Bug 12623
| Summary: | Certain valid numeric strings cause a SIGFPE when attempting | ||
|---|---|---|---|
| Product: | Base System | Reporter: | m.seaman <m.seaman> |
| Component: | alpha | Assignee: | robert |
| Status: | Closed FIXED | ||
| Severity: | Affects Only Me | ||
| Priority: | Normal | ||
| Version: | 3.2-STABLE | ||
| Hardware: | Any | ||
| OS: | Any | ||
At Tue, 13 Jul 1999 12:27:40 +0100 (BST), m.seaman@inpharmatica.co.uk wrote: > >Description: > > Attempting to convert the string "2.49521e-297" into a double > precision value via strtod(3) results in a floating point exception on > the alphaev56 CPU. The problem doesn't occur on i586. /usr/src/lib/libc/stdlib/strtod.c seems broken on alpha in three ways. 1) IEEE_8087 should be defined instead of IEEE_MC68k. 2) It assumes long is 32bit, but long is 64bit on alpha. s/unsigned long/u_int32_t/, s/long/int32_t/. 3) It generates denormal numbers which can not be treaded by the hardware on alpha. It should be compiled with the option `-mtrap-precision=i -mfp-trap-mode=su' to enable software completion. -- /\ Hidetoshi Shimokawa \/ simokawa@sat.t.u-tokyo.ac.jp PGP public key: finger -l simokawa@sat.t.u-tokyo.ac.jp State Changed From-To: open->patched A fix was committed to -CURRENT. Responsible Changed From-To: freebsd-alpha->robert Robert reported that a fix was committed. State Changed From-To: patched->closed |
Attempting to convert the string "2.49521e-297" into a double precision value via strtod(3) results in a floating point exception on the alphaev56 CPU. The problem doesn't occur on i586. The problem is sensitive to both the mantissa and exponent of the number: the following values are known to trigger the bug: 2.49521e-297 3.35554e-295 1.14662e-292 1.14e-293 2.495e-293 2.49e-294 2.4e-293 1e-305 whereas these do not: 2.4e-292 2.49521e-292 2.4951e-293 1.14e-292 1.15e-293 Fix: not known How-To-Repeat: #include <stdlib.h> int main (int argc, char *argv[]) { char str[32] = "2.49521e-297"; double num; num = strtod (str, NULL); }