gem5/src/dev/copy_engine.cc

779 lines
23 KiB
C++
Raw Normal View History

/*
* Copyright (c) 2012 ARM Limited
* All rights reserved
*
* The license below extends only to copyright in the software and shall
* not be construed as granting a license to any other intellectual
* property including but not limited to intellectual property relating
* to a hardware implementation of the functionality of the software
* licensed hereunder. You may use the software subject to the license
* terms below provided that you ensure that this notice is replicated
* unmodified and in its entirety in all distributions of the software,
* modified or unmodified, in source code or in binary form.
*
* Copyright (c) 2008 The Regents of The University of Michigan
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are
* met: redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer;
* redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution;
* neither the name of the copyright holders nor the names of its
* contributors may be used to endorse or promote products derived from
* this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
* "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
* LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
* A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
* OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
* SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
* LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
* DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
* THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
* OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
*
* Authors: Ali Saidi
*/
/* @file
* Device model for Intel's I/O AT DMA copy engine.
*/
#include <algorithm>
#include "base/cp_annotate.hh"
#include "base/trace.hh"
#include "debug/DMACopyEngine.hh"
#include "debug/Drain.hh"
#include "dev/copy_engine.hh"
#include "mem/packet.hh"
#include "mem/packet_access.hh"
#include "params/CopyEngine.hh"
#include "sim/stats.hh"
#include "sim/system.hh"
using namespace CopyEngineReg;
CopyEngine::CopyEngine(const Params *p)
: PciDevice(p)
{
// All Reg regs are initialized to 0 by default
regs.chanCount = p->ChanCnt;
regs.xferCap = findMsbSet(p->XferCap);
regs.attnStatus = 0;
if (regs.chanCount > 64)
fatal("CopyEngine interface doesn't support more than 64 DMA engines\n");
for (int x = 0; x < regs.chanCount; x++) {
CopyEngineChannel *ch = new CopyEngineChannel(this, x);
chan.push_back(ch);
}
}
CopyEngine::CopyEngineChannel::CopyEngineChannel(CopyEngine *_ce, int cid)
: cePort(_ce, _ce->sys),
ce(_ce), channelId(cid), busy(false), underReset(false),
refreshNext(false), latBeforeBegin(ce->params()->latBeforeBegin),
latAfterCompletion(ce->params()->latAfterCompletion),
completionDataReg(0), nextState(Idle), drainManager(NULL),
fetchCompleteEvent(this), addrCompleteEvent(this),
readCompleteEvent(this), writeCompleteEvent(this),
statusCompleteEvent(this)
{
cr.status.dma_transfer_status(3);
cr.descChainAddr = 0;
cr.completionAddr = 0;
curDmaDesc = new DmaDesc;
memset(curDmaDesc, 0, sizeof(DmaDesc));
copyBuffer = new uint8_t[ce->params()->XferCap];
}
CopyEngine::~CopyEngine()
{
for (int x = 0; x < chan.size(); x++) {
delete chan[x];
}
}
CopyEngine::CopyEngineChannel::~CopyEngineChannel()
{
delete curDmaDesc;
delete [] copyBuffer;
}
BaseMasterPort &
CopyEngine::getMasterPort(const std::string &if_name, PortID idx)
{
if (if_name != "dma") {
// pass it along to our super class
return PciDevice::getMasterPort(if_name, idx);
} else {
if (idx >= static_cast<int>(chan.size())) {
