Bug#849499: llvmlite: ERROR: test_get_host_cpu_features (llvmlite.tests.test_binding.TestMisc)

[Daniel Stender]

Source: llvmlite
Version: 0.14.0-2
Severity: serious
File: llvmlite
Justification: fails to build from source

llvmlite [1] (currently in experimental) has a test failure on some archs (mips, ppc64el, s390x etc.) [2]:

<cut>
======================================================================
ERROR: test_get_host_cpu_features (llvmlite.tests.test_binding.TestMisc)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/«PKGBUILDDIR»/llvmlite/tests/test_binding.py", line 237, in test_get_host_cpu_features
    features = llvm.get_host_cpu_features()
  File "/«PKGBUILDDIR»/llvmlite/binding/targets.py", line 61, in get_host_cpu_features
    outdict[feat[1:]] = flag_map[feat[0]]
IndexError: string index out of range
</cut>

get_host_cpu_features() is defined in llvmlite/binding/targets.py:

<cut>
def get_host_cpu_features():
    """
    Returns a dictionary-like object indicating the CPU features for current
    architecture and whether they are enabled for this CPU.  The key-value pairs
    are the feature name as string and a boolean indicating whether the feature
    is available.  The returned value is an instance of ``FeatureMap`` class,
    which adds a new method ``.flatten()`` for returning a string suitable for
    use as the "features" argument to ``Target.create_target_machine()``.
    """
    with ffi.OutputString() as out:
        ffi.lib.LLVMPY_GetHostCPUFeatures(out)
        outdict = FeatureMap()
        flag_map = {'+': True, '-': False}
        for feat in str(out).split(','):
            outdict[feat[1:]] = flag_map[feat[0]]
        return outdict
</cut>

It appears ffi.lib.LLVMPY_GetHostCPUFeatures() doesn't return anything (e.g. on s390x):

<cut>
Python 2.7.13 (default, Dec 18 2016, 20:19:42) 
[GCC 6.2.1 20161215] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> from llvmlite.binding import ffi
>>> with ffi.OutputString() as out:
...     ffi.lib.LLVMPY_GetHostCPUFeatures(out)
...     print(out)
... 
0
</cut>

Could this be a regression from LLVM?

DS

[1] https://packages.qa.debian.org/l/llvmlite.html

[2] https://buildd.debian.org/status/package.php?p=llvmlite&suite=experimental

-- System Information:
Debian Release: stretch/sid
  APT prefers testing
  APT policy: (500, 'testing'), (1, 'experimental')
Architecture: amd64 (x86_64)

Kernel: Linux 4.8.0-2-amd64 (SMP w/4 CPU cores)
Locale: LANG=de_DE.utf8, LC_CTYPE=de_DE.utf8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash
Init: systemd (via /run/systemd/system)

Versions of packages python-llvmlite depends on:
ii  libc6          2.24-8
ii  libgcc1        1:6.2.1-5
ii  libtinfo5      6.0+20161126-1
ii  llvm-3.8       1:3.8.1-16
ii  python         2.7.11-2
ii  python-enum34  1.1.6-1
ii  python-six     1.10.0-3
pn  python:any     <none>
ii  zlib1g         1:1.2.8.dfsg-4

Versions of packages python-llvmlite recommends:
ii  llvmlite-doc  0.14.0-2

python-llvmlite suggests no packages.

-- no debconf information