Bug 710727. Share cmaps across all fonts. r=jkew, a=tracking-firefox

gfx/thebes/gfxDWriteFontList.cpp

@@ -375,84 +375,77 @@ gfxDWriteFontEntry::ReadCMAP()
     nsresult rv;
 
     // attempt this once, if errors occur leave a blank cmap
-    if (mCmapInitialized)
+    if (mCharacterMap) {
         return NS_OK;
-    mCmapInitialized = true;
+    }
+
+    nsRefPtr<gfxCharacterMap> charmap = new gfxCharacterMap();
 
     // if loading via GDI, just use GetFontTable
     if (mFont && gfxDWriteFontList::PlatformFontList()->UseGDIFontTableAccess()) {
-        const PRUint32 kCmapTag = TRUETYPE_TAG('c','m','a','p');
-        AutoFallibleTArray<PRUint8,16384> buffer;
+        PRUint32 kCMAP = TRUETYPE_TAG('c','m','a','p');
     
-        if (GetFontTable(kCmapTag, buffer) != NS_OK)
-            return NS_ERROR_FAILURE;
-        PRUint8 *cmap = buffer.Elements();
+        AutoFallibleTArray<PRUint8,16384> cmap;
+        rv = GetFontTable(kCMAP, cmap);
     
-        bool          unicodeFont = false, symbolFont = false;
-        rv = gfxFontUtils::ReadCMAP(cmap, buffer.Length(),
-                                    mCharacterMap, mUVSOffset,
-                                    unicodeFont, symbolFont);
-#ifdef PR_LOGGING
-        LOG_FONTLIST(("(fontlist-cmap) name: %s, size: %d\n",
-                      NS_ConvertUTF16toUTF8(mName).get(),
-                      mCharacterMap.SizeOfExcludingThis(moz_malloc_size_of)));
-        if (LOG_CMAPDATA_ENABLED()) {
-            char prefix[256];
-            sprintf(prefix, "(cmapdata) name: %.220s",
-                    NS_ConvertUTF16toUTF8(mName).get());
-            mCharacterMap.Dump(prefix, eGfxLog_cmapdata);
+        bool unicodeFont = false, symbolFont = false; // currently ignored
+    
+        if (NS_SUCCEEDED(rv)) {
+            rv = gfxFontUtils::ReadCMAP(cmap.Elements(), cmap.Length(),
+                                        *charmap, mUVSOffset,
+                                        unicodeFont, symbolFont);
+        }
+    } else {
+        // loading using dwrite, don't use GetFontTable to avoid copy
+        nsRefPtr<IDWriteFontFace> fontFace;
+        rv = CreateFontFace(getter_AddRefs(fontFace));
+    
+        if (NS_SUCCEEDED(rv)) {
+            const PRUint32 kCmapTag = DWRITE_MAKE_OPENTYPE_TAG('c', 'm', 'a', 'p');
+            PRUint8 *tableData;
+            PRUint32 len;
+            void *tableContext = NULL;
+            BOOL exists;
+            hr = fontFace->TryGetFontTable(kCmapTag, (const void**)&tableData,
+                                           &len, &tableContext, &exists);
+
+            if (SUCCEEDED(hr)) {
+                bool isSymbol = fontFace->IsSymbolFont();
+                bool isUnicode = true;
+                if (exists) {
+                    rv = gfxFontUtils::ReadCMAP(tableData, len, *charmap,
+                                                mUVSOffset, isUnicode, 
+                                                isSymbol);
+                }
+                fontFace->ReleaseFontTable(tableContext);
+            } else {
+                rv = NS_ERROR_FAILURE;
+            }
         }
-#endif
-        mHasCmapTable = NS_SUCCEEDED(rv);
-        return rv;
     }
 
-    // loading using dwrite, don't use GetFontTable to avoid copy
-    nsRefPtr<IDWriteFontFace> fontFace;
-    rv = CreateFontFace(getter_AddRefs(fontFace));
-
-    if (NS_FAILED(rv)) {
-        return rv;
+    mHasCmapTable = NS_SUCCEEDED(rv);
+    if (mHasCmapTable) {
+        gfxPlatformFontList *pfl = gfxPlatformFontList::PlatformFontList();
+        mCharacterMap = pfl->FindCharMap(charmap);
+    } else {
+        // if error occurred, initialize to null cmap
+        mCharacterMap = new gfxCharacterMap();
     }
 
