LLVM IR parsing error

As reported in numba/numba#2158, @cgohlke saw an LLVM IR parse error in Numba 0.29 / llvmlite 0.14.0:

X:\Python35\lib\site-packages\numba\targets\codegen.py in add_ir_module(self, ir_module)
    147         assert isinstance(ir_module, llvmir.Module)
    148         ir = cgutils.normalize_ir_text(str(ir_module))
--> 149         ll_module = ll.parse_assembly(ir)
    150         ll_module.name = ir_module.name
    151         ll_module.verify()

X:\Python35\lib\site-packages\llvmlite\binding\module.py in parse_assembly(llvmir)
     20         if errmsg:
     21             mod.close()
---> 22             raise RuntimeError("LLVM IR parsing error\n{0}".format(errmsg))
     23     return mod
     24 

RuntimeError: Failed at object (object mode frontend)
Failed at object (object mode backend)
LLVM IR parsing error
<string>:767:55: error: expected type
  %".600" = call i8* (i64) @"PyLong_FromSsize_t"(i64 (0, 0))

[yuan76316]

Hi I encounter the same issue, when I pass a list which’s length is 1. Here is my code:

class F1Optimizer():
    def __init__(self):
        pass

    @staticmethod
    @jit
    def get_expectations(P, pNone=None):
        expectations = []
        P = np.sort(P)[::-1]

        n = np.array(P).shape[0]
        DP_C = np.zeros((n + 2, n + 1))
        if pNone is None:
            pNone = (1.0 - P).prod()

        DP_C[0][0] = 1.0
        for j in range(1, n):
            DP_C[0][j] = (1.0 - P[j - 1]) * DP_C[0, j - 1]

        for i in range(1, n + 1):
            DP_C[i, i] = DP_C[i - 1, i - 1] * P[i - 1]
            for j in range(i + 1, n + 1):
                DP_C[i, j] = P[j - 1] * DP_C[i - 1, j - 1] + (1.0 - P[j - 1]) * DP_C[i, j - 1]

        DP_S = np.zeros((2 * n + 1,))
        DP_SNone = np.zeros((2 * n + 1,))
        for i in range(1, 2 * n + 1):
            DP_S[i] = 1. / (1. * i)
            DP_SNone[i] = 1. / (1. * i + 1)
        for k in range(n + 1)[::-1]:
            f1 = 0
            f1None = 0
            for k1 in range(n + 1):
                f1 += 2 * k1 * DP_C[k1][k] * DP_S[k + k1]
                f1None += 2 * k1 * DP_C[k1][k] * DP_SNone[k + k1]
            for i in range(1, 2 * k - 1):
                DP_S[i] = (1 - P[k - 1]) * DP_S[i] + P[k - 1] * DP_S[i + 1]
                DP_SNone[i] = (1 - P[k - 1]) * DP_SNone[i] + P[k - 1] * DP_SNone[i + 1]
            expectations.append([f1None + 2 * pNone / (2 + k), f1])

        return np.array(expectations[::-1]).T

    @staticmethod
    @jit
    def maximize_expectation(P, pNone=None):
        expectations = F1Optimizer.get_expectations(P, pNone)

        ix_max = np.unravel_index(expectations.argmax(), expectations.shape)
        max_f1 = expectations[ix_max]

        predNone = True if ix_max[0] == 0 else False
        best_k = ix_max[1]

        return best_k, predNone, max_f1

    @staticmethod
    def _F1(tp, fp, fn):
        return 2 * tp / (2 * tp + fp + fn)

    @staticmethod
    def _Fbeta(tp, fp, fn, beta=1.0):
        beta_squared = beta ** 2
        return (1.0 + beta_squared) * tp / ((1.0 + beta_squared) * tp + fp + beta_squared * fn)

F1Optimizer.maximize_expectation([0.3])

This is the error information:

---

RuntimeError Traceback (most recent call last) <ipython-input-15-b3677380eb6c> in <module>() ----> 1 F1Optimizer.maximize_expectation([0.3])

/root/anaconda3/lib/python3.6/site-packages/numba/dispatcher.py in _compile_for_args(self, *args, **kws) 305 argtypes.append(self.typeof_pyval(a)) 306 try: –> 307 return self.compile(tuple(argtypes)) 308 except errors.TypingError as e: 309 # Intercept typing error that may be due to an argument

/root/anaconda3/lib/python3.6/site-packages/numba/dispatcher.py in compile(self, sig) 656 flags=flags, locals=self.locals, 657 lifted=(), –> 658 lifted_from=self.lifted_from) 659 660 # Check typing error if object mode is used