panic("CopyEngine::getMasterPort: unknown index %d\n", idx);
}
return chan[idx]->getMasterPort();
}
}
BaseMasterPort &
CopyEngine::CopyEngineChannel::getMasterPort()
{
return cePort;
}
void
CopyEngine::CopyEngineChannel::recvCommand()
{
if (cr.command.start_dma()) {
assert(!busy);
cr.status.dma_transfer_status(0);
nextState = DescriptorFetch;
fetchAddress = cr.descChainAddr;
if (ce->getDrainState() == DrainState::Running)
fetchDescriptor(cr.descChainAddr);
} else if (cr.command.append_dma()) {
if (!busy) {
nextState = AddressFetch;
if (ce->getDrainState() == DrainState::Running)
fetchNextAddr(lastDescriptorAddr);
} else
refreshNext = true;
} else if (cr.command.reset_dma()) {
if (busy)
underReset = true;
else {
cr.status.dma_transfer_status(3);
nextState = Idle;
}
} else if (cr.command.resume_dma() || cr.command.abort_dma() ||
cr.command.suspend_dma())
panic("Resume, Abort, and Suspend are not supported\n");
cr.command(0);
}
Tick
CopyEngine::read(PacketPtr pkt)
{
int bar;
Addr daddr;
if (!getBAR(pkt->getAddr(), bar, daddr))
panic("Invalid PCI memory access to unmapped memory.\n");
// Only Memory register BAR is allowed
assert(bar == 0);
int size = pkt->getSize();
if (size != sizeof(uint64_t) && size != sizeof(uint32_t) &&
size != sizeof(uint16_t) && size != sizeof(uint8_t)) {
panic("Unknown size for MMIO access: %d\n", pkt->getSize());
}
DPRINTF(DMACopyEngine, "Read device register %#X size: %d\n", daddr, size);
///
/// Handle read of register here
///
if (daddr < 0x80) {
switch (daddr) {
case GEN_CHANCOUNT:
assert(size == sizeof(regs.chanCount));
pkt->set<uint8_t>(regs.chanCount);
break;
case GEN_XFERCAP:
assert(size == sizeof(regs.xferCap));
pkt->set<uint8_t>(regs.xferCap);
break;
case GEN_INTRCTRL:
assert(size == sizeof(uint8_t));
pkt->set<uint8_t>(regs.intrctrl());
regs.intrctrl.master_int_enable(0);
break;
case GEN_ATTNSTATUS:
assert(size == sizeof(regs.attnStatus));
pkt->set<uint32_t>(regs.attnStatus);
regs.attnStatus = 0;
break;
default:
panic("Read request to unknown register number: %#x\n", daddr);
}
pkt->makeAtomicResponse();
return pioDelay;
}
// Find which channel we're accessing
int chanid = 0;
daddr -= 0x80;
while (daddr >= 0x80) {
chanid++;
daddr -= 0x80;
}
if (chanid >= regs.chanCount)
panic("Access to channel %d (device only configured for %d channels)",
chanid, regs.chanCount);
///
/// Channel registers are handled here
///
chan[chanid]->channelRead(pkt, daddr, size);
pkt->makeAtomicResponse();
return pioDelay;
}
void
CopyEngine::CopyEngineChannel::channelRead(Packet *pkt, Addr daddr, int size)
{
switch (daddr) {
case CHAN_CONTROL:
assert(size == sizeof(uint16_t));
pkt->set<uint16_t>(cr.ctrl());
cr.ctrl.in_use(1);
break;
case CHAN_STATUS:
assert(size == sizeof(uint64_t));
pkt->set<uint64_t>(cr.status() | ~busy);
break;
case CHAN_CHAINADDR:
assert(size == sizeof(uint64_t) || size == sizeof(uint32_t));
if (size == sizeof(uint64_t))
pkt->set<uint64_t>(cr.descChainAddr);
else
pkt->set<uint32_t>(bits(cr.descChainAddr,0,31));
break;
case CHAN_CHAINADDR_HIGH:
assert(size == sizeof(uint32_t));
pkt->set<uint32_t>(bits(cr.descChainAddr,32,63));
break;
case CHAN_COMMAND:
assert(size == sizeof(uint8_t));
pkt->set<uint32_t>(cr.command());
break;
case CHAN_CMPLNADDR:
assert(size == sizeof(uint64_t) || size == sizeof(uint32_t));
if (size == sizeof(uint64_t))
pkt->set<uint64_t>(cr.completionAddr);
else
pkt->set<uint32_t>(bits(cr.completionAddr,0,31));
break;
case CHAN_CMPLNADDR_HIGH:
assert(size == sizeof(uint32_t));
pkt->set<uint32_t>(bits(cr.completionAddr,32,63));
break;
case CHAN_ERROR:
