Gyorgy Sarvari schrieb am Mi 20. Nov, 19:56 (+0100):
> Isn't ldconfig a must for python3-ctypes? It[0] seems to call the ldconfig
> binary directly. I'd expect the package to break (at least for some
> usecases) without ldconfig.
>
> [0]:
> https://github.com/python/cpython/blob/6c1a4fb6d400827155fd70e48d682e35397731a1/Lib/ctypes/util.py#L224
This code in line 224 should not be used on Yocto Linux, because of the
condition in line 205
if sys.platform.startswith(("freebsd", "openbsd", "dragonfly")):
In line 282 the Linux function starts:
def _findSoname_ldconfig(name):
…
try:
with subprocess.Popen(['/sbin/ldconfig', '-p'],
stdin=subprocess.DEVNULL,
stderr=subprocess.DEVNULL,
stdout=subprocess.PIPE,
env={'LC_ALL': 'C', 'LANG': 'C'}) as p:
res = re.search(regex, p.stdout.read())
if res:
return os.fsdecode(res.group(1))
except OSError:
pass
According to GitHub's code search this is the only usage that provides a
proper fallback:
def find_library(name):
# See issue #9998
return _findSoname_ldconfig(name) or \
_get_soname(_findLib_gcc(name)) or
_get_soname(_findLib_ld(name))
I would say, the Python code support the non-ldconfig case.
>From you other post:
> Actually looking at it a bit more, it seems to recover from missing
> ldconfig, and it tries other ways. Sorry for the spam, please ignore.
No, I think you provided valuable information they help to clarify the
requirements for ldconfig. Thanks.
But I'm not a Python expert. At least, I can tell our system works without
ldconfig, but maybe we never trigger the interesting code paths.
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#64319): https://lists.yoctoproject.org/g/yocto/message/64319
Mute This Topic: https://lists.yoctoproject.org/mt/109687333/21656
Group Owner: yocto+ow...@lists.yoctoproject.org
Unsubscribe: https://lists.yoctoproject.org/g/yocto/unsub
[arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-
