On Sat Dec 16 08:13:32 2006, [EMAIL PROTECTED] wrote: > test.exe or config/auto/jit/test_c.in segfault during configuration on > Win32. Note that this happent on > Pentium MMX ( 586 ). The test tests for "fucomip" > cpu instruction, but the 586 generation CPUs don't > have it. It's first introduced in > Pentium PRO (686). The segfault opens a debug > window in win32, > where the user must press OK for the Configure > script to continue. After this the configuration and > compilation are successful, because test.exe returns > false after the segfault and "fucomip" is not used. Thanks for reporting this; kelli++ is looking into using CPUID instead of just trying to use the opcode and seeing if we crash and burn, as happens now.
Jonathan