NTSTATUS NTAPI ReceiveComplete
( PDEVICE_OBJECT DeviceObject,
PIRP Irp,
PVOID Context ) {
PAFD_FCB FCB = (PAFD_FCB)Context;
PLIST_ENTRY NextIrpEntry;
PIRP NextIrp;
PAFD_RECV_INFO RecvReq;
PIO_STACK_LOCATION NextIrpSp;
UNREFERENCED_PARAMETER(DeviceObject);
AFD_DbgPrint(MID_TRACE,("Called\n"));
if( !SocketAcquireStateLock( FCB ) )
return STATUS_FILE_CLOSED;
ASSERT(FCB->ReceiveIrp.InFlightRequest == Irp);
FCB->ReceiveIrp.InFlightRequest = NULL;
if( FCB->State == SOCKET_STATE_CLOSED ) {
/* Cleanup our IRP queue because the FCB is being destroyed */
while( !IsListEmpty( &FCB->PendingIrpList[FUNCTION_RECV] ) ) {
NextIrpEntry = RemoveHeadList(&FCB->PendingIrpList[FUNCTION_RECV]);
NextIrp = CONTAINING_RECORD(NextIrpEntry, IRP, Tail.Overlay.ListEntry);
NextIrpSp = IoGetCurrentIrpStackLocation(NextIrp);
RecvReq = GetLockedData(NextIrp, NextIrpSp);
NextIrp->IoStatus.Status = STATUS_FILE_CLOSED;
NextIrp->IoStatus.Information = 0;
UnlockBuffers(RecvReq->BufferArray, RecvReq->BufferCount, FALSE);
if( NextIrp->MdlAddress ) UnlockRequest( NextIrp, IoGetCurrentIrpStackLocation( NextIrp ) );
(void)IoSetCancelRoutine(NextIrp, NULL);
IoCompleteRequest( NextIrp, IO_NETWORK_INCREMENT );
}
SocketStateUnlock( FCB );
return STATUS_FILE_CLOSED;
} else if( FCB->State == SOCKET_STATE_LISTENING ) {
AFD_DbgPrint(MIN_TRACE,("!!! LISTENER GOT A RECEIVE COMPLETE !!!\n"));
SocketStateUnlock( FCB );
return STATUS_INVALID_PARAMETER;
}
HandleReceiveComplete( FCB, Irp->IoStatus.Status, Irp->IoStatus.Information );
ReceiveActivity( FCB, NULL );
SocketStateUnlock( FCB );
return STATUS_SUCCESS;
}
extern "C" void rglswDrawAnimatic(int px, int py, unsigned char* pb)
{
int y;
BYTE* pSource;
BYTE* pDest;
LockBuffers();
for (y = 0; y < 480; y++)
{
pSource = pb + 2*640*y;
pDest = scrbuf + scrpitch*(y+py) + 2*px;
MEMCPY(pDest, pSource, 2*640);
}
UnlockBuffers();
}
/*
* Backend-shutdown callback. Do cleanup that we want to be sure happens
* before all the supporting modules begin to nail their doors shut via
* their own callbacks. Note that because this has to be registered very
* late in startup, it will not get called if we suffer a failure *during*
* startup.
*
* User-level cleanup, such as temp-relation removal and UNLISTEN, happens
* via separate callbacks that execute before this one. We don't combine the
* callbacks because we still want this one to happen if the user-level
* cleanup fails.
*/
static void
ShutdownPostgres(void)
{
/*
* These operations are really just a minimal subset of
* AbortTransaction(). We don't want to do any inessential cleanup,
* since that just raises the odds of failure --- but there's some
* stuff we need to do.
*
* Release any LW locks and buffer context locks we might be holding.
* This is a kluge to improve the odds that we won't get into a
* self-made stuck-lock scenario while trying to shut down.
*/
LWLockReleaseAll();
AbortBufferIO();
UnlockBuffers();
/*
* In case a transaction is open, delete any files it created. This
* has to happen before bufmgr shutdown, so having smgr register a
* callback for it wouldn't work.
*/
smgrDoPendingDeletes(false); /* delete as though aborting xact */
}
NTSTATUS NTAPI
AfdPacketSocketReadData(PDEVICE_OBJECT DeviceObject, PIRP Irp,
PIO_STACK_LOCATION IrpSp ) {
NTSTATUS Status = STATUS_SUCCESS;
PFILE_OBJECT FileObject = IrpSp->FileObject;
PAFD_FCB FCB = FileObject->FsContext;
PAFD_RECV_INFO_UDP RecvReq;
PLIST_ENTRY ListEntry;
PAFD_STORED_DATAGRAM DatagramRecv;
KPROCESSOR_MODE LockMode;
UNREFERENCED_PARAMETER(DeviceObject);
AFD_DbgPrint(MID_TRACE,("Called on %p\n", FCB));
if( !SocketAcquireStateLock( FCB ) ) return LostSocket( Irp );
FCB->EventSelectDisabled &= ~AFD_EVENT_RECEIVE;
/* Check that the socket is bound */
if( FCB->State != SOCKET_STATE_BOUND )
{
AFD_DbgPrint(MIN_TRACE,("Invalid socket state\n"));
return UnlockAndMaybeComplete(FCB, STATUS_INVALID_PARAMETER, Irp, 0);
}
if (FCB->TdiReceiveClosed)
{
AFD_DbgPrint(MIN_TRACE,("Receive closed\n"));
return UnlockAndMaybeComplete(FCB, STATUS_FILE_CLOSED, Irp, 0);
}
if( !(RecvReq = LockRequest( Irp, IrpSp, FALSE, &LockMode )) )
return UnlockAndMaybeComplete(FCB, STATUS_NO_MEMORY, Irp, 0);
AFD_DbgPrint(MID_TRACE,("Recv flags %x\n", RecvReq->AfdFlags));
RecvReq->BufferArray = LockBuffers( RecvReq->BufferArray,
RecvReq->BufferCount,
RecvReq->Address,
RecvReq->AddressLength,
TRUE, TRUE, LockMode );
if( !RecvReq->BufferArray ) { /* access violation in userspace */
return UnlockAndMaybeComplete(FCB, STATUS_ACCESS_VIOLATION, Irp, 0);
}
if (!IsListEmpty(&FCB->DatagramList))
{
ListEntry = RemoveHeadList(&FCB->DatagramList);
DatagramRecv = CONTAINING_RECORD(ListEntry, AFD_STORED_DATAGRAM, ListEntry);
Status = SatisfyPacketRecvRequest(FCB, Irp, DatagramRecv,
(PUINT)&Irp->IoStatus.Information);
if (RecvReq->TdiFlags & TDI_RECEIVE_PEEK)
{
InsertHeadList(&FCB->DatagramList,
&DatagramRecv->ListEntry);
}
if (!IsListEmpty(&FCB->DatagramList))
{
FCB->PollState |= AFD_EVENT_RECEIVE;
FCB->PollStatus[FD_READ_BIT] = STATUS_SUCCESS;
PollReeval( FCB->DeviceExt, FCB->FileObject );
}
else
FCB->PollState &= ~AFD_EVENT_RECEIVE;
UnlockBuffers(RecvReq->BufferArray, RecvReq->BufferCount, TRUE);
return UnlockAndMaybeComplete(FCB, Status, Irp, Irp->IoStatus.Information);
}
else if (!(RecvReq->AfdFlags & AFD_OVERLAPPED) &&
((RecvReq->AfdFlags & AFD_IMMEDIATE) || (FCB->NonBlocking)))
{
AFD_DbgPrint(MID_TRACE,("Nonblocking\n"));
Status = STATUS_CANT_WAIT;
FCB->PollState &= ~AFD_EVENT_RECEIVE;
UnlockBuffers( RecvReq->BufferArray, RecvReq->BufferCount, TRUE );
return UnlockAndMaybeComplete( FCB, Status, Irp, 0 );
}
else
{
FCB->PollState &= ~AFD_EVENT_RECEIVE;
return LeaveIrpUntilLater( FCB, Irp, FUNCTION_RECV );
}
}
NTSTATUS NTAPI
AfdConnectedSocketReadData(PDEVICE_OBJECT DeviceObject, PIRP Irp,
PIO_STACK_LOCATION IrpSp, BOOLEAN Short) {
NTSTATUS Status = STATUS_INVALID_PARAMETER;
PFILE_OBJECT FileObject = IrpSp->FileObject;
PAFD_FCB FCB = FileObject->FsContext;
PAFD_RECV_INFO RecvReq;
UINT TotalBytesCopied = 0;
PAFD_STORED_DATAGRAM DatagramRecv;
PLIST_ENTRY ListEntry;
KPROCESSOR_MODE LockMode;
UNREFERENCED_PARAMETER(DeviceObject);
UNREFERENCED_PARAMETER(Short);
AFD_DbgPrint(MID_TRACE,("Called on %p\n", FCB));
if( !SocketAcquireStateLock( FCB ) ) return LostSocket( Irp );
FCB->EventSelectDisabled &= ~AFD_EVENT_RECEIVE;
if( !(FCB->Flags & AFD_ENDPOINT_CONNECTIONLESS) &&
FCB->State != SOCKET_STATE_CONNECTED &&
FCB->State != SOCKET_STATE_CONNECTING ) {
AFD_DbgPrint(MIN_TRACE,("Called recv on wrong kind of socket (s%x)\n",
FCB->State));
return UnlockAndMaybeComplete( FCB, STATUS_INVALID_PARAMETER,
Irp, 0 );
}
if( !(RecvReq = LockRequest( Irp, IrpSp, FALSE, &LockMode )) )
return UnlockAndMaybeComplete( FCB, STATUS_NO_MEMORY,
Irp, 0 );
AFD_DbgPrint(MID_TRACE,("Recv flags %x\n", RecvReq->AfdFlags));
RecvReq->BufferArray = LockBuffers( RecvReq->BufferArray,
RecvReq->BufferCount,
NULL, NULL,
TRUE, FALSE, LockMode );
if( !RecvReq->BufferArray ) {
return UnlockAndMaybeComplete( FCB, STATUS_ACCESS_VIOLATION,
Irp, 0 );
}
if( FCB->Flags & AFD_ENDPOINT_CONNECTIONLESS )
{
if (!IsListEmpty(&FCB->DatagramList))
{
ListEntry = RemoveHeadList(&FCB->DatagramList);
DatagramRecv = CONTAINING_RECORD(ListEntry, AFD_STORED_DATAGRAM, ListEntry);
Status = SatisfyPacketRecvRequest(FCB, Irp, DatagramRecv,
(PUINT)&Irp->IoStatus.Information);
if (RecvReq->TdiFlags & TDI_RECEIVE_PEEK)
{
InsertHeadList(&FCB->DatagramList,
&DatagramRecv->ListEntry);
}
if (!IsListEmpty(&FCB->DatagramList))
{
FCB->PollState |= AFD_EVENT_RECEIVE;
FCB->PollStatus[FD_READ_BIT] = STATUS_SUCCESS;
PollReeval( FCB->DeviceExt, FCB->FileObject );
}
else
FCB->PollState &= ~AFD_EVENT_RECEIVE;
UnlockBuffers(RecvReq->BufferArray, RecvReq->BufferCount, FALSE);
return UnlockAndMaybeComplete(FCB, Status, Irp, Irp->IoStatus.Information);
}
else if (!(RecvReq->AfdFlags & AFD_OVERLAPPED) &&
((RecvReq->AfdFlags & AFD_IMMEDIATE) || (FCB->NonBlocking)))
{
AFD_DbgPrint(MID_TRACE,("Nonblocking\n"));
Status = STATUS_CANT_WAIT;
FCB->PollState &= ~AFD_EVENT_RECEIVE;
UnlockBuffers( RecvReq->BufferArray, RecvReq->BufferCount, FALSE );
return UnlockAndMaybeComplete( FCB, Status, Irp, 0 );
}
else
{
FCB->PollState &= ~AFD_EVENT_RECEIVE;
return LeaveIrpUntilLater( FCB, Irp, FUNCTION_RECV );
}
}
Irp->IoStatus.Status = STATUS_PENDING;
Irp->IoStatus.Information = 0;
InsertTailList( &FCB->PendingIrpList[FUNCTION_RECV],
&Irp->Tail.Overlay.ListEntry );
/************ From this point, the IRP is not ours ************/
Status = ReceiveActivity( FCB, Irp );
if( Status == STATUS_PENDING &&
!(RecvReq->AfdFlags & AFD_OVERLAPPED) &&
((RecvReq->AfdFlags & AFD_IMMEDIATE) || (FCB->NonBlocking))) {
AFD_DbgPrint(MID_TRACE,("Nonblocking\n"));
Status = STATUS_CANT_WAIT;
TotalBytesCopied = 0;
RemoveEntryList( &Irp->Tail.Overlay.ListEntry );
UnlockBuffers( RecvReq->BufferArray, RecvReq->BufferCount, FALSE );
return UnlockAndMaybeComplete( FCB, Status, Irp,
TotalBytesCopied );
} else if( Status == STATUS_PENDING ) {
AFD_DbgPrint(MID_TRACE,("Leaving read irp\n"));
IoMarkIrpPending( Irp );
(void)IoSetCancelRoutine(Irp, AfdCancelHandler);
} else {
AFD_DbgPrint(MID_TRACE,("Completed with status %x\n", Status));
}
SocketStateUnlock( FCB );
return Status;
}
NTSTATUS NTAPI
PacketSocketRecvComplete(
PDEVICE_OBJECT DeviceObject,
PIRP Irp,
PVOID Context ) {
NTSTATUS Status = STATUS_SUCCESS;
PAFD_FCB FCB = Context;
PIRP NextIrp;
PIO_STACK_LOCATION NextIrpSp;
PLIST_ENTRY ListEntry;
PAFD_RECV_INFO RecvReq;
PAFD_STORED_DATAGRAM DatagramRecv;
UINT DGSize = Irp->IoStatus.Information + sizeof( AFD_STORED_DATAGRAM );
PLIST_ENTRY NextIrpEntry, DatagramRecvEntry;
UNREFERENCED_PARAMETER(DeviceObject);
AFD_DbgPrint(MID_TRACE,("Called on %p\n", FCB));
if( !SocketAcquireStateLock( FCB ) )
return STATUS_FILE_CLOSED;
ASSERT(FCB->ReceiveIrp.InFlightRequest == Irp);
FCB->ReceiveIrp.InFlightRequest = NULL;
if( FCB->State == SOCKET_STATE_CLOSED ) {
/* Cleanup our IRP queue because the FCB is being destroyed */
while( !IsListEmpty( &FCB->PendingIrpList[FUNCTION_RECV] ) ) {
NextIrpEntry = RemoveHeadList(&FCB->PendingIrpList[FUNCTION_RECV]);
NextIrp = CONTAINING_RECORD(NextIrpEntry, IRP, Tail.Overlay.ListEntry);
NextIrpSp = IoGetCurrentIrpStackLocation( NextIrp );
RecvReq = GetLockedData(NextIrp, NextIrpSp);
NextIrp->IoStatus.Status = STATUS_FILE_CLOSED;
NextIrp->IoStatus.Information = 0;
UnlockBuffers(RecvReq->BufferArray, RecvReq->BufferCount, CheckUnlockExtraBuffers(FCB, NextIrpSp));
if( NextIrp->MdlAddress ) UnlockRequest( NextIrp, IoGetCurrentIrpStackLocation( NextIrp ) );
(void)IoSetCancelRoutine(NextIrp, NULL);
IoCompleteRequest( NextIrp, IO_NETWORK_INCREMENT );
}
/* Free all items on the datagram list */
while( !IsListEmpty( &FCB->DatagramList ) ) {
DatagramRecvEntry = RemoveHeadList(&FCB->DatagramList);
DatagramRecv = CONTAINING_RECORD(DatagramRecvEntry, AFD_STORED_DATAGRAM, ListEntry);
ExFreePool( DatagramRecv->Address );
ExFreePool( DatagramRecv );
}
SocketStateUnlock( FCB );
return STATUS_FILE_CLOSED;
}
