John Regehr (me) did in fact find something like 11 bugs in CompCert, most of which caused it to emit wrong code. But the bugs were not in the proved part.
Sounds interesting -- is there a paper that describes this?
I never wrote up all the details (which seemed somewhat mundane, since the bugs were in unproved code) but see Section 3.1 here:
"The striking thing about our CompCert results is that the middleend bugs we found in all other compilers are absent. As of early 2011, the under-development version of CompCert is the only compiler we have tested for which Csmith cannot find wrong-code errors. This is not for lack of trying: we have devoted about six CPU-years to the task."