https://bugs.openldap.org/show_bug.cgi?id=9203
--- Comment #6 from peter@adpm.de peter@adpm.de --- (In reply to Ondřej Kuzník from comment #5)
It was intended that the m=,t=,p= parameters equal the crypt form for new hashes where possible. IIRC, couldn't find a way to let libsodium know what the desired p= value should be.
If there's any discrepancy between the provided value and the final parameters (except for p= above), that's a bug and should be fixed.
See ITS#9206 to get the m= parameter aligned between libsodium & libargon2.