kerneltest/e32test/misc/t_ipccpy.cpp
changeset 0 a41df078684a
--- /dev/null	Thu Jan 01 00:00:00 1970 +0000
+++ b/kerneltest/e32test/misc/t_ipccpy.cpp	Mon Oct 19 15:55:17 2009 +0100
@@ -0,0 +1,724 @@
+// Copyright (c) 1998-2009 Nokia Corporation and/or its subsidiary(-ies).
+// All rights reserved.
+// This component and the accompanying materials are made available
+// under the terms of the License "Eclipse Public License v1.0"
+// which accompanies this distribution, and is available
+// at the URL "http://www.eclipse.org/legal/epl-v10.html".
+//
+// Initial Contributors:
+// Nokia Corporation - initial contribution.
+//
+// Contributors:
+//
+// Description:
+// e32test\misc\t_ipccpy.cpp
+// Overview:
+// Test and benchmark IPC reading, writing, copying.	
+// API Information:
+// RBusLogicalChannel, DLogicalChannel.	
+// Details:
+// - Load the specified logical device driver, open a channel to it, allocate 
+// a cell of specified size from the current thread's heap, get Kernel HAL
+// memory model information.
+// - Make a synchronous Kernel Executive type request to the logical channel 
+// to write specified data to the buffer, read the data and calculate the 
+// time taken for writing and reading the data. Benchmark the time required 
+// to for 1000 64K user->kernel and kernel->user copies.
+// - Create a chunk, get a pointer to the base of the chunk's reserved region,
+// create a server thread, establish a session with the server, signal 
+// completion of the client's request when message is received, read, 
+// write specified bits and check it is as expected.
+// Platforms/Drives/Compatibility:
+// All.
+// Assumptions/Requirement/Pre-requisites:
+// Failures and causes:
+// Base Port information:
+// 
+//
+
+#include <e32test.h>
+#include "d_ipccpy.h"
+#include "u32std.h"
+#include <e32kpan.h>
+#include "../mmu/mmudetect.h"
+#include <hal.h>
+
+RTest test(_L("T_IPCCPY"));
+TUint8* Buffer;
+TUint8* Disc;
+RIpcCpy Ipccpy;
+TUint32 MainId;
+TUint8 Bss[4096];
+TUint8* Kern;
+TUint8* RamDrive;
+TUint8* Nonexistent;
+TUint8* Unaligned=Bss+1;
+TInt CloseTime;
+TLinAddr HwChunkAddr[RIpcCpy::ENumHwChunkTypes];
+TPtr8 UserDes(Buffer+96,96,96);
+
+void SetupAddresses()
+	{
+	Kern=KernData();
+	TUint32 mm_attr=MemModelAttributes();
+	TUint32 mm_type=mm_attr & EMemModelTypeMask;
+	switch (mm_type)
+		{
+		case EMemModelTypeDirect:
+			RamDrive=(TUint8*)0;	// not used anyway
+			Nonexistent=(TUint8*)0xa8000000;
+			break;
+		case EMemModelTypeMoving:
+			RamDrive=(TUint8*)0x40000000;
+			Nonexistent=(TUint8*)0x60f00000;
+			break;
+		case EMemModelTypeMultiple:
+			RamDrive=(TUint8*)0xa0000000;
+			Nonexistent=(TUint8*)0xfe000000;
+			break;
+		case EMemModelTypeFlexible:
+			RamDrive=(TUint8*)0;
