merge

7 files changed, 265 insertions, 144 deletions

diff --git a/indra/llrender/llfontgl.cpp b/indra/llrender/llfontgl.cpp
index 6c5468edb6..7facfbaf4e 100644
--- a/indra/llrender/llfontgl.cpp
+++ b/indra/llrender/llfontgl.cpp
@@ -277,7 +277,6 @@ S32 LLFontGL::render(const LLWString &wstr, S32 begin_offset, F32 x, F32 y, cons
 		}
 	}
 
-
 	const LLFontGlyphInfo* next_glyph = NULL;
 
 	const S32 GLYPH_BATCH_SIZE = 30;
diff --git a/indra/llrender/llglslshader.cpp b/indra/llrender/llglslshader.cpp
index ca92cb6580..f59bac8c3c 100644
--- a/indra/llrender/llglslshader.cpp
+++ b/indra/llrender/llglslshader.cpp
@@ -61,7 +61,7 @@ BOOL shouldChange(const LLVector4& v1, const LLVector4& v2)
 
 LLShaderFeatures::LLShaderFeatures()
 : calculatesLighting(false), isShiny(false), isFullbright(false), hasWaterFog(false),
-hasTransport(false), hasSkinning(false), hasAtmospherics(false), isSpecular(false),
+hasTransport(false), hasSkinning(false), hasObjectSkinning(false), hasAtmospherics(false), isSpecular(false),
 hasGamma(false), hasLighting(false), calculatesAtmospherics(false)
 {
 }
@@ -124,7 +124,7 @@ BOOL LLGLSLShader::createShader(vector<string> * attributes,
 	{
 		GLhandleARB shaderhandle = LLShaderMgr::instance()->loadShaderFile((*fileIter).first, mShaderLevel, (*fileIter).second);
 		LL_DEBUGS("ShaderLoading") << "SHADER FILE: " << (*fileIter).first << " mShaderLevel=" << mShaderLevel << LL_ENDL;
-		if (mShaderLevel > 0)
+		if (shaderhandle > 0)
 		{
 			attachObject(shaderhandle);
 		}
@@ -709,7 +709,15 @@ GLint LLGLSLShader::getUniformLocation(const string& uniform)
 		std::map<string, GLint>::iterator iter = mUniformMap.find(uniform);
 		if (iter != mUniformMap.end())
 		{
-			llassert(iter->second == glGetUniformLocationARB(mProgramObject, uniform.c_str()));
+			if (gDebugGL)
+			{
+				stop_glerror();
+				if (iter->second != glGetUniformLocationARB(mProgramObject, uniform.c_str()))
+				{
+					llerrs << "Uniform does not match." << llendl;
+				}
+				stop_glerror();
+			}
 			return iter->second;
 		}
 	}
@@ -717,6 +725,18 @@ GLint LLGLSLShader::getUniformLocation(const string& uniform)
 	return -1;
 }
 
+GLint LLGLSLShader::getAttribLocation(U32 attrib)
+{
+	if (attrib < mAttribute.size())
+	{
+		return mAttribute[attrib];
+	}
+	else
+	{
+		return -1;
+	}
+}
+
 void LLGLSLShader::uniform1i(const string& uniform, GLint v)
 {
 	GLint location = getUniformLocation(uniform);
@@ -888,7 +908,9 @@ void LLGLSLShader::uniformMatrix4fv(const string& uniform, U32 count, GLboolean
 				
 	if (location >= 0)
 	{
+		stop_glerror();
 		glUniformMatrix4fvARB(location, count, transpose, v);
+		stop_glerror();
 	}
 }
 
diff --git a/indra/llrender/llglslshader.h b/indra/llrender/llglslshader.h
index 166d4af04c..dc493ba162 100644
--- a/indra/llrender/llglslshader.h
+++ b/indra/llrender/llglslshader.h
@@ -48,6 +48,7 @@ public:
 	bool hasWaterFog; // implies no gamma
 	bool hasTransport; // implies no lighting (it's possible to have neither though)
 	bool hasSkinning;	
+	bool hasObjectSkinning;
 	bool hasAtmospherics;
 	bool hasGamma;
 
@@ -109,7 +110,7 @@ public:
 	void vertexAttrib4fv(U32 index, GLfloat* v);
 	
 	GLint getUniformLocation(const std::string& uniform);
-	
+	GLint getAttribLocation(U32 attrib);
 	GLint mapUniformTextureChannel(GLint location, GLenum type);
 	
 
diff --git a/indra/llrender/llimagegl.cpp b/indra/llrender/llimagegl.cpp
index ab9abf88e8..5e69e9e045 100644
--- a/indra/llrender/llimagegl.cpp
+++ b/indra/llrender/llimagegl.cpp
@@ -1707,6 +1707,7 @@ void LLImageGL::analyzeAlpha(const void* data_in, U32 w, U32 h)
 				sample[asum/(16*4)] += 4;
 			}
 			
