Options in new_messenger() must never be null.

[GrayHatter]

cppcheck complains about being redundant or incorrcet. I'd rather be correct.

---

This change is 

[nurupo]

toxcore/Messenger.c, line 2740 at r1 (raw file):

/*  return size of the messenger data (for saving) */
uint32_t m(const Messenger *m)

Huh?

---

Comments from Reviewable

[GrayHatter]

Review status: 0 of 1 files reviewed at latest revision, 1 unresolved discussion.

---

toxcore/Messenger.c, line 2740 at r1 (raw file):

Previously, nurupo wrote…

Huh?

Done.

---

Comments from Reviewable

[nurupo]

(Doing @GrayHatter's work of writing a proper PR description)

For anyone reading this who didn't follow our IRC conversation, cppcheck and clang scan-build started to warn about potential null pointer deference in new_messenger() [cppcheck], [scan-build]. These complains are caused by the 82515f9 commit. It has introduced a

if (options && options->log_callback) {

check to new_messenger(), which makes static analyzers consider the case that options might actually be NULL. There are execution paths in new_messenger() where options is dereferenced without checking if it is NULL or not, which causes the warnings.

This PR adds a function-wide check to make sure options is not NULL. This will remove the confusion static analyzers are having, as well as anyone reading new_messenger() might have, since static analyzers complain for a valid reason as this is a really smelly code.

[nurupo]

---

Reviewed 1 of 1 files at r2.
Review status: all files reviewed at latest revision, all discussions resolved.

---

Comments from Reviewable