[Python-modules-team] Bug#892175: pycryptodome FTBFS on 32bit big endian: src/montgomery.c:245: mont_mult: Assertion `t[2*abn_words] <= 1' failed

[James Cowgill]

On Tue, 06 Mar 2018 14:26:31 +0200 Adrian Bunk <bunk at debian.org> wrote:
> Source: pycryptodome
> Version: 3.4.11-1
> Severity: serious
> 
> https://buildd.debian.org/status/package.php?p=pycryptodome&suite=sid
> 
> ...
>    debian/rules override_dh_auto_test
> make[1]: Entering directory '/<<PKGBUILDDIR>>'
> PYBUILD_SYSTEM=custom \
>         PYBUILD_TEST_ARGS="python{version} -m Cryptodome.SelfTest {build_dir}/" dh_auto_test
> I: pybuild base:184: python2.7 -m Cryptodome.SelfTest /<<PKGBUILDDIR>>/.pybuild/pythonX.Y_2.7/build/
> Skipping AESNI tests
> python2.7: src/montgomery.c:245: mont_mult: Assertion `t[2*abn_words] <= 1' failed.
> Aborted

I've only had a brief look, but there are some (uint64_t*) to
(uint32_t*) casts in src/multiply_32.c addmul128 which look like
undefined behavior to me (strict aliasing rule).

I notice BEBIT which looks like an attempt at big-endian support which
clearly hasn't worked.

A recent commit I noticed:
https://github.com/Legrandin/pycryptodome/commit/aa5fac10a1c3a5f2b98c35954ae74207d545ec13

James

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: OpenPGP digital signature
URL: <http://lists.alioth.debian.org/pipermail/python-modules-team/attachments/20180306/d5d0d4ba/attachment-0001.sig>