chiark / gitweb /
Mark Wooding [Sat, 16 Nov 2019 18:53:24 +0000 (18:53 +0000)]
t/: Add a test suite.
It's fairly substantial, but far from complete. It's also a little
strange in places because it's been sent backwards in time from the
future.
When building the Debian package, run the tests verbosely, so that if
the test crashes I stand a chance of figuring out where.
Mark Wooding [Sun, 24 Nov 2019 22:30:48 +0000 (22:30 +0000)]
ec.c: Don't lose error status when constructing points from a sequence.
The code would call `ecptxl_2' or `ecptxl_3' as appropriate, stash the
error status in `rc', and then fall out of the `if/else if' ladder to
`ok' -- which clobbers `rc', losing the error. This is unfortunate if
the point has been partially filled in.
The right fix is to go unconditionally to `end'. We already have `rc'
set appropriately, so `ok' isn't necessary; `ecptxl_2' has already
converted the point to internal form, and `ecptxl_3' constructs the
point in internal form anyway, so it would be wrong to drop into `fix'.
Mark Wooding [Sun, 24 Nov 2019 22:16:53 +0000 (22:16 +0000)]
ec.c: Free partially constructed points coordinatewise.
The `EC_DESTROY' macro assumes that the other coordinates are null if
and only if `p->x' is. This works badly with the current code
structure, which fills in coordinates as it goes along, and I think this
is a better fix than trying to maintain the coordinates in temporaries
until we're done.
Mark Wooding [Sun, 24 Nov 2019 16:36:24 +0000 (16:36 +0000)]
*.c: Be more careful about `PySequence_Size'.
This can be implemented by Python, so it can throw exceptions.
Fortunately, Python checks that the result is nonnegative, so we don't
have to worry about that.
Mark Wooding [Sun, 24 Nov 2019 16:26:22 +0000 (16:26 +0000)]
key.c: Reformat the rest of the `KeyError' constructor.
Mark Wooding [Sun, 24 Nov 2019 16:25:18 +0000 (16:25 +0000)]
key.c: Parse `KeyError' constructor arguments by hand.
This allows additional arguments to be stored, as is conventional for
exception classes.
Mark Wooding [Sun, 24 Nov 2019 14:59:35 +0000 (14:59 +0000)]
catacomb-python.h: Add a macro for raising `OverflowError'.
Mark Wooding [Sun, 24 Nov 2019 16:21:09 +0000 (16:21 +0000)]
key.c: Collect `KeyError' argument count as a separate step.
Mark Wooding [Sun, 24 Nov 2019 16:19:14 +0000 (16:19 +0000)]
key.c: Use tuple functions on `KeyError' argument tuple.
Mark Wooding [Sun, 24 Nov 2019 14:54:21 +0000 (14:54 +0000)]
key.c: Rename sad-path label to `end'.
Now we can use our exception-raising macros.
Mark Wooding [Sun, 24 Nov 2019 14:48:27 +0000 (14:48 +0000)]
key.c: Delete duplicate setting of `errstring'.
This code is a total mess, and doesn't work anyway. Let's start
cleaning it up.
Mark Wooding [Sun, 24 Nov 2019 15:07:08 +0000 (15:07 +0000)]
util.c (mkexc): Populate dictionary before constructing exception class.
It turns out that the dictionary contents are copied into the new class
rather than being used by reference, so populating it later doesn't
actually do anything.
Mark Wooding [Sun, 24 Nov 2019 14:50:01 +0000 (14:50 +0000)]
key.c: Only set the error code.
This briefly makes the error string unavailable, but only because the
actual constructor isn't wired up -- and if it were wired up then it
would fail immediately because it's programmed to accept only a single
argument.
Mark Wooding [Sun, 24 Nov 2019 15:11:46 +0000 (15:11 +0000)]
catacomb.c, util.c: Publish negative constants correctly.
The various `KERR_...' constants and `PGEN_ABORT' are canonically
negative integers, but weren't published correctly. Add a flags word to
`struct nameval' to identify constants which should really be signed,
and some fancy footwork to convert unsigned integers back to negative
values without upsetting C. Adjust the `C' macro to spot negative
constants and mark them specially.
