[Ada] fix handling of fpt on AAMP target

[Arnaud Charlet <charlet at adacore dot com>]

Tested on i686-linux, committed on trunk.

The implementation of initial values was never right for floating point
variables on the AAMP target, but that was only recently noticed due to new
warnings indicating a size mismatch with the byte variables overlaid with
the floating point variables declared in the package spec. The size of the
overlays for types Long_Float and Long_Long_Float is now parameterized to
handle AAMP 48-bit extended float and initial values are set appropriately
for all AAMP float types.

2006-10-31  Gary Dismukes  <dismukes@adacore.com>

	* s-scaval.adb (Initialize): Add new Boolean flag AFloat that is set
	True when AAMP extended floating-point is in use (48-bit). Change type
	ByteLF to ByteLLF, add new array type ByteLF and condition the size of
	the two byte array types on AFloat. Change type of IV_Ilf overlay
	variable from Byte8 to ByteLF. Add appropriate initializations of
	floating-point overlays for AAMP cases.

Attachment: difs
Description: Text document