source-engine/materialsystem/shaderapidx9/dynamicvb.h

1099 lines
31 KiB
C
Raw Normal View History

2020-04-22 16:56:21 +00:00
//========= Copyright Valve Corporation, All rights reserved. ============//
//
// Purpose:
//
// $NoKeywords: $
//
//===========================================================================//
#ifndef DYNAMICVB_H
#define DYNAMICVB_H
#ifdef _WIN32
#pragma once
#endif
#include "locald3dtypes.h"
#include "recording.h"
#include "shaderapidx8_global.h"
#include "shaderapidx8.h"
#include "imeshdx8.h"
#include "materialsystem/ivballoctracker.h"
#include "gpubufferallocator.h"
#include "tier1/utllinkedlist.h"
#include "tier0/dbg.h"
#include "tier1/memstack.h"
/////////////////////////////
// D. Sim Dietrich Jr.
// sim.dietrich@nvidia.com
//////////////////////
// Helper function to unbind an vertex buffer
void Unbind( IDirect3DVertexBuffer9 *pVertexBuffer );
#define X360_VERTEX_BUFFER_SIZE_MULTIPLIER 2.0 //minimum of 1, only affects dynamic buffers
//#define X360_BLOCK_ON_VB_FLUSH //uncomment to block until all data is consumed when a flush is requested. Otherwise we only block when absolutely necessary
//#define SPEW_VERTEX_BUFFER_STALLS //uncomment to allow buffer stall spewing.
class CVertexBuffer
{
public:
CVertexBuffer( IDirect3DDevice9 * pD3D, VertexFormat_t fmt, DWORD theFVF, int vertexSize,
int theVertexCount, const char *pTextureBudgetName, bool bSoftwareVertexProcessing, bool dynamic = false );
#ifdef _X360
CVertexBuffer();
void Init( IDirect3DDevice9 * pD3D, VertexFormat_t fmt, DWORD theFVF, uint8 *pVertexData, int vertexSize, int theVertexCount );
#endif
~CVertexBuffer();
LPDIRECT3DVERTEXBUFFER GetInterface() const
{
// If this buffer still exists, then Late Creation didn't happen. Best case: we'll render the wrong image. Worst case: Crash.
Assert( !m_pSysmemBuffer );
return m_pVB;
}
// Use at beginning of frame to force a flush of VB contents on first draw
void FlushAtFrameStart() { m_bFlush = true; }
// lock, unlock
unsigned char* Lock( int numVerts, int& baseVertexIndex );
unsigned char* Modify( bool bReadOnly, int firstVertex, int numVerts );
void Unlock( int numVerts );
void HandleLateCreation( );
// Vertex size
int VertexSize() const { return m_VertexSize; }
// Vertex count
int VertexCount() const { return m_VertexCount; }
#ifdef _X360
// For some VBs, memory allocation is managed by CGPUBufferAllocator, via ShaderAPI
const GPUBufferHandle_t *GetBufferAllocationHandle( void );
void SetBufferAllocationHandle( const GPUBufferHandle_t &bufferAllocationHandle );
bool IsPooled( void ) { creturn m_GPUBufferHandle.IsValid(); }
// Expose the data pointer for read-only CPU access to the data
// (double-indirection supports relocation of the data by CGPUBufferAllocator)
const byte **GetBufferDataPointerAddress( void );
#endif // _X360
static int BufferCount()
{
#ifdef _DEBUG
return s_BufferCount;
#else
return 0;
#endif
}
// UID
unsigned int UID() const
{
#ifdef RECORDING
return m_UID;
#else
return 0;
#endif
}
void HandlePerFrameTextureStats( int frame )
{
#ifdef VPROF_ENABLED
if ( m_Frame != frame && !m_bDynamic )
{
m_Frame = frame;
m_pFrameCounter += m_nBufferSize;
}
#endif
}
// Do we have enough room without discarding?
bool HasEnoughRoom( int numVertices ) const;
// Is this dynamic?