-    PRUint8 *tableData;
-    PRUint32 len;
-    void *tableContext = NULL;
-    BOOL exists;
-    hr = fontFace->TryGetFontTable(DWRITE_MAKE_OPENTYPE_TAG('c', 'm', 'a', 'p'),
-                                   (const void**)&tableData,
-                                   &len,
-                                   &tableContext,
-                                   &exists);
-    if (FAILED(hr)) {
-        return NS_ERROR_FAILURE;
-    }
-
-    bool isSymbol = fontFace->IsSymbolFont();
-    bool isUnicode = true;
-    if (exists) {
-        rv = gfxFontUtils::ReadCMAP(tableData,
-                                    len,
-                                    mCharacterMap,
-                                    mUVSOffset,
-                                    isUnicode,
-                                    isSymbol);
-    }
-    fontFace->ReleaseFontTable(tableContext);
-
 #ifdef PR_LOGGING
-    LOG_FONTLIST(("(fontlist-cmap) name: %s, size: %d\n",
+    LOG_FONTLIST(("(fontlist-cmap) name: %s, size: %d hash: %8.8x%s\n",
                   NS_ConvertUTF16toUTF8(mName).get(),
-                  mCharacterMap.SizeOfExcludingThis(moz_malloc_size_of)));
+                  charmap->SizeOfIncludingThis(moz_malloc_size_of),
+                  charmap->mHash, mCharacterMap == charmap ? " new" : ""));
     if (LOG_CMAPDATA_ENABLED()) {
         char prefix[256];
         sprintf(prefix, "(cmapdata) name: %.220s",
                 NS_ConvertUTF16toUTF8(mName).get());
-        mCharacterMap.Dump(prefix, eGfxLog_cmapdata);
+        charmap->Dump(prefix, eGfxLog_cmapdata);
     }
 #endif
 
-    mHasCmapTable = NS_SUCCEEDED(rv);
     return rv;
 }
 

gfx/thebes/gfxFT2FontList.cpp

@@ -343,12 +343,11 @@ FT2FontEntry::CairoFontFace()
 nsresult
 FT2FontEntry::ReadCMAP()
 {
-    if (mCmapInitialized) {
+    if (mCharacterMap) {
         return NS_OK;
     }
 
-    // attempt this once, if errors occur leave a blank cmap
-    mCmapInitialized = true;
+    nsRefPtr<gfxCharacterMap> charmap = new gfxCharacterMap();
 
     AutoFallibleTArray<PRUint8,16384> buffer;
     nsresult rv = GetFontTable(TTAG_cmap, buffer);
@@ -357,11 +356,18 @@ FT2FontEntry::ReadCMAP()
         bool unicodeFont;
         bool symbolFont;
         rv = gfxFontUtils::ReadCMAP(buffer.Elements(), buffer.Length(),
-                                    mCharacterMap, mUVSOffset,
+                                    *charmap, mUVSOffset,
                                     unicodeFont, symbolFont);
     }
 
     mHasCmapTable = NS_SUCCEEDED(rv);
+    if (mHasCmapTable) {
+        gfxPlatformFontList *pfl = gfxPlatformFontList::PlatformFontList();
+        mCharacterMap = pfl->FindCharMap(charmap);
+    } else {
+        // if error occurred, initialize to null cmap
+        mCharacterMap = new gfxCharacterMap();
+    }
     return rv;
 }
 

gfx/thebes/gfxFont.cpp

@@ -103,6 +103,16 @@ static PRUint32 gGlyphExtentsSetupLazyTight = 0;
 static PRUint32 gGlyphExtentsSetupFallBackToTight = 0;
 #endif
 
+void
+gfxCharacterMap::NotifyReleased()
+{
+    gfxPlatformFontList *fontlist = gfxPlatformFontList::PlatformFontList();
+    if (mShared) {
+        fontlist->RemoveCmap(this);
+    }
+    delete this;
+}
+
 gfxFontEntry::~gfxFontEntry() 
 {
     delete mUserFontData;
@@ -115,18 +125,20 @@ bool gfxFontEntry::IsSymbolFont()
 
 bool gfxFontEntry::TestCharacterMap(PRUint32 aCh)
 {
-    if (!mCmapInitialized) {
+    if (!mCharacterMap) {
         ReadCMAP();
+        NS_ASSERTION(mCharacterMap, "failed to initialize character map");
     }
-    return mCharacterMap.test(aCh);
+    return mCharacterMap->test(aCh);
 }
 
 nsresult gfxFontEntry::InitializeUVSMap()
 {
     // mUVSOffset will not be initialized
     // until cmap is initialized.
-    if (!mCmapInitialized) {
+    if (!mCharacterMap) {
         ReadCMAP();
+        NS_ASSERTION(mCharacterMap, "failed to initialize character map");
     }
 
     if (!mUVSOffset) {
@@ -170,7 +182,8 @@ PRUint16 gfxFontEntry::GetUVSGlyph(PRUint32 aCh, PRUint32 aVS)
 
 nsresult gfxFontEntry::ReadCMAP()
 {
-    mCmapInitialized = true;
+    NS_ASSERTION(false, "using default no-op implementation of ReadCMAP");
+    mCharacterMap = new gfxCharacterMap();
     return NS_OK;
 }
 