assert(size == sizeof(uint32_t));
pkt->set<uint32_t>(cr.error());
break;
default:
panic("Read request to unknown channel register number: (%d)%#x\n",
channelId, daddr);
}
}
Tick
CopyEngine::write(PacketPtr pkt)
{
int bar;
Addr daddr;
if (!getBAR(pkt->getAddr(), bar, daddr))
panic("Invalid PCI memory access to unmapped memory.\n");
// Only Memory register BAR is allowed
assert(bar == 0);
int size = pkt->getSize();
///
/// Handle write of register here
///
if (size == sizeof(uint64_t)) {
uint64_t val M5_VAR_USED = pkt->get<uint64_t>();
DPRINTF(DMACopyEngine, "Wrote device register %#X value %#X\n", daddr, val);
} else if (size == sizeof(uint32_t)) {
uint32_t val M5_VAR_USED = pkt->get<uint32_t>();
DPRINTF(DMACopyEngine, "Wrote device register %#X value %#X\n", daddr, val);
} else if (size == sizeof(uint16_t)) {
uint16_t val M5_VAR_USED = pkt->get<uint16_t>();
DPRINTF(DMACopyEngine, "Wrote device register %#X value %#X\n", daddr, val);
} else if (size == sizeof(uint8_t)) {
uint8_t val M5_VAR_USED = pkt->get<uint8_t>();
DPRINTF(DMACopyEngine, "Wrote device register %#X value %#X\n", daddr, val);
} else {
panic("Unknown size for MMIO access: %d\n", size);
}
if (daddr < 0x80) {
switch (daddr) {
case GEN_CHANCOUNT:
case GEN_XFERCAP:
case GEN_ATTNSTATUS:
DPRINTF(DMACopyEngine, "Warning, ignorning write to register %x\n",
daddr);
break;
case GEN_INTRCTRL:
regs.intrctrl.master_int_enable(bits(pkt->get<uint8_t>(),0,1));
break;
default:
panic("Read request to unknown register number: %#x\n", daddr);
}
pkt->makeAtomicResponse();
return pioDelay;
}
// Find which channel we're accessing
int chanid = 0;
daddr -= 0x80;
while (daddr >= 0x80) {
chanid++;
daddr -= 0x80;
}
if (chanid >= regs.chanCount)
panic("Access to channel %d (device only configured for %d channels)",
chanid, regs.chanCount);
///
/// Channel registers are handled here
///
chan[chanid]->channelWrite(pkt, daddr, size);
pkt->makeAtomicResponse();
return pioDelay;
}
void
CopyEngine::CopyEngineChannel::channelWrite(Packet *pkt, Addr daddr, int size)
{
switch (daddr) {
case CHAN_CONTROL:
assert(size == sizeof(uint16_t));
int old_int_disable;
old_int_disable = cr.ctrl.interrupt_disable();
cr.ctrl(pkt->get<uint16_t>());
if (cr.ctrl.interrupt_disable())
cr.ctrl.interrupt_disable(0);
else
cr.ctrl.interrupt_disable(old_int_disable);
break;
case CHAN_STATUS:
assert(size == sizeof(uint64_t));
DPRINTF(DMACopyEngine, "Warning, ignorning write to register %x\n",
daddr);
break;
case CHAN_CHAINADDR:
assert(size == sizeof(uint64_t) || size == sizeof(uint32_t));
if (size == sizeof(uint64_t))
cr.descChainAddr = pkt->get<uint64_t>();
else
cr.descChainAddr = (uint64_t)pkt->get<uint32_t>() |
(cr.descChainAddr & ~mask(32));
DPRINTF(DMACopyEngine, "Chain Address %x\n", cr.descChainAddr);
break;
case CHAN_CHAINADDR_HIGH:
assert(size == sizeof(uint32_t));
cr.descChainAddr = ((uint64_t)pkt->get<uint32_t>() <<32) |
(cr.descChainAddr & mask(32));
DPRINTF(DMACopyEngine, "Chain Address %x\n", cr.descChainAddr);
break;
case CHAN_COMMAND:
assert(size == sizeof(uint8_t));
cr.command(pkt->get<uint8_t>());
recvCommand();
break;
case CHAN_CMPLNADDR:
assert(size == sizeof(uint64_t) || size == sizeof(uint32_t));
if (size == sizeof(uint64_t))
cr.completionAddr = pkt->get<uint64_t>();
else
cr.completionAddr = pkt->get<uint32_t>() |
(cr.completionAddr & ~mask(32));
break;
case CHAN_CMPLNADDR_HIGH:
assert(size == sizeof(uint32_t));
cr.completionAddr = ((uint64_t)pkt->get<uint32_t>() <<32) |
(cr.completionAddr & mask(32));
break;
case CHAN_ERROR:
assert(size == sizeof(uint32_t));
