When compiling for aosp_fugu-userdebug, clang aborted in pass:
Running pass 'Function Pass Manager' on module 'system/core/libmincrypt/p256_ec.c'.
Running pass 'X86 DAG->DAG Instruction Selection' on function '@felem_square'
BUG: 18790528
Change-Id: I416c07cae4c1a9743655bc45bd33c10a5cc9b41f
The declaration is useless and triggers a warning with clang 3.5 (in
code that's built with -Werror)
Change-Id: I8fc42178df264509dcce6e1f1ba318bf849307e3
Signed-off-by: Bernhard Rosenkraenzer <Bernhard.Rosenkranzer@linaro.org>
This adds a small EC library that is capable of verifying a signature of
SHA256 with ECDSA on the NIST P-256 curve.
Change-Id: I2a16639c92a77e8e4783c47ffbc56676de56eb59
The e=3 and e=65537 implementations are nearly identical, refactor
them down into one.
(cherry picked from commit 8fdbf97652)
Change-Id: I9f858204d52a0ae880170d955bf8f29fed57ee94
- adds a library to compute the SHA-256 hash
- updates the RSA verifier to take an argument specifying either SHA-1
or SHA-256
- updates DumpPublicKey to with new "key" version numbers for
specifying SHA-256
- adds new argument to adb auth code to maintain existing behavior
(cherry picked from commit 515e1639ef)
Change-Id: Ib35643b3d864742e817ac6e725499b451e45afcf
DumpPublicKey doesn't need a base64 encoder for anything; remove the
import to avoid deprecation warnings.
Change-Id: I7e01b9d77bb66bb99cb3357c886cc4e2e63de6ae
The output produced by DumpPublicKey now has a version tag on each
line (ie, each key). The existing keys are retroactively dubbed
"version 1", and we add a version 2 for 2048-bit e=65537 keys.
Change-Id: I204ec615c8f2346670220a1aeb99269e4abd5f81
Write the 32-bit chunks of N and RR as unsigned, so we get:
{64,0xc926ad21,{1795090719,2141396315,950055447,2581568430,4268923165,...
instead of:
{64,0xc926ad21,{1795090719,2141396315,950055447,-1713398866,-26044131,...
Change-Id: I575224fd7f7c34e06a1b6ae976eaa0bef41d2942
This optimized implementation of the SHA1 algorithm is about 28%
faster than the old one (on sapphire hardware) but assumes
little-endianness. Add it, but continue using the old implementation
on big-endian hardware.
Chih-Hung Hsieh