Not all is well. In particular, the `KEXP_...' constants should /not/
be published as signed values, even if `time_t' is signed on the target
platform, because the Python bindings handle them exclusively as
`unsigned long' values. To make this work, also introduce `CF' which
allows us to set the flags explicitly, and use it for `KEXP_...'.
Mark Wooding [Sat, 23 Nov 2019 16:04:29 +0000 (16:04 +0000)]
field.c: Delete the completely unused `getfe' function.
It was introduced in the original commit, and never used at all.
Mark Wooding [Thu, 21 Nov 2019 18:23:56 +0000 (18:23 +0000)]
key.c (convfilter): Fix sense of error tests.
So this never worked either.
Mark Wooding [Thu, 21 Nov 2019 19:53:22 +0000 (19:53 +0000)]
buffer.c, ec.c: Fix required size for EC `buffer' encoding.
The problem is zero coordinates: the point at infinity is encoded as a
zero length word, so zero coordinates must be encoded as a single zero
byte, preceded by a length word of 1 -- which overruns the output buffer
provided, unless we take special care, which we haven't.
Mark Wooding [Tue, 19 Nov 2019 22:33:12 +0000 (22:33 +0000)]
algorithms.c: Fix `max' property name in docstrings.
How embarrassing.
Mark Wooding [Sun, 17 Nov 2019 22:57:47 +0000 (22:57 +0000)]
catacomb/__init__.py (_HashBase): Check that integers are within bounds.
The version of the `struct' module included with Python 2.5 writes a
warning to `stderr' and truncates rather than raising an exception. Do
the work ourselves.
Mark Wooding [Sun, 17 Nov 2019 04:23:00 +0000 (04:23 +0000)]
debian/rules: Build using the provided Makefile.
There were several problems with the old system. Firstly, it only
actually worked by accident: setting an explicit build directory caused
Debhelper to search there for a makefile (which it didn't find, for
obvious reasons) and fall back to the Python distutils system. That
would be OK, but only because there's only been one Python version I've
wanted to support in Debian for ages; but that's likely to change. And
it won't run tests properly.
Instead, delete the `-B' option (which will break the upcoming tests).
Now Debhelper will build via the Makefile. Set the `PYTHONS'
environment variable to the Python versions supported in Debian. And
set the prefix correctly during installation.
Mark Wooding [Mon, 25 Nov 2019 17:30:56 +0000 (17:30 +0000)]
debian/control: Add an explicit build-dependency on `dh-python'.
Mark Wooding [Sun, 17 Nov 2019 03:14:51 +0000 (03:14 +0000)]
catacomb/__init__.py: Make KMAC `copy' work.
Two fixes needed:
* Return the `new' object from `copy'.
* Introduce a `_bare_new' class method to handle constructing the
destination object, because `KMAC...' objects need a key argument.
Mark Wooding [Sun, 17 Nov 2019 03:14:31 +0000 (03:14 +0000)]
catacomb.c: Publish `RAND_IBITS' constant.
This is a limit on how much you can request in `seed', so it's sort of
important.
Mark Wooding [Fri, 15 Nov 2019 18:30:55 +0000 (18:30 +0000)]
rand.c: Fix `TrueRand' constructor so it can possibly work.
Mark Wooding [Fri, 15 Nov 2019 13:30:43 +0000 (13:30 +0000)]
pubkey.c (RSAPriv): Fix segfault if `p' is provided but not `q'.
Stupid typo.
Mark Wooding [Mon, 14 Oct 2019 00:04:46 +0000 (01:04 +0100)]
catacomb/__init__.py (_ShakeBase._copy): Actally copy the hash state.
Rather than sharing because that doesn't end well.
Mark Wooding [Mon, 14 Oct 2019 00:03:58 +0000 (01:03 +0100)]
algorithms.c (ShakeNN.copy): Return the copy rather than the original.
Mark Wooding [Mon, 14 Oct 2019 00:00:20 +0000 (01:00 +0100)]
algorithms.c (KECCAK.extract): Fix method name in keyword-args string.
Mark Wooding [Sun, 13 Oct 2019 23:59:30 +0000 (00:59 +0100)]
algorithms.c (hLATIN_prf): Allow all supported sizes of key.
Mark Wooding [Sun, 20 Oct 2019 00:19:18 +0000 (01:19 +0100)]
algorithms.c: Add missing guard for `del' to property `set' functions.