@@ -433,7 +446,12 @@ gfxFontEntry::SizeOfExcludingThis(nsMallocSizeOfFun aMallocSizeOf,
                                   FontListSizes*    aSizes) const
 {
     aSizes->mFontListSize += mName.SizeOfExcludingThisIfUnshared(aMallocSizeOf);
-    aSizes->mCharMapsSize += mCharacterMap.SizeOfExcludingThis(aMallocSizeOf);
+
+    // cmaps are shared so only non-shared cmaps are included here
+    if (mCharacterMap && mCharacterMap->mBuildOnTheFly) {
+        aSizes->mCharMapsSize +=
+            mCharacterMap->SizeOfIncludingThis(aMallocSizeOf);
+    }
     aSizes->mFontTableCacheSize +=
         mFontTableCache.SizeOfExcludingThis(
             FontTableHashEntry::SizeOfEntryExcludingThis,
@@ -770,7 +788,7 @@ CalcStyleMatch(gfxFontEntry *aFontEntry, const gfxFontStyle *aStyle)
 void
 gfxFontFamily::FindFontForChar(GlobalFontMatch *aMatchData)
 {
-    if (mCharacterMapInitialized && !TestCharacterMap(aMatchData->mCh)) {
+    if (mFamilyCharacterMapInitialized && !TestCharacterMap(aMatchData->mCh)) {
         // none of the faces in the family support the required char,
         // so bail out immediately
         return;
@@ -1028,7 +1046,8 @@ gfxFontFamily::SizeOfExcludingThis(nsMallocSizeOfFun aMallocSizeOf,
 {
     aSizes->mFontListSize +=
         mName.SizeOfExcludingThisIfUnshared(aMallocSizeOf);
-    aSizes->mCharMapsSize += mCharacterMap.SizeOfExcludingThis(aMallocSizeOf);
+    aSizes->mCharMapsSize +=
+        mFamilyCharacterMap.SizeOfExcludingThis(aMallocSizeOf);
 
     aSizes->mFontListSize +=
         mAvailableFonts.SizeOfExcludingThis(aMallocSizeOf);
@@ -3649,7 +3668,9 @@ gfxFontGroup::FindFontForChar(PRUint32 aCh, PRUint32 aPrevCh,
 
         // check other faces of the family
         gfxFontFamily *family = font->GetFontEntry()->Family();
-        if (family && family->TestCharacterMap(aCh)) {
+        if (family && !font->GetFontEntry()->mIsProxy &&
+            family->TestCharacterMap(aCh))
+        {
             GlobalFontMatch matchData(aCh, aRunScript, &mStyle);
             family->SearchAllFontsForChar(&matchData);
             gfxFontEntry *fe = matchData.mBestMatch;

gfx/thebes/gfxFont.h

@@ -198,6 +198,56 @@ struct THEBES_API gfxFontStyle {
     static PRUint32 ParseFontLanguageOverride(const nsString& aLangTag);
 };
 
+class gfxCharacterMap : public gfxSparseBitSet {
+public:
+    nsrefcnt AddRef() {
+        NS_PRECONDITION(PRInt32(mRefCnt) >= 0, "illegal refcnt");
+        ++mRefCnt;
+        NS_LOG_ADDREF(this, mRefCnt, "gfxCharacterMap", sizeof(*this));
+        return mRefCnt;
+    }
+
+    nsrefcnt Release() {
+        NS_PRECONDITION(0 != mRefCnt, "dup release");
+        --mRefCnt;
+        NS_LOG_RELEASE(this, mRefCnt, "gfxCharacterMap");
+        if (mRefCnt == 0) {
+            NotifyReleased();
+            // |this| has been deleted.
+            return 0;
+        }
+        return mRefCnt;
+    }
+
+    gfxCharacterMap() :
+        mHash(0), mBuildOnTheFly(false), mShared(false)
+    { }
+
+    void CalcHash() { mHash = GetChecksum(); }
+
+    size_t SizeOfExcludingThis(nsMallocSizeOfFun aMallocSizeOf) const {
+        return gfxSparseBitSet::SizeOfExcludingThis(aMallocSizeOf);
+    }
+
+    // hash of the cmap bitvector
+    PRUint32 mHash;
+
+    // if cmap is built on the fly it's never shared
+    bool mBuildOnTheFly;
+
+    // cmap is shared globally
+    bool mShared;
+
+protected:
+    void NotifyReleased();
+
+    nsAutoRefCnt mRefCnt;
+
+private:
+    gfxCharacterMap(const gfxCharacterMap&);
+    gfxCharacterMap& operator=(const gfxCharacterMap&);
+};
+
 class gfxFontEntry {
 public:
     NS_INLINE_DECL_REFCOUNTING(gfxFontEntry)
@@ -216,7 +266,6 @@ public:
         mCheckedForGraphiteTables(false),
 #endif
         mHasCmapTable(false),
-        mCmapInitialized(false),
         mUVSOffset(0), mUVSData(nsnull),
         mUserFontData(nsnull),
         mLanguageOverride(NO_FONT_LANGUAGE_OVERRIDE),
@@ -259,16 +308,17 @@ public:
 #endif
 
     inline bool HasCmapTable() {
-        if (!mCmapInitialized) {
+        if (!mCharacterMap) {
             ReadCMAP();
+            NS_ASSERTION(mCharacterMap, "failed to initialize character map");
         }
         return mHasCmapTable;
     }
 
     inline bool HasCharacter(PRUint32 ch) {
-        if (mCharacterMap.test(ch))
+        if (mCharacterMap && mCharacterMap->test(ch)) {
             return true;
-
+        }
         return TestCharacterMap(ch);
     }
 
