Michael Hamburg
a69002875c
reverse convention for constant_time_select, resolving a TODO
9 years ago
Michael Hamburg
4de70b837c
separate out strobe and spongerng from shake. strobe is experimental. spongerng is experimental internally but the interface should be pretty good (except for any camelCase vs snake_case issues). shake should be stable
9 years ago
Michael Hamburg
94262dbecb
minor
9 years ago
Michael Hamburg
2e23ac747b
move (some of the?) per-field code to src/per_field.c
9 years ago
Michael Hamburg
825751d034
one gf_strong_reduce to rule them all
9 years ago
Michael Hamburg
38455f34f2
one ser/deser to rule them all (TODO test on NEON and other places with LIMBPERM)
9 years ago
Mike Hamburg
a5bed6b351
Working on getting cross-arch working again. Several TODOs.
Currently compiles and passes tests on x86_64 with arch_32 and
DECAF_FORCE_32_BIT=1 (as well as the native settigs of course),
so that's a start.
Want to make serialization routine cross-arch. Need to check that
perf is good enough (likely). Current routine in p25519/arch_32
is almost cross-arch, but has known bugs (FIXMEs). Needs to take
into account separate p and, for NEON, the LIMBPERM.
Want to decouple arches for each curve/field. Currently the split
between decaf_word_t and word_t makes this fraught with peril. Fix
is probably to rename decaf_word_t to decaf_api_word_t and fix it
to either uint32 or uint64, then make internal things separate per
field. That way we don't have to try arch detection in the header,
which is nice.
Need to make decaf_gen_tables use SC_LIMB. Might as well get rid
of API_NS there too.
9 years ago
Michael Hamburg
6bc97fb756
need an include/arch_*/arch_intrinsics.h for other arches
9 years ago
Michael Hamburg
233f8453c7
further reduce the code in f_impl.h
9 years ago
Michael Hamburg
b4c402c238
further reduce the code in f_impl.h
9 years ago
Michael Hamburg
2402788996
further reduce the code in f_impl.h
9 years ago
Michael Hamburg
4218223dd7
generate most of f_impl.h. Not tested on most arches yet :-(
9 years ago
Michael Hamburg
294eabaec7
generate f_field.h
9 years ago
Michael Hamburg
81403de10c
knock out a couple TODOs
9 years ago
Mike Hamburg
c6d7fdec35
some alignment and gcc-related fixes. Unfortunately, the python ffi still dies for alignment-related reasons :-(
9 years ago
Michael Hamburg
40cc1d24c7
recategorize a couple of todos and unbreak arch_ref64
9 years ago
Michael Hamburg
bdf4375b52
normalize some of the word and limb size declarations, still not entirely happy on that front
9 years ago
Michael Hamburg
36b54ca3db
clear an include ordering fixme
9 years ago
Michael Hamburg
23726f4f73
clear a todo from f_arithmetic.c
10 years ago
Michael Hamburg
cdab495338
Cross-curve compilation working! Still a bunch of FIXMEs though
10 years ago
Michael Hamburg
719fcacc58
working on multicurve build system
10 years ago
Michael Hamburg
5af980b85a
wipe out the multiple layers of rename between decaf_fast and field. still some serious HACKs in the include prio to avoid multiple definition of struct gf
10 years ago
Mike Hamburg
36380f3e2a
one more namespacing change
10 years ago
Mike Hamburg
93edb223bb
remove inverse from f_field.h
10 years ago
Mike Hamburg
0b59ddc2bf
move config to prime directories
10 years ago
Mike Hamburg
f825fd9b75
gcc compat changes
10 years ago
Michael Hamburg
e273155e35
switch to 64-bit shift; still slower than donna?
10 years ago
Michael Hamburg
ba9f201901
faster mulw?
10 years ago
Michael Hamburg
8202c43eba
remove ah = a*19
10 years ago
Michael Hamburg
42a561d018
some accel in for curve25519
10 years ago
Michael Hamburg
f68833cd1e
passing everything but elligator with curve25519! though there are probably some lurking bugs...
10 years ago
Michael Hamburg
4b0bf31fc9
progress checkin. compiles. working on point decode. have switched some of the Ds over. BTW, you can see that this is using PinkBikeShed instead of the real Curve25519; this is temporary
10 years ago
Michael Hamburg
03ecad0551
it compiles, but it certainly doesnt work yet
10 years ago
Mike Hamburg
40b1f8b85e
initial replace 448->255; doesnt compile yet
10 years ago