Continuing the work of
9444777c58c60253afdabf5b45011440845770e0.
Mark Wooding [Sun, 17 Nov 2019 01:34:12 +0000 (01:34 +0000)]
util.c: Make `default' arguments optional in `get', `setdefault', `pop'.
Alas, `pop' is a little trickier than the others.
Mark Wooding [Sun, 17 Nov 2019 01:33:34 +0000 (01:33 +0000)]
util.c: Fix sense of error check in `popitem'.
Mark Wooding [Fri, 9 Nov 2018 17:23:00 +0000 (17:23 +0000)]
rand.c, pgen.c: Invalidate random generators from pgen events.
I've not actually seen a crash from a Python program which keeps hold of
the random generator from a prime-generation event and tries to use it
after the operation has finished, but it was certainly possible.
Arrange for the event object to retain the random generator object (so
it always hands out the same one when requested), and invalidates it
when the event is itself invalidated.
This also involves messing with the `grand' code to cope with the idea
of invalidated random generators.
(cherry picked from commit
d65d80d7c096e6afc500270ee29909f64869e5dc)
Mark Wooding [Sat, 16 Nov 2019 22:42:32 +0000 (22:42 +0000)]
ec.c: Pass a dummy length argument to `PyObject_AsReadBuffer'.
Otherwise it just raises a `SystemException'. So this code never
worked.
Mark Wooding [Thu, 14 Nov 2019 19:08:53 +0000 (19:08 +0000)]
catacomb-python.h, *.c: Fix how Python `pgen' handlers handle exceptions.
Oh, this was a mess. The old code would convert an exception from a
Python handler into `PGEN_ABORT', and hope that the exception state was
still available when the overall operation ended.
This doesn't work. In particular, steppers and testers are finalized by
calling them with `PGEN_DONE', and the interpreter doesn't like
re-entering Python with an exception set. (In debug builds, this is an
assertion failure.)
Overhaul all of this nonsense.
* Add a collection of utilities for saving and restoring the exception
state.
* Add a hook, in the `catacomb' module, for reporting `lost'
exceptions, for the case where further exceptions are raised while
responding to a first exception.
* Use a larger `pypgev' structure to track the state of a Python event
handler through the framework. This structure holds a reference to
the Python object itself, and a slot for recording an exception.
* When a Python handler fails, stash the exception state in the slot
provided by the `pypgev' structure if there isn't one already, and
clear the pending exception. If there is already an exception in
the slot, then report the new exception through the hook described
above.
* Once a `pgen' operation completes, if it raised any exceptions at
all, then the first of these is left in the exception slot. If it
fails otherwise, then we supply a generic exception.
Mark Wooding [Thu, 14 Nov 2019 18:59:24 +0000 (18:59 +0000)]
pgen.c (pgev_python): Delete pointless refcount manipulation.
Mark Wooding [Fri, 14 Jul 2017 22:19:41 +0000 (23:19 +0100)]
catacomb/__init__.py: Awful bodge for symbol conflict.
It seems that, in Debian jessie and later, the main Python binary now
exports `md5_init' and friends. Unfortunately, this overrides
Catacomb's existing `md5_init' with a rather different version, and the
result is a segfault (on i386) or wrong answers (on amd64).
So, as an unpleasant bodge (while this broken thing makes its way
through Debian, see bug #868366), try to force the `RTLD_DEEPBIND' flag
when loading the module. This is unfortunate, because Python doesn't
actually advertise this flag, at least in my version.
(cherry picked from commit
a3ae4a9f590ef84d8e6eac0bc94873a9fd943073)
Mark Wooding [Fri, 14 Jul 2017 22:18:13 +0000 (23:18 +0100)]
catacomb/__init__.py: Rearrange the imports.
Sort the ordinary Python imports into alphabetical order (by module
name, rather than by import, if I'm only taking one or two symbols).
Move the main extension import into its own section, because it's going
to grow.
(cherry picked from commit
378ceeef4e0663d913cb448c32022522d39e7848)
Mark Wooding [Fri, 14 Jul 2017 22:16:02 +0000 (23:16 +0100)]
catacomb/__init__.py: Import `sys' as a whole.
We're only using `argv' in one place, so this isn't a significant
hardship. And I'll want more things from `sys' soon.
