-
Notifications
You must be signed in to change notification settings - Fork 219
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
libcrypt-2-*.12.so not found during auditwheel repair #411
Comments
Using manylinux2014 does not help, with the following auditwheel repair error:
The manylinux2014 build/repair log is here : |
I have removed the dependency on libcurl so libcrzpt is not linked any more and the auditwheel repair operation continues and succeeds |
Closing, anyone in need may reopen this. |
I believe the issue is that
(Taken from inside the manylinux 2014 docker image) So as a packager, I can't avoid this problem without avoiding libcurl or building it from source myself such that it links against When I run This makes sense because libcrypt doesn't show as a direct dependency via. |
This is now much clearer to me thank you.
So... compiling from source would be the way possibly..
po 24. 2. 2020 v 19:37 odesílatel Daniel Alley <[email protected]>
napsal:
… I believe the issue is that libcurl inside this environment is
dynamically linked against libcrypt.so.1
Installed:
libcurl-devel.x86_64 0:7.29.0-54.el7_7.2
Dependency Updated:
curl.x86_64 0:7.29.0-54.el7_7.2 libcurl.x86_64 0:7.29.0-54.el7_7.2
Complete!
+ ldd /lib64/libcurl.so.4
+ grep libcrypt
libcrypto.so.10 => /lib64/libcrypto.so.10 (0x00007f41742d8000)
libcrypt.so.1 => /lib64/libcrypt.so.1 (0x00007f4173428000)
(Taken from inside the manylinux 2014 docker image)
So as a packager, I can't avoid this problem without avoiding libcurl or
building it from source.
The .so inside my own package, which was built inside the docker image,
shows different results for ldd when I run ldd inside the docker image,
or on the Fedora host system. Because on my host system, libcurl
dynamically links against libcrypt.so.2.
—
You are receiving this because you modified the open/close state.
Reply to this email directly, view it on GitHub
<#411?email_source=notifications&email_token=AAJU5QXT7PMF7PDFEDNR6C3REQHXLA5CNFSM4J6MKBD2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEMZBDBQ#issuecomment-590483846>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AAJU5QQ2CBATT6I44AZJMMDREQHXLANCNFSM4J6MKBDQ>
.
|
Hopefully this will resolve the problem properly: pypa/auditwheel#230 |
Should I reopen this issue until pypa/auditwheel#230 is resolved? |
That would be reasonable, but do whatever you like! |
i hit this kind of error when trying to build a wheel that utilized libcrypt.so.1. i traced it back to Centos 6's glibc shipping a libcrypto.so.1 that patchelf does not handle correctly and ends up corrupting it in a way that tools get confused when they try to process the ELF later on. details of the patchelf bug can be found at NixOS/patchelf#197. this impacts the manylinux2010 container only. the manylinux2014 container doesn't seem to have a weird libcrypto.so.1 file. i hacked around it by slightly modifying the librypt.so.1 in the image. this makes patchelf pick the correct section header table when it rewrites the soname. $ printf '\x1a' | dd of=/lib64/libcrypt-2.12.so count=1 bs=1 conv=notrunc seek=$((0x9f68)) still experimenting with this. |
Hello,
This was possibly dealt with in #305 on this tracker, though I do not understand how to fix/mitigate the issue.
In the manylinux2010_x86_64 python 3.7 auditwheel repair step after building a cpython/c++ binding module wheel, I have the following log:
https://github.com/dtschump/gmic-py/runs/358967116
..with especially this non-funny failing auditwheel repair linking error:
I am already pulling the :latest Docker image tag.
Should I yum install libxcrypt or libcrypt-compat or so first?
Should I consider manylinux2014 instead? (my project does not plan to support very old linuxes for now).
Here is the original issue link on our side:
https://github.com/dtschump/gmic-py/issues/9
Thank you for your help :)
The text was updated successfully, but these errors were encountered: