Deprecate DSA_sign_setup() in the documentation

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6460)

diff --git a/doc/man3/DSA_sign.pod b/doc/man3/DSA_sign.pod
index 458e16a3c6adb29fb59bb526dfc3cd82c4c2789e..4ca7756db3eff145a15b48f039ab758403175ca0 100644 (file)
--- a/doc/man3/DSA_sign.pod
+++ b/doc/man3/DSA_sign.pod
@@ -23,13 +23,12 @@ digest B<dgst> using the private key B<dsa> and places its ASN.1 DER
 encoding at B<sigret>. The length of the signature is places in
 *B<siglen>. B<sigret> must point to DSA_size(B<dsa>) bytes of memory.
 
-DSA_sign_setup() may be used to precompute part of the signing
-operation in case signature generation is time-critical. It expects
-B<dsa> to contain DSA parameters. It places the precomputed values
-in newly allocated B<BIGNUM>s at *B<kinvp> and *B<rp>, after freeing
-the old ones unless *B<kinvp> and *B<rp> are NULL. These values may
-be passed to DSA_sign() in B<dsa-E<gt>kinv> and B<dsa-E<gt>r>.
-B<ctx> is a pre-allocated B<BN_CTX> or NULL.
+DSA_sign_setup() is defined only for backward binary compatibility and
+should not be used.
+Since OpenSSL 1.1.0 the DSA type is opaque and the output of
+DSA_sign_setup() cannot be used anyway: calling this function will only
+cause overhead, and does not affect the actual signature
+(pre-)computation.
 
 DSA_verify() verifies that the signature B<sigbuf> of size B<siglen>
 matches a given message digest B<dgst> of size B<len>.