@@ -344,8 +394,7 @@ public:
     bool             mCheckedForGraphiteTables;
 #endif
     bool             mHasCmapTable;
-    bool             mCmapInitialized;
-    gfxSparseBitSet  mCharacterMap;
+    nsRefPtr<gfxCharacterMap> mCharacterMap;
     PRUint32         mUVSOffset;
     nsAutoArrayPtr<PRUint8> mUVSData;
     gfxUserFontData* mUserFontData;
@@ -375,7 +424,6 @@ protected:
         mCheckedForGraphiteTables(false),
 #endif
         mHasCmapTable(false),
-        mCmapInitialized(false),
         mUVSOffset(0), mUVSData(nsnull),
         mUserFontData(nsnull),
         mLanguageOverride(NO_FONT_LANGUAGE_OVERRIDE),
@@ -529,7 +577,7 @@ public:
         mHasStyles(false),
         mIsSimpleFamily(false),
         mIsBadUnderlineFamily(false),
-        mCharacterMapInitialized(false)
+        mFamilyCharacterMapInitialized(false)
         { }
 
     virtual ~gfxFontFamily() {
@@ -607,26 +655,27 @@ public:
         PRUint32 i, numFonts = mAvailableFonts.Length();
         for (i = 0; i < numFonts; i++) {
             gfxFontEntry *fe = mAvailableFonts[i];
-            if (!fe) {
+            // don't try to load cmaps for downloadable fonts not yet loaded
+            if (!fe || fe->mIsProxy) {
                 continue;
             }
             fe->ReadCMAP();
-            mCharacterMap.Union(fe->mCharacterMap);
+            mFamilyCharacterMap.Union(*(fe->mCharacterMap));
         }
-        mCharacterMap.Compact();
-        mCharacterMapInitialized = true;
+        mFamilyCharacterMap.Compact();
+        mFamilyCharacterMapInitialized = true;
     }
 
     bool TestCharacterMap(PRUint32 aCh) {
-        if (!mCharacterMapInitialized) {
+        if (!mFamilyCharacterMapInitialized) {
             ReadAllCMAPs();
         }
-        return mCharacterMap.test(aCh);
+        return mFamilyCharacterMap.test(aCh);
     }
 
     void ResetCharacterMap() {
-        mCharacterMap.reset();
-        mCharacterMapInitialized = false;
+        mFamilyCharacterMap.reset();
+        mFamilyCharacterMapInitialized = false;
     }
 
     // mark this family as being in the "bad" underline offset blacklist
@@ -675,14 +724,14 @@ protected:
 
     nsString mName;
     nsTArray<nsRefPtr<gfxFontEntry> >  mAvailableFonts;
-    gfxSparseBitSet mCharacterMap;
-    bool mOtherFamilyNamesInitialized;
-    bool mHasOtherFamilyNames;
-    bool mFaceNamesInitialized;
-    bool mHasStyles;
-    bool mIsSimpleFamily;
-    bool mIsBadUnderlineFamily;
-    bool mCharacterMapInitialized;
+    gfxSparseBitSet mFamilyCharacterMap;
+    bool mOtherFamilyNamesInitialized : 1;
+    bool mHasOtherFamilyNames : 1;
+    bool mFaceNamesInitialized : 1;
+    bool mHasStyles : 1;
+    bool mIsSimpleFamily : 1;
+    bool mIsBadUnderlineFamily : 1;
+    bool mFamilyCharacterMapInitialized : 1;
 
     enum {
         // for "simple" families, the faces are stored in mAvailableFonts

gfx/thebes/gfxFontUtils.h

@@ -58,6 +58,8 @@
 #include "nsAutoPtr.h"
 #include "nsIStreamBufferAccess.h"
 
+#include "zlib.h"
+
 /* Bug 341128 - w32api defines min/max which causes problems with <bitset> */
 #ifdef __MINGW32__
 #undef min
@@ -87,6 +89,28 @@ public:
                 mBlocks[i] = new Block(*block);
         }
     }
+
+    bool Equals(const gfxSparseBitSet *aOther) const {
+        if (mBlocks.Length() != aOther->mBlocks.Length()) {
+            return false;
+        }
+        size_t n = mBlocks.Length();
+        for (size_t i = 0; i < n; ++i) {
+            const Block *b1 = mBlocks[i];
+            const Block *b2 = aOther->mBlocks[i];
+            if (!b1 != !b2) {
+                return false;
+            }
+            if (!b1) {
+                continue;
+            }
+            if (memcmp(&b1->mBits, &b2->mBits, BLOCK_SIZE) != 0) {
+                return false;
+            }
+        }
+        return true;
+    }
+
     bool test(PRUint32 aIndex) const {
         NS_ASSERTION(mBlocks.DebugGetHeader(), "mHdr is null, this is bad");
         PRUint32 blockIndex = aIndex/BLOCK_SIZE_BITS;
@@ -321,6 +345,18 @@ public:
         mBlocks.Compact();
     }
 