cr.error(~pkt->get<uint32_t>() & cr.error());
break;
default:
panic("Read request to unknown channel register number: (%d)%#x\n",
channelId, daddr);
}
}
void
CopyEngine::regStats()
{
using namespace Stats;
bytesCopied
.init(regs.chanCount)
.name(name() + ".bytes_copied")
.desc("Number of bytes copied by each engine")
.flags(total)
;
copiesProcessed
.init(regs.chanCount)
.name(name() + ".copies_processed")
.desc("Number of copies processed by each engine")
.flags(total)
;
}
void
CopyEngine::CopyEngineChannel::fetchDescriptor(Addr address)
{
anDq();
anBegin("FetchDescriptor");
DPRINTF(DMACopyEngine, "Reading descriptor from at memory location %#x(%#x)\n",
address, ce->platform->pciToDma(address));
assert(address);
busy = true;
DPRINTF(DMACopyEngine, "dmaAction: %#x, %d bytes, to addr %#x\n",
ce->platform->pciToDma(address), sizeof(DmaDesc), curDmaDesc);
cePort.dmaAction(MemCmd::ReadReq, ce->platform->pciToDma(address),
sizeof(DmaDesc), &fetchCompleteEvent,
(uint8_t*)curDmaDesc, latBeforeBegin);
lastDescriptorAddr = address;
}
void
CopyEngine::CopyEngineChannel::fetchDescComplete()
{
DPRINTF(DMACopyEngine, "Read of descriptor complete\n");
if ((curDmaDesc->command & DESC_CTRL_NULL)) {
DPRINTF(DMACopyEngine, "Got NULL descriptor, skipping\n");
assert(!(curDmaDesc->command & DESC_CTRL_CP_STS));
if (curDmaDesc->command & DESC_CTRL_CP_STS) {
panic("Shouldn't be able to get here\n");
nextState = CompletionWrite;
if (inDrain()) return;
writeCompletionStatus();
} else {
anBegin("Idle");
anWait();
busy = false;
nextState = Idle;
inDrain();
}
return;
}
if (curDmaDesc->command & ~DESC_CTRL_CP_STS)
panic("Descriptor has flag other that completion status set\n");
nextState = DMARead;
if (inDrain()) return;
readCopyBytes();
}
void
CopyEngine::CopyEngineChannel::readCopyBytes()
{
anBegin("ReadCopyBytes");
DPRINTF(DMACopyEngine, "Reading %d bytes from buffer to memory location %#x(%#x)\n",
curDmaDesc->len, curDmaDesc->dest,
ce->platform->pciToDma(curDmaDesc->src));
cePort.dmaAction(MemCmd::ReadReq, ce->platform->pciToDma(curDmaDesc->src),
curDmaDesc->len, &readCompleteEvent, copyBuffer, 0);
}
void
CopyEngine::CopyEngineChannel::readCopyBytesComplete()
{
DPRINTF(DMACopyEngine, "Read of bytes to copy complete\n");
nextState = DMAWrite;
if (inDrain()) return;
writeCopyBytes();
}
void
CopyEngine::CopyEngineChannel::writeCopyBytes()
{
anBegin("WriteCopyBytes");
DPRINTF(DMACopyEngine, "Writing %d bytes from buffer to memory location %#x(%#x)\n",
curDmaDesc->len, curDmaDesc->dest,
ce->platform->pciToDma(curDmaDesc->dest));
cePort.dmaAction(MemCmd::WriteReq, ce->platform->pciToDma(curDmaDesc->dest),
curDmaDesc->len, &writeCompleteEvent, copyBuffer, 0);
ce->bytesCopied[channelId] += curDmaDesc->len;
ce->copiesProcessed[channelId]++;
}
void
CopyEngine::CopyEngineChannel::writeCopyBytesComplete()
{
DPRINTF(DMACopyEngine, "Write of bytes to copy complete user1: %#x\n",
curDmaDesc->user1);
cr.status.compl_desc_addr(lastDescriptorAddr >> 6);
completionDataReg = cr.status() | 1;
anQ("DMAUsedDescQ", channelId, 1);
anQ("AppRecvQ", curDmaDesc->user1, curDmaDesc->len);
if (curDmaDesc->command & DESC_CTRL_CP_STS) {
nextState = CompletionWrite;
if (inDrain()) return;
writeCompletionStatus();
return;
}
continueProcessing();
}
void
CopyEngine::CopyEngineChannel::continueProcessing()
{
busy = false;
if (underReset) {
anBegin("Reset");
anWait();
underReset = false;
refreshNext = false;
busy = false;
nextState = Idle;
return;
}
if (curDmaDesc->next) {
nextState = DescriptorFetch;
fetchAddress = curDmaDesc->next;
if (inDrain()) return;