if (Irp->IoStatus.Status != STATUS_SUCCESS)
{
SocketStateUnlock(FCB);
return Irp->IoStatus.Status;
}
if (FCB->TdiReceiveClosed)
{
SocketStateUnlock(FCB);
return STATUS_FILE_CLOSED;
}
DatagramRecv = ExAllocatePool( NonPagedPool, DGSize );
if( DatagramRecv ) {
DatagramRecv->Len = Irp->IoStatus.Information;
RtlCopyMemory( DatagramRecv->Buffer, FCB->Recv.Window,
DatagramRecv->Len );
AFD_DbgPrint(MID_TRACE,("Received (A %p)\n",
FCB->AddressFrom->RemoteAddress));
DatagramRecv->Address =
TaCopyTransportAddress( FCB->AddressFrom->RemoteAddress );
if( !DatagramRecv->Address ) Status = STATUS_NO_MEMORY;
} else Status = STATUS_NO_MEMORY;
if( !NT_SUCCESS( Status ) ) {
if( DatagramRecv ) ExFreePool( DatagramRecv );
SocketStateUnlock( FCB );
return Status;
} else {
FCB->Recv.Content += DatagramRecv->Len;
InsertTailList( &FCB->DatagramList, &DatagramRecv->ListEntry );
}
/* Satisfy as many requests as we can */
while( !IsListEmpty( &FCB->DatagramList ) &&
!IsListEmpty( &FCB->PendingIrpList[FUNCTION_RECV] ) ) {
AFD_DbgPrint(MID_TRACE,("Looping trying to satisfy request\n"));
ListEntry = RemoveHeadList( &FCB->DatagramList );
DatagramRecv = CONTAINING_RECORD( ListEntry, AFD_STORED_DATAGRAM,
ListEntry );
ListEntry = RemoveHeadList( &FCB->PendingIrpList[FUNCTION_RECV] );
NextIrp = CONTAINING_RECORD( ListEntry, IRP, Tail.Overlay.ListEntry );
NextIrpSp = IoGetCurrentIrpStackLocation( NextIrp );
RecvReq = GetLockedData(NextIrp, NextIrpSp);
AFD_DbgPrint(MID_TRACE,("RecvReq: %p, DatagramRecv: %p\n",
RecvReq, DatagramRecv));
AFD_DbgPrint(MID_TRACE,("Satisfying\n"));
Status = SatisfyPacketRecvRequest
( FCB, NextIrp, DatagramRecv,
(PUINT)&NextIrp->IoStatus.Information );
if (RecvReq->TdiFlags & TDI_RECEIVE_PEEK)
{
InsertHeadList(&FCB->DatagramList,
&DatagramRecv->ListEntry);
}
AFD_DbgPrint(MID_TRACE,("Unlocking\n"));
UnlockBuffers( RecvReq->BufferArray, RecvReq->BufferCount, CheckUnlockExtraBuffers(FCB, NextIrpSp) );
if ( NextIrp->MdlAddress ) UnlockRequest( NextIrp, IoGetCurrentIrpStackLocation( NextIrp ) );
AFD_DbgPrint(MID_TRACE,("Completing\n"));
(void)IoSetCancelRoutine(NextIrp, NULL);
NextIrp->IoStatus.Status = Status;
IoCompleteRequest( NextIrp, IO_NETWORK_INCREMENT );
}
if( !IsListEmpty( &FCB->DatagramList ) && IsListEmpty(&FCB->PendingIrpList[FUNCTION_RECV]) ) {
AFD_DbgPrint(MID_TRACE,("Signalling\n"));
FCB->PollState |= AFD_EVENT_RECEIVE;
FCB->PollStatus[FD_READ_BIT] = STATUS_SUCCESS;
PollReeval( FCB->DeviceExt, FCB->FileObject );
} else
FCB->PollState &= ~AFD_EVENT_RECEIVE;
if( NT_SUCCESS(Irp->IoStatus.Status) ) {
/* Now relaunch the datagram request */
Status = TdiReceiveDatagram
( &FCB->ReceiveIrp.InFlightRequest,
FCB->AddressFile.Object,
0,
FCB->Recv.Window,
FCB->Recv.Size,
FCB->AddressFrom,
&FCB->ReceiveIrp.Iosb,
PacketSocketRecvComplete,
FCB );
}
SocketStateUnlock( FCB );
return STATUS_SUCCESS;
}
static NTSTATUS ReceiveActivity( PAFD_FCB FCB, PIRP Irp ) {
PLIST_ENTRY NextIrpEntry;
PIRP NextIrp;
PIO_STACK_LOCATION NextIrpSp;
PAFD_RECV_INFO RecvReq;
UINT TotalBytesCopied = 0, RetBytesCopied = 0;
NTSTATUS Status = STATUS_SUCCESS, RetStatus = STATUS_PENDING;
AFD_DbgPrint(MID_TRACE,("%p %p\n", FCB, Irp));
AFD_DbgPrint(MID_TRACE,("FCB %p Receive data waiting %u\n",
FCB, FCB->Recv.Content));
if( CantReadMore( FCB ) ) {
/* Success here means that we got an EOF. Complete a pending read
* with zero bytes if we haven't yet overread, then kill the others.
*/
while( !IsListEmpty( &FCB->PendingIrpList[FUNCTION_RECV] ) ) {
NextIrpEntry = RemoveHeadList(&FCB->PendingIrpList[FUNCTION_RECV]);
NextIrp = CONTAINING_RECORD(NextIrpEntry, IRP, Tail.Overlay.ListEntry);
NextIrpSp = IoGetCurrentIrpStackLocation( NextIrp );
RecvReq = GetLockedData(NextIrp, NextIrpSp);
AFD_DbgPrint(MID_TRACE,("Completing recv %p (%u)\n", NextIrp,
TotalBytesCopied));
UnlockBuffers( RecvReq->BufferArray,
RecvReq->BufferCount, FALSE );
if (FCB->Overread && FCB->LastReceiveStatus == STATUS_SUCCESS)
{
/* Overread after a graceful disconnect so complete with an error */
Status = STATUS_FILE_CLOSED;
}
else
{
/* Unexpected disconnect by the remote host or initial read after a graceful disconnnect */
Status = FCB->LastReceiveStatus;
}
NextIrp->IoStatus.Status = Status;
NextIrp->IoStatus.Information = 0;
if( NextIrp == Irp ) RetStatus = Status;
if( NextIrp->MdlAddress ) UnlockRequest( NextIrp, IoGetCurrentIrpStackLocation( NextIrp ) );
(void)IoSetCancelRoutine(NextIrp, NULL);
IoCompleteRequest( NextIrp, IO_NETWORK_INCREMENT );
FCB->Overread = TRUE;
}
} else {
/* Kick the user that receive would be possible now */
/* XXX Not implemented yet */
AFD_DbgPrint(MID_TRACE,("FCB %p Receive data waiting %u\n",
FCB, FCB->Recv.Content));
/*OskitDumpBuffer( FCB->Recv.Window, FCB->Recv.Content );*/
/* Try to clear some requests */
while( !IsListEmpty( &FCB->PendingIrpList[FUNCTION_RECV] ) ) {
NextIrpEntry = RemoveHeadList(&FCB->PendingIrpList[FUNCTION_RECV]);
NextIrp = CONTAINING_RECORD(NextIrpEntry, IRP, Tail.Overlay.ListEntry);
NextIrpSp = IoGetCurrentIrpStackLocation( NextIrp );
RecvReq = GetLockedData(NextIrp, NextIrpSp);
AFD_DbgPrint(MID_TRACE,("RecvReq @ %p\n", RecvReq));
Status = TryToSatisfyRecvRequestFromBuffer
( FCB, RecvReq, &TotalBytesCopied );
if( Status == STATUS_PENDING ) {
AFD_DbgPrint(MID_TRACE,("Ran out of data for %p\n", NextIrp));
InsertHeadList(&FCB->PendingIrpList[FUNCTION_RECV],
&NextIrp->Tail.Overlay.ListEntry);
break;
} else {
AFD_DbgPrint(MID_TRACE,("Completing recv %p (%u)\n", NextIrp,
TotalBytesCopied));
UnlockBuffers( RecvReq->BufferArray,
RecvReq->BufferCount, FALSE );
NextIrp->IoStatus.Status = Status;
NextIrp->IoStatus.Information = TotalBytesCopied;
if( NextIrp == Irp ) {
RetStatus = Status;
RetBytesCopied = TotalBytesCopied;
}
if( NextIrp->MdlAddress ) UnlockRequest( NextIrp, IoGetCurrentIrpStackLocation( NextIrp ) );
(void)IoSetCancelRoutine(NextIrp, NULL);
IoCompleteRequest( NextIrp, IO_NETWORK_INCREMENT );
}
}
}
if( FCB->Recv.Content - FCB->Recv.BytesUsed &&
IsListEmpty(&FCB->PendingIrpList[FUNCTION_RECV]) ) {
FCB->PollState |= AFD_EVENT_RECEIVE;
FCB->PollStatus[FD_READ_BIT] = STATUS_SUCCESS;
PollReeval( FCB->DeviceExt, FCB->FileObject );
}
else
{
FCB->PollState &= ~AFD_EVENT_RECEIVE;
}
/* Signal FD_CLOSE if no buffered data remains and the socket can't receive any more */
if (CantReadMore(FCB))
{
if (FCB->LastReceiveStatus == STATUS_SUCCESS)
{
FCB->PollState |= AFD_EVENT_DISCONNECT;
}
else
{
FCB->PollState |= AFD_EVENT_CLOSE;
}
FCB->PollStatus[FD_CLOSE_BIT] = FCB->LastReceiveStatus;
PollReeval(FCB->DeviceExt, FCB->FileObject);
}
AFD_DbgPrint(MID_TRACE,("RetStatus for irp %p is %x\n", Irp, RetStatus));
/* Sometimes we're called with a NULL Irp */
if( Irp ) {
Irp->IoStatus.Status = RetStatus;
Irp->IoStatus.Information = RetBytesCopied;
}
return RetStatus;
}
NTSTATUS NTAPI
AfdConnectedSocketWriteData(PDEVICE_OBJECT DeviceObject, PIRP Irp,
PIO_STACK_LOCATION IrpSp, BOOLEAN Short) {
NTSTATUS Status = STATUS_SUCCESS;
PFILE_OBJECT FileObject = IrpSp->FileObject;
PAFD_FCB FCB = FileObject->FsContext;
PAFD_SEND_INFO SendReq;
UINT TotalBytesCopied = 0, i, SpaceAvail = 0;
BOOLEAN NoSpace = FALSE;
AFD_DbgPrint(MID_TRACE,("Called on %x\n", FCB));
if( !SocketAcquireStateLock( FCB ) ) return LostSocket( Irp );
if( FCB->Flags & AFD_ENDPOINT_CONNECTIONLESS )
{
PAFD_SEND_INFO_UDP SendReq;
PTDI_CONNECTION_INFORMATION TargetAddress;
/* Check that the socket is bound */
if( FCB->State != SOCKET_STATE_BOUND || !FCB->RemoteAddress )
{
AFD_DbgPrint(MIN_TRACE,("Invalid parameter\n"));
return UnlockAndMaybeComplete( FCB, STATUS_INVALID_PARAMETER, Irp,
0 );
}
if( !(SendReq = LockRequest( Irp, IrpSp )) )
return UnlockAndMaybeComplete( FCB, STATUS_NO_MEMORY, Irp, 0 );
/* Must lock buffers before handing off user data */
SendReq->BufferArray = LockBuffers( SendReq->BufferArray,
SendReq->BufferCount,
NULL, NULL,
FALSE, FALSE );
if( !SendReq->BufferArray ) {
return UnlockAndMaybeComplete( FCB, STATUS_ACCESS_VIOLATION,
Irp, 0 );
}
Status = TdiBuildConnectionInfo( &TargetAddress, FCB->RemoteAddress );
if( NT_SUCCESS(Status) ) {
FCB->EventSelectDisabled &= ~AFD_EVENT_SEND;
FCB->PollState &= ~AFD_EVENT_SEND;
Status = QueueUserModeIrp(FCB, Irp, FUNCTION_SEND);
if (Status == STATUS_PENDING)
{
TdiSendDatagram(&FCB->SendIrp.InFlightRequest,
FCB->AddressFile.Object,
SendReq->BufferArray[0].buf,
SendReq->BufferArray[0].len,
TargetAddress,
&FCB->SendIrp.Iosb,
PacketSocketSendComplete,
FCB);
}
ExFreePool( TargetAddress );
SocketStateUnlock(FCB);
return STATUS_PENDING;
}
else
{
UnlockBuffers(SendReq->BufferArray, SendReq->BufferCount, FALSE);
return UnlockAndMaybeComplete( FCB, Status, Irp, 0 );
}
}
if (FCB->PollState & AFD_EVENT_CLOSE)
{
AFD_DbgPrint(MIN_TRACE,("Connection reset by remote peer\n"));
/* This is an unexpected remote disconnect */
return UnlockAndMaybeComplete(FCB, FCB->PollStatus[FD_CLOSE_BIT], Irp, 0);
}
if (FCB->PollState & AFD_EVENT_ABORT)
{
AFD_DbgPrint(MIN_TRACE,("Connection aborted\n"));
/* This is an abortive socket closure on our side */
return UnlockAndMaybeComplete(FCB, FCB->PollStatus[FD_CLOSE_BIT], Irp, 0);
}
if (FCB->SendClosed)
{
AFD_DbgPrint(MIN_TRACE,("No more sends\n"));
/* This is a graceful send closure */
return UnlockAndMaybeComplete(FCB, STATUS_FILE_CLOSED, Irp, 0);
}
if( !(SendReq = LockRequest( Irp, IrpSp )) )
return UnlockAndMaybeComplete
( FCB, STATUS_NO_MEMORY, Irp, TotalBytesCopied );
SendReq->BufferArray = LockBuffers( SendReq->BufferArray,
SendReq->BufferCount,
NULL, NULL,
FALSE, FALSE );
if( !SendReq->BufferArray ) {
return UnlockAndMaybeComplete( FCB, STATUS_ACCESS_VIOLATION,
Irp, 0 );
}
AFD_DbgPrint(MID_TRACE,("Socket state %d\n", FCB->State));
if( FCB->State != SOCKET_STATE_CONNECTED ) {
if( (SendReq->AfdFlags & AFD_IMMEDIATE) || (FCB->NonBlocking) ) {
AFD_DbgPrint(MID_TRACE,("Nonblocking\n"));
UnlockBuffers( SendReq->BufferArray, SendReq->BufferCount, FALSE );
return UnlockAndMaybeComplete
( FCB, STATUS_CANT_WAIT, Irp, 0 );
} else {
AFD_DbgPrint(MID_TRACE,("Queuing request\n"));
return LeaveIrpUntilLater( FCB, Irp, FUNCTION_SEND );
}
}
AFD_DbgPrint(MID_TRACE,("FCB->Send.BytesUsed = %d\n",
FCB->Send.BytesUsed));
SpaceAvail = FCB->Send.Size - FCB->Send.BytesUsed;
AFD_DbgPrint(MID_TRACE,("We can accept %d bytes\n",
SpaceAvail));
for( i = 0; FCB->Send.BytesUsed < FCB->Send.Size &&
i < SendReq->BufferCount; i++ ) {
if (SpaceAvail < SendReq->BufferArray[i].len)
{
if (TotalBytesCopied + SendReq->BufferArray[i].len > FCB->Send.Size)
{
UnlockBuffers( SendReq->BufferArray, SendReq->BufferCount, FALSE );
return UnlockAndMaybeComplete(FCB, STATUS_BUFFER_OVERFLOW, Irp, 0);
}
SpaceAvail += TotalBytesCopied;
NoSpace = TRUE;
break;
}
AFD_DbgPrint(MID_TRACE,("Copying Buffer %d, %x:%d to %x\n",
i,
SendReq->BufferArray[i].buf,
SendReq->BufferArray[i].len,
FCB->Send.Window + FCB->Send.BytesUsed));
RtlCopyMemory( FCB->Send.Window + FCB->Send.BytesUsed,
SendReq->BufferArray[i].buf,
SendReq->BufferArray[i].len );
TotalBytesCopied += SendReq->BufferArray[i].len;
SpaceAvail -= SendReq->BufferArray[i].len;
}
