The memory address space of the process dumping core is not directly available (in theory one can groak the memory table maintained by dlopen but in reality that information is corrupt). This makes analysis of core generated early in a programs execution or with a corrupt dl-table impossible. Suggest writing out the address map to a separate .note section.
Not sure whats left to be done here. The requestors account has been closed, and I think that the functionality that is required is implemented in 5.3 via the core_patter facility. Does anyone with an interest in this bug have any more input here, or shall we close this as NEXTRELEASE?
closing, no response