+			
 			rowstart += 2 * w * mAlphaStride;
 		}
 		length *= 2; // we sampled everything twice, essentially
diff --git a/indra/llrender/llshadermgr.cpp b/indra/llrender/llshadermgr.cpp
index 1286e91e49..9eb5668d88 100644
--- a/indra/llrender/llshadermgr.cpp
+++ b/indra/llrender/llshadermgr.cpp
@@ -152,6 +152,14 @@ BOOL LLShaderMgr::attachShaderFeatures(LLGLSLShader * shader)
 			return FALSE;
 		}
 	}
+
+	if (features->hasObjectSkinning)
+	{
+		if (!shader->attachObject("avatar/objectSkinV.glsl"))
+		{
+			return FALSE;
+		}
+	}
 	
 	///////////////////////////////////////
 	// Attach Fragment Shader Features Next
@@ -226,7 +234,14 @@ BOOL LLShaderMgr::attachShaderFeatures(LLGLSLShader * shader)
 	else if (features->isFullbright)
 	{
 	
-		if (features->hasWaterFog)
+		if (features->isShiny && features->hasWaterFog)
+		{
+			if (!shader->attachObject("lighting/lightFullbrightShinyWaterF.glsl"))
+			{
+				return FALSE;
+			}
+		}
+		else if (features->hasWaterFog)
 		{
 			if (!shader->attachObject("lighting/lightFullbrightWaterF.glsl"))
 			{
@@ -313,11 +328,14 @@ void LLShaderMgr::dumpObjectLog(GLhandleARB ret, BOOL warns)
 
 GLhandleARB LLShaderMgr::loadShaderFile(const std::string& filename, S32 & shader_level, GLenum type)
 {
-	GLenum error;
-	error = glGetError();
-	if (error != GL_NO_ERROR)
+	GLenum error = GL_NO_ERROR;
+	if (gDebugGL)
 	{
-		LL_WARNS("ShaderLoading") << "GL ERROR entering loadShaderFile(): " << error << LL_ENDL;
+		error = glGetError();
+		if (error != GL_NO_ERROR)
+		{
+			LL_WARNS("ShaderLoading") << "GL ERROR entering loadShaderFile(): " << error << LL_ENDL;
+		}
 	}
 	
 	LL_DEBUGS("ShaderLoading") << "Loading shader file: " << filename << " class " << shader_level << LL_ENDL;
@@ -372,31 +390,39 @@ GLhandleARB LLShaderMgr::loadShaderFile(const std::string& filename, S32 & shade
 
 	//create shader object
 	GLhandleARB ret = glCreateShaderObjectARB(type);
-	error = glGetError();
-	if (error != GL_NO_ERROR)
+	if (gDebugGL)
 	{
-		LL_WARNS("ShaderLoading") << "GL ERROR in glCreateShaderObjectARB: " << error << LL_ENDL;
+		error = glGetError();
+		if (error != GL_NO_ERROR)
+		{
+			LL_WARNS("ShaderLoading") << "GL ERROR in glCreateShaderObjectARB: " << error << LL_ENDL;
+		}
 	}
-	else
+	
+	//load source
+	glShaderSourceARB(ret, count, (const GLcharARB**) text, NULL);
+
+	if (gDebugGL)
 	{
-		//load source
-		glShaderSourceARB(ret, count, (const GLcharARB**) text, NULL);
 		error = glGetError();
 		if (error != GL_NO_ERROR)
 		{
 			LL_WARNS("ShaderLoading") << "GL ERROR in glShaderSourceARB: " << error << LL_ENDL;
 		}
-		else
+	}
+
+	//compile source
+	glCompileShaderARB(ret);
+
+	if (gDebugGL)
+	{
+		error = glGetError();
+		if (error != GL_NO_ERROR)
 		{
-			//compile source
-			glCompileShaderARB(ret);
-			error = glGetError();
-			if (error != GL_NO_ERROR)
-			{
-				LL_WARNS("ShaderLoading") << "GL ERROR in glCompileShaderARB: " << error << LL_ENDL;
-			}
+			LL_WARNS("ShaderLoading") << "GL ERROR in glCompileShaderARB: " << error << LL_ENDL;
 		}
 	}
+		
 	//free memory
 	for (GLuint i = 0; i < count; i++)
 	{
@@ -407,13 +433,16 @@ GLhandleARB LLShaderMgr::loadShaderFile(const std::string& filename, S32 & shade
 		//check for errors
 		GLint success = GL_TRUE;
 		glGetObjectParameterivARB(ret, GL_OBJECT_COMPILE_STATUS_ARB, &success);
-		error = glGetError();
-		if (error != GL_NO_ERROR || success == GL_FALSE) 
+		if (gDebugGL || success == GL_FALSE)
 		{
-			//an error occured, print log
-			LL_WARNS("ShaderLoading") << "GLSL Compilation Error: (" << error << ") in " << filename << LL_ENDL;
-			dumpObjectLog(ret);
-			ret = 0;
+			error = glGetError();
+			if (error != GL_NO_ERROR || success == GL_FALSE) 
+			{
+				//an error occured, print log
+				LL_WARNS("ShaderLoading") << "GLSL Compilation Error: (" << error << ") in " << filename << LL_ENDL;
+				dumpObjectLog(ret);
+				ret = 0;
+			}
 		}
 	}
 	else
diff --git a/indra/llrender/llvertexbuffer.cpp b/indra/llrender/llvertexbuffer.cpp
index 8012bbb6c8..71620feed3 100644
--- a/indra/llrender/llvertexbuffer.cpp
+++ b/indra/llrender/llvertexbuffer.cpp
@@ -31,6 +31,7 @@
  */
 