(cherry picked from commit
c04b289c577a5cdf6d5dd641f7b541d90a93adff)
Mark Wooding [Sun, 10 Nov 2019 22:39:15 +0000 (22:39 +0000)]
mp.c, util.c: Use `Py_ssize_t' for lengths.
Missed some from the earlier pass.
Mark Wooding [Sun, 10 Nov 2019 13:23:49 +0000 (13:23 +0000)]
mp.c: Fix crash converting elliptic curve point-at-infinity to integer.
Mark Wooding [Thu, 14 Nov 2019 18:55:41 +0000 (18:55 +0000)]
pgen.c, rand.c: Check correct variable when rejecting delete operations.
Mark Wooding [Thu, 14 Nov 2019 18:53:56 +0000 (18:53 +0000)]
rand.c (BBSPriv.generate): Release the event handler afterwards.
Mark Wooding [Wed, 13 Nov 2019 02:54:55 +0000 (02:54 +0000)]
mp.c: Return the result of `GFN' transformations as `GF'.
And not `MP', which is simply wrong.
Mark Wooding [Wed, 13 Nov 2019 02:54:29 +0000 (02:54 +0000)]
mp.c: Don't leak the field polynomial.
Mark Wooding [Wed, 13 Nov 2019 02:50:55 +0000 (02:50 +0000)]
mp.c: Release the `GFN' object through channels on error.
If the given element turns out not to actually generate a normal basis
then we have to give up constructing the `GFN' object and raise an
exception. In turns out that debug versions of Python get really
unhappy if you try to free objects which still have nonzero reference
counts, so:
* use `Py_DECREF' to free the object on error; and
* mark the object (by leaving `p' null) so that we don't actually
free the conversion matrices if they're weren't set up.
Mark Wooding [Sun, 10 Nov 2019 22:46:35 +0000 (22:46 +0000)]
mp.c: Check that CRT moduli are pairwise coprime.
Mark Wooding [Sun, 10 Nov 2019 22:41:19 +0000 (22:41 +0000)]
mp.c: Arrange to free `xx' on exit.
Mark Wooding [Sun, 10 Nov 2019 22:55:11 +0000 (22:55 +0000)]
mp.c: Check that CRT moduli are actually positive.
Mark Wooding [Wed, 23 Oct 2019 22:18:00 +0000 (23:18 +0100)]
field.c: Convert external-format field element to hex/octal.
Rather than going through the effort of calculating the external
representation of the field element and then returning the internal
version.
Mark Wooding [Wed, 23 Oct 2019 09:39:23 +0000 (10:39 +0100)]
field.c: Return the binary-field polynomial as, err, a polynomial.
Sharing the extraction code with the prime-field case means it gets
returned as an integer.
Mark Wooding [Fri, 22 Nov 2019 18:30:37 +0000 (18:30 +0000)]
rand.c: Add missing return-value mnemonic in docstring.
Mark Wooding [Wed, 9 Oct 2019 10:45:45 +0000 (11:45 +0100)]
*.c: Consistently show keyword arguments as optional in docstrings.
Mark Wooding [Fri, 22 Nov 2019 18:33:33 +0000 (18:33 +0000)]
pgen.c: Add missing `EV' arg in `PrimeGenEventHandler' method docstrings.
Mark Wooding [Fri, 22 Nov 2019 18:57:22 +0000 (18:57 +0000)]
mp.c: Fix Jacobi symbol notation in docstring.
Mark Wooding [Fri, 22 Nov 2019 18:56:34 +0000 (18:56 +0000)]
mp.c: Describe `MP' and `GF' conversion semantics.
Mark Wooding [Fri, 22 Nov 2019 18:55:46 +0000 (18:55 +0000)]
mp.c: Spell `MP' and `GF' in the correct case in docstrings.
Mark Wooding [Fri, 22 Nov 2019 17:23:15 +0000 (17:23 +0000)]
group.c: Fix capitalization of `ECPt' in docstring.
Mark Wooding [Fri, 22 Nov 2019 17:22:27 +0000 (17:22 +0000)]
ec.c: Add missing optional argument to docstring.
Mark Wooding [Fri, 18 Oct 2019 21:37:00 +0000 (22:37 +0100)]
field.c: Fix misleading docstrings.
