Tested on m-c revision 66e4d53697c2: The methods PL_DHashTableOperate PL_DHashTableFinish PL_DHashTableEnumerate in xpcom/build/pldhash.cpp use fallible allocation and may crash on OOM in the INCREMENT_RECURSION_LEVEL(table) call. Khuey mentioned I should file these although they cannot be easily converted to be infallible.

Actually, they could be. I forgot that these don't live in NSPR. bsmedberg, do you object to making these infallible?

The pldhash allocator is configurable, and I think it's reasonable to expect that some callers will be prepared to deal with OOM, while most won't. So I think it's reasonable to make the *default* behavior be infallible, but there's no reason we shouldn't also make pldhash check for OOM correctly.

Closing because no crash reported since 12 weeks.

Closing because no crash reported since 12 weeks.