+    PRUint32 GetChecksum() const {
+        PRUint32 check = adler32(0, Z_NULL, 0);
+        for (PRUint32 i = 0; i < mBlocks.Length(); i++) {
+            if (mBlocks[i]) {
+                const Block *block = mBlocks[i];
+                check = adler32(check, (PRUint8*) (&i), 4);
+                check = adler32(check, (PRUint8*) block, sizeof(Block));
+            }
+        }
+        return check;
+    }
+
 private:
     nsTArray< nsAutoPtr<Block> > mBlocks;
 };

gfx/thebes/gfxGDIFontList.cpp

@@ -201,7 +201,7 @@ GDIFontEntry::GDIFontEntry(const nsAString& aFaceName,
     : gfxFontEntry(aFaceName),
       mWindowsFamily(0), mWindowsPitch(0),
       mFontType(aFontType),
-      mForceGDI(false), mUnknownCMAP(false),
+      mForceGDI(false),
       mCharset(), mUnicodeRanges()
 {
     mUserFontData = aUserFontData;
@@ -218,43 +218,62 @@ GDIFontEntry::GDIFontEntry(const nsAString& aFaceName,
 nsresult
 GDIFontEntry::ReadCMAP()
 {
+    // attempt this once, if errors occur leave a blank cmap
+    if (mCharacterMap) {
+        return NS_OK;
+    }
+
     // skip non-SFNT fonts completely
     if (mFontType != GFX_FONT_TYPE_PS_OPENTYPE && 
         mFontType != GFX_FONT_TYPE_TT_OPENTYPE &&
         mFontType != GFX_FONT_TYPE_TRUETYPE) 
     {
+        mCharacterMap = new gfxCharacterMap();
         return NS_ERROR_FAILURE;
     }
 
-    // attempt this once, if errors occur leave a blank cmap
-    if (mCmapInitialized)
-        return NS_OK;
-    mCmapInitialized = true;
+    nsRefPtr<gfxCharacterMap> charmap = new gfxCharacterMap();
 
-    const PRUint32 kCmapTag = TRUETYPE_TAG('c','m','a','p');
-    AutoFallibleTArray<PRUint8,16384> buffer;
-    if (GetFontTable(kCmapTag, buffer) != NS_OK)
-        return NS_ERROR_FAILURE;
-    PRUint8 *cmap = buffer.Elements();
+    PRUint32 kCMAP = TRUETYPE_TAG('c','m','a','p');
+    nsresult rv;
 
-    bool          unicodeFont = false, symbolFont = false;
-    nsresult rv = gfxFontUtils::ReadCMAP(cmap, buffer.Length(),
-                                         mCharacterMap, mUVSOffset,
-                                         unicodeFont, symbolFont);
+    AutoFallibleTArray<PRUint8,16384> cmap;
+    rv = GetFontTable(kCMAP, cmap);
+
+    bool unicodeFont = false, symbolFont = false; // currently ignored
+
+    if (NS_SUCCEEDED(rv)) {
+        rv = gfxFontUtils::ReadCMAP(cmap.Elements(), cmap.Length(),
+                                    *charmap, mUVSOffset,
+                                    unicodeFont, symbolFont);
+    }
     mSymbolFont = symbolFont;
+
     mHasCmapTable = NS_SUCCEEDED(rv);
+    if (mHasCmapTable) {
+        gfxPlatformFontList *pfl = gfxPlatformFontList::PlatformFontList();
+        mCharacterMap = pfl->FindCharMap(charmap);
+    } else {
+        // if error occurred, initialize to null cmap
+        mCharacterMap = new gfxCharacterMap();
+        // For fonts where we failed to read the character map,
+        // we can take a slow path to look up glyphs character by character
+        mCharacterMap->mBuildOnTheFly = true;
+    }
 
 #ifdef PR_LOGGING
-    LOG_FONTLIST(("(fontlist-cmap) name: %s, size: %d\n",
+    LOG_FONTLIST(("(fontlist-cmap) name: %s, size: %d hash: %8.8x%s\n",
                   NS_ConvertUTF16toUTF8(mName).get(),
-                  mCharacterMap.SizeOfExcludingThis(moz_malloc_size_of)));
+                  charmap->SizeOfIncludingThis(moz_malloc_size_of),
+                  charmap->mHash, mCharacterMap == charmap ? " new" : ""));
     if (LOG_CMAPDATA_ENABLED()) {
         char prefix[256];
         sprintf(prefix, "(cmapdata) name: %.220s",
                 NS_ConvertUTF16toUTF8(mName).get());
-        mCharacterMap.Dump(prefix, eGfxLog_cmapdata);
+        charmap->Dump(prefix, eGfxLog_cmapdata);
     }
 #endif
+
     return rv;
 }
 
@@ -346,13 +365,12 @@ GDIFontEntry::FillLogFont(LOGFONTW *aLogFont,
 bool 
 GDIFontEntry::TestCharacterMap(PRUint32 aCh)
 {
-    if (ReadCMAP() != NS_OK) {
-        // For fonts where we failed to read the character map,
-        // we can take a slow path to look up glyphs character by character
-        mUnknownCMAP = true;
+    if (!mCharacterMap) {
+        nsresult rv = ReadCMAP();
+        NS_ASSERTION(mCharacterMap, "failed to initialize a character map");
     }
 
-    if (mUnknownCMAP) {
+    if (mCharacterMap->mBuildOnTheFly) {
         if (aCh > 0xFFFF)
             return false;
 