 #include "linden_common.h"
+#include "llmemory.h"
 
 #include <boost/static_assert.hpp>
 
@@ -39,6 +40,7 @@
 #include "llglheaders.h"
 #include "llmemtype.h"
 #include "llrender.h"
+#include "llvector4a.h"
 
 //============================================================================
 
@@ -62,20 +64,23 @@ BOOL LLVertexBuffer::sIBOActive = FALSE;
 U32 LLVertexBuffer::sAllocatedBytes = 0;
 BOOL LLVertexBuffer::sMapped = FALSE;
 BOOL LLVertexBuffer::sUseStreamDraw = TRUE;
+S32	LLVertexBuffer::sWeight4Loc = -1;
 
 std::vector<U32> LLVertexBuffer::sDeleteList;
 
-S32 LLVertexBuffer::sTypeOffsets[LLVertexBuffer::TYPE_MAX] =
+
+S32 LLVertexBuffer::sTypeSize[LLVertexBuffer::TYPE_MAX] =
 {
-	sizeof(LLVector3), // TYPE_VERTEX,
-	sizeof(LLVector3), // TYPE_NORMAL,
+	sizeof(LLVector4), // TYPE_VERTEX,
+	sizeof(LLVector4), // TYPE_NORMAL,
 	sizeof(LLVector2), // TYPE_TEXCOORD0,
 	sizeof(LLVector2), // TYPE_TEXCOORD1,
 	sizeof(LLVector2), // TYPE_TEXCOORD2,
 	sizeof(LLVector2), // TYPE_TEXCOORD3,
 	sizeof(LLColor4U), // TYPE_COLOR,
-	sizeof(LLVector3), // TYPE_BINORMAL,
+	sizeof(LLVector4), // TYPE_BINORMAL,
 	sizeof(F32),	   // TYPE_WEIGHT,
+	sizeof(LLVector4), // TYPE_WEIGHT4,
 	sizeof(LLVector4), // TYPE_CLOTHWEIGHT,
 };
 
@@ -210,18 +215,37 @@ void LLVertexBuffer::setupClientArrays(U32 data_mask)
 			glClientActiveTextureARB(GL_TEXTURE0_ARB);
 		}
 	
+		if (sLastMask & MAP_WEIGHT4)
+		{
+			if (sWeight4Loc < 0)
+			{
+				llerrs << "Weighting disabled but vertex buffer still bound!" << llendl;
+			}
+
+			if (!(data_mask & MAP_WEIGHT4))
+			{ //disable 4-component skin weight			
+				glDisableVertexAttribArrayARB(sWeight4Loc);
+			}
+		}
+		else if (data_mask & MAP_WEIGHT4)
+		{
+			if (sWeight4Loc >= 0)
+			{ //enable 4-component skin weight
+				glEnableVertexAttribArrayARB(sWeight4Loc);
+			}
+		}
+				
+
 		sLastMask = data_mask;
 	}
 }
 
-void LLVertexBuffer::drawRange(U32 mode, U32 start, U32 end, U32 count, U32 indices_offset) const
+void LLVertexBuffer::validateRange(U32 start, U32 end, U32 count, U32 indices_offset) const
 {
-	llassert(mRequestedNumVerts >= 0);
-
 	if (start >= (U32) mRequestedNumVerts ||
 	    end >= (U32) mRequestedNumVerts)
 	{
-		llerrs << "Bad vertex buffer draw range: [" << start << ", " << end << "]" << llendl;
+		llerrs << "Bad vertex buffer draw range: [" << start << ", " << end << "] vs " << mRequestedNumVerts << llendl;
 	}
 
 	llassert(mRequestedNumIndices >= 0);
@@ -232,6 +256,25 @@ void LLVertexBuffer::drawRange(U32 mode, U32 start, U32 end, U32 count, U32 indi
 		llerrs << "Bad index buffer draw range: [" << indices_offset << ", " << indices_offset+count << "]" << llendl;
 	}
 
