[Ecls-list] Segfault while parsing floats

Stas Boukarev stassats at gmail.com
Fri Oct 1 19:33:25 UTC 2010


After the commit
"ecl_parse_number and ecl_parse_integer are splitted into separate files and do not rely on the C library (atoi, strtod, ...) because these functions are affected by the locale."
when entering in ecl_min 0.5 i get the following:

#0  0xb7f92b0f in __gmpn_tdiv_qr () from /usr/lib/libgmp.so.10
#1  0xb7f80fe0 in __gmpz_tdiv_qr () from /usr/lib/libgmp.so.10
#2  0xb7f77263 in __gmpz_fdiv_qr () from /usr/lib/libgmp.so.10
#3  0x080adacd in _ecl_big_floor (a=0x883f4c8, b=0xbf800198, pr=0xbf8001d8) at src/c/big.d:270
#4  0x0808fd85 in ecl_floor2 (x=0x883f4c8, y=0x7) at src/c/num_co.d:248
#5  0x08090dff in ecl_truncate2 (x=0x883f4c8, y=0x7) at src/c/num_co.d:608
#6  0x080894cd in prepare_ratio_to_float (num=0x883f4c8, den=0x7, digits=53, scaleout=0xbf800320) at src/c/number.d:714
#7  0x08089634 in ratio_to_double (num=0x883f4f8, den=0x7) at src/c/number.d:758
#8  0x080896c6 in ecl_to_double (x=0x883f4f8) at src/c/number.d:790
#9  0x08089645 in ratio_to_double (num=0x883f558, den=0x7) at src/c/number.d:759
#10 0x080896c6 in ecl_to_double (x=0x883f558) at src/c/number.d:790
#11 0x08089645 in ratio_to_double (num=0x883f5d0, den=0x7) at src/c/number.d:759
... and thousands of frames of mutual recursion between ratio_to_double
and ecl_to_double.

-- 
With Best Regards, Stas.




More information about the ecl-devel mailing list