From 6bef09f68b81290a69b7a6efb7252bde86669d9e Mon Sep 17 00:00:00 2001
From: Eberhard Beilharz <eb1@sil.org>
Date: Fri, 20 Dec 2024 18:00:25 +0100
Subject: [PATCH] fix(core): work around alignment problem when using
 Emscripten

The data we read from the KMX file is not in places 32-bit aligned, e.g.
`dpGroupArray` will often start at a 16-bit boundary. Emscripten doesn't
like this and will abort the program if `SAFE_HEAP=1` is defined.

This change works around the problem by calling `CopyKeyboard` and thus
expanding the structure.

See #12844.
---
 core/src/kmx/kmx_file.cpp       | 18 ++++++++++--------
 core/src/kmx/kmx_processevent.h |  2 +-
 2 files changed, 11 insertions(+), 9 deletions(-)

diff --git a/core/src/kmx/kmx_file.cpp b/core/src/kmx/kmx_file.cpp
index 2eea0185813..9a4100c754d 100644
--- a/core/src/kmx/kmx_file.cpp
+++ b/core/src/kmx/kmx_file.cpp
@@ -68,11 +68,11 @@ KMX_BOOL KMX_ProcessEvent::LoadKeyboardFromBlob(
 
   *lpKeyboard = NULL;
 
-#ifdef KMX_64BIT
+#if defined(KMX_64BIT) || defined(__EMSCRIPTEN__)
   // allocate enough memory for expanded data structure + original data.
-  // Expanded data structure is double the size of data on disk (8-byte
-  // pointers) - on disk the "pointers" are relative to the beginning of
-  // the file.
+  // Expanded data structure is no more than double the size of data on
+  // disk (8-byte pointers) - on disk the "pointers" are relative to the
+  // beginning of the file.
   // We save the original data at the end of buf; we don't copy strings, so
   // those will remain in the location at the end of the buffer.
   buf = new KMX_BYTE[sz * 3];
@@ -84,7 +84,7 @@ KMX_BOOL KMX_ProcessEvent::LoadKeyboardFromBlob(
     DebugLog("Not allocmem");
     return FALSE;
   }
-#ifdef KMX_64BIT
+#if defined(KMX_64BIT) || defined(__EMSCRIPTEN__)
   filebase = buf + sz * 2;
 #else
   filebase = buf;
@@ -103,7 +103,7 @@ KMX_BOOL KMX_ProcessEvent::LoadKeyboardFromBlob(
     return FALSE;
   }
 
-#ifdef KMX_64BIT
+#if defined(KMX_64BIT) || defined(__EMSCRIPTEN__)
   kbp = CopyKeyboard(buf, filebase);
 #else
   kbp = FixupKeyboard(buf, filebase);
@@ -132,13 +132,15 @@ PKMX_WCHAR KMX_ProcessEvent::StringOffset(PKMX_BYTE base, KMX_DWORD offset)
   return (PKMX_WCHAR)(base + offset);
 }
 
-#ifdef KMX_64BIT
+#if defined(KMX_64BIT) || defined(__EMSCRIPTEN__)
 /**
   CopyKeyboard will copy the data read into bufp from x86-sized structures into
   x64-sized structures starting at `base`
   * After this function finishes, we still need to keep the original data because
     we don't copy the strings
-  This method is used on 64-bit architectures.
+  This method is used on 64-bit architectures as well as with Emscripten.
+  Emscripten requires 32-bit alignment for pointers which we don't always
+  have in the KMX data (see #12844).
 */
 LPKEYBOARD KMX_ProcessEvent::CopyKeyboard(PKMX_BYTE bufp, PKMX_BYTE base)
 {
diff --git a/core/src/kmx/kmx_processevent.h b/core/src/kmx/kmx_processevent.h
index 090488dccb4..202541e4550 100644
--- a/core/src/kmx/kmx_processevent.h
+++ b/core/src/kmx/kmx_processevent.h
@@ -57,7 +57,7 @@ class KMX_ProcessEvent {
   KMX_BOOL LoadKeyboardFromBlob(PKMX_BYTE buf, size_t sz, LPKEYBOARD* lpKeyboard);
   KMX_BOOL VerifyKeyboard(PKMX_BYTE filebase, size_t sz);
   KMX_BOOL VerifyChecksum(PKMX_BYTE buf,  size_t sz);
-#ifdef KMX_64BIT
+#if defined(KMX_64BIT) || defined(__EMSCRIPTEN__)
   LPKEYBOARD CopyKeyboard(PKMX_BYTE bufp, PKMX_BYTE base);
 #else
   LPKEYBOARD FixupKeyboard(PKMX_BYTE bufp, PKMX_BYTE base);