+	if (gDebugGL && !useVBOs())
+	{
+		U16* idx = ((U16*) getIndicesPointer())+indices_offset;
+		for (U32 i = 0; i < count; ++i)
+		{
+			if (idx[i] < start || idx[i] > end)
+			{
+				llerrs << "Index out of range: " << idx[i] << " not in [" << start << ", " << end << "]" << llendl;
+			}
+		}
+	}
+}
+
+void LLVertexBuffer::drawRange(U32 mode, U32 start, U32 end, U32 count, U32 indices_offset) const
+{
+	validateRange(start, end, count, indices_offset);
+
+	llassert(mRequestedNumVerts >= 0);
+
 	if (mGLIndices != sGLRenderIndices)
 	{
 		llerrs << "Wrong index buffer bound." << llendl;
@@ -248,9 +291,11 @@ void LLVertexBuffer::drawRange(U32 mode, U32 start, U32 end, U32 count, U32 indi
 		return;
 	}
 
+	U16* idx = ((U16*) getIndicesPointer())+indices_offset;
+
 	stop_glerror();
 	glDrawRangeElements(sGLMode[mode], start, end, count, GL_UNSIGNED_SHORT, 
-		((U16*) getIndicesPointer()) + indices_offset);
+		idx);
 	stop_glerror();
 }
 
@@ -393,17 +438,24 @@ LLVertexBuffer::LLVertexBuffer(U32 typemask, S32 usage) :
 		mUsage = 0;
 	}
 
-	S32 stride = calcStride(typemask, mOffsets);
+	//zero out offsets
+	for (U32 i = 0; i < TYPE_MAX; i++)
+	{
+		mOffsets[i] = 0;
+	}
 
 	mTypeMask = typemask;
-	mStride = stride;
+	mSize = 0;
+	mAlignedOffset = 0;
+	mAlignedIndexOffset = 0;
+
 	sCount++;
 }
 
 //static
-S32 LLVertexBuffer::calcStride(const U32& typemask, S32* offsets)
+S32 LLVertexBuffer::calcOffsets(const U32& typemask, S32* offsets, S32 num_vertices)
 {
-	S32 stride = 0;
+	S32 offset = 0;
 	for (S32 i=0; i<TYPE_MAX; i++)
 	{
 		U32 mask = 1<<i;
@@ -411,13 +463,35 @@ S32 LLVertexBuffer::calcStride(const U32& typemask, S32* offsets)
 		{
 			if (offsets)
 			{
-				offsets[i] = stride;
+				offsets[i] = offset;
+				offset += LLVertexBuffer::sTypeSize[i]*num_vertices;
+				offset = (offset + 0xF) & ~0xF;
 			}
-			stride += sTypeOffsets[i];
 		}
 	}
 
-	return stride;
+	return offset+16;
+}
+
+//static 
+S32 LLVertexBuffer::calcVertexSize(const U32& typemask)
+{
+	S32 size = 0;
+	for (S32 i = 0; i < TYPE_MAX; i++)
+	{
+		U32 mask = 1<<i;
+		if (typemask & mask)
+		{
+			size += LLVertexBuffer::sTypeSize[i];
+		}
+	}
+
+	return size;
+}
+
+S32 LLVertexBuffer::getSize() const
+{
+	return mSize;
 }
 
 // protected, use unref()
@@ -529,8 +603,7 @@ void LLVertexBuffer::createGLBuffer()
 	{
 		static int gl_buffer_idx = 0;
 		mGLBuffer = ++gl_buffer_idx;
-		mMappedData = new U8[size];
-		memset(mMappedData, 0, size);
+		mMappedData = (U8*) ll_aligned_malloc_16(size);
 	}
 }
 
@@ -551,16 +624,20 @@ void LLVertexBuffer::createGLIndices()
 
 	mEmpty = TRUE;
 
+	//pad by 16 bytes for aligned copies
+	size += 16;
+
 	if (useVBOs())
 	{
+		//pad by another 16 bytes for VBO pointer adjustment
+		size += 16;
 		mMappedIndexData = NULL;
 		genIndices();
 		mResized = TRUE;
 	}
 	else
 	{
-		mMappedIndexData = new U8[size];
-		memset(mMappedIndexData, 0, size);
+		mMappedIndexData = (U8*) ll_aligned_malloc_16(size);
 		static int gl_buffer_idx = 0;
 		mGLIndices = ++gl_buffer_idx;
 	}
@@ -581,7 +658,7 @@ void LLVertexBuffer::destroyGLBuffer()
 		}
 		else
 		{
-			delete [] mMappedData;
+			ll_aligned_free_16(mMappedData);
 			mMappedData = NULL;
 			mEmpty = TRUE;
 		}
@@ -590,7 +667,7 @@ void LLVertexBuffer::destroyGLBuffer()
 	}
 	
 	mGLBuffer = 0;
-	unbind();
+	//unbind();
 }
 
 void LLVertexBuffer::destroyGLIndices()
@@ -608,7 +685,7 @@ void LLVertexBuffer::destroyGLIndices()
 		}
 		else
 		{
-			delete [] mMappedIndexData;
+			ll_aligned_free_16(mMappedIndexData);
 			mMappedIndexData = NULL;
 			mEmpty = TRUE;
 		}
@@ -617,7 +694,7 @@ void LLVertexBuffer::destroyGLIndices()
 	}
 