FCB->EventSelectDisabled &= ~AFD_EVENT_SEND;
if( TotalBytesCopied == 0 ) {
AFD_DbgPrint(MID_TRACE,("Empty send\n"));
UnlockBuffers( SendReq->BufferArray, SendReq->BufferCount, FALSE );
return UnlockAndMaybeComplete
( FCB, STATUS_SUCCESS, Irp, TotalBytesCopied );
}
if (SpaceAvail)
{
FCB->PollState |= AFD_EVENT_SEND;
FCB->PollStatus[FD_WRITE_BIT] = STATUS_SUCCESS;
PollReeval( FCB->DeviceExt, FCB->FileObject );
}
else
{
FCB->PollState &= ~AFD_EVENT_SEND;
}
if (!NoSpace)
{
FCB->Send.BytesUsed += TotalBytesCopied;
AFD_DbgPrint(MID_TRACE,("Copied %d bytes\n", TotalBytesCopied));
Status = QueueUserModeIrp(FCB, Irp, FUNCTION_SEND);
if (Status == STATUS_PENDING && !FCB->SendIrp.InFlightRequest)
{
TdiSend(&FCB->SendIrp.InFlightRequest,
FCB->Connection.Object,
0,
FCB->Send.Window,
FCB->Send.BytesUsed,
&FCB->SendIrp.Iosb,
SendComplete,
FCB);
}
SocketStateUnlock(FCB);
return STATUS_PENDING;
}
else
{
FCB->PollState &= ~AFD_EVENT_SEND;
if( (SendReq->AfdFlags & AFD_IMMEDIATE) || (FCB->NonBlocking) ) {
AFD_DbgPrint(MID_TRACE,("Nonblocking\n"));
UnlockBuffers( SendReq->BufferArray, SendReq->BufferCount, FALSE );
return UnlockAndMaybeComplete
( FCB, STATUS_CANT_WAIT, Irp, 0 );
} else {
AFD_DbgPrint(MID_TRACE,("Queuing request\n"));
return LeaveIrpUntilLater( FCB, Irp, FUNCTION_SEND );
}
}
}
/*
* Main entry point for bgwriter process
*
* This is invoked from BootstrapMain, which has already created the basic
* execution environment, but not enabled signals yet.
*/
void
BackgroundWriterMain(void)
{
sigjmp_buf local_sigjmp_buf;
MemoryContext bgwriter_context;
BgWriterShmem->bgwriter_pid = MyProcPid;
am_bg_writer = true;
/*
* If possible, make this process a group leader, so that the postmaster
* can signal any child processes too. (bgwriter probably never has any
* child processes, but for consistency we make all postmaster child
* processes do this.)
*/
#ifdef HAVE_SETSID
if (setsid() < 0)
elog(FATAL, "setsid() failed: %m");
#endif
/*
* Properly accept or ignore signals the postmaster might send us
*
* Note: we deliberately ignore SIGTERM, because during a standard Unix
* system shutdown cycle, init will SIGTERM all processes at once. We
* want to wait for the backends to exit, whereupon the postmaster will
* tell us it's okay to shut down (via SIGUSR2).
*
* SIGUSR1 is presently unused; keep it spare in case someday we want this
* process to participate in sinval messaging.
*/
pqsignal(SIGHUP, BgSigHupHandler); /* set flag to read config file */
pqsignal(SIGINT, ReqCheckpointHandler); /* request checkpoint */
pqsignal(SIGTERM, SIG_IGN); /* ignore SIGTERM */
pqsignal(SIGQUIT, bg_quickdie); /* hard crash time */
pqsignal(SIGALRM, SIG_IGN);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR1, SIG_IGN); /* reserve for sinval */
pqsignal(SIGUSR2, ReqShutdownHandler); /* request shutdown */
/*
* Reset some signals that are accepted by postmaster but not here
*/
pqsignal(SIGCHLD, SIG_DFL);
pqsignal(SIGTTIN, SIG_DFL);
pqsignal(SIGTTOU, SIG_DFL);
pqsignal(SIGCONT, SIG_DFL);
pqsignal(SIGWINCH, SIG_DFL);
/* We allow SIGQUIT (quickdie) at all times */
#ifdef HAVE_SIGPROCMASK
sigdelset(&BlockSig, SIGQUIT);
#else
BlockSig &= ~(sigmask(SIGQUIT));
#endif
/*
* Initialize so that first time-driven event happens at the correct time.
*/
last_checkpoint_time = last_xlog_switch_time = time(NULL);
/*
* Create a resource owner to keep track of our resources (currently only
* buffer pins).
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "Background Writer");
/*
* Create a memory context that we will do all our work in. We do this so
* that we can reset the context during error recovery and thereby avoid
* possible memory leaks. Formerly this code just ran in
* TopMemoryContext, but resetting that would be a really bad idea.
*/
bgwriter_context = AllocSetContextCreate(TopMemoryContext,
"Background Writer",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
MemoryContextSwitchTo(bgwriter_context);
/*
* If an exception is encountered, processing resumes here.
*
* See notes in postgres.c about the design of this coding.
*/
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
/* Since not using PG_TRY, must reset error stack by hand */
error_context_stack = NULL;
/* Prevent interrupts while cleaning up */
HOLD_INTERRUPTS();
/* Report the error to the server log */
EmitErrorReport();
/*
* These operations are really just a minimal subset of
* AbortTransaction(). We don't have very many resources to worry
* about in bgwriter, but we do have LWLocks, buffers, and temp files.
*/
LWLockReleaseAll();
AbortBufferIO();
UnlockBuffers();
/* buffer pins are released here: */
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
/* we needn't bother with the other ResourceOwnerRelease phases */
AtEOXact_Buffers(false);
AtEOXact_Files();
AtEOXact_HashTables(false);
/* Warn any waiting backends that the checkpoint failed. */
if (ckpt_active)
{
/* use volatile pointer to prevent code rearrangement */
volatile BgWriterShmemStruct *bgs = BgWriterShmem;
SpinLockAcquire(&bgs->ckpt_lck);
bgs->ckpt_failed++;
bgs->ckpt_done = bgs->ckpt_started;
SpinLockRelease(&bgs->ckpt_lck);
ckpt_active = false;
}
/*
* Now return to normal top-level context and clear ErrorContext for
* next time.
*/
MemoryContextSwitchTo(bgwriter_context);
FlushErrorState();
/* Flush any leaked data in the top-level context */
MemoryContextResetAndDeleteChildren(bgwriter_context);
/* Now we can allow interrupts again */
RESUME_INTERRUPTS();
/*
* Sleep at least 1 second after any error. A write error is likely
* to be repeated, and we don't want to be filling the error logs as
* fast as we can.
*/
pg_usleep(1000000L);
/*
* Close all open files after any error. This is helpful on Windows,
* where holding deleted files open causes various strange errors.
* It's not clear we need it elsewhere, but shouldn't hurt.
*/
smgrcloseall();
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
/*
* Unblock signals (they were blocked when the postmaster forked us)
*/
PG_SETMASK(&UnBlockSig);
/*
* Loop forever
*/
for (;;)
{
bool do_checkpoint = false;
int flags = 0;
time_t now;
int elapsed_secs;
/*
* Emergency bailout if postmaster has died. This is to avoid the
* necessity for manual cleanup of all postmaster children.
*/
if (!PostmasterIsAlive(true))
exit(1);
/*
* Process any requests or signals received recently.
*/
AbsorbFsyncRequests();
if (got_SIGHUP)
{
got_SIGHUP = false;
ProcessConfigFile(PGC_SIGHUP);
}
if (checkpoint_requested)
{
checkpoint_requested = false;
do_checkpoint = true;
BgWriterStats.m_requested_checkpoints++;
}
if (shutdown_requested)
{
/*
* From here on, elog(ERROR) should end with exit(1), not send
* control back to the sigsetjmp block above
*/
ExitOnAnyError = true;
/* Close down the database */
ShutdownXLOG(0, 0);
DumpFreeSpaceMap(0, 0);
/* Normal exit from the bgwriter is here */
proc_exit(0); /* done */
}
/*
* Force a checkpoint if too much time has elapsed since the last one.
* Note that we count a timed checkpoint in stats only when this
* occurs without an external request, but we set the CAUSE_TIME flag
* bit even if there is also an external request.
*/
now = time(NULL);
elapsed_secs = now - last_checkpoint_time;
if (elapsed_secs >= CheckPointTimeout)
{
if (!do_checkpoint)
BgWriterStats.m_timed_checkpoints++;
do_checkpoint = true;
flags |= CHECKPOINT_CAUSE_TIME;
}
/*
* Do a checkpoint if requested, otherwise do one cycle of
* dirty-buffer writing.
*/
if (do_checkpoint)
{
/* use volatile pointer to prevent code rearrangement */
volatile BgWriterShmemStruct *bgs = BgWriterShmem;
/*
* Atomically fetch the request flags to figure out what kind of a
* checkpoint we should perform, and increase the started-counter
* to acknowledge that we've started a new checkpoint.
*/
SpinLockAcquire(&bgs->ckpt_lck);
flags |= bgs->ckpt_flags;
bgs->ckpt_flags = 0;
bgs->ckpt_started++;
SpinLockRelease(&bgs->ckpt_lck);
/*
* We will warn if (a) too soon since last checkpoint (whatever
* caused it) and (b) somebody set the CHECKPOINT_CAUSE_XLOG flag
* since the last checkpoint start. Note in particular that this
* implementation will not generate warnings caused by
* CheckPointTimeout < CheckPointWarning.
*/
if ((flags & CHECKPOINT_CAUSE_XLOG) &&
elapsed_secs < CheckPointWarning)
ereport(LOG,
(errmsg("checkpoints are occurring too frequently (%d seconds apart)",
elapsed_secs),
errhint("Consider increasing the configuration parameter \"checkpoint_segments\".")));
/*
* Initialize bgwriter-private variables used during checkpoint.
*/
ckpt_active = true;
ckpt_start_recptr = GetInsertRecPtr();
ckpt_start_time = now;
ckpt_cached_elapsed = 0;
/*
* Do the checkpoint.
*/
CreateCheckPoint(flags);
/*
* After any checkpoint, close all smgr files. This is so we
* won't hang onto smgr references to deleted files indefinitely.
*/
smgrcloseall();
/*
* Indicate checkpoint completion to any waiting backends.
*/
SpinLockAcquire(&bgs->ckpt_lck);
bgs->ckpt_done = bgs->ckpt_started;
SpinLockRelease(&bgs->ckpt_lck);
ckpt_active = false;
/*
* Note we record the checkpoint start time not end time as
* last_checkpoint_time. This is so that time-driven checkpoints
* happen at a predictable spacing.
*/
last_checkpoint_time = now;
}
else
BgBufferSync();
/* Check for archive_timeout and switch xlog files if necessary. */
CheckArchiveTimeout();
/* Nap for the configured time. */
BgWriterNap();
}
}
/*
* Main entry point for bgwriter process
*
* This is invoked from BootstrapMain, which has already created the basic
* execution environment, but not enabled signals yet.
*/
void
BackgroundWriterMain(void)
{
sigjmp_buf local_sigjmp_buf;
MemoryContext bgwriter_context;
am_bg_writer = true;
/*
* If possible, make this process a group leader, so that the postmaster
* can signal any child processes too. (bgwriter probably never has any
* child processes, but for consistency we make all postmaster child
* processes do this.)
*/
#ifdef HAVE_SETSID
if (setsid() < 0)
elog(FATAL, "setsid() failed: %m");
#endif
/*
* Properly accept or ignore signals the postmaster might send us
*
* SIGUSR1 is presently unused; keep it spare in case someday we want this
* process to participate in ProcSignal signalling.
*/
pqsignal(SIGHUP, BgSigHupHandler); /* set flag to read config file */
pqsignal(SIGINT, SIG_IGN); /* as of 9.2 no longer requests checkpoint */
pqsignal(SIGTERM, ReqShutdownHandler); /* shutdown */
pqsignal(SIGQUIT, bg_quickdie); /* hard crash time */
pqsignal(SIGALRM, SIG_IGN);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR1, SIG_IGN); /* reserve for ProcSignal */
pqsignal(SIGUSR2, SIG_IGN);
/*
* Reset some signals that are accepted by postmaster but not here
*/
pqsignal(SIGCHLD, SIG_DFL);
pqsignal(SIGTTIN, SIG_DFL);
pqsignal(SIGTTOU, SIG_DFL);
pqsignal(SIGCONT, SIG_DFL);
pqsignal(SIGWINCH, SIG_DFL);
/* We allow SIGQUIT (quickdie) at all times */
sigdelset(&BlockSig, SIGQUIT);
/*
* Create a resource owner to keep track of our resources (currently only
* buffer pins).
