0003072: Reproducible segmentation fault when choosing fortress location

Notes
(0011934) vyznev (reporter) 2010-08-19 14:37	I couldn't find any way to attach the world directly to this bug report, so I uploaded it to http://dffd.wimbli.com/file.php?id=2973 [^] instead.

(0011939) Logical2u (manager) 2010-08-19 17:56	The world generation parameters would be more helpful, please upload them to a service like Pastebin.

(0012308) vyznev (reporter) 2010-09-01 12:53	The params seem to have been these: http://vyznev.net/misc/region2-world_gen_param.txt [^] However, when I regen the world using these setting (either on the DF I originally genned it on or on a virgin install), it doesn't seem crash anymore. Apparently the save is corrupted in some way, triggering the crash. (I'm not sure what might've corrupted it, though; I don't recall having any data corruption issues on this computer before.) A quick diff -r says: Binary files region1/unit-0.dat and region2/unit-0.dat differ Binary files region1/unit-100.dat and region2/unit-100.dat differ Binary files region1/unit-109.dat and region2/unit-109.dat differ Binary files region1/unit-112.dat and region2/unit-112.dat differ Binary files region1/unit-114.dat and region2/unit-114.dat differ Binary files region1/unit-116.dat and region2/unit-116.dat differ Binary files region1/unit-67.dat and region2/unit-67.dat differ Binary files region1/unit-68.dat and region2/unit-68.dat differ Binary files region1/unit-70.dat and region2/unit-70.dat differ Binary files region1/unit-80.dat and region2/unit-80.dat differ Binary files region1/unit-89.dat and region2/unit-89.dat differ Binary files region1/unit-95.dat and region2/unit-95.dat differ Binary files region1/world.dat and region2/world.dat differ Comparing the file sizes, I get: unit-0.dat: OK = 35677, bad = 35480 unit-100.dat: OK = 15399, bad = 15420 unit-109.dat: OK = 12649, bad = 12689 unit-112.dat: OK = 99254, bad = 99154 unit-114.dat: OK = 81728, bad = 81632 unit-116.dat: OK = 88167, bad = 88187 unit-67.dat: OK = 89626, bad = 89822 unit-68.dat: OK = 15446, bad = 15886 unit-70.dat: OK = 25511, bad = 25536 unit-80.dat: OK = 18582, bad = 18618 unit-89.dat: OK = 9416, bad = 9438 unit-95.dat: OK = 4235, bad = 4419 world.dat: OK = 9913705, bad = 9913976 Without a binary diff tool or any way to parse the .dat files, though, I'm not sure what else to do to diagnose this further.

(0012310) vyznev (reporter) 2010-09-01 13:01	...I also noticed that the crashing version has these extra files: Only in region2: feature-1-5.dat Only in region2: feature-1-6.dat Only in region2: feature-1-7.dat Only in region2: feature-2-3.dat Only in region2: feature-2-4.dat Only in region2: feature-2-5.dat Only in region2: feature-2-6.dat Only in region2: feature-2-7.dat Only in region2: feature-2-8.dat Only in region2: feature-3-3.dat Only in region2: feature-3-4.dat Only in region2: feature-3-5.dat Only in region2: feature-3-6.dat Only in region2: feature-3-7.dat Only in region2: feature-3-8.dat Only in region2: feature-4-3.dat Only in region2: feature-4-4.dat Only in region2: feature-4-5.dat Only in region2: feature-4-6.dat Only in region2: feature-4-7.dat Only in region2: feature-5-3.dat Only in region2: feature-5-4.dat Only in region2: feature-5-5.dat Only in region2: feature-5-6.dat Only in region2: feature-5-7.dat Only in region2: feature-5-8.dat Only in region2: feature-6-3.dat Only in region2: feature-6-4.dat Only in region2: feature-6-5.dat Only in region2: feature-6-6.dat Only in region2: feature-6-8.dat Only in region2: feature-7-3.dat Only in region2: feature-7-4.dat Only in region2: feature-7-5.dat Only in region2: feature-7-6.dat Only in region2: feature-8-4.dat Only in region2: feature-8-5.dat Only in region2: feature-8-6.dat Only in region2: feature-8-7.dat Only in region2: feature-9-6.dat Only in region2: feature-9-7.dat Indeed, I just tried to simply delete these files from the save folder, and it seems this fixes the crash(!). However, I don't understand the DF save format enough to know what these files are actually supposed to be for.

(0012311) vyznev (reporter) 2010-09-01 13:21 edited on: 2010-09-01 13:22	I did some testing by deleting various subsets of the feature-*.dat files, trying to see if I could narrow down which ones were causing the crash. However, it seems that each (or at least most) of them corresponds to a particular map region that one needs to view to make the game crash, so I was unable to significantly narrow it down. I did observe that e.g. feature-3-3.dat alone is sufficient (but not necessary) to reproduce the crash using the exact steps given above.

(0021963) dree12 (reporter) 2012-03-31 17:43 edited on: 2012-03-31 17:44	Recently, some feature corruption bugs have been fixed (0005077 in particular). It might be useful to check this old bug.

(0021989) Footkerchief (manager) 2012-04-01 09:23	Good call, it is indeed corrupted.

Issue History
Date Modified	Username	Field	Change
2010-08-19 14:27	vyznev	New Issue
2010-08-19 14:27	vyznev	Issue Monitored: vyznev
2010-08-19 14:37	vyznev	Note Added: 0011934
2010-08-19 17:56	Logical2u	Note Added: 0011939
2010-09-01 12:53	vyznev	Note Added: 0012308
2010-09-01 13:01	vyznev	Note Added: 0012310
2010-09-01 13:21	vyznev	Note Added: 0012311
2010-09-01 13:22	vyznev	Note Edited: 0012311	View Revisions
2012-03-31 17:43	dree12	Note Added: 0021963
2012-03-31 17:44	dree12	Note Edited: 0021963	View Revisions
2012-03-31 17:44	dree12	Issue Monitored: dree12
2012-04-01 09:23	Footkerchief	Note Added: 0021989
2012-04-01 09:23	Footkerchief	Relationship added	duplicate of 0005077
2012-04-01 09:23	Footkerchief	Status	new => resolved
2012-04-01 09:23	Footkerchief	Resolution	open => duplicate
2012-04-01 09:23	Footkerchief	Assigned To	=> Footkerchief
2012-04-01 17:51	dree12	Issue End Monitor: dree12