 	mGLIndices = 0;
-	unbind();
+	//unbind();
 }
 
 void LLVertexBuffer::updateNumVerts(S32 nverts)
@@ -633,7 +710,7 @@ void LLVertexBuffer::updateNumVerts(S32 nverts)
 	}
 
 	mRequestedNumVerts = nverts;
-	
+
 	if (!mDynamicSize)
 	{
 		mNumVerts = nverts;
@@ -648,7 +725,7 @@ void LLVertexBuffer::updateNumVerts(S32 nverts)
 		}
 		mNumVerts = nverts;
 	}
-
+	mSize = calcOffsets(mTypeMask, mOffsets, mNumVerts);
 }
 
 void LLVertexBuffer::updateNumIndices(S32 nindices)
@@ -679,6 +756,12 @@ void LLVertexBuffer::allocateBuffer(S32 nverts, S32 nindices, bool create)
 {
 	LLMemType mt2(LLMemType::MTYPE_VERTEX_ALLOCATE_BUFFER);
 		
+	if (nverts < 0 || nindices < 0 ||
+		nverts > 65536)
+	{
+		llerrs << "Bad vertex buffer allocation: " << nverts << " : " << nindices << llendl;
+	}
+
 	updateNumVerts(nverts);
 	updateNumIndices(nindices);
 	
@@ -717,9 +800,6 @@ void LLVertexBuffer::resizeBuffer(S32 newnverts, S32 newnindices)
 	{
 		sAllocatedBytes -= getSize() + getIndicesSize();
 		
-		S32 oldsize = getSize();
-		S32 old_index_size = getIndicesSize();
-
 		updateNumVerts(newnverts);		
 		updateNumIndices(newnindices);
 		
@@ -736,26 +816,10 @@ void LLVertexBuffer::resizeBuffer(S32 newnverts, S32 newnindices)
 			}
 			else
 			{
-				//delete old buffer, keep GL buffer for now
 				if (!useVBOs())
 				{
-					U8* old = mMappedData;
-					mMappedData = new U8[newsize];
-					if (old)
-					{	
-						memcpy(mMappedData, old, llmin(newsize, oldsize));
-						if (newsize > oldsize)
-						{
-							memset(mMappedData+oldsize, 0, newsize-oldsize);
-						}
-
-						delete [] old;
-					}
-					else
-					{
-						memset(mMappedData, 0, newsize);
-						mEmpty = TRUE;
-					}
+					ll_aligned_free_16(mMappedData);
+					mMappedData = (U8*) ll_aligned_malloc_16(newsize);
 				}
 				mResized = TRUE;
 			}
@@ -775,24 +839,8 @@ void LLVertexBuffer::resizeBuffer(S32 newnverts, S32 newnindices)
 			{
 				if (!useVBOs())
 				{
-					//delete old buffer, keep GL buffer for now
-					U8* old = mMappedIndexData;
-					mMappedIndexData = new U8[new_index_size];
-					
-					if (old)
-					{	
-						memcpy(mMappedIndexData, old, llmin(new_index_size, old_index_size));
-						if (new_index_size > old_index_size)
-						{
-							memset(mMappedIndexData+old_index_size, 0, new_index_size - old_index_size);
-						}
-						delete [] old;
-					}
-					else
-					{
-						memset(mMappedIndexData, 0, new_index_size);
-						mEmpty = TRUE;
-					}
+					ll_aligned_free_16(mMappedIndexData);
+					mMappedIndexData = (U8*) ll_aligned_malloc_16(new_index_size);
 				}
 				mResized = TRUE;
 			}
@@ -849,12 +897,19 @@ U8* LLVertexBuffer::mapBuffer(S32 access)
 			setBuffer(0);
 			mLocked = TRUE;
 			stop_glerror();	
-			mMappedData = (U8*) glMapBufferARB(GL_ARRAY_BUFFER_ARB, GL_WRITE_ONLY_ARB);
+
+			U8* src = (U8*) glMapBufferARB(GL_ARRAY_BUFFER_ARB, GL_WRITE_ONLY_ARB);
+			mMappedData = LL_NEXT_ALIGNED_ADDRESS<U8>(src);
+			mAlignedOffset = mMappedData - src;
+			
 			stop_glerror();
 		}
 		{
 			LLMemType mt_v(LLMemType::MTYPE_VERTEX_MAP_BUFFER_INDICES);
-			mMappedIndexData = (U8*) glMapBufferARB(GL_ELEMENT_ARRAY_BUFFER_ARB, GL_WRITE_ONLY_ARB);
+			U8* src = (U8*) glMapBufferARB(GL_ELEMENT_ARRAY_BUFFER_ARB, GL_WRITE_ONLY_ARB);
+			mMappedIndexData = LL_NEXT_ALIGNED_ADDRESS<U8>(src);
+			mAlignedIndexOffset = mMappedIndexData - src;
+
 			stop_glerror();
 		}
 