bool IsDynamic() const { return m_bDynamic; }
bool IsExternal() const { return m_bExternalMemory; }
// Block until this part of the vertex buffer is free
void BlockUntilUnused( int nBufferSize );
// used to alter the characteristics after creation
// allows one dynamic vb to be shared for multiple formats
void ChangeConfiguration( int vertexSize, int totalSize )
{
Assert( m_bDynamic && !m_bLocked && vertexSize );
m_VertexSize = vertexSize;
m_VertexCount = m_nBufferSize / vertexSize;
}
// Compute the next offset for the next lock
int NextLockOffset( ) const;
// Returns the allocated size
int AllocationSize() const;
// Returns the number of vertices we have enough room for
int NumVerticesUntilFlush() const
{
#if defined( _X360 )
if( m_AllocationRing.Count() )
{
//Cycle through the ring buffer and see what memory is free now
int iNode = m_AllocationRing.Head();
while( m_AllocationRing.IsValidIndex( iNode ) )
{
if( Dx9Device()->IsFencePending( m_AllocationRing[iNode].m_Fence ) )
break;
iNode = m_AllocationRing.Next( iNode );
}
if( m_AllocationRing.IsValidIndex( iNode ) )
{
int iEndFreeOffset = m_AllocationRing[iNode].m_iEndOffset;
if( iEndFreeOffset < m_Position )
{
//Wrapped. Making the arbitrary decision that the return value for this function *should* handle the singe giant allocation case which requires contiguous memory
if( iEndFreeOffset > (m_iNextBlockingPosition - m_Position) )
return iEndFreeOffset / m_VertexSize;
else
return (m_iNextBlockingPosition - m_Position) / m_VertexSize;
}
}
else
{
//we didn't block on any fence
return m_VertexCount;
}
}
return m_VertexCount;
#else
return (m_nBufferSize - NextLockOffset()) / m_VertexSize;
#endif
}
// Marks a fence indicating when this buffer was used
void MarkUsedInRendering()
{
#ifdef _X360
if ( m_bDynamic && m_pVB )
{
Assert( m_AllocationRing.Count() > 0 );
m_AllocationRing[m_AllocationRing.Tail()].m_Fence = Dx9Device()->GetCurrentFence();
}
#endif
}
private:
void Create( IDirect3DDevice9 *pD3D );
inline void ReallyUnlock( int unlockBytes )
{
#if DX_TO_GL_ABSTRACTION
// Knowing how much data was actually written is critical for performance under OpenGL.
m_pVB->UnlockActualSize( unlockBytes );
#else
unlockBytes; // Unused here
m_pVB->Unlock();
#endif
}
enum LOCK_FLAGS
{
LOCKFLAGS_FLUSH = D3DLOCK_NOSYSLOCK | D3DLOCK_DISCARD,
#if !defined( _X360 )
LOCKFLAGS_APPEND = D3DLOCK_NOSYSLOCK | D3DLOCK_NOOVERWRITE
#else
// X360BUG: forcing all locks to gpu flush, otherwise bizarre mesh corruption on decals
// Currently iterating with microsoft 360 support to track source of gpu corruption
LOCKFLAGS_APPEND = D3DLOCK_NOSYSLOCK
#endif
};
LPDIRECT3DVERTEXBUFFER m_pVB;
#ifdef _X360
struct DynamicBufferAllocation_t
{
DWORD m_Fence; //track whether this memory is safe to use again.
int m_iStartOffset;
int m_iEndOffset;
unsigned int m_iZPassIdx; // The zpass during which this allocation was made
};
int m_iNextBlockingPosition; // m_iNextBlockingPosition >= m_Position where another allocation is still in use.
unsigned char *m_pAllocatedMemory;
int m_iAllocationSize; //Total size of the ring buffer, usually more than what was asked for
IDirect3DVertexBuffer9 m_D3DVertexBuffer; //Only need one shared D3D header for our usage patterns.
CUtlLinkedList<DynamicBufferAllocation_t> m_AllocationRing; //tracks what chunks of our memory are potentially still in use by D3D
GPUBufferHandle_t m_GPUBufferHandle; // Handle to a memory allocation within a shared physical memory pool (see CGPUBufferAllocator)
#endif
VertexFormat_t m_VertexBufferFormat; // yes, Vertex, only used for allocation tracking
int m_nBufferSize;
int m_Position;
int m_VertexCount;
int m_VertexSize;
DWORD m_TheFVF;
byte *m_pSysmemBuffer;
int m_nSysmemBufferStartBytes;
uint m_nLockCount;