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "Background Writer");
/*
* Create a memory context that we will do all our work in. We do this so
* that we can reset the context during error recovery and thereby avoid
* possible memory leaks. Formerly this code just ran in
* TopMemoryContext, but resetting that would be a really bad idea.
*/
bgwriter_context = AllocSetContextCreate(TopMemoryContext,
"Background Writer",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
MemoryContextSwitchTo(bgwriter_context);
/*
* If an exception is encountered, processing resumes here.
*
* See notes in postgres.c about the design of this coding.
*/
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
/* Since not using PG_TRY, must reset error stack by hand */
error_context_stack = NULL;
/* Prevent interrupts while cleaning up */
HOLD_INTERRUPTS();
/* Report the error to the server log */
EmitErrorReport();
/*
* These operations are really just a minimal subset of
* AbortTransaction(). We don't have very many resources to worry
* about in bgwriter, but we do have LWLocks, buffers, and temp files.
*/
LWLockReleaseAll();
AbortBufferIO();
UnlockBuffers();
/* buffer pins are released here: */
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
/* we needn't bother with the other ResourceOwnerRelease phases */
AtEOXact_Buffers(false);
AtEOXact_Files();
AtEOXact_HashTables(false);
/*
* Now return to normal top-level context and clear ErrorContext for
* next time.
*/
MemoryContextSwitchTo(bgwriter_context);
FlushErrorState();
/* Flush any leaked data in the top-level context */
MemoryContextResetAndDeleteChildren(bgwriter_context);
/* Now we can allow interrupts again */
RESUME_INTERRUPTS();
/*
* Sleep at least 1 second after any error. A write error is likely
* to be repeated, and we don't want to be filling the error logs as
* fast as we can.
*/
pg_usleep(1000000L);
/*
* Close all open files after any error. This is helpful on Windows,
* where holding deleted files open causes various strange errors.
* It's not clear we need it elsewhere, but shouldn't hurt.
*/
smgrcloseall();
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
/*
* Unblock signals (they were blocked when the postmaster forked us)
*/
PG_SETMASK(&UnBlockSig);
/*
* Use the recovery target timeline ID during recovery
*/
if (RecoveryInProgress())
ThisTimeLineID = GetRecoveryTargetTLI();
/*
* Loop forever
*/
for (;;)
{
/*
* Emergency bailout if postmaster has died. This is to avoid the
* necessity for manual cleanup of all postmaster children.
*/
if (!PostmasterIsAlive())
exit(1);
if (got_SIGHUP)
{
got_SIGHUP = false;
ProcessConfigFile(PGC_SIGHUP);
/* update global shmem state for sync rep */
}
if (shutdown_requested)
{
/*
* From here on, elog(ERROR) should end with exit(1), not send
* control back to the sigsetjmp block above
*/
ExitOnAnyError = true;
/* Normal exit from the bgwriter is here */
proc_exit(0); /* done */
}
/*
* Do one cycle of dirty-buffer writing.
*/
BgBufferSync();
/* Nap for the configured time. */
BgWriterNap();
}
}
static NTSTATUS NTAPI SendComplete
( PDEVICE_OBJECT DeviceObject,
PIRP Irp,
PVOID Context ) {
NTSTATUS Status = Irp->IoStatus.Status;
PAFD_FCB FCB = (PAFD_FCB)Context;
PLIST_ENTRY NextIrpEntry;
PIRP NextIrp = NULL;
PIO_STACK_LOCATION NextIrpSp;
PAFD_SEND_INFO SendReq = NULL;
PAFD_MAPBUF Map;
UINT TotalBytesCopied = 0, TotalBytesProcessed = 0, SpaceAvail, i;
/*
* The Irp parameter passed in is the IRP of the stream between AFD and
* TDI driver. It's not very usefull to us. We need the IRPs of the stream
* between usermode and AFD. Those are chained from
* FCB->PendingIrpList[FUNCTION_SEND] and you'll see them in the code
* below as "NextIrp" ('cause they are the next usermode IRP to be
* processed).
*/
AFD_DbgPrint(MID_TRACE,("Called, status %x, %d bytes used\n",
Irp->IoStatus.Status,
Irp->IoStatus.Information));
if( !SocketAcquireStateLock( FCB ) )
return STATUS_FILE_CLOSED;
ASSERT(FCB->SendIrp.InFlightRequest == Irp);
FCB->SendIrp.InFlightRequest = NULL;
/* Request is not in flight any longer */
if( FCB->State == SOCKET_STATE_CLOSED ) {
/* Cleanup our IRP queue because the FCB is being destroyed */
while( !IsListEmpty( &FCB->PendingIrpList[FUNCTION_SEND] ) ) {
NextIrpEntry = RemoveHeadList(&FCB->PendingIrpList[FUNCTION_SEND]);
NextIrp = CONTAINING_RECORD(NextIrpEntry, IRP, Tail.Overlay.ListEntry);
NextIrpSp = IoGetCurrentIrpStackLocation( NextIrp );
SendReq = GetLockedData(NextIrp, NextIrpSp);
NextIrp->IoStatus.Status = STATUS_FILE_CLOSED;
NextIrp->IoStatus.Information = 0;
UnlockBuffers(SendReq->BufferArray, SendReq->BufferCount, FALSE);
if( NextIrp->MdlAddress ) UnlockRequest( NextIrp, IoGetCurrentIrpStackLocation( NextIrp ) );
(void)IoSetCancelRoutine(NextIrp, NULL);
IoCompleteRequest( NextIrp, IO_NETWORK_INCREMENT );
}
RetryDisconnectCompletion(FCB);
SocketStateUnlock( FCB );
return STATUS_FILE_CLOSED;
}
if( !NT_SUCCESS(Status) ) {
/* Complete all following send IRPs with error */
while( !IsListEmpty( &FCB->PendingIrpList[FUNCTION_SEND] ) ) {
NextIrpEntry =
RemoveHeadList(&FCB->PendingIrpList[FUNCTION_SEND]);
NextIrp =
CONTAINING_RECORD(NextIrpEntry, IRP, Tail.Overlay.ListEntry);
NextIrpSp = IoGetCurrentIrpStackLocation( NextIrp );
SendReq = GetLockedData(NextIrp, NextIrpSp);
UnlockBuffers( SendReq->BufferArray,
SendReq->BufferCount,
FALSE );
NextIrp->IoStatus.Status = Status;
NextIrp->IoStatus.Information = 0;
if ( NextIrp->MdlAddress ) UnlockRequest( NextIrp, IoGetCurrentIrpStackLocation( NextIrp ) );
(void)IoSetCancelRoutine(NextIrp, NULL);
IoCompleteRequest( NextIrp, IO_NETWORK_INCREMENT );
}
RetryDisconnectCompletion(FCB);
SocketStateUnlock( FCB );
return STATUS_SUCCESS;
}
RtlMoveMemory( FCB->Send.Window,
FCB->Send.Window + FCB->Send.BytesUsed,
FCB->Send.BytesUsed - Irp->IoStatus.Information );
TotalBytesProcessed = 0;
while (!IsListEmpty(&FCB->PendingIrpList[FUNCTION_SEND]) &&
TotalBytesProcessed != Irp->IoStatus.Information) {
NextIrpEntry =
RemoveHeadList(&FCB->PendingIrpList[FUNCTION_SEND]);
NextIrp =
CONTAINING_RECORD(NextIrpEntry, IRP, Tail.Overlay.ListEntry);
NextIrpSp = IoGetCurrentIrpStackLocation( NextIrp );
SendReq = GetLockedData(NextIrp, NextIrpSp);
Map = (PAFD_MAPBUF)(SendReq->BufferArray + SendReq->BufferCount);
TotalBytesCopied = 0;
for( i = 0; i < SendReq->BufferCount; i++ )
TotalBytesCopied += SendReq->BufferArray[i].len;
NextIrp->IoStatus.Status = Irp->IoStatus.Status;
NextIrp->IoStatus.Information = TotalBytesCopied;
TotalBytesProcessed += TotalBytesCopied;
(void)IoSetCancelRoutine(NextIrp, NULL);
UnlockBuffers( SendReq->BufferArray,
SendReq->BufferCount,
FALSE );
if (NextIrp->MdlAddress) UnlockRequest(NextIrp, NextIrpSp);
IoCompleteRequest(NextIrp, IO_NETWORK_INCREMENT);
}
ASSERT(TotalBytesProcessed == Irp->IoStatus.Information);
FCB->Send.BytesUsed -= TotalBytesProcessed;
while( !IsListEmpty( &FCB->PendingIrpList[FUNCTION_SEND] ) ) {
NextIrpEntry =
RemoveHeadList(&FCB->PendingIrpList[FUNCTION_SEND]);
NextIrp =
CONTAINING_RECORD(NextIrpEntry, IRP, Tail.Overlay.ListEntry);
NextIrpSp = IoGetCurrentIrpStackLocation( NextIrp );
SendReq = GetLockedData(NextIrp, NextIrpSp);
Map = (PAFD_MAPBUF)(SendReq->BufferArray + SendReq->BufferCount);
AFD_DbgPrint(MID_TRACE,("SendReq @ %x\n", SendReq));
SpaceAvail = FCB->Send.Size - FCB->Send.BytesUsed;
TotalBytesCopied = 0;
for( i = 0; i < SendReq->BufferCount; i++ ) {
if (SpaceAvail < SendReq->BufferArray[i].len)
{
InsertHeadList(&FCB->PendingIrpList[FUNCTION_SEND],
&NextIrp->Tail.Overlay.ListEntry);
NextIrp = NULL;
break;
}
Map[i].BufferAddress =
MmMapLockedPages( Map[i].Mdl, KernelMode );
RtlCopyMemory( FCB->Send.Window + FCB->Send.BytesUsed,
Map[i].BufferAddress,
SendReq->BufferArray[i].len );
MmUnmapLockedPages( Map[i].BufferAddress, Map[i].Mdl );
TotalBytesCopied += SendReq->BufferArray[i].len;
SpaceAvail -= SendReq->BufferArray[i].len;
}
if (NextIrp != NULL)
{
FCB->Send.BytesUsed += TotalBytesCopied;
}
else
break;
}
if (FCB->Send.Size - FCB->Send.BytesUsed != 0 &&
!FCB->SendClosed)
{
FCB->PollState |= AFD_EVENT_SEND;
FCB->PollStatus[FD_WRITE_BIT] = STATUS_SUCCESS;
PollReeval( FCB->DeviceExt, FCB->FileObject );
}
else
{
FCB->PollState &= ~AFD_EVENT_SEND;
}
/* Some data is still waiting */
if( FCB->Send.BytesUsed )
{
Status = TdiSend( &FCB->SendIrp.InFlightRequest,
FCB->Connection.Object,
0,
FCB->Send.Window,
FCB->Send.BytesUsed,
&FCB->SendIrp.Iosb,
SendComplete,
FCB );
}
else
{
/* Nothing is waiting so try to complete a pending disconnect */
RetryDisconnectCompletion(FCB);
}
SocketStateUnlock( FCB );
return STATUS_SUCCESS;
}
/*
* Main entry point for walwriter process
*
* This is invoked from BootstrapMain, which has already created the basic
* execution environment, but not enabled signals yet.
*/
void
WalWriterMain(void)
{
sigjmp_buf local_sigjmp_buf;
MemoryContext walwriter_context;
/*
* If possible, make this process a group leader, so that the postmaster
* can signal any child processes too. (walwriter probably never has any
* child processes, but for consistency we make all postmaster child
* processes do this.)
*/
#ifdef HAVE_SETSID
if (setsid() < 0)
elog(FATAL, "setsid() failed: %m");
#endif
/*
* Properly accept or ignore signals the postmaster might send us
*
* We have no particular use for SIGINT at the moment, but seems
* reasonable to treat like SIGTERM.
*/
pqsignal(SIGHUP, WalSigHupHandler); /* set flag to read config file */
pqsignal(SIGINT, WalShutdownHandler); /* request shutdown */
pqsignal(SIGTERM, WalShutdownHandler); /* request shutdown */
pqsignal(SIGQUIT, wal_quickdie); /* hard crash time */
pqsignal(SIGALRM, SIG_IGN);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR1, SIG_IGN); /* reserve for ProcSignal */
pqsignal(SIGUSR2, SIG_IGN); /* not used */
/*
* Reset some signals that are accepted by postmaster but not here
*/
pqsignal(SIGCHLD, SIG_DFL);
pqsignal(SIGTTIN, SIG_DFL);
pqsignal(SIGTTOU, SIG_DFL);
pqsignal(SIGCONT, SIG_DFL);
pqsignal(SIGWINCH, SIG_DFL);
/* We allow SIGQUIT (quickdie) at all times */
sigdelset(&BlockSig, SIGQUIT);
/*
* Create a resource owner to keep track of our resources (not clear that
* we need this, but may as well have one).
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "Wal Writer");
/*
* Create a memory context that we will do all our work in. We do this so
* that we can reset the context during error recovery and thereby avoid
* possible memory leaks. Formerly this code just ran in
* TopMemoryContext, but resetting that would be a really bad idea.
*/
walwriter_context = AllocSetContextCreate(TopMemoryContext,
"Wal Writer",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
MemoryContextSwitchTo(walwriter_context);
/*
* If an exception is encountered, processing resumes here.
*
* This code is heavily based on bgwriter.c, q.v.
*/
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
/* Since not using PG_TRY, must reset error stack by hand */
error_context_stack = NULL;
/* Prevent interrupts while cleaning up */
HOLD_INTERRUPTS();
/* Report the error to the server log */
EmitErrorReport();
/*
* These operations are really just a minimal subset of
* AbortTransaction(). We don't have very many resources to worry
* about in walwriter, but we do have LWLocks, and perhaps buffers?
*/
LWLockReleaseAll();
AbortBufferIO();
UnlockBuffers();
/* buffer pins are released here: */
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
/* we needn't bother with the other ResourceOwnerRelease phases */
AtEOXact_Buffers(false);
AtEOXact_Files();
AtEOXact_HashTables(false);
/*
* Now return to normal top-level context and clear ErrorContext for
* next time.
*/
MemoryContextSwitchTo(walwriter_context);
FlushErrorState();
/* Flush any leaked data in the top-level context */
MemoryContextResetAndDeleteChildren(walwriter_context);
/* Now we can allow interrupts again */
RESUME_INTERRUPTS();
/*
* Sleep at least 1 second after any error. A write error is likely
* to be repeated, and we don't want to be filling the error logs as
* fast as we can.
*/
pg_usleep(1000000L);
/*
* Close all open files after any error. This is helpful on Windows,
* where holding deleted files open causes various strange errors.
* It's not clear we need it elsewhere, but shouldn't hurt.
