Incorrect optimalization for float point
Original Reporter info from Mantis: FPCCore
-
Reporter name: FPC core team
Original Reporter info from Mantis: FPCCore
- Reporter name: FPC core team
Description:
If progrem use float point and compiler is set to store variables in processor registers incorrect code is generate. Some valuas are store in FPU registers and are destroy in FPU operation time.
Example :
If in1 and in2 are extended type, temporary store in FPU register and program assign values to this variables, next operation (ex. in1 := in1 + in2) is impossible.
Probably compiler don't check change of position on FPU stack
Without optimalization all is OK.
(sorry for my English) J.T.
Additional information:
Reporter: Janusz
EMail: turek dot fm at poczta dot fm
Mantis conversion info:
- Mantis ID: 5459
- OS: GO32V2
- Version: 1.0.10
- Fixed in version: 1.9.5