The floating-point environment includes exception flags and masks, the current rounding mode, and other architecture-specific settings. However, it does not include the floating-point register file.
The
fegetenv() function stores the current floating-point environment in the object pointed to by
envp, whereas
feholdexcept() saves the current environment, then clears all exception flags and masks all floating-point exceptions.
The
fesetenv() function restores a previously saved environment. The
feupdateenv() function restores a saved environment as well, but it also raises any exceptions that were set in the environment it replaces.
The
feholdexcept() function is often used with
feupdateenv() or
fesetenv() to suppress spurious exceptions that occur as a result of intermediate computations. An example in
fenv(3) demonstrates how to do this.