fetchDescriptor(curDmaDesc->next);
} else if (refreshNext) {
nextState = AddressFetch;
refreshNext = false;
if (inDrain()) return;
fetchNextAddr(lastDescriptorAddr);
} else {
inDrain();
nextState = Idle;
anWait();
anBegin("Idle");
}
}
void
CopyEngine::CopyEngineChannel::writeCompletionStatus()
{
anBegin("WriteCompletionStatus");
DPRINTF(DMACopyEngine, "Writing completion status %#x to address %#x(%#x)\n",
completionDataReg, cr.completionAddr,
ce->platform->pciToDma(cr.completionAddr));
cePort.dmaAction(MemCmd::WriteReq,
ce->platform->pciToDma(cr.completionAddr),
sizeof(completionDataReg), &statusCompleteEvent,
(uint8_t*)&completionDataReg, latAfterCompletion);
}
void
CopyEngine::CopyEngineChannel::writeStatusComplete()
{
DPRINTF(DMACopyEngine, "Writing completion status complete\n");
continueProcessing();
}
void
CopyEngine::CopyEngineChannel::fetchNextAddr(Addr address)
{
anBegin("FetchNextAddr");
DPRINTF(DMACopyEngine, "Fetching next address...\n");
busy = true;
cePort.dmaAction(MemCmd::ReadReq,
ce->platform->pciToDma(address + offsetof(DmaDesc, next)),
sizeof(Addr), &addrCompleteEvent,
(uint8_t*)curDmaDesc + offsetof(DmaDesc, next), 0);
}
void
CopyEngine::CopyEngineChannel::fetchAddrComplete()
{
DPRINTF(DMACopyEngine, "Fetching next address complete: %#x\n",
curDmaDesc->next);
if (!curDmaDesc->next) {
DPRINTF(DMACopyEngine, "Got NULL descriptor, nothing more to do\n");
busy = false;
nextState = Idle;
anWait();
anBegin("Idle");
inDrain();
return;
}
nextState = DescriptorFetch;
fetchAddress = curDmaDesc->next;
if (inDrain()) return;
fetchDescriptor(curDmaDesc->next);
}
bool
CopyEngine::CopyEngineChannel::inDrain()
{
if (ce->getDrainState() == DrainState::Draining) {
DPRINTF(Drain, "CopyEngine done draining, processing drain event\n");
assert(drainManager);
drainManager->signalDrainDone();
drainManager = NULL;
}
return ce->getDrainState() != DrainState::Running;
}
unsigned int
CopyEngine::CopyEngineChannel::drain(DrainManager *dm)
{
if (nextState == Idle || ce->getDrainState() != DrainState::Running)
return 0;
unsigned int count = 1;
count += cePort.drain(dm);
DPRINTF(Drain, "CopyEngineChannel not drained\n");
this->drainManager = dm;
return count;
}
unsigned int
CopyEngine::drain(DrainManager *dm)
{
unsigned int count;
count = pioPort.drain(dm) + dmaPort.drain(dm) + configPort.drain(dm);
for (int x = 0;x < chan.size(); x++)
count += chan[x]->drain(dm);
if (count)
setDrainState(DrainState::Draining);
else
setDrainState(DrainState::Drained);
DPRINTF(Drain, "CopyEngine not drained\n");
return count;
}
void
sim: Refactor the serialization base class Objects that are can be serialized are supposed to inherit from the Serializable class. This class is meant to provide a unified API for such objects. However, so far it has mainly been used by SimObjects due to some fundamental design limitations. This changeset redesigns to the serialization interface to make it more generic and hide the underlying checkpoint storage. Specifically: * Add a set of APIs to serialize into a subsection of the current object. Previously, objects that needed this functionality would use ad-hoc solutions using nameOut() and section name generation. In the new world, an object that implements the interface has the methods serializeSection() and unserializeSection() that serialize into a named /subsection/ of the current object. Calling serialize() serializes an object into the current section. * Move the name() method from Serializable to SimObject as it is no longer needed for serialization. The fully qualified section name is generated by the main serialization code on the fly as objects serialize sub-objects. * Add a scoped ScopedCheckpointSection helper class. Some objects need to serialize data structures, that are not deriving from Serializable, into subsections. Previously, this was done using nameOut() and manual section name generation. To simplify this, this changeset introduces a ScopedCheckpointSection() helper class. When this class is instantiated, it adds a new /subsection/ and subsequent serialization calls during the lifetime of this helper class happen inside this section (or a subsection in case of nested sections). * The serialize() call is now const which prevents accidental state manipulation during serialization. Objects that rely on modifying state can use the serializeOld() call instead. The default implementation simply calls serialize(). Note: The old-style calls need to be explicitly called using the serializeOld()/serializeSectionOld() style APIs. These are used by default when serializing SimObjects. * Both the input and output checkpoints now use their own named types. This hides underlying checkpoint implementation from objects that need checkpointing and makes it easier to change the underlying checkpoint storage code.
2015-07-07 10:51:03 +02:00
CopyEngine::serialize(CheckpointOut &cp) const
{
sim: Refactor the serialization base class Objects that are can be serialized are supposed to inherit from the Serializable class. This class is meant to provide a unified API for such objects. However, so far it has mainly been used by SimObjects due to some fundamental design limitations. This changeset redesigns to the serialization interface to make it more generic and hide the underlying checkpoint storage. Specifically: * Add a set of APIs to serialize into a subsection of the current object. Previously, objects that needed this functionality would use ad-hoc solutions using nameOut() and section name generation. In the new world, an object that implements the interface has the methods serializeSection() and unserializeSection() that serialize into a named /subsection/ of the current object. Calling serialize() serializes an object into the current section. * Move the name() method from Serializable to SimObject as it is no longer needed for serialization. The fully qualified section name is generated by the main serialization code on the fly as objects serialize sub-objects. * Add a scoped ScopedCheckpointSection helper class. Some objects need to serialize data structures, that are not deriving from Serializable, into subsections. Previously, this was done using nameOut() and manual section name generation. To simplify this, this changeset introduces a ScopedCheckpointSection() helper class. When this class is instantiated, it adds a new /subsection/ and subsequent serialization calls during the lifetime of this helper class happen inside this section (or a subsection in case of nested sections). * The serialize() call is now const which prevents accidental state manipulation during serialization. Objects that rely on modifying state can use the serializeOld() call instead. The default implementation simply calls serialize(). Note: The old-style calls need to be explicitly called using the serializeOld()/serializeSectionOld() style APIs. These are used by default when serializing SimObjects. * Both the input and output checkpoints now use their own named types. This hides underlying checkpoint implementation from objects that need checkpointing and makes it easier to change the underlying checkpoint storage code.