unsigned char m_bDynamic : 1;
unsigned char m_bLocked : 1;
unsigned char m_bFlush : 1;
unsigned char m_bExternalMemory : 1;
unsigned char m_bSoftwareVertexProcessing : 1;
unsigned char m_bLateCreateShouldDiscard : 1;
#ifdef VPROF_ENABLED
int m_Frame;
int *m_pFrameCounter;
int *m_pGlobalCounter;
#endif
#ifdef _DEBUG
static int s_BufferCount;
#endif
#ifdef RECORDING
unsigned int m_UID;
#endif
};
#if defined( _X360 )
#include "utlmap.h"
MEMALLOC_DECLARE_EXTERNAL_TRACKING( XMem_CVertexBuffer );
#endif
//-----------------------------------------------------------------------------
// constructor, destructor
//-----------------------------------------------------------------------------
inline CVertexBuffer::CVertexBuffer(IDirect3DDevice9 * pD3D, VertexFormat_t fmt, DWORD theFVF,
int vertexSize, int vertexCount, const char *pTextureBudgetName,
bool bSoftwareVertexProcessing, bool dynamic ) :
m_pVB(0),
m_Position(0),
m_VertexSize(vertexSize),
m_VertexCount(vertexCount),
m_bFlush(true),
m_bLocked(false),
m_bExternalMemory( false ),
m_nBufferSize(vertexSize * vertexCount),
m_TheFVF( theFVF ),
m_bSoftwareVertexProcessing( bSoftwareVertexProcessing ),
m_bDynamic(dynamic),
m_VertexBufferFormat( fmt ),
m_bLateCreateShouldDiscard( false )
#ifdef _X360
,m_pAllocatedMemory(NULL)
,m_iNextBlockingPosition(0)
,m_iAllocationSize(0)
#endif
#ifdef VPROF_ENABLED
,m_Frame( -1 )
#endif
{
MEM_ALLOC_CREDIT_( pTextureBudgetName );
#ifdef RECORDING
// assign a UID
static unsigned int uid = 0;
m_UID = uid++;
#endif
#ifdef _DEBUG
++s_BufferCount;
#endif
#ifdef VPROF_ENABLED
if ( !m_bDynamic )
{
char name[256];
V_strcpy_safe( name, "TexGroup_global_" );
V_strcat_safe( name, pTextureBudgetName, sizeof(name) );
m_pGlobalCounter = g_VProfCurrentProfile.FindOrCreateCounter( name, COUNTER_GROUP_TEXTURE_GLOBAL );
V_strcpy_safe( name, "TexGroup_frame_" );
V_strcat_safe( name, pTextureBudgetName, sizeof(name) );
m_pFrameCounter = g_VProfCurrentProfile.FindOrCreateCounter( name, COUNTER_GROUP_TEXTURE_PER_FRAME );
}
else
{
m_pGlobalCounter = g_VProfCurrentProfile.FindOrCreateCounter( "TexGroup_global_" TEXTURE_GROUP_DYNAMIC_VERTEX_BUFFER, COUNTER_GROUP_TEXTURE_GLOBAL );
}
#endif
if ( !g_pShaderUtil->IsRenderThreadSafe() )
{
m_pSysmemBuffer = ( byte * )MemAlloc_AllocAligned( m_nBufferSize, 16 );
m_nSysmemBufferStartBytes = 0;
}
else
{
m_pSysmemBuffer = NULL;
Create( pD3D );
}
#ifdef VPROF_ENABLED
if ( IsX360() || !m_bDynamic )
{
Assert( m_pGlobalCounter );
*m_pGlobalCounter += m_nBufferSize;
}
#endif
}
void CVertexBuffer::Create( IDirect3DDevice9 *pD3D )
{
D3DVERTEXBUFFER_DESC desc;
memset( &desc, 0x00, sizeof( desc ) );
desc.Format = D3DFMT_VERTEXDATA;
desc.Size = m_nBufferSize;
desc.Type = D3DRTYPE_VERTEXBUFFER;
desc.Pool = m_bDynamic ? D3DPOOL_DEFAULT : D3DPOOL_MANAGED;
desc.FVF = m_TheFVF;
#if defined(IS_WINDOWS_PC) && defined(SHADERAPIDX9)
extern bool g_ShaderDeviceUsingD3D9Ex;
if ( g_ShaderDeviceUsingD3D9Ex )
{
desc.Pool = D3DPOOL_DEFAULT;
}
#endif
desc.Usage = D3DUSAGE_WRITEONLY;
if ( m_bDynamic )
{
desc.Usage |= D3DUSAGE_DYNAMIC;
// Dynamic meshes should never be compressed (slows down writing to them)
Assert( CompressionType( m_TheFVF ) == VERTEX_COMPRESSION_NONE );
}
if ( m_bSoftwareVertexProcessing )
{
desc.Usage |= D3DUSAGE_SOFTWAREPROCESSING;
}
#if !defined( _X360 )
RECORD_COMMAND( DX8_CREATE_VERTEX_BUFFER, 6 );
RECORD_INT( m_UID );
RECORD_INT( m_nBufferSize );
RECORD_INT( desc.Usage );
RECORD_INT( desc.FVF );
RECORD_INT( desc.Pool );
RECORD_INT( m_bDynamic );
HRESULT hr = pD3D->CreateVertexBuffer( m_nBufferSize, desc.Usage, desc.FVF, desc.Pool, &m_pVB, NULL );
if ( hr == D3DERR_OUTOFVIDEOMEMORY || hr == E_OUTOFMEMORY )
{
// Don't have the memory for this. Try flushing all managed resources
// out of vid mem and try again.
