summaryrefslogtreecommitdiff
path: root/c
Commit message (Collapse)AuthorAgeFilesLines
* style issues onlyhppaArmin Rigo2022-12-241-4/+4
|
* Fix the test failures on hppaJohn David Anglin2022-12-181-2/+9
| | | | | | | | | | | | | | | | A couple of years ago the libffi port for hppa-linux was changed from using a trampoline executed on the stack to the function descriptor technique used by ia64. This doesn't require an executable stack and is simpler. However, function pointers need to have the PLABEL bit set in the pointer. As a result, a simple cast can't be used to convert closure pointers to function pointers. python-cffi uses its own closure allocation so the problem can't be fixed in libffi. I added a macro CFFI_FN to do the conversion. It shouldn't affect other architectures. There is a similar define in libffi. Fixes: https://bugs.debian.org/1024271
* Issue 553Armin Rigo2022-12-041-0/+2
| | | | Two missing calls to PyObject_GC_UnTrack()
* Replace py.test usage with pytest, explicitly require py for testspy.testMiro Hrončok2022-11-111-194/+193
| | | | | | | pytest 7.2+ no longer depends on py. It ships py.path and py.error only. See https://docs.pytest.org/en/7.2.x/changelog.html#deprecations The tests use py.code as well, hence we declare and document a test dependency on py.
* Adjust tests for a last minute Python 3.11 change in the traceback formatpython3.11.0b4Miro Hrončok2022-07-151-33/+2
| | | | | See https://github.com/python/cpython/issues/93883 and https://github.com/python/cpython/pull/93994
* musllinux test updates and wheels, py3.11 beta wheelsMatt Davis2022-06-291-2/+13
|
* tweak the test for pypy, where we get an extra null byte after a plain 'char'Armin Rigo2022-06-081-1/+1
|
* dump version to 1.15.1Armin Rigo2022-06-082-2/+2
|
* issue #535Armin Rigo2022-06-082-1/+50
| | | | | Give a warning when we ask a ffi.buffer() that can be proven to overflow. Might help with the confusion on that issue #535.
* #532 Use Py_SET_REFCNT in _cffi_backend.c instead of ob_refcntArmin Rigo2022-04-121-3/+4
|
* Issue #531Armin Rigo2022-03-291-3/+8
| | | | | on ppc64le, libffi requires 16 bytes alignment for at least the return value of functions if that is 'long double'.
* update from defaultMatt Davis2022-02-011-1/+46
|\
| * apply patch from Tomáš on PR 111adapt-tests-for-python3.11Armin Rigo2022-02-011-1/+46
| |
| * Backed out changeset: e184bbd523d6Armin Rigo2022-02-011-7/+0
| |
| * Adapt tests for Python 3.11Tomáš Hrnčiar2022-01-241-0/+7
| | | | | | | | Fix broken tests, Python 3.11 brings in enhanced error locations in tracebacks.
* | fix hardcoded pathMatt Davis2022-02-011-1/+2
| |
* | update libffi for arm64 to release 3.4.2 and add scripts to generate static ↵Niyas Sait2021-10-286-31/+134
|/ | | | libffi static library
* Bump version to 1.15.0Matt Davis2021-10-122-2/+2
|
* Bump version to 1.15.0rc2Matt Davis2021-10-042-2/+2
|
* Bump version to 1.15.0rc1Matt Davis2021-09-172-2/+2
|
* Remove this directory, it was already copied to c/libffi_x86_64Armin Rigo2021-07-2213-2210/+0
|
* added support for building cffi on windows on arm machinesNiyas Sait2021-07-2218-0/+3037
|
* Bump version to 1.14.6Armin Rigo2021-06-202-2/+2
|
* fix tests for py.test running on CPython v3.10.0b3Armin Rigo2021-06-201-0/+2
|
* fix for fdc0cbcdb65f on pre-3.10 pythonsArmin Rigo2021-06-201-2/+2
|
* fix a test that fails in the 3.10 alphas and will start failing on pypy soonfix-attribute-error-test-3.10Carl Friedrich Bolz-Tereick2021-04-241-2/+4
| | | | once I merge that feature: AttributeError is printing extra information
* #492Armin Rigo2021-02-192-1/+92
| | | | | | Python 3.8 adds support for sys.unraisablehook(), but also a new _PyErr_WriteUnraisableMsg() that calls it and is slightly more flexible in displaying the error message. Use that
* Need a cast here because the type ffi_status is unsignedArmin Rigo2021-02-191-2/+2
|
* fix another (very minor) leak, also found by Sebastian BergArmin Rigo2021-02-181-1/+1
|
* BUG: Fix string memory leak in `b_do_dlopen`Sebastian Berg2021-02-171-2/+7
| | | | | | When requesting a string with `es` or `et`, the string has to be freed with `PyMem_Free`. (https://docs.python.org/3/c-api/arg.html#strings-and-buffers)
* Bump version to 1.14.5 and add whatsnew entryArmin Rigo2021-02-112-2/+2
|
* add some more testsArmin Rigo2020-12-211-0/+5
|
* A test that passes on CPython but fails on PyPyArmin Rigo2020-12-211-0/+9
|
* gcc-4.4.7: this pragma must not be inside functionsArmin Rigo2020-11-281-13/+16
|
* bump version to 1.14.4Armin Rigo2020-11-242-2/+2
|
* fix for yet another way in which old libffi's are not compatible with new ↵Armin Rigo2020-11-241-5/+13
| | | | libffi's
* Fix what I think will be a compile-time error in some casesArmin Rigo2020-11-201-10/+20
|
* Issue #475Armin Rigo2020-10-142-2/+3
| | | | | Fix a crash that can occur only if there is an error when building a callback (in very rare cases), or in debug mode.
* py3Armin Rigo2020-10-141-2/+2
|
* Fix building `cffi` for macOS with Apple Silicon support + backwards ↵Max Bélanger2020-10-021-47/+83
| | | | compatibility
* explicit test of unaligned pointer reads/writesArmin Rigo2020-09-231-0/+21
|
* bump version to 1.14.3Armin Rigo2020-09-152-2/+2
|
* this test is fragile on a non-translated pypy, skip itArmin Rigo2020-09-031-2/+2
|
* merge headsArmin Rigo2020-08-311-11/+11
|\
| * let the test run directly on pypy tooArmin Rigo2020-08-301-11/+11
| |
* | Don't call PyEval_InitThreads() on CPython >= 3.7, where it should always be aArmin Rigo2020-08-311-2/+4
|/ | | | no-op. It is explicitly deprecated in 3.9.
* bump version number to 1.14.2Armin Rigo2020-08-152-2/+2
|
* bump version number to 1.14.1Armin Rigo2020-07-262-2/+2
|
* more of the same, for win64Armin Rigo2020-07-211-0/+5
|
* next fix, for win64Armin Rigo2020-07-211-0/+2
|
View Lorry Controller Status