@@ -404,12 +422,12 @@ GDIFontEntry::TestCharacterMap(PRUint32 aCh)
         ReleaseDC(NULL, dc);
 
         if (hasGlyph) {
-            mCharacterMap.set(aCh);
+            mCharacterMap->set(aCh);
             return true;
         }
     } else {
         // font had a cmap so simply check that
-        return mCharacterMap.test(aCh);
+        return mCharacterMap->test(aCh);
     }
 
     return false;

gfx/thebes/gfxGDIFontList.h

@@ -293,7 +293,6 @@ public:
 
     gfxWindowsFontType mFontType;
     bool mForceGDI    : 1;
-    bool mUnknownCMAP : 1;
 
     gfxSparseBitSet mCharset;
     gfxSparseBitSet mUnicodeRanges;

gfx/thebes/gfxMacPlatformFontList.mm

@@ -189,93 +189,98 @@ nsresult
 MacOSFontEntry::ReadCMAP()
 {
     // attempt this once, if errors occur leave a blank cmap
-    if (mCmapInitialized) {
+    if (mCharacterMap) {
         return NS_OK;
     }
-    mCmapInitialized = true;
+
+    nsRefPtr<gfxCharacterMap> charmap = new gfxCharacterMap();
 
     PRUint32 kCMAP = TRUETYPE_TAG('c','m','a','p');
+    nsresult rv;
 
     AutoFallibleTArray<PRUint8,16384> cmap;
-    if (GetFontTable(kCMAP, cmap) != NS_OK) {
-        return NS_ERROR_FAILURE;
+    rv = GetFontTable(kCMAP, cmap);
+
+    bool unicodeFont = false, symbolFont = false; // currently ignored
+
+    if (NS_SUCCEEDED(rv)) {
+        rv = gfxFontUtils::ReadCMAP(cmap.Elements(), cmap.Length(),
+                                    *charmap, mUVSOffset,
+                                    unicodeFont, symbolFont);
     }
   
-    bool          unicodeFont, symbolFont; // currently ignored
-    nsresult rv = gfxFontUtils::ReadCMAP(cmap.Elements(), cmap.Length(),
-                                         mCharacterMap, mUVSOffset,
-                                         unicodeFont, symbolFont);
-    if (NS_FAILED(rv)) {
-        mCharacterMap.reset();
-        return rv;
-    }
-    mHasCmapTable = true;
+    if (NS_SUCCEEDED(rv)) {
+        // for layout support, check for the presence of mort/morx and/or
+        // opentype layout tables
+        bool hasAATLayout = HasFontTable(TRUETYPE_TAG('m','o','r','x')) ||
+                              HasFontTable(TRUETYPE_TAG('m','o','r','t'));
+        bool hasGSUB = HasFontTable(TRUETYPE_TAG('G','S','U','B'));
+        bool hasGPOS = HasFontTable(TRUETYPE_TAG('G','P','O','S'));
     
-    CGFontRef fontRef = GetFontRef();
-    if (!fontRef) {
-        return NS_ERROR_FAILURE;
-    }
+        if (hasAATLayout && !(hasGSUB || hasGPOS)) {
+            mRequiresAAT = true; // prefer CoreText if font has no OTL tables
+        }
     
-    // for layout support, check for the presence of mort/morx and/or
-    // opentype layout tables
-    bool hasAATLayout = HasFontTable(TRUETYPE_TAG('m','o','r','x')) ||
-                          HasFontTable(TRUETYPE_TAG('m','o','r','t'));
-    bool hasGSUB = HasFontTable(TRUETYPE_TAG('G','S','U','B'));
-    bool hasGPOS = HasFontTable(TRUETYPE_TAG('G','P','O','S'));
+        PRUint32 numScripts =
+            sizeof(gScriptsThatRequireShaping) / sizeof(ScriptRange);
     
-    if (hasAATLayout && !(hasGSUB || hasGPOS)) {
-        mRequiresAAT = true; // prefer CoreText if font has no OTL tables
-    }
+        for (PRUint32 s = 0; s < numScripts; s++) {
+            eComplexScript  whichScript = gScriptsThatRequireShaping[s].script;
     
-    PRUint32 numScripts =
-        sizeof(gScriptsThatRequireShaping) / sizeof(ScriptRange);
+            // check to see if the cmap includes complex script codepoints
+            if (charmap->TestRange(gScriptsThatRequireShaping[s].rangeStart,
+                                   gScriptsThatRequireShaping[s].rangeEnd)) {
+                bool omitRange = true;
     
