tree 45d8e5c660fd485adfd22a8df794a4465c590618
parent 4a653e082de47bd45fb9dcc730f71559c6d804a3
author David Benjamin <davidben@google.com> 1730145430 -0400
committer Boringssl LUCI CQ <boringssl-scoped@luci-project-accounts.iam.gserviceaccount.com> 1730270507 +0000

runner: Remove explicit seq parameter to encrypt and decrypt

It was always derived from hc.seq. The only subtleties are:

1. The DTLS code relies on record header parsing zooming the sequence
   number forward. (It was already doing this.)

2. Outgoing records need to accomodate the goofy SequenceNumberMapping
   feature.

3. The funny sequence number business in DTLS 1.2 vs 1.3 was previously
   handled at the header parser for incoming records and at encrypt()
   for outgoing records. Unify everything on doing it at
   encrypt/decrypt.

I added this parameter in
https://boringssl-review.googlesource.com/c/boringssl/+/71407, but I
think that was a mistake. We (mostly) always know the expected sequence
number, and this is one more field we can derive from the epoch.

Change-Id: I00124aee57618dfbde5e458d0f9572d16946c0bc
Reviewed-on: https://boringssl-review.googlesource.com/c/boringssl/+/72648
Reviewed-by: Nick Harper <nharper@chromium.org>
Commit-Queue: David Benjamin <davidben@google.com>