Andi [:andi] Assignee
	Description • 9 years ago

The Static Analysis tool Coverity added lazySticky is not initialized in the constructor of RegExpStatics.

Looking through code lazySticky  is used in several places such:

1.
>>    static size_t offsetOfLazySticky() {
>>        return offsetof(RegExpStatics, lazySticky);
>>    }

2. 
>>    RootedLinearString input(cx, matchesInput);
>>    RegExpRunStatus status = g->execute(cx, input, lazyIndex, lazySticky, &this->matches, nullptr);
>>    if (status == RegExpRunStatus_Error)
>>        return false;

I don't know for sure that the initialization of the variable that takes place here:

>>    lazySource = shared->source;
>>    lazyFlags = shared->flags;
>>    lazyIndex = lastIndex;
>>    lazySticky = sticky;
>>    pendingLazyEvaluation = 1;

is done prior of using it.  I consider that there is no harm in initializing it with false in the RegExpStatics::clear function.

	Andi [:andi] Assignee
	Comment 1 • 9 years ago

Review commit: https://reviewboard.mozilla.org/r/30527/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/30527/

	Jason Orendorff [:jorendorff]
	Comment 2 • 8 years ago

Comment on attachment 8706871 [details]
MozReview Request: Bug 1238917 - initialize lazySticky in clear function. r?jorendorff@mozilla.com

https://reviewboard.mozilla.org/r/30527/#review35487

Sorry for the ridiculously slow review here.

For what it's worth, the existing code is correct: all the lazy fields are meaningless (and they won't be read) unless pendingLastEvaluation is 1.

But the constructor initializes all the other fields, so it's super weird not to be initializing this one too.

	Pulsebot
	Comment 3 • 8 years ago

https://hg.mozilla.org/integration/mozilla-inbound/rev/5d450b4ccb14

	Carsten Book [:Tomcat]
	Comment 4 • 8 years ago
bugherder

https://hg.mozilla.org/mozilla-central/rev/5d450b4ccb14