// FIXME: need to record this
pD3D->EvictManagedResources();
pD3D->CreateVertexBuffer( m_nBufferSize, desc.Usage, desc.FVF, desc.Pool, &m_pVB, NULL );
}
#ifdef _DEBUG
if ( hr != D3D_OK )
{
switch ( hr )
{
case D3DERR_INVALIDCALL:
Assert( !"D3DERR_INVALIDCALL" );
break;
case D3DERR_OUTOFVIDEOMEMORY:
Assert( !"D3DERR_OUTOFVIDEOMEMORY" );
break;
case E_OUTOFMEMORY:
Assert( !"E_OUTOFMEMORY" );
break;
default:
Assert( 0 );
break;
}
}
#endif
Assert( m_pVB );
#else
// _X360
if ( m_bDynamic )
{
m_iAllocationSize = m_nBufferSize * X360_VERTEX_BUFFER_SIZE_MULTIPLIER;
Assert( m_iAllocationSize >= m_nBufferSize );
m_pAllocatedMemory = (unsigned char*)XPhysicalAlloc( m_iAllocationSize, MAXULONG_PTR, 0, PAGE_READWRITE | MEM_LARGE_PAGES | PAGE_WRITECOMBINE );
}
else if ( MeshMgr()->AllocatePooledVB( this, m_nBufferSize, pTextureBudgetName ) )
{
// Successfully allocated in a shared ShaderAPI memory pool (SetBufferAllocationHandle will have been called to set the pointer and stream offset)
m_iAllocationSize = m_nBufferSize;
Assert( m_pAllocatedMemory );
}
else
{
// Fall back to allocating a standalone VB
// NOTE: write-combining (PAGE_WRITECOMBINE) is deliberately not used, since it slows down CPU access to the data (decals+defragmentation)
m_iAllocationSize = m_nBufferSize;
m_pAllocatedMemory = (unsigned char*)XPhysicalAlloc( m_iAllocationSize, MAXULONG_PTR, 0, PAGE_READWRITE );
}
if ( m_pAllocatedMemory && !IsPooled() )
{
MemAlloc_RegisterExternalAllocation( XMem_CVertexBuffer, m_pAllocatedMemory, XPhysicalSize( m_pAllocatedMemory ) );
if ( !m_bDynamic )
{
// Track non-pooled physallocs, to help tune CGPUBufferAllocator usage
g_SizeIndividualVBPhysAllocs += XPhysicalSize( m_pAllocatedMemory );
g_NumIndividualVBPhysAllocs++;
}
}
m_iNextBlockingPosition = m_iAllocationSize;
#endif
#ifdef MEASURE_DRIVER_ALLOCATIONS
int nMemUsed = 1024;
VPROF_INCREMENT_GROUP_COUNTER( "vb count", COUNTER_GROUP_NO_RESET, 1 );
VPROF_INCREMENT_GROUP_COUNTER( "vb driver mem", COUNTER_GROUP_NO_RESET, nMemUsed );
VPROF_INCREMENT_GROUP_COUNTER( "total driver mem", COUNTER_GROUP_NO_RESET, nMemUsed );
#endif
// Track VB allocations
#if !defined( _X360 )
g_VBAllocTracker->CountVB( m_pVB, m_bDynamic, m_nBufferSize, m_VertexSize, m_VertexBufferFormat );
#else
g_VBAllocTracker->CountVB( this, m_bDynamic, m_iAllocationSize, m_VertexSize, m_VertexBufferFormat );
#endif
}
#ifdef _X360
void *AllocateTempBuffer( size_t nSizeInBytes );
//-----------------------------------------------------------------------------
// This variant is for when we already have the data in physical memory
//-----------------------------------------------------------------------------
inline CVertexBuffer::CVertexBuffer( ) :
m_pVB( 0 ),
m_Position( 0 ),
m_VertexSize( 0 ),
m_VertexCount( 0 ),
m_bFlush( false ),
m_bLocked( false ),
m_bExternalMemory( true ),
m_nBufferSize( 0 ),
m_bDynamic( false )
#ifdef VPROF_ENABLED
,m_Frame( -1 )
#endif
{
m_iAllocationSize = 0;
m_pAllocatedMemory = 0;
m_iNextBlockingPosition = 0;
}
#include "tier0/memdbgoff.h"
inline void CVertexBuffer::Init( IDirect3DDevice9 *pD3D, VertexFormat_t fmt, DWORD theFVF, uint8 *pVertexData, int vertexSize, int vertexCount )
{
m_nBufferSize = vertexSize * vertexCount;
m_Position = m_Position;
m_VertexSize = vertexSize;
m_VertexCount = vertexCount;
m_iAllocationSize = m_nBufferSize;
m_pAllocatedMemory = pVertexData;
m_iNextBlockingPosition = m_iAllocationSize;
m_pVB = new( AllocateTempBuffer( sizeof( IDirect3DVertexBuffer9 ) ) ) IDirect3DVertexBuffer9;
XGSetVertexBufferHeader( m_nBufferSize, 0, 0, 0, m_pVB );
XGOffsetResourceAddress( m_pVB, pVertexData );
}
#include "tier0/memdbgon.h"
#endif // _X360
inline CVertexBuffer::~CVertexBuffer()
{
// Track VB allocations (even if pooled)
