THULAC
THULAC copied to clipboard
SEGV signal occurred when running program thulac
When I try to run thulac and thulac_test program. I found this :
ASAN:SIGSEGV
=================================================================
==12976==ERROR: AddressSanitizer: SEGV on unknown address 0x000000000000 (pc 0x7fbf3a4841ba bp 0x000000000000 sp 0x7ffc53739440 T0)
#0 0x7fbf3a4841b9 in _IO_fread (/lib/x86_64-linux-gnu/libc.so.6+0x6e1b9)
#1 0x442c38 in fread /usr/include/x86_64-linux-gnu/bits/stdio2.h:295
#2 0x442c38 in permm::BasicModel<int>::BasicModel(char const*) include/cb_model.h:89
#3 0x436b5b in THULAC::init(char const*, char const*, int, int, int, char) include/thulac.h:157
#4 0x404962 in main src/thulac.cc:80
#5 0x7fbf3a43682f in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x2082f)
#6 0x4054e8 in _start (/home/mfc_fuzz/newprogram/THULAC/thulac+0x4054e8)
I believe another issue occurred when I running train_c. The input file has been put at :https://github.com/fCorleone/fuzz_programs/blob/master/thulac/issue2_testcase. It's another SEGV signal issue:
ASAN:SIGSEGV
=================================================================
==13658==ERROR: AddressSanitizer: SEGV on unknown address 0x61c00008e89c (pc 0x000000413848 bp 0x61c00000f080 sp 0x7ffdf4de1450 T0)
#0 0x413847 in thulac::NGramFeature::find_bases(int, int, int, int&, int&) include/cb_ngram_feature.h:248
#1 0x413847 in thulac::NGramFeature::put_values(int*, int) include/cb_ngram_feature.h:118
#2 0x413847 in thulac::TaggingDecoder::put_values() include/cb_tagging_decoder.h:387
#3 0x42c888 in thulac::TaggingLearner::train(char const*, char const*, char const*, char const*) include/cb_tagging_learner.h:305
#4 0x404239 in main src/train_c.cc:62
#5 0x7f54f512582f in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x2082f)
#6 0x404c98 in _start (/home/mfc_fuzz/newprogram/THULAC/train_c+0x404c98)