+			Nonexistent=(TUint8*)0x8ff00000;
+			break;
+		case EMemModelTypeEmul:
+			RamDrive=(TUint8*)0;	// not used anyway
+			Nonexistent=(TUint8*)0xf0000000;
+			break;
+		default:
+			test(0);
+			break;
+		}
+	new (&UserDes) TPtr8(Buffer+96,96,96);
+	Ipccpy.HardwareChunks(HwChunkAddr,UserDes);
+	test.Printf(_L("Buffer=%08x\n"),Buffer);
+	test.Printf(_L("Bss=%08x\n"),Bss);
+	test.Printf(_L("Kern=%08x\n"),Kern);
+	test.Printf(_L("RamDrive=%08x\n"),RamDrive);
+	test.Printf(_L("Nonexistent=%08x\n"),Nonexistent);
+	test.Printf(_L("Unaligned=%08x\n"),Unaligned);
+	test.Printf(_L("HwChunkSupRw=%08x\n"),HwChunkAddr[RIpcCpy::EHwChunkSupRw]);
+	test.Printf(_L("HwChunkUserRw=%08x\n"),HwChunkAddr[RIpcCpy::EHwChunkUserRw]);
+	test.Printf(_L("HwChunkUserRo=%08x\n"),HwChunkAddr[RIpcCpy::EHwChunkUserRo]);
+	}
+
+_LIT(KLitKernExec,"KERN-EXEC");
+
+void TestEq(TInt a, TInt b, TInt l);
+void Test(TBool c, TInt l);
+
+#define TESTEQ(a,b)		TestEq((a),(b),__LINE__)
+#define TEST(c)			Test((c),__LINE__)
+
+void TestEq(TInt a, TInt b, TInt l)
+	{
+	if (a!=b)
+		{
+		if (TUint32(RThread().Id())==MainId)
+			{
+			test.Printf(_L("Line %d a=%d, b=%d\n"),l,a,b);
+			test(0);
+			}
+		else
+			User::Panic(_L("TESTEQ"),l);
+		}
+	}
+
+void Test(TBool c, TInt l)
+	{
+	if (!c)
+		{
+		if (TUint32(RThread().Id())==MainId)
+			{
+			test.Printf(_L("Line %d FAIL\n"),l);
+			test(0);
+			}
+		else
+			User::Panic(_L("TEST"),l);
+		}
+	}
+
+struct SIpcTestInfo
+	{
+	const TAny* iLocal;
+	const TAny* iRemote;
+	TInt iOffset;
+	TInt iMode;			// bit 0 = 1 for 16 bit, bit 1 = 1 for write
+	};
+
+class RLocalSession : public RSessionBase
+	{
+public:
+	TInt Connect(RServer2 aSrv,TRequestStatus* aStat)
+		{return CreateSession(aSrv,TVersion(),-1,EIpcSession_Unsharable,0,aStat);}
+	void Test(const TAny* aRemote)
+		{Send(0,TIpcArgs((const TDesC8*)aRemote,(const TDesC16*)aRemote,(TDes8*)aRemote,(TDes16*)aRemote));}
+	void Wait()
+		{SendReceive(1);}
+	};
+
+RServer2 IpcServer;
+
+TInt IpcTestFn(TAny* aInfo)
+	{
+	SIpcTestInfo& i=*(SIpcTestInfo*)aInfo;
+
+	if (IpcServer.Handle())
+		IpcServer.Close();
+	
+	TESTEQ(IpcServer.CreateGlobal(KNullDesC),KErrNone);
+	RLocalSession sess;
+	TRequestStatus stat;
+	TESTEQ(sess.Connect(IpcServer,&stat),KErrNone);
+	RMessage2 m;
+	IpcServer.Receive(m);
+	m.Complete(KErrNone);	// connect
+	User::WaitForRequest(stat);	// connection message report
+	sess.Test(i.iRemote);
+	IpcServer.Receive(m);
+
+	TInt r=KMinTInt;
+	switch (i.iMode)
+		{
+		case 0:
+			{	// read 8 bit
+			TDesC8* pR=(TDesC8*)i.iRemote;
+			TDes8* pL=(TDes8*)i.iLocal;
+			r=m.Read(0,*pL,i.iOffset);
+			if (r==KErrNone)
+				{
+				TESTEQ(pL->Length(),pR->Length()-i.iOffset);
+				TEST(*pL==pR->Mid(i.iOffset));
+				}
+			break;
+			}
+		case 1:
+			{	// read 16 bit
