|
From: Julian S. <js...@ac...> - 2006-02-09 23:46:40
|
> Yes , getting rid of that code and adding just the have_fp =have_vmx = > False; works. > It seems that the test in the original code is taking the wrong path and > it fails. No; signal handling in your Linux distro, or cross-compiling arrangement, or something, is broken. This code works fine on multiple different PowerPC platforms we test on. > Where is the __builtin_setjmp defined???? I could not find it). It's a function that gcc 'knows' about without needing a prototype. > Also, are you trying to defect if you have fp or not at runtime?? Yes, also trying to detect presence/absence of Altivec. J |