Fix potential null pointer deref.

[jplevyak]

Signed-off-by: John Plevyak jplevyak@gmail.com

For an explanation of how to fill out the fields, please see the relevant section
in PULL_REQUESTS.md

Description:
Risk Level:
Testing:
Docs Changes:
Release Notes:
[Optional Fixes #Issue]
[Optional Deprecated:]

[lizan]

@jplevyak Great finding, can we do a full cherry-pick of envoyproxy#8106? which seems contains same change for warming clusters as well, and come with tests.

[duderino]

@lizan thanks for the tip. @jplevyak thanks for the PR.

Closing this in favor of the upstream cherry pick

[jplevyak]

This doesn't cherrypick cleanly. My tiny patch is the same exact fix as in upstream minus the tests and other baggage. Do we want to do a more complex merge or go with the simpler fix?

[lizan]

We should at least also cherry-pick the regression test from upstream. The else branch doesn't do anything other than ASSERT but no hurt either.

[jplevyak]

That is where the merge problems are. So full merge it is.

[jplevyak]

This is 1.1 and there has been significant divergence. Consequentially there is no easy merge here. Several core objects have been altered. Getting the test to work would require substantial changes. Given that this change is completely safe (it only verifies that a pointer is not invalid before using it) and given that there are tests are in the other branches I think we should just proceed with the simple fix.

[lizan]

@jplevyak sure that's fair.