*/
smgrcloseall();
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
/*
* Unblock signals (they were blocked when the postmaster forked us)
*/
PG_SETMASK(&UnBlockSig);
/*
* Loop forever
*/
for (;;)
{
long udelay;
/*
* Emergency bailout if postmaster has died. This is to avoid the
* necessity for manual cleanup of all postmaster children.
*/
if (!PostmasterIsAlive(true))
exit(1);
/*
* Process any requests or signals received recently.
*/
if (got_SIGHUP)
{
got_SIGHUP = false;
ProcessConfigFile(PGC_SIGHUP);
}
if (shutdown_requested)
{
/* Normal exit from the walwriter is here */
proc_exit(0); /* done */
}
/*
* Do what we're here for...
*/
XLogBackgroundFlush();
/*
* Delay until time to do something more, but fall out of delay
* reasonably quickly if signaled.
*/
udelay = WalWriterDelay * 1000L;
while (udelay > 999999L)
{
if (got_SIGHUP || shutdown_requested)
break;
pg_usleep(1000000L);
udelay -= 1000000L;
}
if (!(got_SIGHUP || shutdown_requested))
pg_usleep(udelay);
}
}
NTSTATUS NTAPI
AfdPacketSocketWriteData(PDEVICE_OBJECT DeviceObject, PIRP Irp,
PIO_STACK_LOCATION IrpSp) {
NTSTATUS Status = STATUS_SUCCESS;
PTDI_CONNECTION_INFORMATION TargetAddress;
PFILE_OBJECT FileObject = IrpSp->FileObject;
PAFD_FCB FCB = FileObject->FsContext;
PAFD_SEND_INFO_UDP SendReq;
KPROCESSOR_MODE LockMode;
UNREFERENCED_PARAMETER(DeviceObject);
AFD_DbgPrint(MID_TRACE,("Called on %p\n", FCB));
if( !SocketAcquireStateLock( FCB ) ) return LostSocket( Irp );
FCB->EventSelectDisabled &= ~AFD_EVENT_SEND;
/* Check that the socket is bound */
if( FCB->State != SOCKET_STATE_BOUND &&
FCB->State != SOCKET_STATE_CREATED)
{
AFD_DbgPrint(MIN_TRACE,("Invalid socket state\n"));
return UnlockAndMaybeComplete(FCB, STATUS_INVALID_PARAMETER, Irp, 0);
}
if (FCB->SendClosed)
{
AFD_DbgPrint(MIN_TRACE,("No more sends\n"));
return UnlockAndMaybeComplete(FCB, STATUS_FILE_CLOSED, Irp, 0);
}
if( !(SendReq = LockRequest( Irp, IrpSp, FALSE, &LockMode )) )
return UnlockAndMaybeComplete(FCB, STATUS_NO_MEMORY, Irp, 0);
if (FCB->State == SOCKET_STATE_CREATED)
{
if( FCB->LocalAddress ) ExFreePool( FCB->LocalAddress );
FCB->LocalAddress =
TaBuildNullTransportAddress( ((PTRANSPORT_ADDRESS)SendReq->TdiConnection.RemoteAddress)->
Address[0].AddressType );
if( FCB->LocalAddress ) {
Status = WarmSocketForBind( FCB, AFD_SHARE_WILDCARD );
if( NT_SUCCESS(Status) )
FCB->State = SOCKET_STATE_BOUND;
else
return UnlockAndMaybeComplete( FCB, Status, Irp, 0 );
} else
return UnlockAndMaybeComplete
( FCB, STATUS_NO_MEMORY, Irp, 0 );
}
SendReq->BufferArray = LockBuffers( SendReq->BufferArray,
SendReq->BufferCount,
NULL, NULL,
FALSE, FALSE, LockMode );
if( !SendReq->BufferArray )
return UnlockAndMaybeComplete( FCB, STATUS_ACCESS_VIOLATION,
Irp, 0 );
AFD_DbgPrint
(MID_TRACE,("RemoteAddress #%d Type %u\n",
((PTRANSPORT_ADDRESS)SendReq->TdiConnection.RemoteAddress)->
TAAddressCount,
((PTRANSPORT_ADDRESS)SendReq->TdiConnection.RemoteAddress)->
Address[0].AddressType));
Status = TdiBuildConnectionInfo( &TargetAddress,
((PTRANSPORT_ADDRESS)SendReq->TdiConnection.RemoteAddress) );
/* Check the size of the Address given ... */
if( NT_SUCCESS(Status) ) {
FCB->PollState &= ~AFD_EVENT_SEND;
Status = QueueUserModeIrp(FCB, Irp, FUNCTION_SEND);
if (Status == STATUS_PENDING)
{
TdiSendDatagram(&FCB->SendIrp.InFlightRequest,
FCB->AddressFile.Object,
SendReq->BufferArray[0].buf,
SendReq->BufferArray[0].len,
TargetAddress,
PacketSocketSendComplete,
FCB);
}
ExFreePool(TargetAddress);
SocketStateUnlock(FCB);
return STATUS_PENDING;
}
else
{
UnlockBuffers(SendReq->BufferArray, SendReq->BufferCount, FALSE);
return UnlockAndMaybeComplete( FCB, Status, Irp, 0 );
}
}
/*
* Main entry point for checkpointer process
*
* This is invoked from AuxiliaryProcessMain, which has already created the
* basic execution environment, but not enabled signals yet.
*/
void
CheckpointerMain(void)
{
sigjmp_buf local_sigjmp_buf;
MemoryContext checkpointer_context;
CheckpointerShmem->checkpointer_pid = MyProcPid;
/*
* Properly accept or ignore signals the postmaster might send us
*
* Note: we deliberately ignore SIGTERM, because during a standard Unix
* system shutdown cycle, init will SIGTERM all processes at once. We
* want to wait for the backends to exit, whereupon the postmaster will
* tell us it's okay to shut down (via SIGUSR2).
*/
pqsignal(SIGHUP, ChkptSigHupHandler); /* set flag to read config
* file */
pqsignal(SIGINT, ReqCheckpointHandler); /* request checkpoint */
pqsignal(SIGTERM, SIG_IGN); /* ignore SIGTERM */
pqsignal(SIGQUIT, chkpt_quickdie); /* hard crash time */
pqsignal(SIGALRM, SIG_IGN);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR1, chkpt_sigusr1_handler);
pqsignal(SIGUSR2, ReqShutdownHandler); /* request shutdown */
/*
* Reset some signals that are accepted by postmaster but not here
*/
pqsignal(SIGCHLD, SIG_DFL);
pqsignal(SIGTTIN, SIG_DFL);
pqsignal(SIGTTOU, SIG_DFL);
pqsignal(SIGCONT, SIG_DFL);
pqsignal(SIGWINCH, SIG_DFL);
/* We allow SIGQUIT (quickdie) at all times */
sigdelset(&BlockSig, SIGQUIT);
/*
* Initialize so that first time-driven event happens at the correct time.
*/
last_checkpoint_time = last_xlog_switch_time = (pg_time_t) time(NULL);
/*
* Create a resource owner to keep track of our resources (currently only
* buffer pins).
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "Checkpointer");
/*
* Create a memory context that we will do all our work in. We do this so
* that we can reset the context during error recovery and thereby avoid
* possible memory leaks. Formerly this code just ran in
* TopMemoryContext, but resetting that would be a really bad idea.
*/
checkpointer_context = AllocSetContextCreate(TopMemoryContext,
"Checkpointer",
ALLOCSET_DEFAULT_SIZES);
MemoryContextSwitchTo(checkpointer_context);
/*
* If an exception is encountered, processing resumes here.
*
* See notes in postgres.c about the design of this coding.
*/
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
/* Since not using PG_TRY, must reset error stack by hand */
error_context_stack = NULL;
/* Prevent interrupts while cleaning up */
HOLD_INTERRUPTS();
/* Report the error to the server log */
EmitErrorReport();
/*
* These operations are really just a minimal subset of
* AbortTransaction(). We don't have very many resources to worry
* about in checkpointer, but we do have LWLocks, buffers, and temp
* files.
*/
LWLockReleaseAll();
ConditionVariableCancelSleep();
pgstat_report_wait_end();
AbortBufferIO();
UnlockBuffers();
/* buffer pins are released here: */
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
/* we needn't bother with the other ResourceOwnerRelease phases */
AtEOXact_Buffers(false);
AtEOXact_SMgr();
AtEOXact_Files();
AtEOXact_HashTables(false);
/* Warn any waiting backends that the checkpoint failed. */
if (ckpt_active)
{
SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
CheckpointerShmem->ckpt_failed++;
CheckpointerShmem->ckpt_done = CheckpointerShmem->ckpt_started;
SpinLockRelease(&CheckpointerShmem->ckpt_lck);
ckpt_active = false;
}
/*
* Now return to normal top-level context and clear ErrorContext for
* next time.
*/
MemoryContextSwitchTo(checkpointer_context);
FlushErrorState();
/* Flush any leaked data in the top-level context */
MemoryContextResetAndDeleteChildren(checkpointer_context);
/* Now we can allow interrupts again */
RESUME_INTERRUPTS();
/*
* Sleep at least 1 second after any error. A write error is likely
* to be repeated, and we don't want to be filling the error logs as
* fast as we can.
*/
pg_usleep(1000000L);
/*
* Close all open files after any error. This is helpful on Windows,
* where holding deleted files open causes various strange errors.
* It's not clear we need it elsewhere, but shouldn't hurt.
*/
smgrcloseall();
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
/*
* Unblock signals (they were blocked when the postmaster forked us)
*/
PG_SETMASK(&UnBlockSig);
/*
* Ensure all shared memory values are set correctly for the config. Doing
* this here ensures no race conditions from other concurrent updaters.
*/
UpdateSharedMemoryConfig();
/*
* Advertise our latch that backends can use to wake us up while we're
* sleeping.
*/
ProcGlobal->checkpointerLatch = &MyProc->procLatch;
/*
* Loop forever
*/
for (;;)
{
bool do_checkpoint = false;
int flags = 0;
pg_time_t now;
int elapsed_secs;
int cur_timeout;
int rc;
/* Clear any already-pending wakeups */
ResetLatch(MyLatch);
/*
* Process any requests or signals received recently.
*/
AbsorbFsyncRequests();
if (got_SIGHUP)
{
got_SIGHUP = false;
ProcessConfigFile(PGC_SIGHUP);
/*
* Checkpointer is the last process to shut down, so we ask it to
* hold the keys for a range of other tasks required most of which
* have nothing to do with checkpointing at all.
*
* For various reasons, some config values can change dynamically
* so the primary copy of them is held in shared memory to make
* sure all backends see the same value. We make Checkpointer
* responsible for updating the shared memory copy if the
* parameter setting changes because of SIGHUP.
*/
UpdateSharedMemoryConfig();
}
if (checkpoint_requested)
{
checkpoint_requested = false;
do_checkpoint = true;
BgWriterStats.m_requested_checkpoints++;
}
if (shutdown_requested)
{
/*
* From here on, elog(ERROR) should end with exit(1), not send
* control back to the sigsetjmp block above
*/
ExitOnAnyError = true;
/* Close down the database */
ShutdownXLOG(0, 0);
/* Normal exit from the checkpointer is here */
proc_exit(0); /* done */
}
/*
* Force a checkpoint if too much time has elapsed since the last one.
* Note that we count a timed checkpoint in stats only when this
* occurs without an external request, but we set the CAUSE_TIME flag
* bit even if there is also an external request.
*/
now = (pg_time_t) time(NULL);
elapsed_secs = now - last_checkpoint_time;
if (elapsed_secs >= CheckPointTimeout)
{
if (!do_checkpoint)
BgWriterStats.m_timed_checkpoints++;
do_checkpoint = true;
flags |= CHECKPOINT_CAUSE_TIME;
}
/*
* Do a checkpoint if requested.
*/
if (do_checkpoint)
{
bool ckpt_performed = false;
bool do_restartpoint;
/*
* Check if we should perform a checkpoint or a restartpoint. As a
* side-effect, RecoveryInProgress() initializes TimeLineID if
* it's not set yet.
*/
do_restartpoint = RecoveryInProgress();
/*
* Atomically fetch the request flags to figure out what kind of a
* checkpoint we should perform, and increase the started-counter
* to acknowledge that we've started a new checkpoint.
*/
SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
flags |= CheckpointerShmem->ckpt_flags;
CheckpointerShmem->ckpt_flags = 0;
CheckpointerShmem->ckpt_started++;
SpinLockRelease(&CheckpointerShmem->ckpt_lck);
/*
* The end-of-recovery checkpoint is a real checkpoint that's
* performed while we're still in recovery.
*/
if (flags & CHECKPOINT_END_OF_RECOVERY)
do_restartpoint = false;
/*
* We will warn if (a) too soon since last checkpoint (whatever
* caused it) and (b) somebody set the CHECKPOINT_CAUSE_XLOG flag
* since the last checkpoint start. Note in particular that this
* implementation will not generate warnings caused by
* CheckPointTimeout < CheckPointWarning.
*/
if (!do_restartpoint &&
(flags & CHECKPOINT_CAUSE_XLOG) &&
elapsed_secs < CheckPointWarning)
ereport(LOG,
(errmsg_plural("checkpoints are occurring too frequently (%d second apart)",
"checkpoints are occurring too frequently (%d seconds apart)",
elapsed_secs,
elapsed_secs),
errhint("Consider increasing the configuration parameter \"max_wal_size\".")));
/*
* Initialize checkpointer-private variables used during
* checkpoint.
*/
ckpt_active = true;
if (do_restartpoint)
ckpt_start_recptr = GetXLogReplayRecPtr(NULL);
else
ckpt_start_recptr = GetInsertRecPtr();
ckpt_start_time = now;
ckpt_cached_elapsed = 0;
/*
* Do the checkpoint.
*/
if (!do_restartpoint)
{
CreateCheckPoint(flags);
ckpt_performed = true;
}
else
ckpt_performed = CreateRestartPoint(flags);
/*
* After any checkpoint, close all smgr files. This is so we
* won't hang onto smgr references to deleted files indefinitely.
*/
smgrcloseall();
/*
* Indicate checkpoint completion to any waiting backends.
*/
SpinLockAcquire(&CheckpointerShmem->ckpt_lck);
CheckpointerShmem->ckpt_done = CheckpointerShmem->ckpt_started;
SpinLockRelease(&CheckpointerShmem->ckpt_lck);
if (ckpt_performed)
{
/*
* Note we record the checkpoint start time not end time as
* last_checkpoint_time. This is so that time-driven
* checkpoints happen at a predictable spacing.
*/
last_checkpoint_time = now;
}
else
{
/*
* We were not able to perform the restartpoint (checkpoints
* throw an ERROR in case of error). Most likely because we
* have not received any new checkpoint WAL records since the
* last restartpoint. Try again in 15 s.