@@ -960,8 +1015,8 @@ template <class T,S32 type> struct VertexBufferStrider
 		}
 		else if (vbo.hasDataType(type))
 		{
-			S32 stride = vbo.getStride();
-			strider = (T*)(vbo.getMappedData() + vbo.getOffset(type) + index*stride);
+			S32 stride = LLVertexBuffer::sTypeSize[type];
+			strider = (T*)(vbo.getMappedData() + vbo.getOffset(type)+index*stride);
 			strider.setStride(stride);
 			return TRUE;
 		}
@@ -973,7 +1028,6 @@ template <class T,S32 type> struct VertexBufferStrider
 	}
 };
 
-
 bool LLVertexBuffer::getVertexStrider(LLStrider<LLVector3>& strider, S32 index)
 {
 	return VertexBufferStrider<LLVector3,TYPE_VERTEX>::get(*this, strider, index);
@@ -1014,28 +1068,15 @@ bool LLVertexBuffer::getWeightStrider(LLStrider<F32>& strider, S32 index)
 {
 	return VertexBufferStrider<F32,TYPE_WEIGHT>::get(*this, strider, index);
 }
-bool LLVertexBuffer::getClothWeightStrider(LLStrider<LLVector4>& strider, S32 index)
+
+bool LLVertexBuffer::getWeight4Strider(LLStrider<LLVector4>& strider, S32 index)
 {
-	return VertexBufferStrider<LLVector4,TYPE_CLOTHWEIGHT>::get(*this, strider, index);
+	return VertexBufferStrider<LLVector4,TYPE_WEIGHT4>::get(*this, strider, index);
 }
 
-void LLVertexBuffer::setStride(S32 type, S32 new_stride)
+bool LLVertexBuffer::getClothWeightStrider(LLStrider<LLVector4>& strider, S32 index)
 {
-	LLMemType mt2(LLMemType::MTYPE_VERTEX_SET_STRIDE);
-	if (mNumVerts)
-	{
-		llerrs << "LLVertexBuffer::setOffset called with mNumVerts = " << mNumVerts << llendl;
-	}
-	// This code assumes that setStride() will only be called once per VBO per type.
-	S32 delta = new_stride - sTypeOffsets[type];
-	for (S32 i=type+1; i<TYPE_MAX; i++)
-	{
-		if (mTypeMask & (1<<i))
-		{
-			mOffsets[i] += delta;
-		}
-	}
-	mStride += delta;
+	return VertexBufferStrider<LLVector4,TYPE_CLOTHWEIGHT>::get(*this, strider, index);
 }
 
 //----------------------------------------------------------------------------
@@ -1234,8 +1275,7 @@ void LLVertexBuffer::setupVertexBuffer(U32 data_mask) const
 {
 	LLMemType mt2(LLMemType::MTYPE_VERTEX_SETUP_VERTEX_BUFFER);
 	stop_glerror();
-	U8* base = useVBOs() ? NULL : mMappedData;
-	S32 stride = mStride;
+	U8* base = useVBOs() ? (U8*) mAlignedOffset : mMappedData;
 
 	if ((data_mask & mTypeMask) != data_mask)
 	{
@@ -1244,52 +1284,58 @@ void LLVertexBuffer::setupVertexBuffer(U32 data_mask) const
 