It seems that the `value' and `_value' properties have always been able
to return `GF' for binary field elements.
Mark Wooding [Fri, 11 Oct 2019 10:06:05 +0000 (11:06 +0100)]
buffer.c: Fix docstrings.
Mark Wooding [Sat, 19 Oct 2019 19:33:28 +0000 (20:33 +0100)]
util.c: Fix docstrings for generic-map iterator classes.
Mark Wooding [Sat, 19 Oct 2019 19:32:51 +0000 (20:32 +0100)]
group.c: Fix docstring keyword for `G.checkgroup'.
Mark Wooding [Tue, 15 Oct 2019 11:09:44 +0000 (12:09 +0100)]
ec.c (ecpt_pyrichcompare): Fix point comparisons.
Previously we'd just reject comparisons of points with different curves.
Instead, support comparing curveless points with curvy ones by just
comparing the points coordinatewise.
Unfortunately, to make equality be transitive, this means permitting
comparisons between points on different curves, which is unpleasant.
Mark Wooding [Tue, 15 Oct 2019 11:02:59 +0000 (12:02 +0100)]
ec.c (ecpt_pyhash): Fix hashing.
Previously, hashing a curveless point would just crash, which is
surprisingly bad form. Replace this mess with a simpler thing which
just converts the point to external form and hashes the coordinates.
Mark Wooding [Tue, 15 Oct 2019 10:48:39 +0000 (11:48 +0100)]
field.c, mp.c: Hash `GF' and `FE' objects the same as `MP'.
They can be compared for equality, and so they must hash the same way.
Mark Wooding [Tue, 15 Oct 2019 10:48:12 +0000 (11:48 +0100)]
mp.c: Factor out and export `mphash'.
Mark Wooding [Sat, 12 Oct 2019 12:44:54 +0000 (13:44 +0100)]
ec.c (ec2osp, os2ecp): Collect flags correctly.
Previously `ec2osp' collected an `int', which probably wasn't completely
terrible, and `os2ecp' collected a float, which probably was.
Mark Wooding [Fri, 11 Oct 2019 09:00:46 +0000 (10:00 +0100)]
ec.c: Fix reported function name in `ec2osp'.
Mark Wooding [Fri, 11 Oct 2019 09:00:19 +0000 (10:00 +0100)]
ec.c: Fix keyword-argument list for `os2ecp'.
Mark Wooding [Fri, 18 Oct 2019 21:15:46 +0000 (22:15 +0100)]
ec.c (eccurve_pyrichcompare): Check that second operand has correct type.
A segfault waiting to happen, which has been lurking since the
beginning.
Mark Wooding [Fri, 18 Oct 2019 20:57:12 +0000 (21:57 +0100)]
ec.c (ecpt_pymul): Don't leak the scalar value.
Mark Wooding [Fri, 18 Oct 2019 20:17:37 +0000 (21:17 +0100)]
ec.c: Don't drop through into an error case.
Mark Wooding [Fri, 18 Oct 2019 20:16:24 +0000 (21:16 +0100)]
ec.c: Fix three-argument point construction.
This has been wrong literally forever.
Mark Wooding [Sat, 19 Oct 2019 19:25:39 +0000 (20:25 +0100)]
buffer.c: Don't advertise `WBUF.putecpt' as a keyword method.
Mark Wooding [Sat, 19 Oct 2019 19:23:46 +0000 (20:23 +0100)]
catacomb/__init__.py (BaseRat, MP, GF): Add missing true-division methods.
Since these all produce exact (rational) results, they satisfy the true-
division requirements.
Mark Wooding [Sat, 19 Oct 2019 19:21:02 +0000 (20:21 +0100)]
catacomb/__init__.py (BaseRat): Add missing reverse-multiplication method.
Mark Wooding [Sun, 20 Oct 2019 01:27:53 +0000 (02:27 +0100)]
catacomb/__init__.py (BaseRat): Make comparisons actually work.
This was broken in
83c77564338b3e410eb2ca2db3d35173dd6666cc.
Mark Wooding [Sat, 12 Oct 2019 20:07:05 +0000 (21:07 +0100)]
buffer.c (wbmeth_putblkN): Check input block size.
Otherwise the C code fails an assertion.
Mark Wooding [Fri, 11 Oct 2019 09:01:11 +0000 (10:01 +0100)]
buffer.c: Return 64-bit values, even if they're too big for `getulong'.