/root/anaconda3/lib/python3.6/site-packages/numba/compiler.py in compile_ir(typingctx, targetctx, func_ir, args, return_type, flags, locals, lifted, lifted_from, library) 752 args, return_type, flags, locals) 753 return pipeline.compile_ir(func_ir=func_ir, lifted=lifted, –> 754 lifted_from=lifted_from) 755 756

/root/anaconda3/lib/python3.6/site-packages/numba/compiler.py in compile_ir(self, func_ir, lifted, lifted_from) 366 367 self._set_and_check_ir(func_ir) –> 368 return self._compile_ir() 369 370 def stage_analyze_bytecode(self):

/root/anaconda3/lib/python3.6/site-packages/numba/compiler.py in _compile_ir(self) 704 “”" 705 assert self.func_ir is not None –> 706 return self._compile_core() 707 708

/root/anaconda3/lib/python3.6/site-packages/numba/compiler.py in _compile_core(self) 684 685 pm.finalize() –> 686 res = pm.run(self.status) 687 if res is not None: 688 # Early pipeline completion

/root/anaconda3/lib/python3.6/site-packages/numba/compiler.py in run(self, status) 244 # No more fallback pipelines? 245 if is_final_pipeline: –> 246 raise patched_exception 247 # Go to next fallback pipeline 248 else:

/root/anaconda3/lib/python3.6/site-packages/numba/compiler.py in run(self, status) 236 try: 237 event(stage_name) –> 238 stage() 239 except _EarlyPipelineCompletion as e: 240 return e.result

/root/anaconda3/lib/python3.6/site-packages/numba/compiler.py in stage_objectmode_backend(self) 597 “”" 598 lowerfn = self.backend_object_mode –> 599 self._backend(lowerfn, objectmode=True) 600 601 # Warn if compiled function in object mode and force_pyobject not set

/root/anaconda3/lib/python3.6/site-packages/numba/compiler.py in _backend(self, lowerfn, objectmode) 574 self.library.enable_object_caching() 575 –> 576 lowered = lowerfn() 577 signature = typing.signature(self.return_type, *self.args) 578 self.cr = compile_result(typing_context=self.typingctx,

/root/anaconda3/lib/python3.6/site-packages/numba/compiler.py in backend_object_mode(self) 548 self.library, 549 self.func_ir, –> 550 self.flags) 551 552 def backend_nopython_mode(self):

/root/anaconda3/lib/python3.6/site-packages/numba/compiler.py in py_lowering_stage(targetctx, library, interp, flags) 880 fndesc = funcdesc.PythonFunctionDescriptor.from_object_mode_function(interp) 881 lower = objmode.PyLower(targetctx, library, fndesc, interp) –> 882 lower.lower() 883 if not flags.no_cpython_wrapper: 884 lower.create_cpython_wrapper()

/root/anaconda3/lib/python3.6/site-packages/numba/lowering.py in lower(self) 158 159 # Materialize LLVM Module –> 160 self.library.add_ir_module(self.module) 161 162 def extract_function_arguments(self):

/root/anaconda3/lib/python3.6/site-packages/numba/targets/codegen.py in add_ir_module(self, ir_module) 183 ir = cgutils.normalize_ir_text(str(ir_module)) 184 with llvmts.lock_llvm: –> 185 ll_module = llvmts.parse_assembly(ir) 186 ll_module.name = ir_module.name 187 ll_module.verify()

/root/anaconda3/lib/python3.6/site-packages/numba/llvmthreadsafe.py in wrapped(*args, **kwargs) 28 def wrapped(*args, **kwargs): 29 with self: —> 30 return fn(*args, **kwargs) 31 return wrapped 32

/root/anaconda3/lib/python3.6/site-packages/numba/llvmthreadsafe.py in wrapper(*args, **kwargs) 57 @functools.wraps(fn) 58 def wrapper(*args, **kwargs): —> 59 return _patch_dispose(fn(*args, **kwargs)) 60 return wrapper 61

/root/anaconda3/lib/python3.6/site-packages/llvmlite/binding/module.py in parse_assembly(llvmir) 20 if errmsg: 21 mod.close() —> 22 raise RuntimeError(“LLVM IR parsing error\n{0}”.format(errmsg)) 23 return mod 24

RuntimeError: Failed at object (object mode backend) LLVM IR parsing error <string>:1257:48: error: expected value token %“.845” = call i8* @“PyLong_FromSsize_t”(i64 slice(None, None, -1)) ^

However, when I pass [0.3,02] to the function, it works perfectly fine. Did I miss something?

[stuartarchibald]

I’m closing this as the reproducer here https://github.com/numba/llvmlite/issues/218#issuecomment-257287340 no longer has problems. Further https://github.com/numba/llvmlite/issues/218#issuecomment-319007108 also works fine (Numba 0.40, llvmlite 0.25) .