C1's PatchingStub copies bytes from the method's regular code area to a dedicated patching template which gets copied back when the patching work is completed. Currently, C1's PatchingStub with PatchID "access_field_id" copies all bytes between the PatchingStub's instanciation and the patching_epilog call. This is not only more than necessary, it can cause _bytes_to_copy to exceed its one byte limitation: emit_int8(_bytes_to_copy). On some platforms, code for VerifyOops is part of the code which gets copied forth and back. This is unnecessary and should be avoided. Note that the template for PatchID "access_field_id" never gets executed, so there's no reason for copying more than the code to be patched. So there are basically 2 issues: - Possible size overflow (depending on platform and activated features) - Wasted code cache space