+			TDesC16* pR=(TDesC16*)i.iRemote;
+			TDes16* pL=(TDes16*)i.iLocal;
+			r=m.Read(1,*pL,i.iOffset);
+			if (r==KErrNone)
+				{
+				TESTEQ(pL->Length(),pR->Length()-i.iOffset);
+				TEST(*pL==pR->Mid(i.iOffset));
+				}
+			break;
+			}
+		case 2:
+			{	// write 8 bit
+			TDes8* pR=(TDes8*)i.iRemote;
+			TDesC8* pL=(TDesC8*)i.iLocal;
+			r=m.Write(2,*pL,i.iOffset);
+			if (r==KErrNone)
+				{
+				TESTEQ(pR->Length(),pL->Length()+i.iOffset);
+				TEST(*pL==pR->Mid(i.iOffset));
+				}
+			break;
+			}
+		case 3:
+			{	// write 16 bit
+			TDes16* pR=(TDes16*)i.iRemote;
+			TDesC16* pL=(TDesC16*)i.iLocal;
+			r=m.Write(3,*pL,i.iOffset);
+			if (r==KErrNone)
+				{
+				TESTEQ(pR->Length(),pL->Length()+i.iOffset);
+				TEST(*pL==pR->Mid(i.iOffset));
+				}
+			break;
+			}
+		default:
+			User::Panic(_L("MODE"),i.iMode);
+		}
+	m.Complete(0);
+	sess.Close();
+	IpcServer.Close();
+
+	return r;
+	}
+
+void _DoIpcTest(const TAny* aLocal, const TAny* aRemote, TInt aOffset, TInt aMode, const TDesC* aPanicCat, TInt aResult, TInt aLine)
+	{
+	test.Printf(_L("Line %d\n"),aLine);
+	SIpcTestInfo info;
+	info.iLocal=aLocal;
+	info.iRemote=aRemote;
+	info.iOffset=aOffset;
+	info.iMode=aMode;
+	if (!aPanicCat)
+		{
+		// do test in this thread
+		TInt r=IpcTestFn(&info);
+		TESTEQ(r,aResult);
+		return;
+		}
+	TBool jit=User::JustInTime();
+	RThread t;
+	TInt r=t.Create(KNullDesC(),IpcTestFn,0x2000,NULL,&info);
+	test(r==KErrNone);
+	TRequestStatus s;
+	t.Logon(s);
+	User::SetJustInTime(EFalse);
+	t.Resume();
+	User::WaitForRequest(s);
+	User::SetJustInTime(jit);
+	test(t.ExitType()==EExitPanic);
+	test(t.ExitCategory()==*aPanicCat);
+	TESTEQ(t.ExitReason(),aResult);
+	t.Close();
+	}
+
+void DoIpcTest(const TUint8* aLocal, const TUint8* aRemote, TInt aLength, TInt aMode, const TDesC* aPanicCat, TInt aResult, TInt aLine)
+	{
+	TPtr8 local((TUint8*)aLocal,aLength,aLength);
+	TPtr8 remote((TUint8*)aRemote,aLength,aLength);
+	_DoIpcTest(&local,&remote,0,aMode,aPanicCat,aResult,aLine);
+	}
+
+void DoIpcTest(const TUint8* aLocal, const TDesC8& aRemote, TInt aLength, TInt aMode, const TDesC* aPanicCat, TInt aResult, TInt aLine)
+	{
+	TPtr8 local((TUint8*)aLocal,aLength,aLength);
+	_DoIpcTest(&local,&aRemote,0,aMode,aPanicCat,aResult,aLine);
+	}
+
+void TestIpcCopyErrors()
+	{
+	RChunk c;
+	TInt r=c.CreateDisconnectedLocal(0,0,0x500000);
+	test(r==KErrNone);
+	r=c.Commit(0,0x1000);
+	test(r==KErrNone);
+	r=c.Commit(0x2000,0x1000);
+	test(r==KErrNone);
+	r=c.Commit(0x3ff000,0x1000);
+	test(r==KErrNone);
+	Disc=c.Base();
+	test.Printf(_L("Disc=%08x\n"),Disc);
+	DoIpcTest(Buffer,(const TUint8*)&TestEq,100,0,NULL,KErrNone,__LINE__);
+	DoIpcTest(Buffer,(const TUint8*)&TestEq,100,2,NULL,KErrBadDescriptor,__LINE__);
+	DoIpcTest((const TUint8*)&TestEq,Buffer,100,2,NULL,KErrNone,__LINE__);
+	DoIpcTest((const TUint8*)&TestEq,Buffer,100,0,&KLitKernExec,ECausedException,__LINE__);
+	DoIpcTest(Buffer,Nonexistent,100,0,NULL,KErrBadDescriptor,__LINE__);
+	DoIpcTest(Buffer,Nonexistent,100,2,NULL,KErrBadDescriptor,__LINE__);
+	DoIpcTest(Nonexistent,Buffer,100,2,&KLitKernExec,ECausedException,__LINE__);
+	DoIpcTest(Nonexistent,Buffer,100,0,&KLitKernExec,ECausedException,__LINE__);
+	DoIpcTest(Buffer,Unaligned,100,0,NULL,KErrNone,__LINE__);
+	DoIpcTest(Buffer,Unaligned,100,2,NULL,KErrNone,__LINE__);
+	DoIpcTest(Unaligned,Buffer,100,2,NULL,KErrNone,__LINE__);
+	DoIpcTest(Unaligned,Buffer,100,0,NULL,KErrNone,__LINE__);
+
+	DoIpcTest(Disc+4001,Buffer,95,0,NULL,KErrNone,__LINE__);
+	if (HaveVirtMem())
+		DoIpcTest(Disc+4001,Buffer,96,0,&KLitKernExec,ECausedException,__LINE__);
+	DoIpcTest(Buffer,Disc+4001,95,0,NULL,KErrNone,__LINE__);
+	if (HaveVirtMem())
+		DoIpcTest(Buffer,Disc+4001,96,0,NULL,KErrBadDescriptor,__LINE__);
+
+	TPtr8* pdes;
+	if (HaveVirtMem())
+		{
+		// test descriptor stored stradling chunk end...
+		pdes = (TPtr8*)(Disc+0x3ffff4);
+		memcpy(pdes,&UserDes,12);
+		DoIpcTest(Buffer,*pdes,pdes->Size(),0,NULL,KErrNone,__LINE__);
+		pdes = (TPtr8*)(Disc+0x3ffff8);
+		memcpy(pdes,&UserDes,8);
+		DoIpcTest(Buffer,*pdes,pdes->Size(),0,NULL,KErrBadDescriptor,__LINE__);
+		pdes = (TPtr8*)(Disc+0x3ffffc);
+		memcpy(pdes,&UserDes,4);
+		DoIpcTest(Buffer,*pdes,pdes->Size(),0,NULL,KErrBadDescriptor,__LINE__);
+		r=c.Commit(0x400000,0x1000);
+		test(r==KErrNone);
+		pdes = (TPtr8*)(Disc+0x3ffff4);
+		memcpy(pdes,&UserDes,12);
+		DoIpcTest(Buffer,*pdes,pdes->Size(),0,NULL,KErrNone,__LINE__);
+		pdes = (TPtr8*)(Disc+0x3ffff8);
+		memcpy(pdes,&UserDes,12);
+		DoIpcTest(Buffer,*pdes,pdes->Size(),0,NULL,KErrNone,__LINE__);
+		pdes = (TPtr8*)(Disc+0x3ffffc);
+		memcpy(pdes,&UserDes,12);
+		DoIpcTest(Buffer,*pdes,pdes->Size(),0,NULL,KErrNone,__LINE__);
+		}
+
+	if (HaveMultAddr())
+		{
+		if(RamDrive)
+			{
+			DoIpcTest(Disc+0x100000,Buffer,96,0,&KLitKernExec,ECausedException,__LINE__);
+			DoIpcTest(Buffer,Disc+0x100000,96,0,NULL,KErrBadDescriptor,__LINE__);
+			DoIpcTest(RamDrive,Buffer,4,0,&KLitKernExec,ECausedException,__LINE__);
+			DoIpcTest(Buffer,RamDrive,4,0,NULL,KErrBadDescriptor,__LINE__);
+			DoIpcTest(RamDrive,Buffer,4,2,&KLitKernExec,ECausedException,__LINE__);
+			DoIpcTest(Buffer,RamDrive,4,2,NULL,KErrBadDescriptor,__LINE__);
+			}
+
+		// if memory alising happens during IPC then the memory at 'Disc' would be aliased
+		// at KIPCAliasAddress and so would not be protected by MMU permission checks.
+		// However, the kernel should still prevent this, to avoid degrading process
+		// protection for memory in other parts of the alias region.
