cffi issue on NetBSD #3372
Comments
|
Does it always fail at the same point? Could you run |
|
Seems quite repeatable, yes, I tried three times and it was always the Output with |
|
Okay, this script should replicate the segfault for you then: Could you confirm that? And then, just for good measure run these variants and tell me what you get? |
|
Scripts 1, 3, and 4 give me When I add a print line like the other scripts have, it also dumps core. |
|
Yep, just wanted to confirm that the actual callback was causing the crash. Interesting that the crash occurs even when the memory size is sufficiently large to handle it. Okay, I need to head out for a bit but would you mind cloning current master and installing it in a virtualenv to see if the crash occurs there? |
|
Ok, I did: and this mostly ran through, except for py-cffi in pkgsrc has one patch: Not sure what else would explain the test success. |
jsonn
pushed a commit
to jsonn/pkgsrc
that referenced
this issue
Jan 28, 2017
Add more test dependencies. Self tests cause a python core dump, see pyca/cryptography#3372 1.7.2 - 2017-01-27 ~~~~~~~~~~~~~~~~~~ * Updated Windows and macOS wheels to be compiled against OpenSSL 1.0.2k.
|
The problem is not related to the py-cffi patch, removing that does not change the situation. I made a package for current git head, to make sure the environments for both tests are the same, and I cannot reproduce the problem there. Do you have an idea which commit(s) might have fixed the problem? |
0-wiz-0
added a commit
to NetBSD/pkgsrc-wip
that referenced
this issue
Jan 28, 2017
|
It's almost certainly #3350 which changes py3.x to use a different cffi callback model. It's possible that there's a bug in the old-style cffi callbacks on NetBSD I suppose, but to check that we'd need to build a reduced test case. |
|
I don't think that's it, or not only that. Also, I'm testing with python-3.6, so |
|
Prior to that PR it checked to see if it was >= (3, 5) to disable static callbacks. You'll need to recompile/reinstall after applying that patch to verify the behavioral change. If you did that and it's still crashing then that's surprising 😮 |
jsonn
pushed a commit
to jsonn/pkgsrc
that referenced
this issue
Jan 28, 2017
pyca/cryptography#3350 Bump PKGREVISION. Identified by @reaperhulk in pyca/cryptography#3372
|
Closing since this is effectively resolved (and will be in the next release) |
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
py-acme does not work on NetBSD. Tracking down the cause I arrived at py-cryptography. It fails the self tests with:
gdb gives:
I don't know enough about cffi to debug this, can you help please?
The text was updated successfully, but these errors were encountered: