commit c0c65387ae9819c50eee4f548bf7360a4ca2b9d1
author Jonas Oreland <jonaso@webrtc.org> Mon Dec 05 09:40:09 2022
committer WebRTC LUCI CQ <webrtc-scoped@luci-project-accounts.iam.gserviceaccount.com> Mon Dec 05 10:01:01 2022
tree 287d8c4e3d43fcc66da9c06338743b16a7e692ae
parent b6e8c2e393c5f876f11a839ecdb41f7011896038

AndroidNetworkMonitor - loosen assumptions even more

This cl/ attempts to fix (rather) rare crashes in
OnNetworkConnected_n by loosening the assumptions
that a network handle will keep it's network name.

With this cl/ it is possible that a NetworkHandle
can call OnNetworkConnected_n with one interface name
and then directly afterwards call it with another (
w/o an OnNetworkDisconnected_n inbetween).

This is the only scenario in which I could see the previous
crash occurring.

i.e
OnNetworkConnected(handle, "some-if-name")
OnNetworkConnected(handle, "some-other-name")

- previously this caused crash,
- now this is treated as if there was an OnNetworkDisconnected(handle) in between.

---

Also 1: shamelessly copy TYPE_MOBILE_DUN & TYPE_MOBILE_HIPRI from chromium: https://eureka-internal.git.corp.google.com/chromium/src/+/87987f0e76e5c045d7196b5da904dc17ce254ba3?pli=1

Also 2: Modify testcase not to use real interface names, so I can ran them on personal test phone w/o the real networks interfering.

Bug: webrtc:13741
Change-Id: I5480d5ce7031c2b5c09b958064076d02b3db1248
Reviewed-on: https://webrtc-review.googlesource.com/c/src/+/285980
Commit-Queue: Jonas Oreland <jonaso@webrtc.org>
Reviewed-by: Harald Alvestrand <hta@webrtc.org>
Cr-Commit-Position: refs/heads/main@{#38808}