+#ifdef __CPU_X86
+		const TUint8* KIPCAliasAddress;
+		if((MemModelAttributes()&EMemModelTypeMask) == EMemModelTypeFlexible)
+			KIPCAliasAddress = (TUint8*)0x7e000000;
+		else
+			KIPCAliasAddress = (TUint8*)0xc0400000;
+#else
+		const TUint8* KIPCAliasAddress = (TUint8*)0x00200000;
+#endif
+		DoIpcTest(KIPCAliasAddress,Disc,4,0,&KLitKernExec,ECausedException,__LINE__);
+		DoIpcTest(Disc,KIPCAliasAddress,4,0,NULL,KErrBadDescriptor,__LINE__);
+		DoIpcTest(KIPCAliasAddress,Disc,4,2,&KLitKernExec,ECausedException,__LINE__);
+		DoIpcTest(Disc,KIPCAliasAddress,4,2,NULL,KErrBadDescriptor,__LINE__);
+		}
+
+	if (HaveIPCKernProt())
+		{
+		DoIpcTest(Kern,Buffer,96,0,&KLitKernExec,ECausedException,__LINE__);
+		DoIpcTest(Buffer,Kern,96,0,NULL,KErrBadDescriptor,__LINE__);
+		TUint8* addrRW = (TUint8*)HwChunkAddr[RIpcCpy::EHwChunkSupRw];
+		if(addrRW)
+			{
+			DoIpcTest(Buffer,*(TDes8*)addrRW,96,0,NULL,KErrBadDescriptor,__LINE__);
+			DoIpcTest(Buffer,*(TDes8*)addrRW,96,2,NULL,KErrBadDescriptor,__LINE__);
+			DoIpcTest(addrRW+96,Buffer,96,0,&KLitKernExec,ECausedException,__LINE__);
+			DoIpcTest(Buffer,addrRW,96,0,NULL,KErrBadDescriptor,__LINE__);
+			DoIpcTest(addrRW+96,Buffer,96,2,&KLitKernExec,ECausedException,__LINE__);
+			DoIpcTest(Buffer,addrRW,96,2,NULL,KErrBadDescriptor,__LINE__);
+			}
+		}
+
+	if((MemModelAttributes()&EMemModelTypeMask) == EMemModelTypeMultiple
+		|| (MemModelAttributes()&EMemModelTypeMask) == EMemModelTypeFlexible
+		)
+		{
+		// On multiple memory model, test IPC to Hardware Chunks.
+		// IPC to hardware chunks not supported on Moving Memory
+		TUint8* addrRW = (TUint8*)HwChunkAddr[RIpcCpy::EHwChunkUserRw];
+		if(addrRW)
+			{
+			DoIpcTest(Buffer,*(TDes8*)addrRW,96,0,NULL,KErrNone,__LINE__);
+			DoIpcTest(Buffer,*(TDes8*)addrRW,96,2,NULL,KErrNone,__LINE__);
+			DoIpcTest(addrRW+96,Buffer,96,0,NULL,KErrNone,__LINE__);
+			DoIpcTest(Buffer,addrRW,96,0,NULL,KErrNone,__LINE__);
+			DoIpcTest(addrRW+96,Buffer,96,2,NULL,KErrNone,__LINE__);
+			DoIpcTest(Buffer,addrRW,96,2,NULL,KErrNone,__LINE__);
+			DoIpcTest(addrRW+96,addrRW,96,0,NULL,KErrNone,__LINE__);
+			DoIpcTest(addrRW+96,addrRW,96,2,NULL,KErrNone,__LINE__);
+			}
+		TUint8* addrRO = (TUint8*)HwChunkAddr[RIpcCpy::EHwChunkUserRo];
+		if(addrRO && HaveWriteProt())
+			{
+			DoIpcTest(Buffer,*(TDes8*)addrRO,96,0,NULL,KErrNone,__LINE__);
+			DoIpcTest(Buffer,*(TDes8*)addrRO,96,2,&KLitKernExec,EBadIpcDescriptor,__LINE__);
+			DoIpcTest(addrRO+96,Buffer,96,0,&KLitKernExec,ECausedException,__LINE__);
+			DoIpcTest(Buffer,addrRO,96,0,NULL,KErrNone,__LINE__);
+			DoIpcTest(addrRO+96,Buffer,96,2,NULL,KErrNone,__LINE__);
+			DoIpcTest(Buffer,addrRO,96,2,NULL,KErrBadDescriptor,__LINE__);
+			DoIpcTest(addrRW+96,addrRO,96,0,NULL,KErrNone,__LINE__);
+			DoIpcTest(addrRW+96,addrRW,96,2,NULL,KErrNone,__LINE__);
+			DoIpcTest(addrRO+96,addrRO,96,0,&KLitKernExec,ECausedException,__LINE__);
+			DoIpcTest(addrRO+96,addrRW,96,2,NULL,KErrNone,__LINE__);
+			}
+		}
+
+	c.Close();
+	}
+
+RMessage2 Msg1, Msg2;
+
+TInt SendAndExit(TAny* aPtr)
+	{
+	RLocalSession sess;
+	TInt r=sess.Connect(IpcServer,NULL);
+	if (r!=KErrNone)
+		return r;
+	sess.Test(aPtr);
+	sess.Wait();
+	sess.Close();
+	User::AfterHighRes(1000*CloseTime);
+	Msg1.Complete(0);		// complete my own message! - this removes message reference to thread
+	return 0;
+	}
+
+void TestIpcAsyncClose()
+	{
+
+	// Create a 16MB chunk
+	const TInt desSize = 8*1024*1024;
+	RChunk chunk;
+	test(chunk.CreateLocal(2 * desSize, 2 * desSize) == KErrNone);
+	test(chunk.Adjust(2 * desSize) == KErrNone);
+
+	TUint8* bigBuf=chunk.Base();
+	test(bigBuf!=NULL);
+	TUint8* bigBuf2=chunk.Base() + desSize;
+	test(bigBuf2!=NULL);
+	TPtr8 bigBufPtr(bigBuf, desSize, desSize);
+	TPtr8 bigBufPtr2(bigBuf2, 0, desSize);
+
+	if (IpcServer.Handle())
+		IpcServer.Close();
+	TESTEQ(IpcServer.CreateGlobal(KNullDesC),KErrNone);
+
+	RThread t;
+	TInt r=t.Create(KNullDesC,SendAndExit,0x1000,NULL,&bigBufPtr);
+	test(r==KErrNone);
+	TFullName fn(t.FullName());
+	TRequestStatus s;
+	t.Logon(s);
+	t.SetPriority(EPriorityMuchMore);
+	t.Resume();
+
+	IpcServer.Receive(Msg1);	// connect
+	Msg1.Complete(KErrNone);
+	IpcServer.Receive(Msg1);	// test message
+	IpcServer.Receive(Msg2);	// wait/synch message
+	TUint32 initial = User::NTickCount();
+	r=Msg1.Read(2,bigBufPtr2,0);	// arg2 is writable 8 bit descriptor
+	TUint32 final = User::NTickCount();
+	TUint32 elapsed = final - initial;
+	if (elapsed<3)
+		test.Printf(_L("*** WARNING! The big IPC only took %dms, which means the next test might fail! \n"),elapsed);
+	else
+		test.Printf(_L("Big IPC took %dms\n"),elapsed);
+	CloseTime = (TInt)(elapsed>>2);
+	Msg2.Complete(0);
+	IpcServer.Receive(Msg2);	// disconnect
+	TUint32 disconnect = User::NTickCount();
+	
+	// We expect this IPC read to fail part way through
+	r=Msg1.Read(2,bigBufPtr2,0);	// arg2 is writable 8 bit descriptor
+	test.Printf(_L("counters: initial=%d final=%d disconnect=%d current=%d\n"),initial,final,disconnect,User::NTickCount());
+	test.Printf(_L("2nd Big IPC returned %d\n"),r);
+	test(r==KErrDied);
+	test(Msg1.IsNull());
+	Msg2.Complete(0);		// complete session closure as well
+	User::WaitForRequest(s);
+	test(s==KErrNone);
+	CLOSE_AND_WAIT(t);
+	test(t.Open(fn)==KErrNotFound);
+	IpcServer.Close();
+
+	// t already closed
+//	User::Free(bigBuf);
+//	User::Free(bigBuf2);
+	chunk.Close();
+	}
+
+void BenchmarkTest()
+	{
+	TAny* bigbuf = User::Alloc(65536);
+	test(bigbuf != NULL);
+	TInt i;
+	TUint32 initial, final;
+	initial = User::NTickCount();
+	for (i=0; i<1000; ++i)
+		Ipccpy.BigWrite(bigbuf, 0);
+	final = User::NTickCount();
+	TUint32 wcal = final - initial;
+	initial = User::NTickCount();
+	for (i=0; i<1000; ++i)
+		Ipccpy.BigWrite(bigbuf, 65536);
+	final = User::NTickCount();
+	TUint32 write = final - initial;
+	test.Printf(_L("64K user->kernel copy takes %d us\n"), write - wcal);
+	initial = User::NTickCount();
+	for (i=0; i<1000; ++i)
+		Ipccpy.BigRead(bigbuf, 0);
+	final = User::NTickCount();
+	TUint32 rcal = final - initial;
+	initial = User::NTickCount();
+	for (i=0; i<1000; ++i)
+		Ipccpy.BigRead(bigbuf, 65536);
+	final = User::NTickCount();
+	TUint32 read = final - initial;
+	test.Printf(_L("64K kernel->user copy takes %d us\n"), read - rcal);
+	User::Free(bigbuf);
+//	User::After(10*1000*1000);
+	}
+
+
+RMessage2 IpcMesage;
+const TInt KTestChunkSize = 1024*1024;
+const TInt KReadSize = 4096;
+
+TInt IpcMultipleAliasesThread(TAny* aBuffer)
+	{
+	TBuf8<KReadSize> data;
+	TAny** dataStart = (TAny**)data.Ptr();
+	TAny** dataEnd = (TAny**)(data.Ptr()+KReadSize-sizeof(TAny*));
+	for(;;)
+		{
+		TInt offset;
+		for(offset=0; offset<KTestChunkSize; offset+=KReadSize)
+			{
+			TInt r = IpcMesage.Read(0,data,offset);
+			if(r!=KErrNone)
+				return r;
+			if(data.Size()!=KReadSize)
+				return 1;
+			TAny* expected = (TAny*)((TInt)aBuffer+offset);
+			if(*dataStart != expected)
+				{
+				RDebug::Printf("Offset=%x, expected %x but read %x",offset,expected,*dataStart);
+				return 2;
+				}
+			expected = (TAny*)((TInt)aBuffer+offset+KReadSize-sizeof(TAny*));
+			if(*dataEnd != expected)
+				{
+				RDebug::Printf("Offset=%x, expected %x but read %x",offset,expected,*dataEnd);
+				return 3;
+				}
+			}
+		}
+	}
+
+/*
+This tests exercises the situation where multiple threads are doing IPC simultaneousely.
+On the Multiple Memory Model, this aims to test the per-thread memory aliasing code.
+(DMemModelThread::Alias and company)
+*/
+void TestIpcMultipleThreads()
+	{
+	test.Start(_L("Test Multiple Threads IPC"));
+
+	// create chunk for threads to do IPC from...
+	RChunk chunk;
+	TESTEQ(chunk.CreateLocal(KTestChunkSize,KTestChunkSize),KErrNone);
+	TAny** buffer = (TAny**)chunk.Base();
+	TAny** bufferEnd = (TAny**)((TInt)buffer+KTestChunkSize);
+	for(; buffer<bufferEnd; ++buffer)
+		*buffer=buffer;
+
+	// create a server message which test threads can use to do IPC memory operations
+	if (IpcServer.Handle())
+		IpcServer.Close();	
+	TESTEQ(IpcServer.CreateGlobal(KNullDesC),KErrNone);
+	RLocalSession sess;
+	TRequestStatus stat;
+	TESTEQ(sess.Connect(IpcServer,&stat),KErrNone);
+	RMessage2 m;
+	IpcServer.Receive(m);
+	m.Complete(KErrNone);	// connect
+	User::WaitForRequest(stat);	// connection message report
+	TAny* ptrMem = User::Alloc(0x2000);
+	TPtr8* pptr = (TPtr8*)(((TInt)ptrMem&~0xfff)+0x1000-sizeof(TInt));
+	new (pptr) TPtr8(chunk.Base(),KTestChunkSize,KTestChunkSize); // create a TPtr8 which straddles a page boundary
+	sess.Test(pptr);
+	IpcServer.Receive(IpcMesage);
+
+	// create some test threads...
