tree bcbdee79dee8c1baf4af70b5488d737f7da7e58c
parent 5c6393a8d590d2df35fd245153653f6d2d4239d8
author Jerry Johns <johnsj@google.com> 1663940319 -0700
committer GitHub <noreply@github.com> 1663940319 -0400
gpgsig -----BEGIN PGP SIGNATURE-----
 
 wsBcBAABCAAQBQJjLbbfCRBK7hj4Ov3rIwAAKVwIAJ5xWExvCTGX8XE15OVgXm2z
 5KMVIHsQgBJ4wae24XKqLroUrL24xlojsJu7qNMYeWs8q1UAPcOcLKtiHPflu0uw
 dYEisPKpzVlRsc9x+bDvRGZXVwmo53CgOOJ8g1VwEdVIM45JKcMTclzADgCKPEot
 /Jyc5ZZ9i6ChIVikM4MXOcAz0WznCLNuIH7rQTz90/b61s6nUpn+LrcIr31yTgmV
 XsThbYcYyJOt0HKBNo8mepN1u3O3hp87/l/sfzIHNe03oet6OVodhGNjNAfoU9YN
 0kfmwRgrPfesD9c21vOqX49ZvV2IedtkQ/2Uh431T653Q0m/nMYODKfnquc4sAk=
 =n/nn
 -----END PGP SIGNATURE-----
 

Fix KeepSubscription flag handling (#22805)

* Fix KeepSubscription flag handling

This fixes the KeepSubscription flag handling to be done right at the
onset of processing a SubscribeRequest message. This ensures that
matching existing subscriptions are evicted before we attempt to
continue further processing and allocation of ReadHandlers.

Testing:

In addition to the new test in TestRead, validated setting up a sub in
chip-tool and then cancelling it by sending another sub to an invalid
endpoint:

basic subscribe location  0 5 2 0
any subscribe-by-id 0xFFFFFFFF 0xFFFFFFFF 0 5 2 100

Also tested by sending an empty subscribe request with the REPL.

* Fixup printf specifier