#if !defined( _X360 )
if ( m_pVB != NULL )
{
g_VBAllocTracker->UnCountVB( m_pVB );
}
#else
if ( m_pVB && m_pVB->IsSet( Dx9Device() ) )
{
Unbind( m_pVB );
}
if ( !m_bExternalMemory )
{
g_VBAllocTracker->UnCountVB( this );
}
#endif
if ( !m_bExternalMemory )
{
#ifdef MEASURE_DRIVER_ALLOCATIONS
int nMemUsed = 1024;
VPROF_INCREMENT_GROUP_COUNTER( "vb count", COUNTER_GROUP_NO_RESET, -1 );
VPROF_INCREMENT_GROUP_COUNTER( "vb driver mem", COUNTER_GROUP_NO_RESET, -nMemUsed );
VPROF_INCREMENT_GROUP_COUNTER( "total driver mem", COUNTER_GROUP_NO_RESET, -nMemUsed );
#endif
#ifdef VPROF_ENABLED
if ( IsX360() || !m_bDynamic )
{
Assert( m_pGlobalCounter );
*m_pGlobalCounter -= m_nBufferSize;
}
#endif
#ifdef _DEBUG
--s_BufferCount;
#endif
}
Unlock( 0 );
if ( m_pSysmemBuffer )
{
MemAlloc_FreeAligned( m_pSysmemBuffer );
m_pSysmemBuffer = NULL;
}
#if !defined( _X360 )
if ( m_pVB )
{
RECORD_COMMAND( DX8_DESTROY_VERTEX_BUFFER, 1 );
RECORD_INT( m_UID );
m_pVB->Release();
}
#else
if ( m_pAllocatedMemory && !m_bExternalMemory )
{
if ( IsPooled() )
{
MeshMgr()->DeallocatePooledVB( this );
}
else
{
MemAlloc_RegisterExternalDeallocation( XMem_CVertexBuffer, m_pAllocatedMemory, XPhysicalSize( m_pAllocatedMemory ) );
if ( !m_bDynamic )
{
// Track non-pooled physallocs, to help tune CGPUBufferAllocator usage
g_SizeIndividualVBPhysAllocs -= XPhysicalSize( m_pAllocatedMemory );
g_NumIndividualVBPhysAllocs--;
}
XPhysicalFree( m_pAllocatedMemory );
}
}
m_pAllocatedMemory = NULL;
m_pVB = NULL;
#endif // _X360
}
#ifdef _X360
//-----------------------------------------------------------------------------
// Get memory allocation data
//-----------------------------------------------------------------------------
inline const GPUBufferHandle_t *CVertexBuffer::GetBufferAllocationHandle( void )
{
Assert( IsPooled() );
return ( IsPooled() ? &m_GPUBufferHandle : NULL );
}
//-----------------------------------------------------------------------------
// Update memory allocation data
//-----------------------------------------------------------------------------
inline void CVertexBuffer::SetBufferAllocationHandle( const GPUBufferHandle_t &bufferAllocationHandle )
{
// This VB's memory has been reallocated or freed, update our cached pointer and the D3D header
// NOTE: this should never be called while any rendering is in flight!
Assert( ( m_pAllocatedMemory == NULL ) || IsPooled() );
if ( ( m_pAllocatedMemory == NULL ) || IsPooled() )
{
m_GPUBufferHandle = bufferAllocationHandle;
m_pAllocatedMemory = m_GPUBufferHandle.pMemory;
if ( m_pVB )
{
XGSetVertexBufferHeader( m_nBufferSize, 0, D3DPOOL_DEFAULT, 0, m_pVB );
XGOffsetResourceAddress( m_pVB, m_pAllocatedMemory );
}
}
}
//-----------------------------------------------------------------------------
// Expose the data pointer for read-only CPU access to the data
//-----------------------------------------------------------------------------
inline const byte **CVertexBuffer::GetBufferDataPointerAddress( void )
{
if ( m_bDynamic /* FIXME: || m_bExternalMemory*/ )
return NULL;
return (const byte **)&m_pAllocatedMemory;
}
#endif // _X360
//-----------------------------------------------------------------------------
// Compute the next offset for the next lock
//-----------------------------------------------------------------------------
inline int CVertexBuffer::NextLockOffset( ) const
{
#if !defined( _X360 )
int nNextOffset = ( m_Position + m_VertexSize - 1 ) / m_VertexSize;
nNextOffset *= m_VertexSize;
return nNextOffset;
#else
return m_Position; //position is already aligned properly on unlocks for 360.
#endif
}
//-----------------------------------------------------------------------------
// Do we have enough room without discarding?
