILING_WHITESPACE_RE.sub(r'', asm)
342  return asm
343
344def get_triple_from_march(march):
345  triples = {
346      'amdgcn': 'amdgcn',
347      'r600': 'r600',
348      'mips': 'mips',
349      'sparc': 'sparc',
350      'hexagon': 'hexagon',
351  }
352  for prefix, triple in triples.items():
353    if march.startswith(prefix):
354      return triple
355  print("Cannot find a triple. Assume 'x86'", file=sys.stderr)
356  return 'x86'
357
358def get_run_handler(triple):
359  target_handlers = {
360      'i686': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
361      'x86': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
362      'i386': (scrub_asm_x86, ASM_FUNCTION_X86_RE),
363      'aarch64': (scrub_asm_arm_eabi, ASM_FUNCTION_AARCH64_RE),
364      'aarch64-apple-darwin': (scrub_asm_arm_eabi, ASM_FUNCTION_AARCH64_DARWIN_RE),
365      'hexagon': (scrub_asm_hexagon, ASM_FUNCTION_HEXAGON_RE),
366      'r600': (scrub_asm_amdgpu, ASM_FUNCTION_AMDGPU_RE),
367      'amdgcn': (scrub_asm_amdgpu, ASM_FUNCTION_AMDGPU_RE),
368      'arm': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
369      'arm64': (scrub_asm_arm_eabi, ASM_FUNCTION_AARCH64_RE),
370      'arm64e': (scrub_asm_arm_eabi, ASM_FUNCTION_AARCH64_DARWIN_RE),
371      'arm64-apple-ios': (scrub_asm_arm_eabi, ASM_FUNCTION_AARCH64_DARWIN_RE),
372      'armv7-apple-ios' : (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_IOS_RE),
373      'armv7-apple-darwin': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_DARWIN_RE),
374      'thumb': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_RE),
375      'thumb-macho': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_MACHO_RE),
376      'thumbv5-macho': (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_MACHO_RE),
377      'thumbv7-apple-ios' : (scrub_asm_arm_eabi, ASM_FUNCTION_ARM_IOS_RE),
378      'm68k': (scrub_asm_m68k, ASM_FUNCTION_M68K_RE),
379      'mips': (scrub_asm_mips, ASM_FUNCTION_MIPS_RE),
380      'msp430': (scrub_asm_msp430, ASM_FUNCTION_MSP430_RE),
381      'avr': (scrub_asm_avr, ASM_FUNCTION_AVR_RE),
382      'ppc32': (scrub_asm_powerpc, ASM_FUNCTION_PPC_RE),
383      'powerpc': (scrub_asm_powerpc, ASM_FUNCTION_PPC_RE),
384      'riscv32': (scrub_asm_riscv, ASM_FUNCTION_RISCV_RE),
385      'riscv64': (scrub_asm_riscv, ASM_FUNCTION_RISCV_RE),
386      'lanai': (scrub_asm_lanai, ASM_FUNCTION_LANAI_RE),
387      'sparc': (scrub_asm_sparc, ASM_FUNCTION_SPARC_RE),
388      's390x': (scrub_asm_systemz, ASM_FUNCTION_SYSTEMZ_RE),
389      'wasm32': (scrub_asm_wasm32, ASM_FUNCTION_WASM32_RE),
390  }
391  handler = None
392  best_prefix = ''
393  for prefix, s in target_handlers.items():
394    if triple.startswith(prefix) and len(prefix) > len(best_prefix):
395      handler = s
396      best_prefix = prefix
397
398  if handler is None:
399    raise KeyError('Triple %r is not supported' % (triple))
400
401  return handler
402
403##### Generator of assembly CHECK lines
404
405def add_asm_checks(output_lines, comment_marker, prefix_list, func_dict, func_name):
406  # Label format is based on ASM string.
407  check_label_format = '{} %s-LABEL: %s%s:'.format(comment_marker)
408  global_vars_seen_dict = {}
409  common.add_checks(output_lines, comment_marker, prefix_list, func_dict, func_name, check_label_format, True, False, global_vars_seen_dict)
410