GitLab

Used to be implicitly done by oid_get_pk_alg().

openssl pkcs8 -topk8 -in ec_prv.sec1.pem -passout pass:polar \
    -v1 PBE-SHA1-RC4-128 -outform der -out ec_prv.pk8.pw.der
openssl pkcs8 -topk8 -in ec_prv.sec1.pem -passout pass:polar \
    -v1 PBE-SHA1-RC4-128 -outform pem -out ec_prv.pk8.pw.pem

Used to work only for RSAPrivateKey content, now accepts ECPrivateKey too,
and may even work with similar enough structures when they appear.

Test files were generated as follows:

openssl ecparam -name prime192v1 -genkey > key.pem

openssl ec -in key.pem -pubout -outform PEM > pub.pem
openssl ec -in key.pem -pubout -outform DER > pub.der

openssl ec -in key.pem -outform pem > prv.sec1.pem
openssl ec -in key.pem -outform der > prv.sec1.der
openssl ec -in key.pem -des -passout pass:polar -outform pem > prv.sec1.pw.pem

openssl pkcs8 -topk8 -in key.pem -nocrypt -outform pem > prv.pk8.pem
openssl pkcs8 -topk8 -in key.pem -nocrypt -outform der > prv.pk8.der
openssl pkcs8 -topk8 -in key.pem -passout pass:polar -outform der \
    > prv.pk8.pw.der
openssl pkcs8 -topk8 -in key.pem -passout pass:polar -outform pem \
    > prv.pk8.pw.pem

Warning: due to a bug in oid_descriptor_from_buf(), keys associated to some
curves (secp224r1, secp384r1, secp521r1) are incorrectly rejected,
since their namedCurve OID contains a nul byte.

Overflow check was done too early, causing many false positives.

Also group key checking and generation functions in ecp.h and ecp.c.

Can now run without need of time() / localtime() and gettimeofday()

Eat your own dog-food..

The SHA4 name was not clear with regards to the new SHA-3 standard. So
SHA2 and SHA4 have been renamed to better represent what they are:
SHA256 and SHA512 modules.