*/
last_checkpoint_time = now - CheckPointTimeout + 15;
}
ckpt_active = false;
}
/* Check for archive_timeout and switch xlog files if necessary. */
CheckArchiveTimeout();
/*
* Send off activity statistics to the stats collector. (The reason
* why we re-use bgwriter-related code for this is that the bgwriter
* and checkpointer used to be just one process. It's probably not
* worth the trouble to split the stats support into two independent
* stats message types.)
*/
pgstat_send_bgwriter();
/*
* Sleep until we are signaled or it's time for another checkpoint or
* xlog file switch.
*/
now = (pg_time_t) time(NULL);
elapsed_secs = now - last_checkpoint_time;
if (elapsed_secs >= CheckPointTimeout)
continue; /* no sleep for us ... */
cur_timeout = CheckPointTimeout - elapsed_secs;
if (XLogArchiveTimeout > 0 && !RecoveryInProgress())
{
elapsed_secs = now - last_xlog_switch_time;
if (elapsed_secs >= XLogArchiveTimeout)
continue; /* no sleep for us ... */
cur_timeout = Min(cur_timeout, XLogArchiveTimeout - elapsed_secs);
}
rc = WaitLatch(MyLatch,
WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH,
cur_timeout * 1000L /* convert to ms */,
WAIT_EVENT_CHECKPOINTER_MAIN);
/*
* Emergency bailout if postmaster has died. This is to avoid the
* necessity for manual cleanup of all postmaster children.
*/
if (rc & WL_POSTMASTER_DEATH)
exit(1);
}
}
NTSTATUS NTAPI
AfdConnectedSocketWriteData(PDEVICE_OBJECT DeviceObject, PIRP Irp,
PIO_STACK_LOCATION IrpSp, BOOLEAN Short) {
NTSTATUS Status = STATUS_SUCCESS;
PFILE_OBJECT FileObject = IrpSp->FileObject;
PAFD_FCB FCB = FileObject->FsContext;
PAFD_SEND_INFO SendReq;
UINT TotalBytesCopied = 0, i, SpaceAvail = 0, BytesCopied, SendLength;
KPROCESSOR_MODE LockMode;
UNREFERENCED_PARAMETER(DeviceObject);
UNREFERENCED_PARAMETER(Short);
AFD_DbgPrint(MID_TRACE,("Called on %p\n", FCB));
if( !SocketAcquireStateLock( FCB ) ) return LostSocket( Irp );
FCB->EventSelectDisabled &= ~AFD_EVENT_SEND;
if( FCB->Flags & AFD_ENDPOINT_CONNECTIONLESS )
{
PAFD_SEND_INFO_UDP SendReq;
PTDI_CONNECTION_INFORMATION TargetAddress;
/* Check that the socket is bound */
if( FCB->State != SOCKET_STATE_BOUND || !FCB->RemoteAddress )
{
AFD_DbgPrint(MIN_TRACE,("Invalid parameter\n"));
return UnlockAndMaybeComplete( FCB, STATUS_INVALID_PARAMETER, Irp,
0 );
}
if( !(SendReq = LockRequest( Irp, IrpSp, FALSE, &LockMode )) )
return UnlockAndMaybeComplete( FCB, STATUS_NO_MEMORY, Irp, 0 );
/* Must lock buffers before handing off user data */
SendReq->BufferArray = LockBuffers( SendReq->BufferArray,
SendReq->BufferCount,
NULL, NULL,
FALSE, FALSE, LockMode );
if( !SendReq->BufferArray ) {
return UnlockAndMaybeComplete( FCB, STATUS_ACCESS_VIOLATION,
Irp, 0 );
}
Status = TdiBuildConnectionInfo( &TargetAddress, FCB->RemoteAddress );
if( NT_SUCCESS(Status) ) {
FCB->PollState &= ~AFD_EVENT_SEND;
Status = QueueUserModeIrp(FCB, Irp, FUNCTION_SEND);
if (Status == STATUS_PENDING)
{
TdiSendDatagram(&FCB->SendIrp.InFlightRequest,
FCB->AddressFile.Object,
SendReq->BufferArray[0].buf,
SendReq->BufferArray[0].len,
TargetAddress,
PacketSocketSendComplete,
FCB);
}
ExFreePool( TargetAddress );
SocketStateUnlock(FCB);
return STATUS_PENDING;
}
else
{
UnlockBuffers(SendReq->BufferArray, SendReq->BufferCount, FALSE);
return UnlockAndMaybeComplete( FCB, Status, Irp, 0 );
}
}
if (FCB->PollState & AFD_EVENT_CLOSE)
{
AFD_DbgPrint(MIN_TRACE,("Connection reset by remote peer\n"));
/* This is an unexpected remote disconnect */
return UnlockAndMaybeComplete(FCB, FCB->PollStatus[FD_CLOSE_BIT], Irp, 0);
}
if (FCB->PollState & AFD_EVENT_ABORT)
{
AFD_DbgPrint(MIN_TRACE,("Connection aborted\n"));
/* This is an abortive socket closure on our side */
return UnlockAndMaybeComplete(FCB, FCB->PollStatus[FD_CLOSE_BIT], Irp, 0);
}
if (FCB->SendClosed)
{
AFD_DbgPrint(MIN_TRACE,("No more sends\n"));
/* This is a graceful send closure */
return UnlockAndMaybeComplete(FCB, STATUS_FILE_CLOSED, Irp, 0);
}
if( !(SendReq = LockRequest( Irp, IrpSp, FALSE, &LockMode )) )
return UnlockAndMaybeComplete
( FCB, STATUS_NO_MEMORY, Irp, 0 );
SendReq->BufferArray = LockBuffers( SendReq->BufferArray,
SendReq->BufferCount,
NULL, NULL,
FALSE, FALSE, LockMode );
if( !SendReq->BufferArray ) {
return UnlockAndMaybeComplete( FCB, STATUS_ACCESS_VIOLATION,
Irp, 0 );
}
AFD_DbgPrint(MID_TRACE,("Socket state %u\n", FCB->State));
if( FCB->State != SOCKET_STATE_CONNECTED ) {
if (!(SendReq->AfdFlags & AFD_OVERLAPPED) &&
((SendReq->AfdFlags & AFD_IMMEDIATE) || (FCB->NonBlocking))) {
AFD_DbgPrint(MID_TRACE,("Nonblocking\n"));
UnlockBuffers( SendReq->BufferArray, SendReq->BufferCount, FALSE );
return UnlockAndMaybeComplete( FCB, STATUS_CANT_WAIT, Irp, 0 );
} else {
AFD_DbgPrint(MID_TRACE,("Queuing request\n"));
return LeaveIrpUntilLater( FCB, Irp, FUNCTION_SEND );
}
}
AFD_DbgPrint(MID_TRACE,("FCB->Send.BytesUsed = %u\n",
FCB->Send.BytesUsed));
SpaceAvail = FCB->Send.Size - FCB->Send.BytesUsed;
AFD_DbgPrint(MID_TRACE,("We can accept %u bytes\n",
SpaceAvail));
/* Count the total transfer size */
SendLength = 0;
for (i = 0; i < SendReq->BufferCount; i++)
{
SendLength += SendReq->BufferArray[i].len;
}
/* Make sure we've got the space */
if (SendLength > SpaceAvail)
{
/* Blocking sockets have to wait here */
if (SendLength <= FCB->Send.Size && !((SendReq->AfdFlags & AFD_IMMEDIATE) || (FCB->NonBlocking)))
{
FCB->PollState &= ~AFD_EVENT_SEND;
return LeaveIrpUntilLater(FCB, Irp, FUNCTION_SEND);
}
/* Check if we can send anything */
if (SpaceAvail == 0)
{
FCB->PollState &= ~AFD_EVENT_SEND;
/* Non-overlapped sockets will fail if we can send nothing */
if (!(SendReq->AfdFlags & AFD_OVERLAPPED))
{
UnlockBuffers( SendReq->BufferArray, SendReq->BufferCount, FALSE );
return UnlockAndMaybeComplete( FCB, STATUS_CANT_WAIT, Irp, 0 );
}
else
{
/* Overlapped sockets just pend */
return LeaveIrpUntilLater(FCB, Irp, FUNCTION_SEND);
}
}
}
for ( i = 0; SpaceAvail > 0 && i < SendReq->BufferCount; i++ )
{
BytesCopied = MIN(SendReq->BufferArray[i].len, SpaceAvail);
AFD_DbgPrint(MID_TRACE,("Copying Buffer %u, %p:%u to %p\n",
i,
SendReq->BufferArray[i].buf,
BytesCopied,
FCB->Send.Window + FCB->Send.BytesUsed));
RtlCopyMemory(FCB->Send.Window + FCB->Send.BytesUsed,
SendReq->BufferArray[i].buf,
BytesCopied);
TotalBytesCopied += BytesCopied;
SpaceAvail -= BytesCopied;
FCB->Send.BytesUsed += BytesCopied;
}
Irp->IoStatus.Information = TotalBytesCopied;
if( TotalBytesCopied == 0 ) {
AFD_DbgPrint(MID_TRACE,("Empty send\n"));
UnlockBuffers( SendReq->BufferArray, SendReq->BufferCount, FALSE );
return UnlockAndMaybeComplete
( FCB, STATUS_SUCCESS, Irp, TotalBytesCopied );
}
if (SpaceAvail)
{
FCB->PollState |= AFD_EVENT_SEND;
FCB->PollStatus[FD_WRITE_BIT] = STATUS_SUCCESS;
PollReeval( FCB->DeviceExt, FCB->FileObject );
}
else
{
FCB->PollState &= ~AFD_EVENT_SEND;
}
/* We use the IRP tail for some temporary storage here */
Irp->Tail.Overlay.DriverContext[3] = (PVOID)Irp->IoStatus.Information;
Status = QueueUserModeIrp(FCB, Irp, FUNCTION_SEND);
if (Status == STATUS_PENDING && !FCB->SendIrp.InFlightRequest)
{
TdiSend(&FCB->SendIrp.InFlightRequest,
FCB->Connection.Object,
0,
FCB->Send.Window,
FCB->Send.BytesUsed,
SendComplete,
FCB);
}
SocketStateUnlock(FCB);
return STATUS_PENDING;
}
static NTSTATUS NTAPI PacketSocketSendComplete
( PDEVICE_OBJECT DeviceObject,
PIRP Irp,
PVOID Context ) {
PAFD_FCB FCB = (PAFD_FCB)Context;
PLIST_ENTRY NextIrpEntry;
PIRP NextIrp;
PAFD_SEND_INFO SendReq;
UNREFERENCED_PARAMETER(DeviceObject);
AFD_DbgPrint(MID_TRACE,("Called, status %x, %u bytes used\n",
Irp->IoStatus.Status,
Irp->IoStatus.Information));
if( !SocketAcquireStateLock( FCB ) )
return STATUS_FILE_CLOSED;
ASSERT(FCB->SendIrp.InFlightRequest == Irp);
FCB->SendIrp.InFlightRequest = NULL;
/* Request is not in flight any longer */
if( FCB->State == SOCKET_STATE_CLOSED ) {
/* Cleanup our IRP queue because the FCB is being destroyed */
while( !IsListEmpty( &FCB->PendingIrpList[FUNCTION_SEND] ) ) {
NextIrpEntry = RemoveHeadList(&FCB->PendingIrpList[FUNCTION_SEND]);
NextIrp = CONTAINING_RECORD(NextIrpEntry, IRP, Tail.Overlay.ListEntry);
SendReq = GetLockedData(NextIrp, IoGetCurrentIrpStackLocation(NextIrp));
NextIrp->IoStatus.Status = STATUS_FILE_CLOSED;
NextIrp->IoStatus.Information = 0;
(void)IoSetCancelRoutine(NextIrp, NULL);
UnlockBuffers(SendReq->BufferArray, SendReq->BufferCount, FALSE);
UnlockRequest( NextIrp, IoGetCurrentIrpStackLocation( NextIrp ) );
IoCompleteRequest( NextIrp, IO_NETWORK_INCREMENT );
}
SocketStateUnlock( FCB );
return STATUS_FILE_CLOSED;
}
ASSERT(!IsListEmpty(&FCB->PendingIrpList[FUNCTION_SEND]));
/* TDI spec guarantees FIFO ordering on IRPs */
NextIrpEntry = RemoveHeadList(&FCB->PendingIrpList[FUNCTION_SEND]);
NextIrp = CONTAINING_RECORD(NextIrpEntry, IRP, Tail.Overlay.ListEntry);
SendReq = GetLockedData(NextIrp, IoGetCurrentIrpStackLocation(NextIrp));
NextIrp->IoStatus.Status = Irp->IoStatus.Status;
NextIrp->IoStatus.Information = Irp->IoStatus.Information;
(void)IoSetCancelRoutine(NextIrp, NULL);
UnlockBuffers(SendReq->BufferArray, SendReq->BufferCount, FALSE);
UnlockRequest(NextIrp, IoGetCurrentIrpStackLocation(NextIrp));
IoCompleteRequest(NextIrp, IO_NETWORK_INCREMENT);
FCB->PollState |= AFD_EVENT_SEND;
FCB->PollStatus[FD_WRITE_BIT] = STATUS_SUCCESS;
PollReeval(FCB->DeviceExt, FCB->FileObject);
SocketStateUnlock(FCB);
return STATUS_SUCCESS;
}
static NTSTATUS NTAPI SendComplete
( PDEVICE_OBJECT DeviceObject,
PIRP Irp,
PVOID Context ) {
NTSTATUS Status = Irp->IoStatus.Status;
PAFD_FCB FCB = (PAFD_FCB)Context;
PLIST_ENTRY NextIrpEntry;
PIRP NextIrp = NULL;
PIO_STACK_LOCATION NextIrpSp;
PAFD_SEND_INFO SendReq = NULL;
PAFD_MAPBUF Map;
UINT TotalBytesCopied = 0, TotalBytesProcessed = 0, SpaceAvail, i;
UINT SendLength, BytesCopied;
BOOLEAN HaltSendQueue;
UNREFERENCED_PARAMETER(DeviceObject);
/*
* The Irp parameter passed in is the IRP of the stream between AFD and
* TDI driver. It's not very usefull to us. We need the IRPs of the stream
* between usermode and AFD. Those are chained from
* FCB->PendingIrpList[FUNCTION_SEND] and you'll see them in the code
* below as "NextIrp" ('cause they are the next usermode IRP to be
* processed).
