From 3526060a19e0e8cfad7383b98a05484793e6efda Mon Sep 17 00:00:00 2001
From: slederer <dev@insignificance.de>
Date: Mon, 7 Apr 2025 00:25:32 +0200
Subject: [PATCH] sasm: bugfix for LBRANCH/LCBRANCH size changes

- The LBRANCH and LCBRANCH directives
  create different instruction sequences
  depending on the jump distance.
  So the code size can shrink during the first pass
  when the jump distance can be determined (when the
  label that is the jump destination is parsed).
  In the long form, LOADREL/JUMP is used, which might
  or might not need 2 bytes of padding. With this bugfix,
  the padding is always added, either before or after
  the LOADREL indirect operand, so that the code
  size does not change depending on the padding
  required. Otherwise the code might shrink further
  on the second pass because a LBRANCH/LCBRANCH
  instruction no longer needs padding due to an
  earlier code size change.
---
 lib/runtime.s  |  2 +-
 pcomp/sasm.pas | 28 +++++++++++++++++++++-------
 2 files changed, 22 insertions(+), 8 deletions(-)

diff --git a/lib/runtime.s b/lib/runtime.s
index f4ad875..6f248b9 100644
--- a/lib/runtime.s
+++ b/lib/runtime.s
@@ -1616,7 +1616,7 @@ MEM_FREE_XT:
 ; Since the return stack is no longer valid afterwards, directly
 ; jumps to _MAIN instead of using RET.
 
-; parameters: [ _MAIN entry point, start of heap address ]
+; parameters: [ start of heap address ]
 _MEM_INIT:
 	; initialize anchor chunk with start of heap address
 	; and heap size - header size
diff --git a/pcomp/sasm.pas b/pcomp/sasm.pas
index ab5230b..7b01f32 100644
--- a/pcomp/sasm.pas
+++ b/pcomp/sasm.pas
@@ -1515,9 +1515,7 @@ begin
 		end;
 
 		if dref^.symboldata.value <> value then
-			(* writeln('////// label changed value ', keyword, ' ',
-				dref^.symboldata.value, ' -> ', value); *)
-		dref^.symboldata.value := value;
+			dref^.symboldata.value := value;
 	end;
 end;
 
@@ -1872,9 +1870,9 @@ begin
 		if (pc and 3) = 0 then	(* no padding *)
 		begin
 			pad := false;
-			(* total size 8 bytes *)
+			(* total size 10 bytes *)
 			offset := 4;	(* offset for LOADREL *)
-			shrinkage := 6; (* difference to short form size *)
+			shrinkage := 8; (* difference to short form size *)
 		end
 		else
 		begin
@@ -1892,9 +1890,17 @@ begin
 		encodeInstruction('JUMP', 0, encoded);
 		emitInstructionWord(encoded);
 
+		(* Padding is always done, so the length of the sequence does
+			not change depending on the position. The pad is either
+			placed before the address to align it on a word address,
+			or after, if it is already aligned. *)
 		if pad then
 			emitInstructionWord(0);
+
 		emitWord(value);
+
+		if not pad then
+			emitInstructionWord(0);
 	end
 	else
 	begin
@@ -1937,9 +1943,9 @@ begin
 		if (pc and 3) = 2 then (* no padding *)
 		begin
 			pad := false;
-			(* total size 10 bytes *)
+			(* total size 12 bytes *)
 			offset := 4;	(* offset for LOADREL *)
-			shrinkage := 8; (* difference to short form size *)
+			shrinkage := 10; (* difference to short form size *)
 		end
 		else
 		begin
@@ -1967,9 +1973,17 @@ begin
 		encodeInstruction('JUMP', 0, encoded);
 		emitInstructionWord(encoded);
 
+		(* Padding is always done, so the length of the sequence does
+			not change depending on the position. The pad is either
+			placed before the address to align it on a word address,
+			or after, if it is already aligned. *)
 		if pad then
 			emitInstructionWord(0);
+
 		emitWord(value);
+
+		if not pad then
+			emitInstructionWord(0);
 	end
 	else
 	begin