paweld/fpc
1
0
Fork 0
mirror of https://gitlab.com/freepascal.org/fpc/source.git synced 2025-04-06 14:47:55 +02:00
Code Issues Packages Projects Releases Wiki Activity

Use round to convert float constant into s64comp type for LLVM compiler as is done in general case

Browse Source
This commit is contained in:
Pierre Muller 2022-05-12 22:32:26 +02:00
parent deec6fc4ba
commit 69d40dd17a
1 changed files with 1 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
compiler/llvm/nllvmcon.pas
View File

@ -199,9 +199,7 @@ implementation
s32real,s64real:
current_asmdata.CurrAsmList.concat(taillvm.op_reg_size_fpconst_size(la_bitcast,location.register,resultdef,value_real,resultdef));
{ comp and currency are handled as int64 at the llvm level }
s64comp:
{ sc80floattype instead of resultdef, see comment in thlcgllvm.a_loadfpu_ref_reg }
current_asmdata.CurrAsmList.concat(taillvm.op_reg_size_const_size(la_sitofp,location.register,s64inttype,trunc(value_real),sc80floattype));
s64comp,
s64currency:
{ sc80floattype instead of resultdef, see comment in thlcgllvm.a_loadfpu_ref_reg }
current_asmdata.CurrAsmList.concat(taillvm.op_reg_size_const_size(la_sitofp,location.register,s64inttype,round(value_real),sc80floattype));