 	if (data_mask & MAP_NORMAL)
 	{
-		glNormalPointer(GL_FLOAT, stride, (void*)(base + mOffsets[TYPE_NORMAL]));
+		glNormalPointer(GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_NORMAL], (void*)(base + mOffsets[TYPE_NORMAL]));
 	}
 	if (data_mask & MAP_TEXCOORD3)
 	{
 		glClientActiveTextureARB(GL_TEXTURE3_ARB);
-		glTexCoordPointer(2,GL_FLOAT, stride, (void*)(base + mOffsets[TYPE_TEXCOORD3]));
+		glTexCoordPointer(2,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD3], (void*)(base + mOffsets[TYPE_TEXCOORD3]));
 		glClientActiveTextureARB(GL_TEXTURE0_ARB);
 	}
 	if (data_mask & MAP_TEXCOORD2)
 	{
 		glClientActiveTextureARB(GL_TEXTURE2_ARB);
-		glTexCoordPointer(2,GL_FLOAT, stride, (void*)(base + mOffsets[TYPE_TEXCOORD2]));
+		glTexCoordPointer(2,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD2], (void*)(base + mOffsets[TYPE_TEXCOORD2]));
 		glClientActiveTextureARB(GL_TEXTURE0_ARB);
 	}
 	if (data_mask & MAP_TEXCOORD1)
 	{
 		glClientActiveTextureARB(GL_TEXTURE1_ARB);
-		glTexCoordPointer(2,GL_FLOAT, stride, (void*)(base + mOffsets[TYPE_TEXCOORD1]));
+		glTexCoordPointer(2,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD1], (void*)(base + mOffsets[TYPE_TEXCOORD1]));
 		glClientActiveTextureARB(GL_TEXTURE0_ARB);
 	}
 	if (data_mask & MAP_BINORMAL)
 	{
 		glClientActiveTextureARB(GL_TEXTURE2_ARB);
-		glTexCoordPointer(3,GL_FLOAT, stride, (void*)(base + mOffsets[TYPE_BINORMAL]));
+		glTexCoordPointer(3,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_BINORMAL], (void*)(base + mOffsets[TYPE_BINORMAL]));
 		glClientActiveTextureARB(GL_TEXTURE0_ARB);
 	}
 	if (data_mask & MAP_TEXCOORD0)
 	{
-		glTexCoordPointer(2,GL_FLOAT, stride, (void*)(base + mOffsets[TYPE_TEXCOORD0]));
+		glTexCoordPointer(2,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_TEXCOORD0], (void*)(base + mOffsets[TYPE_TEXCOORD0]));
 	}
 	if (data_mask & MAP_COLOR)
 	{
-		glColorPointer(4, GL_UNSIGNED_BYTE, stride, (void*)(base + mOffsets[TYPE_COLOR]));
+		glColorPointer(4, GL_UNSIGNED_BYTE, LLVertexBuffer::sTypeSize[TYPE_COLOR], (void*)(base + mOffsets[TYPE_COLOR]));
 	}
 	
 	if (data_mask & MAP_WEIGHT)
 	{
-		glVertexAttribPointerARB(1, 1, GL_FLOAT, FALSE, stride, (void*)(base + mOffsets[TYPE_WEIGHT]));
+		glVertexAttribPointerARB(1, 1, GL_FLOAT, FALSE, LLVertexBuffer::sTypeSize[TYPE_WEIGHT], (void*)(base + mOffsets[TYPE_WEIGHT]));
 	}
+
+	if (data_mask & MAP_WEIGHT4 && sWeight4Loc != -1)
+	{
+		glVertexAttribPointerARB(sWeight4Loc, 4, GL_FLOAT, FALSE, LLVertexBuffer::sTypeSize[TYPE_WEIGHT4], (void*)(base+mOffsets[TYPE_WEIGHT4]));
+	}
+
 	if (data_mask & MAP_CLOTHWEIGHT)
 	{
-		glVertexAttribPointerARB(4, 4, GL_FLOAT, TRUE,  stride, (void*)(base + mOffsets[TYPE_CLOTHWEIGHT]));
+		glVertexAttribPointerARB(4, 4, GL_FLOAT, TRUE,  LLVertexBuffer::sTypeSize[TYPE_CLOTHWEIGHT], (void*)(base + mOffsets[TYPE_CLOTHWEIGHT]));
 	}
 	if (data_mask & MAP_VERTEX)
 	{
-		glVertexPointer(3,GL_FLOAT, stride, (void*)(base + 0));
+		glVertexPointer(3,GL_FLOAT, LLVertexBuffer::sTypeSize[TYPE_VERTEX], (void*)(base + 0));
 	}
 
 	llglassertok();
diff --git a/indra/llrender/llvertexbuffer.h b/indra/llrender/llvertexbuffer.h
index ef0bdb21b4..a6173636bf 100644
--- a/indra/llrender/llvertexbuffer.h
+++ b/indra/llrender/llvertexbuffer.h
@@ -74,16 +74,29 @@ protected:
 
 
 //============================================================================
-// base class
+// base class 
 
 class LLVertexBuffer : public LLRefCount
 {
 public:
+	LLVertexBuffer(const LLVertexBuffer& rhs)
+	{
+		*this = rhs;
+	}
+
+	const LLVertexBuffer& operator=(const LLVertexBuffer& rhs)
+	{
+		llerrs << "Illegal operation!" << llendl;
+		return *this;
+	}
+
 	static LLVBOPool sStreamVBOPool;
 	static LLVBOPool sDynamicVBOPool;
 	static LLVBOPool sStreamIBOPool;
 	static LLVBOPool sDynamicIBOPool;
 
+	static S32	sWeight4Loc;
+
 	static BOOL	sUseStreamDraw;
 
 	static void initClass(bool use_vbo);
@@ -93,11 +106,14 @@ public:
 	static void unbind(); //unbind any bound vertex buffer
 
 	//get the size of a vertex with the given typemask
-	//if offsets is not NULL, its contents will be filled
-	//with the offset of each vertex component in the buffer, 
+	static S32 calcVertexSize(const U32& typemask);
+
+	//get the size of a buffer with the given typemask and vertex count
+	//fill offsets with the offset of each vertex component array into the buffer
 	// indexed by the following enum
-	static S32 calcStride(const U32& typemask, S32* offsets = NULL); 										
+	static S32 calcOffsets(const U32& typemask, S32* offsets, S32 num_vertices);		
 
