The had have construction in an if-clause is an example of what is commonly called the double perfect. This extract from grammar.about.com cites The Cambridge Grammar of the English Language's discussion of the usage, which it calls non-standard:
"Instead of an ordinary preterite perfect, a non-standard 'double
perfect' is often found:
(9i) If it had've come yesterday he would surely have told her.
(9ii) I wish he hadn't've left.
This is largely restricted to speech (or the written representation of speech). It appears to be increasing in frequency, and though it is not as yet
established as a standard form, it is used by many who in general
speak standard English. From a grammatical point of view, it is a
curious construction in that had is followed by a plain form, not a
past participle; it is normally reduced to /əv/, and in writing is
sometimes spelled of, indicating that it is not perceived as a form of
have."
(Rodney Huddleston and Geoffrey K. Pullum, The Cambridge
Grammar of the English Language, Cambridge University Press, 2002)
There's an extended discussion of the issue in Counterfactuality in non-standard subjunctive conditionals. The authors claim that there is a semantic difference between the standard and non-standard forms. Here is their opening section which offers references to possible explanations for the 'extra piece of morphology'.
- On standard and non-standard subjunctive conditionals: The data
English subjunctive conditionals can adopt different forms. The
standard form of subjunctive conditionals (SC) is illustrated in (1).
However, speakers of some dialects of English also accept some of the
variants in (2) as subjunctive conditionals. The forms in (2) differ
from (1) in that they include: would or had.
(1) If Jones had taken arsenic, he would have shown some symptoms. [SC]
(2) If Jones would have/ had have/ would’ve/ had’ve/ woulda’/ hada’/ would of / had of taken arsenic, he would have shown some symptoms. [NSSC]
The non-standard forms (NSSC) exemplified in (2) appeared in the 15th C,
at the time when the subjunctive voice in English disappeared
(Huddleston & Pullum 2002). While the NSSCs mark different registers,
no differences have been observed in terms of interpretation between
an NSSC and an SC. The extra piece of morphology has usually been
considered vacuous: a redundant repetition (Wilson 1993), a
phonological harmony effect (Molencki 2000), or a psychological effect
(Boyland 1995).
The aim of this paper is two-fold: a) we present a novel empirical
observation: we claim there is a contrast in meaning between (1) and (2); and b) we offer an analysis that predicts the differences in meaning.