From 7dc12ac44f3a03348230f563e8409a9e64adf7b5 Mon Sep 17 00:00:00 2001
From: David Given <dg@cowlark.com>
Date: Mon, 26 Feb 2024 21:46:34 +0100
Subject: [PATCH] Don't write garbage after strings when initialising arrays to
 a string which is shorter than the array itself. When writing garbage, make
 sure it's actually escaped properly.

---
 mach/proto/mcg/data.c     | 11 ++++++++---
 mach/proto/mcg/mcg.h      |  2 +-
 mach/proto/mcg/parse_em.c |  2 +-
 3 files changed, 10 insertions(+), 5 deletions(-)

diff --git a/mach/proto/mcg/data.c b/mach/proto/mcg/data.c
index dfd3651d4..a137196e9 100644
--- a/mach/proto/mcg/data.c
+++ b/mach/proto/mcg/data.c
@@ -81,10 +81,10 @@ static bool istext(char c)
     return isprint(c) && (c != '"');
 }
 
-void data_block(const uint8_t* data, size_t size, bool is_ro)
+void data_block(const uint8_t* data, size_t datalen, size_t size, bool is_ro)
 {
     const uint8_t* start = data;
-    const uint8_t* end = data + size;
+    const uint8_t* end = data + datalen;
     const uint8_t* p = data;
 
 	emit_header(is_ro ? SECTION_ROM : SECTION_DATA);
@@ -100,7 +100,9 @@ void data_block(const uint8_t* data, size_t size, bool is_ro)
             fprintf(outputfile, "\t.ascii \"");
             while (start < p)
             {
-                fprintf(outputfile, "%c", *start);
+                if ((*start == '\\') || (*start == '"'))
+                    fputc('\\', outputfile);
+                fputc(*start, outputfile);
                 start++;
             }
             fprintf(outputfile, "\"\n");
@@ -125,6 +127,9 @@ void data_block(const uint8_t* data, size_t size, bool is_ro)
             fprintf(outputfile, "\n");
         }
     }
+
+    for (size_t zeroes = 0; zeroes < (size - datalen); zeroes++)
+        fprintf(outputfile, "\t.data1 0\n");
 }
 
 void data_offset(const char* label, arith offset, bool is_ro)
diff --git a/mach/proto/mcg/mcg.h b/mach/proto/mcg/mcg.h
index 69f194e85..a0e5436c7 100644
--- a/mach/proto/mcg/mcg.h
+++ b/mach/proto/mcg/mcg.h
@@ -95,7 +95,7 @@ extern struct symbol* symbol_walk(symbol_walker_t* walker, void* user);
 extern void data_label(const char* name);
 extern void data_int(arith data, size_t size, bool is_ro);
 extern void data_float(const char* data, size_t size, bool is_ro);
-extern void data_block(const uint8_t* data, size_t size, bool is_ro);
+extern void data_block(const uint8_t* data, size_t datalen, size_t size, bool is_ro);
 extern void data_offset(const char* label, arith offset, bool is_ro);
 extern void data_bss(arith size, int init);
 
diff --git a/mach/proto/mcg/parse_em.c b/mach/proto/mcg/parse_em.c
index e8859d2fb..8411d13d5 100644
--- a/mach/proto/mcg/parse_em.c
+++ b/mach/proto/mcg/parse_em.c
@@ -281,7 +281,7 @@ static void parse_pseu(void)
                 }
 
 				case str_ptyp:
-                    data_block((const uint8_t*) strdup(em.em_string), em.em_size, ro);
+                    data_block((const uint8_t*) em.em_string, strlen(em.em_string), em.em_size, ro);
 					break;
 
                 case cst_ptyp: