i960 Processor Compiler User's Guide
2-42
2
mdouble4 Generate code so that the size and alignment of
double is the same as float.
mlong-double4 Generate code so that the size and alignment of
long double is the same as float.
NOTE. The mdouble4 and mlong-double4 options force floating-point
arguments to be passed in single-precision format. When your source
program explicitly calls functions (such as
sin and printf) that require
double-precision or extended-precision arguments, the arguments passed
to these functions are incorrect.
mi960_align=
n
Aligns struct data on the byte boundary that is
a multiple of
n
. (Legal values are 1, 2, 4, 8, 16.)
mic-compat Use ic960 R2.0's rules for size and alignment of
mic2.0-compat types. This option also causes the compiler to
use the ic960 compiler's rules for promotion of
char, unsigned char, short, and unsigned
short
types at function call and return.
mic3.0-compat Use ic960 R3.0's rules for size and alignment of
types and other conventions. These are largely
the same as gcc960's, but ic960 R3.0 selects the
size of
enums based on their value. Additionally,
ic960 R3.0 assumes that type char is signed by
default, whereas gcc960 assumes it is unsigned.
The
mic3.0-compat option emulates ic960's
behavior.
mleaf-procedures | Generate/do not generate output that contains
mno-leaf-procedures leaf procedures: these are procedures that may be
entered with the
bal instruction rather than
call. The linker automatically promotes call
instructions into bal instructions when
To Next Page
Loading...
