Hi, I remember there was a read beyond allocated area somewhere. This happens in very rare cases iirc. Only had this once yet.
I am not sure - maybe I already submitted a bug report with fix to the kraken bugtracker? Maybe I can check in the evening where this problem came from. BR, Georg Zitat von mad <m...@auth.se>: > Hi kraken developers, hi list! > > I experience a segmentation fault when cracking a specific single burst. > That happens using current git version, A5Cpu.so as well as A5Ati.so > on Ubuntu 11.04 64-bit, compiled with AMD App SDK 2.4. > In cpu mode kraken crashes after a found state is displayed, in gpu > mode without printing but at the analogous moment, short before the > search cycle usually ends. > It's only that specific burst I came by (yet) that leads to this, > but it should be a more general, if rare occuring problem. > > Included a valgrind output on that crack command. > I hope that's sufficient information for you to debug and someone > could take a look at this, thank you. > > > Regards, > Mad > > > -------8<------8<------ > Kraken> crack > 010011101001111010101110100000001101111100010011000010011110011110000110010111110101000101000101000000011000000000 > > Cracking > 010011101001111010101110100000001101111100010011000010011110011110000110010111110101000101000101000000011000000000 > ==2932== Thread 2: > ==2932== Conditional jump or move depends on uninitialised value(s) > ==2932== at 0x4070C9: NcqDevice::WorkerThread() (in > /home/mad/Kraken/kraken) > ==2932== by 0x40713D: NcqDevice::thread_stub(void*) (in > /home/mad/Kraken/kraken) > ==2932== by 0x4E35D8B: start_thread (pthread_create.c:304) > ==2932== by 0x5AD804C: clone (clone.S:112) > ==2932== > Found 19b62fb1e7a227b2 @ 1 #0 (table:436) > ==2932== Thread 1: > ==2932== Invalid read of size 4 > ==2932== at 0x407A92: > DeltaLookup::StartEndpointSearch(NcqRequestor*, unsigned long, > unsigned long&) (in /home/mad/Kraken/kraken) > ==2932== by 0x408269: Fragment::handleSearchResult(unsigned long, > int) (in /home/mad/Kraken/kraken) > ==2932== by 0x40449B: Kraken::Tick() (in /home/mad/Kraken/kraken) > ==2932== by 0x405051: main (in /home/mad/Kraken/kraken) > ==2932== Address 0x28d0f458 is 0 bytes after a block of size > 40,961,048 alloc'd > ==2932== at 0x4C28658: operator new[](unsigned long) > (vg_replace_malloc.c:305) > ==2932== by 0x4076AB: DeltaLookup::DeltaLookup(NcqDevice*, > std::string) (in /home/mad/Kraken/kraken) > ==2932== by 0x4038CD: Kraken::Kraken(char const*, int) (in > /home/mad/Kraken/kraken) > ==2932== by 0x405026: main (in /home/mad/Kraken/kraken) > ==2932== > ==2932== Conditional jump or move depends on uninitialised value(s) > ==2932== at 0x407AA3: > DeltaLookup::StartEndpointSearch(NcqRequestor*, unsigned long, > unsigned long&) (in /home/mad/Kraken/kraken) > ==2932== by 0x408269: Fragment::handleSearchResult(unsigned long, > int) (in /home/mad/Kraken/kraken) > ==2932== by 0x40449B: Kraken::Tick() (in /home/mad/Kraken/kraken) > ==2932== by 0x405051: main (in /home/mad/Kraken/kraken) > ==2932== > ==2932== > ==2932== More than 10000000 total errors detected. I'm not > reporting any more. > ==2932== Final error counts will be inaccurate. Go fix your program! > ==2932== Rerun with --error-limit=no to disable this cutoff. Note > ==2932== that errors may occur in your program without prior warning from > ==2932== Valgrind, because errors are no longer being displayed. > ==2932== > ==2932== > ==2932== Process terminating with default action of signal 11 (SIGSEGV) > ==2932== Access not within mapped region at address 0x37F46000 > ==2932== at 0x407A92: > DeltaLookup::StartEndpointSearch(NcqRequestor*, unsigned long, > unsigned long&) (in /home/mad/Kraken/kraken) > ==2932== by 0x408269: Fragment::handleSearchResult(unsigned long, > int) (in /home/mad/Kraken/kraken) > ==2932== by 0x40449B: Kraken::Tick() (in /home/mad/Kraken/kraken) > ==2932== by 0x405051: main (in /home/mad/Kraken/kraken) > ==2932== If you believe this happened as a result of a stack > ==2932== overflow in your program's main thread (unlikely but > ==2932== possible), you can try to increase the size of the > ==2932== main thread stack using the --main-stacksize= flag. > ==2932== The main thread stack size used in this run was 8388608. > ==2932== > ==2932== HEAP SUMMARY: > ==2932== in use at exit: 1,650,537,400 bytes in 2,865 blocks > ==2932== total heap usage: 36,296 allocs, 33,431 frees, > 1,653,870,945 bytes allocated > ==2932== > ==2932== LEAK SUMMARY: > ==2932== definitely lost: 0 bytes in 0 blocks > ==2932== indirectly lost: 0 bytes in 0 blocks > ==2932== possibly lost: 3,239 bytes in 11 blocks > ==2932== still reachable: 1,650,534,161 bytes in 2,854 blocks > ==2932== suppressed: 0 bytes in 0 blocks > ==2932== Rerun with --leak-check=full to see details of leaked memory > ==2932== > ==2932== For counts of detected and suppressed errors, rerun with: -v > ==2932== Use --track-origins=yes to see where uninitialised values come from > ==2932== ERROR SUMMARY: 10000000 errors from 3 contexts (suppressed: > 4 from 4) > > _______________________________________________ > A51 mailing list > A51@lists.reflextor.com > http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51 > ---------------------------------------------------------------- This message was sent using IMP, the Internet Messaging Program. _______________________________________________ A51 mailing list A51@lists.reflextor.com http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51