Follow-up Comment #10, bug #13498 (project freeciv):

Hmm well, I managed to get one crash but then never
again, even retracing the same steps from the same
savegame with the same binary. Furthermore the back-
trace showed a slightly different code path, but
still basically a map redraw from a gui callback.
Without a consistent way to reproduce the bug I
cannot isolate the original cause; it surely is
some corruption of ptile->worked, but whether this
occurs due to incorrect behavior of handle_city
_info or some other client function, or even due
to something on the server-side, this is unknown.

Maybe civilwar is involved somehow (or maybe it
just causes more changes at the same time, raising
the probability that the bug occurs).

Anyway, as a temporary stopgap measure the attached
patch checks for a corrupted ptile->worked pointer
in find_city_or_settler_near_tile(), and if found
ignores it. In trunk the worked field handling is
already different (there in handle_tile_info for
some reason, along with the strange "invisible
cities" hack), so I will leave that version alone
for the moment.

Please test and report if it at least avoids the
client crash (and any other bad behavior).


(file #5791)

Additional Item Attachment:

File name: avoid_mysterious_ptile_worked_crash.diff Size:0 KB


Reply to this item at:


  Message sent via/by Gna!

Freeciv-dev mailing list

Reply via email to