Author: Robert Hyatt
Date: 08:05:03 07/12/01
Go up one level in this thread
On July 12, 2001 at 10:00:34, Gian-Carlo Pascutto wrote: >Hi all, > >I've known for a while that the latest release of Sjeng had >some crashing bugs, but I've never been able to reliably >reproduce any of them. > >The coredumps left behind on the unix system always were bogus, >so that was no help too. > >Today I finally found a position where the bugs are reproducible. > >My problem is, that neither gcc/gdb/Linux nor MSVC/WinME >give out any usefull information whatsoever. > >I either get a sigsegv (linux) or illegal instruction error >(msvc) at the same (bogus) address 0x75. > >I have no idea how to debug this. It occurs quite deep in the >search and only seems to trigger in freak conditions (once in >about 150-200 games on the server). > >My guess is that I am overwriting the stack somewhere, but how >do I find out where? > >If someone has the equipement or knowledge to help out, here >is how to trigger the bug: > >Download the latest Sjeng release (10.0) from my webpage (www.sjeng.org), >either the tarball or the windows port. > >Fire up Win/XBoard with the -variant crazyhouse parameter and >load the following game: > >[Event "ICS Rated crazyhouse match"] >[Site "freechess.org"] >[Date "2001.07.12"] >[Round "-"] >[White "Sjeng"] >[Black "BEO"] >[Result "*"] >[WhiteElo "2398"] >[BlackElo "2254"] >[TimeControl "180"] >[Variant "crazyhouse"] > >1. e4 e6 2. d4 d5 3. exd5 exd5 4. Bb5+ c6 5. Qe2+ Be7 6. Bf4 cxb5 7. Qxb5+ >Nc6 8. P@c7 Qd7 9. P@e3 Nf6 10. Nd2 O-O 11. Rc1 B@a5 12. c3 Bxc7 13. Bxc7 >Qxc7 14. B@g3 Qd8 15. Bh4 B@a6 16. Qb3 Ne4 17. Bxe7 Nxe7 18. Nxe4 dxe4 19. >N@c5 N@d3+ 20. Nxd3 exd3 21. B@d2 N@c2+ 22. Kf1 P@e2+ 23. Nxe2 dxe2+ 24. >Kg1 B@d1 25. N@h6+ gxh6 26. Qxc2 Bxc2 27. N@f6+ Kh8 28. c4 Kg7 29. Nh5+ Kg8 >30. Nf6+ Kh8 31. h4 Bxc4 32. P@g3 P@g7 33. e4 gxf6 34. g4 Bxe4 35. Kh2 >N@f1+ 36. Rhxf1 exf1=N+ 37. Rxf1 Bxf1 38. P@g7+ Kxg7 39. Bxh6+ Kxh6 40. >P@g5+ fxg5 41. hxg5+ Kxg5 42. N@h3+ Kxg4 43. f3+ Kh5 >* > >Skip to the end of the game and start analysis mode. Sjeng will crash >somewhere in ply 5. > >If someone can advise, please do so. > >-- >GCP Have you tried it by compiling with -g, then running under gdb and when it crashes, typing "where"? That should show you what called what, unless you are blowing out memory which might wipe out the ability to trace back thru the stack frames to see how the calls were done...
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.