https://sourceware.org/bugzilla/show_bug.cgi?id=32260
--- Comment #1 from Michael Matz <matz at suse dot de> --- Welcome back big testcase :-) > XXX bfdtab->count=1751 table->nbuckets=524288 > XXX bfdtab->count=1752 table->nbuckets=0 So, something between those two wants to reallocate the hash table to have more buckets than can fit into an uint. Like in that other bug report, maybe also print the sizes of sections or the number of presumed entries that it wants to maximally add to get a feel for how much this is overestimating stuff. (E.g. in sec_merge_maybe_resize to print the 'added' param from inside the if condition there; or sec->size in the sole caller of that function) It may be that this is again a too aggressive overestimation. But it may also be real, in which case we really need to audit the code paths that are supposed to handle the cannot-enlarge-hashtable-anymore case (what essentially is supposed to happen is that these overly large input sections simply aren't deduplicated but pasted into the output as is; obviously something goes wrong for that to happen). -- You are receiving this mail because: You are on the CC list for the bug.