//-----------------------------------------------------------------------------
inline bool CVertexBuffer::HasEnoughRoom( int numVertices ) const
{
#if defined( _X360 )
return numVertices <= m_VertexCount; //the ring buffer will free room as needed
#else
return (NextLockOffset() + (numVertices * m_VertexSize)) <= m_nBufferSize;
#endif
}
//-----------------------------------------------------------------------------
// Block until this part of the index buffer is free
//-----------------------------------------------------------------------------
inline void CVertexBuffer::BlockUntilUnused( int nBufferSize )
{
Assert( nBufferSize <= m_nBufferSize );
#ifdef _X360
int nLockOffset = NextLockOffset();
Assert( (m_AllocationRing.Count() != 0) || ((m_Position == 0) && (m_iNextBlockingPosition == m_iAllocationSize)) );
if ( (m_iNextBlockingPosition - nLockOffset) >= nBufferSize )
return;
Assert( (m_AllocationRing[m_AllocationRing.Head()].m_iStartOffset == 0) || ((m_iNextBlockingPosition == m_AllocationRing[m_AllocationRing.Head()].m_iStartOffset) && (m_Position <= m_iNextBlockingPosition)) );
int iMinBlockPosition = nLockOffset + nBufferSize;
if( iMinBlockPosition > m_iAllocationSize )
{
//Allocation requires us to wrap
iMinBlockPosition = nBufferSize;
m_Position = 0;
//modify the last allocation so that it uses up the whole tail end of the buffer. Makes other code simpler
Assert( m_AllocationRing.Count() != 0 );
m_AllocationRing[m_AllocationRing.Tail()].m_iEndOffset = m_iAllocationSize;
//treat all allocations between the current position and the tail end of the ring as freed since they will be before we unblock
while( m_AllocationRing.Count() )
{
unsigned int head = m_AllocationRing.Head();
if( m_AllocationRing[head].m_iStartOffset == 0 )
break;
m_AllocationRing.Remove( head );
}
}
//now we go through the allocations until we find the last fence we care about. Treat everything up until that fence as freed.
DWORD FinalFence = 0;
unsigned int iFinalAllocationZPassIdx = 0;
while( m_AllocationRing.Count() )
{
unsigned int head = m_AllocationRing.Head();
if( m_AllocationRing[head].m_iEndOffset >= iMinBlockPosition )
{
//When this frees, we'll finally have enough space for the allocation
FinalFence = m_AllocationRing[head].m_Fence;
iFinalAllocationZPassIdx = m_AllocationRing[head].m_iZPassIdx;
m_iNextBlockingPosition = m_AllocationRing[head].m_iEndOffset;
m_AllocationRing.Remove( head );
break;
}
m_AllocationRing.Remove( head );
}
Assert( FinalFence != 0 );
if( Dx9Device()->IsFencePending( FinalFence ) )
{
#ifdef SPEW_VERTEX_BUFFER_STALLS
float st = Plat_FloatTime();
#endif
if ( ( Dx9Device()->GetDeviceState() & D3DDEVICESTATE_ZPASS_BRACKET ) &&
( iFinalAllocationZPassIdx == ShaderAPI()->Get360ZPassCounter() ) )
{
// We're about to overrun our VB ringbuffer in a single Z prepass. To avoid rendering corruption, close out the
// Z prepass and continue. This will reduce early-Z rejection efficiency and could cause a momentary framerate drop,
// but it's better than rendering corruption.
Warning( "Dynamic VB ring buffer overrun in Z Prepass. Tell Thorsten.\n" );
ShaderAPI()->End360ZPass();
}
Dx9Device()->BlockOnFence( FinalFence );
#ifdef SPEW_VERTEX_BUFFER_STALLS
float dt = Plat_FloatTime() - st;
Warning( "Blocked locking dynamic vertex buffer for %f ms!\n", 1000.0 * dt );
#endif
}
#endif
}
//-----------------------------------------------------------------------------
// lock, unlock
//-----------------------------------------------------------------------------
inline unsigned char* CVertexBuffer::Lock( int numVerts, int& baseVertexIndex )
{
#if defined( _X360 )
if ( m_pVB && m_pVB->IsSet( Dx9Device() ) )
{
Unbind( m_pVB );
}
#endif
m_nLockCount = numVerts;
unsigned char* pLockedData = 0;
baseVertexIndex = 0;
int nBufferSize = numVerts * m_VertexSize;
Assert( IsPC() || ( IsX360() && !m_bLocked ) );
// Ensure there is enough space in the VB for this data
if ( numVerts > m_VertexCount )
{
Assert( 0 );
return 0;
}
if ( !IsX360() && !m_pVB && !m_pSysmemBuffer )
return 0;
DWORD dwFlags;
if ( m_bDynamic )
{
dwFlags = LOCKFLAGS_APPEND;
#if !defined( _X360 )
// If either the user forced us to flush,
// or there is not enough space for the vertex data,
// then flush the buffer contents
if ( !m_Position || m_bFlush || !HasEnoughRoom(numVerts) )
{
if ( m_pSysmemBuffer || !g_pShaderUtil->IsRenderThreadSafe() )
m_bLateCreateShouldDiscard = true;
m_bFlush = false;
m_Position = 0;
dwFlags = LOCKFLAGS_FLUSH;
}
#else
if( m_bFlush )
{
# if ( defined( X360_BLOCK_ON_VB_FLUSH ) )
{
if( m_AllocationRing.Count() )
{
DWORD FinalFence = m_AllocationRing[m_AllocationRing.Tail()].m_Fence;
m_AllocationRing.RemoveAll();
m_Position = 0;
m_iNextBlockingPosition = m_iAllocationSize;
# if ( defined( SPEW_VERTEX_BUFFER_STALLS ) )
if( Dx9Device()->IsFencePending( FinalFence ) )
{
float st = Plat_FloatTime();
# endif
Dx9Device()->BlockOnFence( FinalFence );
# if ( defined ( SPEW_VERTEX_BUFFER_STALLS ) )
float dt = Plat_FloatTime() - st;
Warning( "Blocked FLUSHING dynamic vertex buffer for %f ms!\n", 1000.0 * dt );
}
# endif
}
}
# endif
m_bFlush = false;
}
#endif
}
else
{
// Since we are a static VB, always lock the beginning of the buffer.