Mark Wooding [Sat, 19 Oct 2019 19:02:08 +0000 (20:02 +0100)]
util.c: Use Python's machinery for handling 64-bit integers.
Where available.
Mark Wooding [Sat, 19 Oct 2019 18:59:59 +0000 (19:59 +0100)]
util.c: Augment `convu64' to collect its argument via `kludge64'.
Otherwise integers too large for `unsigned long' get rejected with an
error, even though they ought to be acceptable as a `uint64'.
Mark Wooding [Sat, 19 Oct 2019 17:19:07 +0000 (18:19 +0100)]
key.c, pgen.c: Add missing guards for `del' to property `set' functions.
Mark Wooding [Sat, 19 Oct 2019 16:10:41 +0000 (17:10 +0100)]
*.py: Use `str.replace' rather than `str.translate'.
It seems that the `None' argument to `str.translate' was a brief
experiment added in 2.6 which didn't survive into 3.0.
Mark Wooding [Fri, 14 Jul 2017 22:13:56 +0000 (23:13 +0100)]
buffer.c: Fix typoed variable name in `assert'.
Evidently I've never actually compiled this code with the assertions
turned on before.
(cherry picked from commit
a11849068dd55f5997365984cce759c0f2d7caeb)
Mark Wooding [Mon, 24 Dec 2018 15:21:16 +0000 (15:21 +0000)]
Release 1.2.1.1.
Mark Wooding [Wed, 16 Aug 2017 03:54:26 +0000 (04:54 +0100)]
debian/control: Build-depend on `python-all-dev', not explicit versions.
Mark Wooding [Wed, 16 Aug 2017 03:54:08 +0000 (04:54 +0100)]
debian/: Use `dh_python2' for packaging.
Mark Wooding [Fri, 28 Jul 2017 23:58:08 +0000 (00:58 +0100)]
setup.py: Fix the advertised package URL.
Unfortunately the old URL was terrible both because it was poorly chosen
and because it was broken long ago (if indeed it ever worked in the
first place).
Mark Wooding [Fri, 7 Jul 2017 20:18:42 +0000 (21:18 +0100)]
Merge remote-tracking branch 'origin/HEAD'
* origin/HEAD:
catacomb/pwsafe.py: Fix stupid error which breaks `delete'.
Mark Wooding [Sat, 1 Jul 2017 09:43:44 +0000 (10:43 +0100)]
catacomb/pwsafe.py: Fix stupid error which breaks `delete'.
Mark Wooding [Fri, 16 Jun 2017 00:01:24 +0000 (01:01 +0100)]
Release 1.2.1.
Mark Wooding [Sun, 28 May 2017 18:03:08 +0000 (19:03 +0100)]
pubkey.c (dsa_setup): Make sure `u' is None or an MP object.
Don't just store the caller's object and hope for the best.
Mark Wooding [Sun, 28 May 2017 18:03:08 +0000 (19:03 +0100)]
pubkey.c: Fix keyword-argument order for KCDSAPriv constructor.
Mark Wooding [Sun, 28 May 2017 18:03:08 +0000 (19:03 +0100)]
ec.c: Fix embarrassing use-after-free in EC point hashing.
The hashed data is sometimes (unpredictably) mangled by freeing causing
hash mismatches, which is annoying. Also, obviously incorrect.
Mark Wooding [Thu, 11 May 2017 09:42:15 +0000 (10:42 +0100)]
Release 1.2.0.
Mark Wooding [Thu, 11 May 2017 09:42:15 +0000 (10:42 +0100)]
pubkey.c: Allow RSA key generation with user-chosen public exponent.
New feature in the underlying library.
Mark Wooding [Thu, 11 May 2017 09:42:15 +0000 (10:42 +0100)]
mp.c: Add binding for `leastcongruent' function.
Mark Wooding [Thu, 11 May 2017 09:42:15 +0000 (10:42 +0100)]
pubkey.c: Add support for Ed448 signatures, following RFC8032.
Mark Wooding [Thu, 11 May 2017 09:42:15 +0000 (10:42 +0100)]
catacomb/__init__.py: Settle on SHAKE256 for X448 box-key generation.
This matches Ed448 hashing, which is probably a good thing.