-    for (PRUint32 s = 0; s < numScripts; s++) {
-        eComplexScript  whichScript = gScriptsThatRequireShaping[s].script;
-
-        // check to see if the cmap includes complex script codepoints
-        if (mCharacterMap.TestRange(gScriptsThatRequireShaping[s].rangeStart,
-                                    gScriptsThatRequireShaping[s].rangeEnd)) {
-            bool omitRange = true;
-
-            if (hasAATLayout) {
-                omitRange = false;
-                // prefer CoreText for Apple's complex-script fonts,
-                // even if they also have some OpenType tables
-                // (e.g. Geeza Pro Bold on 10.6; see bug 614903)
-                mRequiresAAT = true;
-            } else if (whichScript == eComplexScriptArabic) {
-                // special-case for Arabic:
-                // even if there's no morph table, CoreText can shape Arabic
-                // using OpenType layout; or if it's a downloaded font,
-                // assume the site knows what it's doing (as harfbuzz will
-                // be able to shape even though the font itself lacks tables
-                // stripped during sanitization).
-                // We check for GSUB here, as GPOS alone would not be ok
-                // for Arabic shaping.
-                if (hasGSUB || (mIsUserFont && !mIsLocalUserFont)) {
-                    // TODO: to be really thorough, we could check that the
-                    // GSUB table actually supports the 'arab' script tag.
+                if (hasAATLayout) {
                     omitRange = false;
+                    // prefer CoreText for Apple's complex-script fonts,
+                    // even if they also have some OpenType tables
+                    // (e.g. Geeza Pro Bold on 10.6; see bug 614903)
+                    mRequiresAAT = true;
+                } else if (whichScript == eComplexScriptArabic) {
+                    // special-case for Arabic:
+                    // even if there's no morph table, CoreText can shape Arabic
+                    // using OpenType layout; or if it's a downloaded font,
+                    // assume the site knows what it's doing (as harfbuzz will
+                    // be able to shape even though the font itself lacks tables
+                    // stripped during sanitization).
+                    // We check for GSUB here, as GPOS alone would not be ok
+                    // for Arabic shaping.
+                    if (hasGSUB || (mIsUserFont && !mIsLocalUserFont)) {
+                        // TODO: to be really thorough, we could check that the
+                        // GSUB table actually supports the 'arab' script tag.
+                        omitRange = false;
+                    }
                 }
-            }
     
-            if (omitRange) {
-                mCharacterMap.ClearRange(gScriptsThatRequireShaping[s].rangeStart,
-                                         gScriptsThatRequireShaping[s].rangeEnd);
+                if (omitRange) {
+                    charmap->ClearRange(gScriptsThatRequireShaping[s].rangeStart,
+                                        gScriptsThatRequireShaping[s].rangeEnd);
+                }
             }
         }
     }
 
+    mHasCmapTable = NS_SUCCEEDED(rv);
+    if (mHasCmapTable) {
+        gfxPlatformFontList *pfl = gfxPlatformFontList::PlatformFontList();
+        mCharacterMap = pfl->FindCharMap(charmap);
+    } else {
+        // if error occurred, initialize to null cmap
+        mCharacterMap = new gfxCharacterMap();
+    }
+
 #ifdef PR_LOGGING
-    LOG_FONTLIST(("(fontlist-cmap) name: %s, size: %d\n",
+    LOG_FONTLIST(("(fontlist-cmap) name: %s, size: %d hash: %8.8x%s\n",
                   NS_ConvertUTF16toUTF8(mName).get(),
-                  mCharacterMap.SizeOfExcludingThis(moz_malloc_size_of)));
+                  charmap->SizeOfIncludingThis(moz_malloc_size_of),
+                  charmap->mHash, mCharacterMap == charmap ? " new" : ""));
     if (LOG_CMAPDATA_ENABLED()) {
         char prefix[256];
         sprintf(prefix, "(cmapdata) name: %.220s",
                 NS_ConvertUTF16toUTF8(mName).get());
-        mCharacterMap.Dump(prefix, eGfxLog_cmapdata);
+        charmap->Dump(prefix, eGfxLog_cmapdata);
     }
 #endif
 

gfx/thebes/gfxPlatformFontList.cpp

@@ -213,10 +213,13 @@ gfxPlatformFontList::gfxPlatformFontList(bool aNeedFullnamePostscriptNames)
     gFontListPrefObserver = new gfxFontListPrefObserver();
     NS_ADDREF(gFontListPrefObserver);
     Preferences::AddStrongObservers(gFontListPrefObserver, kObservedPrefs);
+
+    mSharedCmaps.Init(16);
 }
 
 gfxPlatformFontList::~gfxPlatformFontList()
 {
+    mSharedCmaps.Clear();
     NS_ASSERTION(gFontListPrefObserver, "There is no font list pref observer");
     Preferences::RemoveObservers(gFontListPrefObserver, kObservedPrefs);
     NS_RELEASE(gFontListPrefObserver);
@@ -714,6 +717,41 @@ gfxPlatformFontList::GetStandardFamilyName(const nsAString& aFontName, nsAString
     return !aFamilyName.IsEmpty();
 }
 
+gfxCharacterMap*
+gfxPlatformFontList::FindCharMap(gfxCharacterMap *aCmap)
+{
+    aCmap->CalcHash();
+    gfxCharacterMap *cmap = AddCmap(aCmap);
+    cmap->mShared = true;
+    return cmap;
+}
+
+// add a cmap to the shared cmap set
+gfxCharacterMap*
+gfxPlatformFontList::AddCmap(const gfxCharacterMap* aCharMap)
+{
+    CharMapHashKey *found =
+        mSharedCmaps.PutEntry(const_cast<gfxCharacterMap*>(aCharMap));
+    return found->GetKey();
+}
+
+// remove the cmap from the shared cmap set
+void
+gfxPlatformFontList::RemoveCmap(const gfxCharacterMap* aCharMap)
+{
+    // skip lookups during teardown
+    if (mSharedCmaps.Count() == 0) {
+        return;
+    }
+
+    // cmap needs to match the entry *and* be the same ptr before removing
+    CharMapHashKey *found =
+        mSharedCmaps.GetEntry(const_cast<gfxCharacterMap*>(aCharMap));
+    if (found && found->GetKey() == aCharMap) {
+        mSharedCmaps.RemoveEntry(const_cast<gfxCharacterMap*>(aCharMap));
+    }
+}
+
 void 
 gfxPlatformFontList::InitLoader()
 {
@@ -830,6 +868,21 @@ SizeOfStringEntryExcludingThis(nsStringHashKey*  aHashEntry,
     return aHashEntry->GetKey().SizeOfExcludingThisIfUnshared(aMallocSizeOf);
 }
 