dwFlags = D3DLOCK_NOSYSLOCK;
m_Position = 0;
}
if ( IsX360() && m_bDynamic )
{
// Block until we have enough room in the buffer, this affects the result of NextLockOffset() in wrap conditions.
BlockUntilUnused( nBufferSize );
m_pVB = NULL;
}
int nLockOffset = NextLockOffset( );
RECORD_COMMAND( DX8_LOCK_VERTEX_BUFFER, 4 );
RECORD_INT( m_UID );
RECORD_INT( nLockOffset );
RECORD_INT( nBufferSize );
RECORD_INT( dwFlags );
#if !defined( _X360 )
// If the caller isn't in the thread that owns the render lock, need to return a system memory pointer--cannot talk to GL from
// the non-current thread.
if ( !m_pSysmemBuffer && !g_pShaderUtil->IsRenderThreadSafe() )
{
m_pSysmemBuffer = ( byte * )MemAlloc_AllocAligned( m_nBufferSize, 16 );
m_nSysmemBufferStartBytes = nLockOffset;
Assert( ( m_nSysmemBufferStartBytes % m_VertexSize ) == 0 );
}
if ( m_pSysmemBuffer != NULL )
{
// Ensure that we're never moving backwards in a buffer--this code would need to be rewritten if so.
// We theorize this can happen if you hit the end of a buffer and then wrap before drawing--but
// this would probably break in other places as well.
Assert( nLockOffset >= m_nSysmemBufferStartBytes );
pLockedData = m_pSysmemBuffer + nLockOffset;
}
else
{
m_pVB->Lock( nLockOffset,
nBufferSize,
reinterpret_cast< void** >( &pLockedData ),
dwFlags );
}
#else
pLockedData = m_pAllocatedMemory + nLockOffset;
#endif
Assert( pLockedData != 0 );
m_bLocked = true;
if ( !IsX360() )
{
baseVertexIndex = nLockOffset / m_VertexSize;
}
else
{
baseVertexIndex = 0;
}
return pLockedData;
}
inline unsigned char* CVertexBuffer::Modify( bool bReadOnly, int firstVertex, int numVerts )
{
unsigned char* pLockedData = 0;
// D3D still returns a pointer when you call lock with 0 verts, so just in
// case it's actually doing something, don't even try to lock the buffer with 0 verts.
if ( numVerts == 0 )
return NULL;
m_nLockCount = numVerts;
// If this hits, m_pSysmemBuffer logic needs to be added to this code.
Assert( g_pShaderUtil->IsRenderThreadSafe() );
Assert( !m_pSysmemBuffer ); // if this hits, then we need to add code to handle it
Assert( m_pVB && !m_bDynamic );
if ( firstVertex + numVerts > m_VertexCount )
{
Assert( 0 );
return NULL;
}
DWORD dwFlags = D3DLOCK_NOSYSLOCK;
if ( bReadOnly )
{
dwFlags |= D3DLOCK_READONLY;
}
RECORD_COMMAND( DX8_LOCK_VERTEX_BUFFER, 4 );
RECORD_INT( m_UID );
RECORD_INT( firstVertex * m_VertexSize );
RECORD_INT( numVerts * m_VertexSize );
RECORD_INT( dwFlags );
// mmw: for forcing all dynamic... LOCKFLAGS_FLUSH );
#if !defined( _X360 )
m_pVB->Lock(
firstVertex * m_VertexSize,
numVerts * m_VertexSize,
reinterpret_cast< void** >( &pLockedData ),
dwFlags );
#else
if ( m_pVB->IsSet( Dx9Device() ) )
{
Unbind( m_pVB );
}
pLockedData = m_pAllocatedMemory + (firstVertex * m_VertexSize);
#endif
m_Position = firstVertex * m_VertexSize;
Assert( pLockedData != 0 );
m_bLocked = true;
return pLockedData;
}
inline void CVertexBuffer::Unlock( int numVerts )
{
if ( !m_bLocked )
return;
if ( !IsX360() && !m_pVB && !m_pSysmemBuffer )
return;
int nLockOffset = NextLockOffset();
int nBufferSize = numVerts * m_VertexSize;
RECORD_COMMAND( DX8_UNLOCK_VERTEX_BUFFER, 1 );
RECORD_INT( m_UID );
#if !defined( _X360 )
if ( m_pSysmemBuffer != NULL )
{
}
else
{
#if DX_TO_GL_ABSTRACTION
Assert( numVerts <= (int)m_nLockCount );
int unlockBytes = ( m_bDynamic ? nBufferSize : ( m_nLockCount * m_VertexSize ) );
#else
int unlockBytes = 0;
#endif
ReallyUnlock( unlockBytes );
}
m_Position = nLockOffset + nBufferSize;
#else
if ( m_bDynamic )
{
if ( numVerts > 0 )
{
DynamicBufferAllocation_t LockData;
LockData.m_Fence = Dx9Device()->GetCurrentFence(); //This isn't the correct fence, but it's all we have access to for now and it'll provide marginal safety if something goes really wrong.