+	const TInt KNumIpcThreads = 10;
+	RThread threads[KNumIpcThreads];
+	TRequestStatus stats[KNumIpcThreads];
+	TInt i;
+	for(i=0; i<KNumIpcThreads; i++)
+		{
+		TESTEQ(threads[i].Create(KNullDesC,IpcMultipleAliasesThread,KReadSize+0x1000,&User::Allocator(),chunk.Base()),KErrNone);
+		threads[i].Logon(stats[i]);
+		}
+	test.Printf(_L("Resuming threads...\n"));
+	for(i=0; i<KNumIpcThreads; i++)
+		threads[i].Resume();
+
+	User::After(10*1000000);
+	for(i=0; i<KNumIpcThreads; i++)
+		{
+		test(stats[i]==KRequestPending); // theads should still be running
+		}
+
+	// close chunk whilst test threads are still doing IPC...
+	test.Printf(_L("Closing chunk...\n"));
+	chunk.Close();
+	for(i=0; i<KNumIpcThreads; i++)
+		{
+		User::WaitForRequest(stats[i]);
+		TInt r=stats[i].Int();
+		test.Printf(_L("Thread %d result = %d\n"),i,r);
+		test(r==KErrBadDescriptor);
+		}
+
+	IpcServer.Close();
+	User::Free(ptrMem);
+	test.End();
+	}
+
+GLDEF_C TInt E32Main()
+	{
+	MainId=TUint32(RThread().Id());
+//	RThread().SetPriority(EPriorityAbsoluteForeground);
+	test.Title();
+	test.Start(_L("Load LDD"));
+	TInt r=User::LoadLogicalDevice(_L("D_IPCCPY"));
+	test(r==KErrNone || r==KErrAlreadyExists);
+	test.Next(_L("Open channel"));
+	r=Ipccpy.Open();
+	test(r==KErrNone);
+	test.Next(_L("Allocate heap buffer"));
+	Buffer=(TUint8*)User::Alloc(4096);
+	test(Buffer!=NULL);
+	SetupAddresses();
+
+	BenchmarkTest();
+
+	TestIpcCopyErrors();
+	TestIpcAsyncClose();
+	TestIpcMultipleThreads();
+
+	FOREVER
+		{
+		TRequestStatus s;
+		Mem::Fill(Buffer,272,0xcd);
+		TPtr8 ptr(Buffer,0,272);
+		Ipccpy.IpcCpy(s,ptr);
+		User::WaitForRequest(s);
+		TInt x=s.Int();
+		if (x<0)
+			{
+			test.Printf(_L("Error %d\n"),x);
+			test(0);
+			}
+		TInt src_offset=x&3;
+		TInt dest_offset=(x>>2)&3;
+		TInt length=(x>>4)+1;
+		TInt err=-1;
+		TInt i;
+		for (i=0; i<dest_offset && err<0; ++i)
+			{
+			if (Buffer[i]!=0xcd)
+				err=i;
+			}
+		TUint8 v=(TUint8)src_offset;
+		for (i=0; i<length && err<0; ++i)
+			{
+			++v;
+			if (Buffer[i+dest_offset]!=v)
+				err=i+dest_offset;
+			}
+		for (i=dest_offset+length; i<272 && err<0; ++i)
+			{
+			if (Buffer[i]!=0xcd)
+				err=i;
+			}
+		if (err>=0)
+			{
+			test.Printf(_L("Sequence number %03x\nSrcOffset %d, DestOffset %d, Length %d\n"),x,src_offset,dest_offset,length);
+			test.Printf(_L("First error at %d"),err);
+			for (i=0; i<272; i+=16)
+				{
+				TInt j;
+				test.Printf(_L("%03x:"),i);
+				for (j=0; j<16; ++j)
+					{
+					test.Printf(_L(" %02x"),Buffer[i+j]);
+					}
+				}
+			test(0);
+			}
+		if (x==4095)
+			break;
+		}
+	Ipccpy.Close();
+	test.End();
+	return KErrNone;
+	}