*/
AFD_DbgPrint(MID_TRACE,("Called, status %x, %u bytes used\n",
Irp->IoStatus.Status,
Irp->IoStatus.Information));
if( !SocketAcquireStateLock( FCB ) )
return STATUS_FILE_CLOSED;
ASSERT(FCB->SendIrp.InFlightRequest == Irp);
FCB->SendIrp.InFlightRequest = NULL;
/* Request is not in flight any longer */
if( FCB->State == SOCKET_STATE_CLOSED ) {
/* Cleanup our IRP queue because the FCB is being destroyed */
while( !IsListEmpty( &FCB->PendingIrpList[FUNCTION_SEND] ) ) {
NextIrpEntry = RemoveHeadList(&FCB->PendingIrpList[FUNCTION_SEND]);
NextIrp = CONTAINING_RECORD(NextIrpEntry, IRP, Tail.Overlay.ListEntry);
NextIrpSp = IoGetCurrentIrpStackLocation( NextIrp );
SendReq = GetLockedData(NextIrp, NextIrpSp);
NextIrp->IoStatus.Status = STATUS_FILE_CLOSED;
NextIrp->IoStatus.Information = 0;
UnlockBuffers(SendReq->BufferArray, SendReq->BufferCount, FALSE);
if( NextIrp->MdlAddress ) UnlockRequest( NextIrp, IoGetCurrentIrpStackLocation( NextIrp ) );
(void)IoSetCancelRoutine(NextIrp, NULL);
IoCompleteRequest( NextIrp, IO_NETWORK_INCREMENT );
}
RetryDisconnectCompletion(FCB);
SocketStateUnlock( FCB );
return STATUS_FILE_CLOSED;
}
if( !NT_SUCCESS(Status) ) {
/* Complete all following send IRPs with error */
while( !IsListEmpty( &FCB->PendingIrpList[FUNCTION_SEND] ) ) {
NextIrpEntry =
RemoveHeadList(&FCB->PendingIrpList[FUNCTION_SEND]);
NextIrp =
CONTAINING_RECORD(NextIrpEntry, IRP, Tail.Overlay.ListEntry);
NextIrpSp = IoGetCurrentIrpStackLocation( NextIrp );
SendReq = GetLockedData(NextIrp, NextIrpSp);
UnlockBuffers( SendReq->BufferArray,
SendReq->BufferCount,
FALSE );
NextIrp->IoStatus.Status = Status;
NextIrp->IoStatus.Information = 0;
if ( NextIrp->MdlAddress ) UnlockRequest( NextIrp, IoGetCurrentIrpStackLocation( NextIrp ) );
(void)IoSetCancelRoutine(NextIrp, NULL);
IoCompleteRequest( NextIrp, IO_NETWORK_INCREMENT );
}
RetryDisconnectCompletion(FCB);
SocketStateUnlock( FCB );
return STATUS_SUCCESS;
}
RtlMoveMemory( FCB->Send.Window,
FCB->Send.Window + Irp->IoStatus.Information,
FCB->Send.BytesUsed - Irp->IoStatus.Information );
TotalBytesProcessed = 0;
SendLength = Irp->IoStatus.Information;
HaltSendQueue = FALSE;
while (!IsListEmpty(&FCB->PendingIrpList[FUNCTION_SEND]) && SendLength > 0) {
NextIrpEntry = RemoveHeadList(&FCB->PendingIrpList[FUNCTION_SEND]);
NextIrp = CONTAINING_RECORD(NextIrpEntry, IRP, Tail.Overlay.ListEntry);
NextIrpSp = IoGetCurrentIrpStackLocation( NextIrp );
SendReq = GetLockedData(NextIrp, NextIrpSp);
Map = (PAFD_MAPBUF)(SendReq->BufferArray + SendReq->BufferCount);
TotalBytesCopied = (ULONG_PTR)NextIrp->Tail.Overlay.DriverContext[3];
ASSERT(TotalBytesCopied != 0);
/* If we didn't get enough, keep waiting */
if (TotalBytesCopied > SendLength)
{
/* Update the bytes left to copy */
TotalBytesCopied -= SendLength;
NextIrp->Tail.Overlay.DriverContext[3] = (PVOID)TotalBytesCopied;
/* Update the state variables */
FCB->Send.BytesUsed -= SendLength;
TotalBytesProcessed += SendLength;
SendLength = 0;
/* Pend the IRP */
InsertHeadList(&FCB->PendingIrpList[FUNCTION_SEND],
&NextIrp->Tail.Overlay.ListEntry);
HaltSendQueue = TRUE;
break;
}
ASSERT(NextIrp->IoStatus.Information != 0);
NextIrp->IoStatus.Status = Irp->IoStatus.Status;
FCB->Send.BytesUsed -= TotalBytesCopied;
TotalBytesProcessed += TotalBytesCopied;
SendLength -= TotalBytesCopied;
(void)IoSetCancelRoutine(NextIrp, NULL);
UnlockBuffers( SendReq->BufferArray,
SendReq->BufferCount,
FALSE );
if (NextIrp->MdlAddress) UnlockRequest(NextIrp, NextIrpSp);
IoCompleteRequest(NextIrp, IO_NETWORK_INCREMENT);
}
ASSERT(SendLength == 0);
if ( !HaltSendQueue && !IsListEmpty( &FCB->PendingIrpList[FUNCTION_SEND] ) ) {
NextIrpEntry = FCB->PendingIrpList[FUNCTION_SEND].Flink;
NextIrp = CONTAINING_RECORD(NextIrpEntry, IRP, Tail.Overlay.ListEntry);
NextIrpSp = IoGetCurrentIrpStackLocation( NextIrp );
SendReq = GetLockedData(NextIrp, NextIrpSp);
Map = (PAFD_MAPBUF)(SendReq->BufferArray + SendReq->BufferCount);
AFD_DbgPrint(MID_TRACE,("SendReq @ %p\n", SendReq));
SpaceAvail = FCB->Send.Size - FCB->Send.BytesUsed;
TotalBytesCopied = 0;
/* Count the total transfer size */
SendLength = 0;
for (i = 0; i < SendReq->BufferCount; i++)
{
SendLength += SendReq->BufferArray[i].len;
}
/* Make sure we've got the space */
if (SendLength > SpaceAvail)
{
/* Blocking sockets have to wait here */
if (SendLength <= FCB->Send.Size && !((SendReq->AfdFlags & AFD_IMMEDIATE) || (FCB->NonBlocking)))
{
FCB->PollState &= ~AFD_EVENT_SEND;
NextIrp = NULL;
}
/* Check if we can send anything */
if (SpaceAvail == 0)
{
FCB->PollState &= ~AFD_EVENT_SEND;
/* We should never be non-overlapped and get to this point */
ASSERT(SendReq->AfdFlags & AFD_OVERLAPPED);
NextIrp = NULL;
}
}
if (NextIrp != NULL)
{
for( i = 0; i < SendReq->BufferCount; i++ ) {
BytesCopied = MIN(SendReq->BufferArray[i].len, SpaceAvail);
Map[i].BufferAddress =
MmMapLockedPages( Map[i].Mdl, KernelMode );
RtlCopyMemory( FCB->Send.Window + FCB->Send.BytesUsed,
Map[i].BufferAddress,
BytesCopied );
MmUnmapLockedPages( Map[i].BufferAddress, Map[i].Mdl );
TotalBytesCopied += BytesCopied;
SpaceAvail -= BytesCopied;
FCB->Send.BytesUsed += BytesCopied;
}
NextIrp->IoStatus.Information = TotalBytesCopied;
NextIrp->Tail.Overlay.DriverContext[3] = (PVOID)NextIrp->IoStatus.Information;
}
}
if (FCB->Send.Size - FCB->Send.BytesUsed != 0 && !FCB->SendClosed &&
IsListEmpty(&FCB->PendingIrpList[FUNCTION_SEND]))
{
FCB->PollState |= AFD_EVENT_SEND;
FCB->PollStatus[FD_WRITE_BIT] = STATUS_SUCCESS;
PollReeval( FCB->DeviceExt, FCB->FileObject );
}
else
{
FCB->PollState &= ~AFD_EVENT_SEND;
}
/* Some data is still waiting */
if( FCB->Send.BytesUsed )
{
Status = TdiSend( &FCB->SendIrp.InFlightRequest,
FCB->Connection.Object,
0,
FCB->Send.Window,
FCB->Send.BytesUsed,
SendComplete,
FCB );
}
else
{
/* Nothing is waiting so try to complete a pending disconnect */
RetryDisconnectCompletion(FCB);
}
SocketStateUnlock( FCB );
return STATUS_SUCCESS;
}
/*
* Main entry point for walwriter process
*
* This is invoked from AuxiliaryProcessMain, which has already created the
* basic execution environment, but not enabled signals yet.
*/
void
WalWriterMain(void)
{
sigjmp_buf local_sigjmp_buf;
MemoryContext walwriter_context;
int left_till_hibernate;
bool hibernating;
/*
* Properly accept or ignore signals the postmaster might send us
*
* We have no particular use for SIGINT at the moment, but seems
* reasonable to treat like SIGTERM.
*/
pqsignal(SIGHUP, WalSigHupHandler); /* set flag to read config file */
pqsignal(SIGINT, WalShutdownHandler); /* request shutdown */
pqsignal(SIGTERM, WalShutdownHandler); /* request shutdown */
pqsignal(SIGQUIT, wal_quickdie); /* hard crash time */
pqsignal(SIGALRM, SIG_IGN);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR1, walwriter_sigusr1_handler);
pqsignal(SIGUSR2, SIG_IGN); /* not used */
/*
* Reset some signals that are accepted by postmaster but not here
*/
pqsignal(SIGCHLD, SIG_DFL);
pqsignal(SIGTTIN, SIG_DFL);
pqsignal(SIGTTOU, SIG_DFL);
pqsignal(SIGCONT, SIG_DFL);
pqsignal(SIGWINCH, SIG_DFL);
/* We allow SIGQUIT (quickdie) at all times */
sigdelset(&BlockSig, SIGQUIT);
/*
* Create a resource owner to keep track of our resources (not clear that
* we need this, but may as well have one).
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "Wal Writer");
/*
* Create a memory context that we will do all our work in. We do this so
* that we can reset the context during error recovery and thereby avoid
* possible memory leaks. Formerly this code just ran in
* TopMemoryContext, but resetting that would be a really bad idea.
*/
walwriter_context = AllocSetContextCreate(TopMemoryContext,
"Wal Writer",
ALLOCSET_DEFAULT_SIZES);
MemoryContextSwitchTo(walwriter_context);
/*
* If an exception is encountered, processing resumes here.
*
* This code is heavily based on bgwriter.c, q.v.
*/
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
/* Since not using PG_TRY, must reset error stack by hand */
error_context_stack = NULL;
/* Prevent interrupts while cleaning up */
HOLD_INTERRUPTS();
/* Report the error to the server log */
EmitErrorReport();
/*
* These operations are really just a minimal subset of
* AbortTransaction(). We don't have very many resources to worry
* about in walwriter, but we do have LWLocks, and perhaps buffers?
*/
LWLockReleaseAll();
ConditionVariableCancelSleep();
pgstat_report_wait_end();
AbortBufferIO();
UnlockBuffers();
/* buffer pins are released here: */
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
/* we needn't bother with the other ResourceOwnerRelease phases */
AtEOXact_Buffers(false);
AtEOXact_SMgr();
AtEOXact_Files();
AtEOXact_HashTables(false);
/*
* Now return to normal top-level context and clear ErrorContext for
* next time.
*/
MemoryContextSwitchTo(walwriter_context);
FlushErrorState();
/* Flush any leaked data in the top-level context */
MemoryContextResetAndDeleteChildren(walwriter_context);
/* Now we can allow interrupts again */
RESUME_INTERRUPTS();
/*
* Sleep at least 1 second after any error. A write error is likely
* to be repeated, and we don't want to be filling the error logs as
* fast as we can.
*/
pg_usleep(1000000L);
/*
* Close all open files after any error. This is helpful on Windows,
* where holding deleted files open causes various strange errors.
* It's not clear we need it elsewhere, but shouldn't hurt.
*/
smgrcloseall();
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
/*
* Unblock signals (they were blocked when the postmaster forked us)
*/
PG_SETMASK(&UnBlockSig);
/*
* Reset hibernation state after any error.
*/
left_till_hibernate = LOOPS_UNTIL_HIBERNATE;
hibernating = false;
SetWalWriterSleeping(false);
/*
* Advertise our latch that backends can use to wake us up while we're
* sleeping.
*/
ProcGlobal->walwriterLatch = &MyProc->procLatch;
/*
* Loop forever
*/
for (;;)
{
long cur_timeout;
int rc;
/*
* Advertise whether we might hibernate in this cycle. We do this
* before resetting the latch to ensure that any async commits will
* see the flag set if they might possibly need to wake us up, and
* that we won't miss any signal they send us. (If we discover work
* to do in the last cycle before we would hibernate, the global flag
* will be set unnecessarily, but little harm is done.) But avoid
* touching the global flag if it doesn't need to change.
*/
if (hibernating != (left_till_hibernate <= 1))
{
hibernating = (left_till_hibernate <= 1);
SetWalWriterSleeping(hibernating);
}
/* Clear any already-pending wakeups */
ResetLatch(MyLatch);
/*
* Process any requests or signals received recently.
*/
if (got_SIGHUP)
{
got_SIGHUP = false;
ProcessConfigFile(PGC_SIGHUP);
}
if (shutdown_requested)
{
/* Normal exit from the walwriter is here */
proc_exit(0); /* done */
}
/*
* Do what we're here for; then, if XLogBackgroundFlush() found useful
* work to do, reset hibernation counter.
*/
if (XLogBackgroundFlush())
left_till_hibernate = LOOPS_UNTIL_HIBERNATE;
else if (left_till_hibernate > 0)
left_till_hibernate--;
/*
* Sleep until we are signaled or WalWriterDelay has elapsed. If we
* haven't done anything useful for quite some time, lengthen the
* sleep time so as to reduce the server's idle power consumption.
*/
if (left_till_hibernate > 0)
cur_timeout = WalWriterDelay; /* in ms */
else
cur_timeout = WalWriterDelay * HIBERNATE_FACTOR;
rc = WaitLatch(MyLatch,
WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH,
cur_timeout,
WAIT_EVENT_WAL_WRITER_MAIN);
/*
* Emergency bailout if postmaster has died. This is to avoid the
* necessity for manual cleanup of all postmaster children.
*/
if (rc & WL_POSTMASTER_DEATH)
exit(1);
}
}
/*
* AbortTransaction
*/
static void
AbortTransaction(void)
{
TransactionState s = CurrentTransactionState;
/* Prevent cancel/die interrupt while cleaning up */
HOLD_INTERRUPTS();
/*
* Release any LW locks we might be holding as quickly as possible.
* (Regular locks, however, must be held till we finish aborting.)
* Releasing LW locks is critical since we might try to grab them
* again while cleaning up!
*/
LWLockReleaseAll();
/* Clean up buffer I/O and buffer context locks, too */
AbortBufferIO();
UnlockBuffers();
/*
* Also clean up any open wait for lock, since the lock manager will
* choke if we try to wait for another lock before doing this.
*/
LockWaitCancel();
/*
* check the current transaction state
*
* reduced to DEBUG2 because this is expected when rejecting an
* invalidly-encoded query outside a transaction block. PG 8.0
* and up fix it better, but it's not worth back-porting those
* changes to 7.4.