LockData.m_iStartOffset = nLockOffset;
LockData.m_iEndOffset = LockData.m_iStartOffset + nBufferSize;
LockData.m_iZPassIdx = ( Dx9Device()->GetDeviceState() & D3DDEVICESTATE_ZPASS_BRACKET ) ? ShaderAPI()->Get360ZPassCounter() : 0;
// Round dynamic locks to 4k boundaries for GPU cache reasons
LockData.m_iEndOffset = ALIGN_VALUE( LockData.m_iEndOffset, 4096 );
if( LockData.m_iEndOffset > m_iAllocationSize )
LockData.m_iEndOffset = m_iAllocationSize;
m_AllocationRing.AddToTail( LockData );
m_Position = LockData.m_iEndOffset;
void* pLockedData = m_pAllocatedMemory + LockData.m_iStartOffset;
//Always re-use the same vertex buffer header based on the assumption that D3D copies it off in the draw calls.
m_pVB = &m_D3DVertexBuffer;
XGSetVertexBufferHeader( nBufferSize, 0, D3DPOOL_DEFAULT, 0, m_pVB );
XGOffsetResourceAddress( m_pVB, pLockedData );
// Invalidate the GPU caches for this memory.
Dx9Device()->InvalidateGpuCache( pLockedData, nBufferSize, 0 );
}
}
else
{
if ( !m_pVB )
{
m_pVB = &m_D3DVertexBuffer;
XGSetVertexBufferHeader( m_nBufferSize, 0, D3DPOOL_DEFAULT, 0, m_pVB );
XGOffsetResourceAddress( m_pVB, m_pAllocatedMemory );
}
m_Position = nLockOffset + nBufferSize;
// Invalidate the GPU caches for this memory.
Dx9Device()->InvalidateGpuCache( m_pAllocatedMemory, m_nBufferSize, 0 );
}
#endif
m_bLocked = false;
}
inline void CVertexBuffer::HandleLateCreation( )
{
if ( !m_pSysmemBuffer )
{
return;
}
if( !m_pVB )
{
bool bPrior = g_VBAllocTracker->TrackMeshAllocations( "HandleLateCreation" );
Create( Dx9Device() );
if ( !bPrior )
{
g_VBAllocTracker->TrackMeshAllocations( NULL );
}
}
void* pWritePtr = NULL;
const int dataToWriteBytes = m_bDynamic ? ( m_Position - m_nSysmemBufferStartBytes ) : ( m_nLockCount * m_VertexSize );
DWORD dwFlags = D3DLOCK_NOSYSLOCK;
if ( m_bDynamic )
{
dwFlags |= ( m_bLateCreateShouldDiscard ? D3DLOCK_DISCARD : D3DLOCK_NOOVERWRITE );
}
// Always clear this.
m_bLateCreateShouldDiscard = false;
// Don't use the Lock function, it does a bunch of stuff we don't want.
HRESULT hr = m_pVB->Lock( m_nSysmemBufferStartBytes,
dataToWriteBytes,
&pWritePtr,
dwFlags);
// If this fails we're about to crash. Consider skipping the update and leaving
// m_pSysmemBuffer around to try again later. (For example in case of device loss)
Assert( SUCCEEDED( hr ) ); hr;
memcpy( pWritePtr, m_pSysmemBuffer + m_nSysmemBufferStartBytes, dataToWriteBytes );
ReallyUnlock( dataToWriteBytes );
MemAlloc_FreeAligned( m_pSysmemBuffer );
m_pSysmemBuffer = NULL;
}
// Returns the allocated size
inline int CVertexBuffer::AllocationSize() const
{
#ifdef _X360
return m_iAllocationSize;
#else
return m_VertexCount * m_VertexSize;
#endif
}
#endif // DYNAMICVB_H