+	
 	enum {
 		TYPE_VERTEX,
 		TYPE_NORMAL,
@@ -109,6 +125,7 @@ public:
 		// These use VertexAttribPointer and should possibly be made generic
 		TYPE_BINORMAL,
 		TYPE_WEIGHT,
+		TYPE_WEIGHT4,
 		TYPE_CLOTHWEIGHT,
 		TYPE_MAX,
 		TYPE_INDEX,
@@ -124,6 +141,7 @@ public:
 		// These use VertexAttribPointer and should possibly be made generic
 		MAP_BINORMAL = (1<<TYPE_BINORMAL),
 		MAP_WEIGHT = (1<<TYPE_WEIGHT),
+		MAP_WEIGHT4 = (1<<TYPE_WEIGHT4),
 		MAP_CLOTHWEIGHT = (1<<TYPE_CLOTHWEIGHT),
 	};
 	
@@ -173,6 +191,7 @@ public:
 	bool getBinormalStrider(LLStrider<LLVector3>& strider, S32 index=0);
 	bool getColorStrider(LLStrider<LLColor4U>& strider, S32 index=0);
 	bool getWeightStrider(LLStrider<F32>& strider, S32 index=0);
+	bool getWeight4Strider(LLStrider<LLVector4>& strider, S32 index=0);
 	bool getClothWeightStrider(LLStrider<LLVector4>& strider, S32 index=0);
 	
 	BOOL isEmpty() const					{ return mEmpty; }
@@ -182,33 +201,37 @@ public:
 	S32 getRequestedVerts() const			{ return mRequestedNumVerts; }
 	S32 getRequestedIndices() const			{ return mRequestedNumIndices; }
 
-	U8* getIndicesPointer() const			{ return useVBOs() ? NULL : mMappedIndexData; }
-	U8* getVerticesPointer() const			{ return useVBOs() ? NULL : mMappedData; }
-	S32 getStride() const					{ return mStride; }
-	S32 getTypeMask() const					{ return mTypeMask; }
+	U8* getIndicesPointer() const			{ return useVBOs() ? (U8*) mAlignedIndexOffset : mMappedIndexData; }
+	U8* getVerticesPointer() const			{ return useVBOs() ? (U8*) mAlignedOffset : mMappedData; }
+	U32 getTypeMask() const					{ return mTypeMask; }
 	BOOL hasDataType(S32 type) const		{ return ((1 << type) & getTypeMask()) ? TRUE : FALSE; }
-	S32 getSize() const						{ return mNumVerts*mStride; }
+	S32 getSize() const;
 	S32 getIndicesSize() const				{ return mNumIndices * sizeof(U16); }
 	U8* getMappedData() const				{ return mMappedData; }
 	U8* getMappedIndices() const			{ return mMappedIndexData; }
 	S32 getOffset(S32 type) const			{ return mOffsets[type]; }
 	S32 getUsage() const					{ return mUsage; }
 
-	void setStride(S32 type, S32 new_stride);
-	
 	void markDirty(U32 vert_index, U32 vert_count, U32 indices_index, U32 indices_count);
 
 	void draw(U32 mode, U32 count, U32 indices_offset) const;
 	void drawArrays(U32 mode, U32 offset, U32 count) const;
 	void drawRange(U32 mode, U32 start, U32 end, U32 count, U32 indices_offset) const;
 
+	//for debugging, validate data in given range is valid
+	void validateRange(U32 start, U32 end, U32 count, U32 offset) const;
+
+	
+
 protected:	
 	S32		mNumVerts;		// Number of vertices allocated
 	S32		mNumIndices;	// Number of indices allocated
 	S32		mRequestedNumVerts;  // Number of vertices requested
 	S32		mRequestedNumIndices;  // Number of indices requested
 
-	S32		mStride;
+	ptrdiff_t mAlignedOffset;
+	ptrdiff_t mAlignedIndexOffset;
+	S32		mSize;
 	U32		mTypeMask;
 	S32		mUsage;			// GL usage
 	U32		mGLBuffer;		// GL VBO handle
@@ -222,7 +245,7 @@ protected:
 	S32		mOffsets[TYPE_MAX];
 	BOOL	mResized;		// if TRUE, client buffer has been resized and GL buffer has not
 	BOOL	mDynamicSize;	// if TRUE, buffer has been resized at least once (and should be padded)
-
+	
 	class DirtyRegion
 	{
 	public:
@@ -247,7 +270,7 @@ public:
 	typedef std::list<LLVertexBuffer*> buffer_list_t;
 		
 	static BOOL sEnableVBOs;
-	static S32 sTypeOffsets[TYPE_MAX];
+	static S32 sTypeSize[TYPE_MAX];
 	static U32 sGLMode[LLRender::NUM_MODES];
 	static U32 sGLRenderBuffer;
 	static U32 sGLRenderIndices;