*/
if (s->state != TRANS_INPROGRESS)
elog(DEBUG2, "AbortTransaction and not in in-progress state");
/*
* set the current transaction state information appropriately during
* the abort processing
*/
s->state = TRANS_ABORT;
/* Make sure we are in a valid memory context */
AtAbort_Memory();
/*
* Reset user id which might have been changed transiently
*/
SetUserId(GetSessionUserId());
/*
* do abort processing
*/
DeferredTriggerAbortXact();
AtAbort_Portals();
lo_commit(false); /* 'false' means it's abort */
AtAbort_Notify();
AtEOXact_UpdatePasswordFile(false);
/* Advertise the fact that we aborted in pg_clog. */
RecordTransactionAbort();
/*
* Let others know about no transaction in progress by me. Note that
* this must be done _before_ releasing locks we hold and _after_
* RecordTransactionAbort.
*/
if (MyProc != (PGPROC *) NULL)
{
/* Lock SInvalLock because that's what GetSnapshotData uses. */
LWLockAcquire(SInvalLock, LW_EXCLUSIVE);
MyProc->xid = InvalidTransactionId;
MyProc->xmin = InvalidTransactionId;
LWLockRelease(SInvalLock);
}
/*
* Post-abort cleanup. See notes in CommitTransaction() concerning
* ordering.
*/
smgrDoPendingDeletes(false);
AtAbort_Cache();
AtEOXact_Buffers(false);
smgrabort();
AtAbort_Locks();
CallEOXactCallbacks(false);
AtEOXact_GUC(false);
AtEOXact_SPI();
AtEOXact_gist();
AtEOXact_hash();
AtEOXact_nbtree();
AtEOXact_rtree();
AtEOXact_on_commit_actions(false);
AtEOXact_Namespace(false);
AtEOXact_CatCache(false);
AtEOXact_Files();
SetReindexProcessing(InvalidOid, InvalidOid);
pgstat_count_xact_rollback();
/*
* State remains TRANS_ABORT until CleanupTransaction().
*/
RESUME_INTERRUPTS();
}
void
FileRepSubProcess_Main()
{
const char *statmsg;
MemoryContext fileRepSubProcessMemoryContext;
sigjmp_buf local_sigjmp_buf;
MyProcPid = getpid();
MyStartTime = time(NULL);
/*
* Create a PGPROC so we can use LWLocks in FileRep sub-processes. The
* routine also register clean up at process exit
*/
InitAuxiliaryProcess();
InitBufferPoolBackend();
FileRepSubProcess_ConfigureSignals();
/*
* If an exception is encountered, processing resumes here.
*
* See notes in postgres.c about the design of this coding.
*/
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
/* Prevents interrupts while cleaning up */
HOLD_INTERRUPTS();
/* Report the error to the server log */
EmitErrorReport();
LWLockReleaseAll();
if (FileRepPrimary_IsResyncManagerOrWorker())
{
LockReleaseAll(DEFAULT_LOCKMETHOD, false);
}
if (FileRepIsBackendSubProcess(fileRepProcessType))
{
AbortBufferIO();
UnlockBuffers();
/* buffer pins are released here: */
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
}
/*
* We can now go away. Note that because we'll call InitProcess, a
* callback will be registered to do ProcKill, which will clean up
* necessary state.
*/
proc_exit(0);
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
PG_SETMASK(&UnBlockSig);
/*
* Identify myself via ps
*/
statmsg = FileRepProcessTypeToString[fileRepProcessType];
init_ps_display(statmsg, "", "", "");
/* Create the memory context where cross-transaction state is stored */
fileRepSubProcessMemoryContext = AllocSetContextCreate(TopMemoryContext,
"filerep subprocess memory context",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
MemoryContextSwitchTo(fileRepSubProcessMemoryContext);
stateChangeRequestCounter++;
FileRepSubProcess_ProcessSignals();
switch (fileRepProcessType)
{
case FileRepProcessTypePrimarySender:
FileRepPrimary_StartSender();
break;
case FileRepProcessTypeMirrorReceiver:
FileRepMirror_StartReceiver();
break;
case FileRepProcessTypeMirrorConsumer:
case FileRepProcessTypeMirrorConsumerWriter:
case FileRepProcessTypeMirrorConsumerAppendOnly1:
FileRepMirror_StartConsumer();
break;
case FileRepProcessTypeMirrorSenderAck:
FileRepAckMirror_StartSender();
break;
case FileRepProcessTypePrimaryReceiverAck:
FileRepAckPrimary_StartReceiver();
break;
case FileRepProcessTypePrimaryConsumerAck:
FileRepAckPrimary_StartConsumer();
break;
case FileRepProcessTypePrimaryRecovery:
FileRepSubProcess_InitProcess();
/*
* At this point, database is starting up and xlog is not yet
* replayed. Initializing relcache now is dangerous, a sequential
* scan of catalog tables may end up with incorrect hint bits.
* E.g. a committed transaction's dirty heap pages made it to disk
* but pg_clog update was still in memory and we crashed. If a
* tuple inserted by this transaction is read during relcache
* initialization, status of the tuple's xmin will be incorrectly
* determined as "not commited" from pg_clog. And
* HEAP_XMIN_INVALID hint bit will be set, rendering the tuple
* perpetually invisible. Relcache initialization must be
* deferred to only after all of xlog has been replayed.
*/
FileRepPrimary_StartRecovery();
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
break;
case FileRepProcessTypeResyncManager:
FileRepSubProcess_InitProcess();
FileRepPrimary_StartResyncManager();
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
break;
case FileRepProcessTypeResyncWorker1:
case FileRepProcessTypeResyncWorker2:
case FileRepProcessTypeResyncWorker3:
case FileRepProcessTypeResyncWorker4:
FileRepSubProcess_InitProcess();
FileRepPrimary_StartResyncWorker();
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
break;
default:
elog(PANIC, "unrecognized process type: %s(%d)",
statmsg, fileRepProcessType);
break;
}
switch (FileRepSubProcess_GetState())
{
case FileRepStateShutdown:
case FileRepStateReady:
proc_exit(0);
break;
default:
proc_exit(2);
break;
}
}
/*
* Main entry point for bgwriter process
*
* This is invoked from AuxiliaryProcessMain, which has already created the
* basic execution environment, but not enabled signals yet.
*/
void
BackgroundWriterMain(void)
{
sigjmp_buf local_sigjmp_buf;
MemoryContext bgwriter_context;
bool prev_hibernate;
/*
* Properly accept or ignore signals the postmaster might send us.
*
* bgwriter doesn't participate in ProcSignal signalling, but a SIGUSR1
* handler is still needed for latch wakeups.
*/
pqsignal(SIGHUP, BgSigHupHandler); /* set flag to read config file */
pqsignal(SIGINT, SIG_IGN);
pqsignal(SIGTERM, ReqShutdownHandler); /* shutdown */
pqsignal(SIGQUIT, bg_quickdie); /* hard crash time */
pqsignal(SIGALRM, SIG_IGN);
pqsignal(SIGPIPE, SIG_IGN);
pqsignal(SIGUSR1, bgwriter_sigusr1_handler);
pqsignal(SIGUSR2, SIG_IGN);
/*
* Reset some signals that are accepted by postmaster but not here
*/
pqsignal(SIGCHLD, SIG_DFL);
pqsignal(SIGTTIN, SIG_DFL);
pqsignal(SIGTTOU, SIG_DFL);
pqsignal(SIGCONT, SIG_DFL);
pqsignal(SIGWINCH, SIG_DFL);
/* We allow SIGQUIT (quickdie) at all times */
sigdelset(&BlockSig, SIGQUIT);
/*
* Create a resource owner to keep track of our resources (currently only
* buffer pins).
*/
CurrentResourceOwner = ResourceOwnerCreate(NULL, "Background Writer");
/*
* We just started, assume there has been either a shutdown or
* end-of-recovery snapshot.
*/
last_snapshot_ts = GetCurrentTimestamp();
/*
* Create a memory context that we will do all our work in. We do this so
* that we can reset the context during error recovery and thereby avoid
* possible memory leaks. Formerly this code just ran in
* TopMemoryContext, but resetting that would be a really bad idea.
*/
bgwriter_context = AllocSetContextCreate(TopMemoryContext,
"Background Writer",
ALLOCSET_DEFAULT_MINSIZE,
ALLOCSET_DEFAULT_INITSIZE,
ALLOCSET_DEFAULT_MAXSIZE);
MemoryContextSwitchTo(bgwriter_context);
/*
* If an exception is encountered, processing resumes here.
*
* See notes in postgres.c about the design of this coding.
*/
if (sigsetjmp(local_sigjmp_buf, 1) != 0)
{
/* Since not using PG_TRY, must reset error stack by hand */
error_context_stack = NULL;
/* Prevent interrupts while cleaning up */
HOLD_INTERRUPTS();
/* Report the error to the server log */
EmitErrorReport();
/*
* These operations are really just a minimal subset of
* AbortTransaction(). We don't have very many resources to worry
* about in bgwriter, but we do have LWLocks, buffers, and temp files.
*/
LWLockReleaseAll();
AbortBufferIO();
UnlockBuffers();
/* buffer pins are released here: */
ResourceOwnerRelease(CurrentResourceOwner,
RESOURCE_RELEASE_BEFORE_LOCKS,
false, true);
/* we needn't bother with the other ResourceOwnerRelease phases */
AtEOXact_Buffers(false);
AtEOXact_SMgr();
AtEOXact_Files();
AtEOXact_HashTables(false);
/*
* Now return to normal top-level context and clear ErrorContext for
* next time.
*/
MemoryContextSwitchTo(bgwriter_context);
FlushErrorState();
/* Flush any leaked data in the top-level context */
MemoryContextResetAndDeleteChildren(bgwriter_context);
/* Now we can allow interrupts again */
RESUME_INTERRUPTS();
/*
* Sleep at least 1 second after any error. A write error is likely
* to be repeated, and we don't want to be filling the error logs as
* fast as we can.
*/
pg_usleep(1000000L);
/*
* Close all open files after any error. This is helpful on Windows,
* where holding deleted files open causes various strange errors.
* It's not clear we need it elsewhere, but shouldn't hurt.
*/
smgrcloseall();
/* Report wait end here, when there is no further possibility of wait */
pgstat_report_wait_end();
}
/* We can now handle ereport(ERROR) */
PG_exception_stack = &local_sigjmp_buf;
/*
* Unblock signals (they were blocked when the postmaster forked us)
*/
PG_SETMASK(&UnBlockSig);
/*
* Reset hibernation state after any error.
*/
prev_hibernate = false;
/*
* Loop forever
*/
for (;;)
{
bool can_hibernate;
int rc;
/* Clear any already-pending wakeups */
ResetLatch(MyLatch);
if (got_SIGHUP)
{
got_SIGHUP = false;
ProcessConfigFile(PGC_SIGHUP);
}
if (shutdown_requested)
{
/*
* From here on, elog(ERROR) should end with exit(1), not send
* control back to the sigsetjmp block above
*/
ExitOnAnyError = true;
/* Normal exit from the bgwriter is here */
proc_exit(0); /* done */
}
/*
* Do one cycle of dirty-buffer writing.
*/
can_hibernate = BgBufferSync();
/*
* Send off activity statistics to the stats collector
*/
pgstat_send_bgwriter();
if (FirstCallSinceLastCheckpoint())
{
/*
* After any checkpoint, close all smgr files. This is so we
* won't hang onto smgr references to deleted files indefinitely.
*/
smgrcloseall();
}
/*
* Log a new xl_running_xacts every now and then so replication can
* get into a consistent state faster (think of suboverflowed
* snapshots) and clean up resources (locks, KnownXids*) more
* frequently. The costs of this are relatively low, so doing it 4
* times (LOG_SNAPSHOT_INTERVAL_MS) a minute seems fine.
*
* We assume the interval for writing xl_running_xacts is
* significantly bigger than BgWriterDelay, so we don't complicate the
* overall timeout handling but just assume we're going to get called
* often enough even if hibernation mode is active. It's not that
* important that log_snap_interval_ms is met strictly. To make sure
* we're not waking the disk up unnecessarily on an idle system we
* check whether there has been any WAL inserted since the last time
* we've logged a running xacts.
*
* We do this logging in the bgwriter as its the only process that is
* run regularly and returns to its mainloop all the time. E.g.
* Checkpointer, when active, is barely ever in its mainloop and thus
* makes it hard to log regularly.
*/
if (XLogStandbyInfoActive() && !RecoveryInProgress())
{
TimestampTz timeout = 0;
TimestampTz now = GetCurrentTimestamp();
timeout = TimestampTzPlusMilliseconds(last_snapshot_ts,
LOG_SNAPSHOT_INTERVAL_MS);
/*
* only log if enough time has passed and some xlog record has
* been inserted.
*/
if (now >= timeout &&
last_snapshot_lsn != GetXLogInsertRecPtr())
{
last_snapshot_lsn = LogStandbySnapshot();
last_snapshot_ts = now;
}
}
/*
* Sleep until we are signaled or BgWriterDelay has elapsed.
*
* Note: the feedback control loop in BgBufferSync() expects that we
* will call it every BgWriterDelay msec. While it's not critical for
* correctness that that be exact, the feedback loop might misbehave
* if we stray too far from that. Hence, avoid loading this process
* down with latch events that are likely to happen frequently during
* normal operation.
*/
rc = WaitLatch(MyLatch,
WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH,
BgWriterDelay /* ms */ );
/*
* If no latch event and BgBufferSync says nothing's happening, extend
* the sleep in "hibernation" mode, where we sleep for much longer
* than bgwriter_delay says. Fewer wakeups save electricity. When a
* backend starts using buffers again, it will wake us up by setting
* our latch. Because the extra sleep will persist only as long as no
* buffer allocations happen, this should not distort the behavior of
* BgBufferSync's control loop too badly; essentially, it will think
* that the system-wide idle interval didn't exist.
*
* There is a race condition here, in that a backend might allocate a
* buffer between the time BgBufferSync saw the alloc count as zero
* and the time we call StrategyNotifyBgWriter. While it's not
* critical that we not hibernate anyway, we try to reduce the odds of
* that by only hibernating when BgBufferSync says nothing's happening
* for two consecutive cycles. Also, we mitigate any possible
* consequences of a missed wakeup by not hibernating forever.
*/
if (rc == WL_TIMEOUT && can_hibernate && prev_hibernate)
{
/* Ask for notification at next buffer allocation */
StrategyNotifyBgWriter(MyProc->pgprocno);
/* Sleep ... */
rc = WaitLatch(MyLatch,
WL_LATCH_SET | WL_TIMEOUT | WL_POSTMASTER_DEATH,
BgWriterDelay * HIBERNATE_FACTOR);
/* Reset the notification request in case we timed out */
StrategyNotifyBgWriter(-1);
}
/*
* Emergency bailout if postmaster has died. This is to avoid the
* necessity for manual cleanup of all postmaster children.
*/
if (rc & WL_POSTMASTER_DEATH)
exit(1);
prev_hibernate = can_hibernate;
}
}
