paweld/fpc
1
0
Fork 0
mirror of https://gitlab.com/freepascal.org/fpc/source.git synced 2025-08-18 06:09:22 +02:00
Code Issues Packages Projects Releases Wiki Activity

+ comment about badly placed release_loc calls for a_mul which

Browse Source 
causes wrong regdeallocations. Don't know how to fix :(
This commit is contained in:
Jonas Maebe 2000-01-09 16:35:39 +00:00
parent 7b87725809
commit 2ce42d351a
1 changed files with 9 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

10
compiler/cg386add.pas
View File

@ -1024,6 +1024,10 @@ implementation
popeax:=false; popeax:=false;
popedx:=false; popedx:=false;
{ here you need to free the symbol first } { here you need to free the symbol first }
{ This is VERY bad!!!! Now the optimizer assumes the current values of }
{ those registers are NOT needed anymore, which causes bugs! The }
{ optimizer now contains a workaround which tries to fix all wrong }
{ deallocations, but this kind of things should NOT be done (JM) }
clear_location(p^.location); clear_location(p^.location);
release_loc(p^.right^.location); release_loc(p^.right^.location);
release_loc(p^.left^.location); release_loc(p^.left^.location);
@ -2206,7 +2210,11 @@ implementation
end. end.
{ {
$Log$ $Log$
Revision 1.86 2000-01-09 12:34:59 jonas Revision 1.87 2000-01-09 16:35:39 jonas
+ comment about badly placed release_loc calls for a_mul which
causes wrong regdeallocations. Don't know how to fix :(
Revision 1.86 2000/01/09 12:34:59 jonas
* changed edi allocation to use getexplicitregister32/ungetregister * changed edi allocation to use getexplicitregister32/ungetregister
(adapted tgeni386 a bit for this) and enabled it by default (adapted tgeni386 a bit for this) and enabled it by default
* fixed very big and stupid bug of mine in cg386mat that broke the * fixed very big and stupid bug of mine in cg386mat that broke the