+static size_t
+SizeOfSharedCmapExcludingThis(CharMapHashKey*   aHashEntry,
+                              nsMallocSizeOfFun aMallocSizeOf,
+                              void*             aUserArg)
+{
+    FontListSizes *sizes = static_cast<FontListSizes*>(aUserArg);
+
+    PRUint32 size = aHashEntry->GetKey()->SizeOfIncludingThis(aMallocSizeOf);
+    sizes->mCharMapsSize += size;
+
+    // we return zero here because the measurements have been added directly
+    // to the relevant fields of the FontListSizes record
+    return 0;
+}
+
 void
 gfxPlatformFontList::SizeOfExcludingThis(nsMallocSizeOfFun aMallocSizeOf,
                                          FontListSizes*    aSizes) const
@@ -865,6 +918,10 @@ gfxPlatformFontList::SizeOfExcludingThis(nsMallocSizeOfFun aMallocSizeOf,
     aSizes->mFontListSize +=
         mBadUnderlineFamilyNames.SizeOfExcludingThis(SizeOfStringEntryExcludingThis,
                                                      aMallocSizeOf);
+
+    aSizes->mFontListSize +=
+        mSharedCmaps.SizeOfExcludingThis(SizeOfSharedCmapExcludingThis,
+                                         aMallocSizeOf, aSizes);
 }
 
 void

gfx/thebes/gfxPlatformFontList.h

@@ -50,6 +50,53 @@
 #include "nsIMemoryReporter.h"
 #include "mozilla/FunctionTimer.h"
 
+class CharMapHashKey : public PLDHashEntryHdr
+{
+public:
+    typedef gfxCharacterMap* KeyType;
+    typedef const gfxCharacterMap* KeyTypePointer;
+
+    CharMapHashKey(const gfxCharacterMap *aCharMap) :
+        mCharMap(const_cast<gfxCharacterMap*>(aCharMap))
+    {
+        MOZ_COUNT_CTOR(CharMapHashKey);
+    }
+    CharMapHashKey(const CharMapHashKey& toCopy) :
+        mCharMap(toCopy.mCharMap)
+    {
+        MOZ_COUNT_CTOR(CharMapHashKey);
+    }
+    ~CharMapHashKey()
+    {
+        MOZ_COUNT_DTOR(CharMapHashKey);
+    }
+
+    gfxCharacterMap* GetKey() const { return mCharMap; }
+
+    bool KeyEquals(const gfxCharacterMap *aCharMap) const {
+        NS_ASSERTION(!aCharMap->mBuildOnTheFly && !mCharMap->mBuildOnTheFly,
+                     "custom cmap used in shared cmap hashtable");
+        // cmaps built on the fly never match
+        if (aCharMap->mHash != mCharMap->mHash)
+        {
+            return false;
+        }
+        return mCharMap->Equals(aCharMap);
+    }
+
+    static const gfxCharacterMap* KeyToPointer(gfxCharacterMap *aCharMap) {
+        return aCharMap;
+    }
+    static PLDHashNumber HashKey(const gfxCharacterMap *aCharMap) {
+        return aCharMap->mHash;
+    }
+
+    enum { ALLOW_MEMMOVE = true };
+
+protected:
+    gfxCharacterMap *mCharMap;
+};
+
 // gfxPlatformFontList is an abstract class for the global font list on the system;
 // concrete subclasses for each platform implement the actual interface to the system fonts.
 // This class exists because we cannot rely on the platform font-finding APIs to behave
@@ -155,6 +202,16 @@ public:
     virtual void SizeOfIncludingThis(nsMallocSizeOfFun aMallocSizeOf,
                                      FontListSizes*    aSizes) const;
 
+    // search for existing cmap that matches the input
+    // return the input if no match is found
+    gfxCharacterMap* FindCharMap(gfxCharacterMap *aCmap);
+
+    // add a cmap to the shared cmap set
+    gfxCharacterMap* AddCmap(const gfxCharacterMap *aCharMap);
+
+    // remove the cmap from the shared cmap set
+    void RemoveCmap(const gfxCharacterMap *aCharMap);
+
 protected:
     class MemoryReporter
         : public nsIMemoryMultiReporter
@@ -264,6 +321,10 @@ protected:
 
     nsTHashtable<nsStringHashKey> mBadUnderlineFamilyNames;
 
+    // character map data shared across families
+    // contains weak ptrs to cmaps shared by font entry objects
+    nsTHashtable<CharMapHashKey> mSharedCmaps;
+
     // data used as part of the font cmap loading process
     nsTArray<nsRefPtr<gfxFontFamily> > mFontFamiliesToLoad;
     PRUint32 mStartIndex;