2015-07-07 10:51:03 +02:00
PciDevice::serialize(cp);
regs.serialize(cp);
for (int x =0; x < chan.size(); x++)
chan[x]->serializeSection(cp, csprintf("channel%d", x));
}
void
sim: Refactor the serialization base class Objects that are can be serialized are supposed to inherit from the Serializable class. This class is meant to provide a unified API for such objects. However, so far it has mainly been used by SimObjects due to some fundamental design limitations. This changeset redesigns to the serialization interface to make it more generic and hide the underlying checkpoint storage. Specifically: * Add a set of APIs to serialize into a subsection of the current object. Previously, objects that needed this functionality would use ad-hoc solutions using nameOut() and section name generation. In the new world, an object that implements the interface has the methods serializeSection() and unserializeSection() that serialize into a named /subsection/ of the current object. Calling serialize() serializes an object into the current section. * Move the name() method from Serializable to SimObject as it is no longer needed for serialization. The fully qualified section name is generated by the main serialization code on the fly as objects serialize sub-objects. * Add a scoped ScopedCheckpointSection helper class. Some objects need to serialize data structures, that are not deriving from Serializable, into subsections. Previously, this was done using nameOut() and manual section name generation. To simplify this, this changeset introduces a ScopedCheckpointSection() helper class. When this class is instantiated, it adds a new /subsection/ and subsequent serialization calls during the lifetime of this helper class happen inside this section (or a subsection in case of nested sections). * The serialize() call is now const which prevents accidental state manipulation during serialization. Objects that rely on modifying state can use the serializeOld() call instead. The default implementation simply calls serialize(). Note: The old-style calls need to be explicitly called using the serializeOld()/serializeSectionOld() style APIs. These are used by default when serializing SimObjects. * Both the input and output checkpoints now use their own named types. This hides underlying checkpoint implementation from objects that need checkpointing and makes it easier to change the underlying checkpoint storage code.
2015-07-07 10:51:03 +02:00
CopyEngine::unserialize(CheckpointIn &cp)
{
sim: Refactor the serialization base class Objects that are can be serialized are supposed to inherit from the Serializable class. This class is meant to provide a unified API for such objects. However, so far it has mainly been used by SimObjects due to some fundamental design limitations. This changeset redesigns to the serialization interface to make it more generic and hide the underlying checkpoint storage. Specifically: * Add a set of APIs to serialize into a subsection of the current object. Previously, objects that needed this functionality would use ad-hoc solutions using nameOut() and section name generation. In the new world, an object that implements the interface has the methods serializeSection() and unserializeSection() that serialize into a named /subsection/ of the current object. Calling serialize() serializes an object into the current section. * Move the name() method from Serializable to SimObject as it is no longer needed for serialization. The fully qualified section name is generated by the main serialization code on the fly as objects serialize sub-objects. * Add a scoped ScopedCheckpointSection helper class. Some objects need to serialize data structures, that are not deriving from Serializable, into subsections. Previously, this was done using nameOut() and manual section name generation. To simplify this, this changeset introduces a ScopedCheckpointSection() helper class. When this class is instantiated, it adds a new /subsection/ and subsequent serialization calls during the lifetime of this helper class happen inside this section (or a subsection in case of nested sections). * The serialize() call is now const which prevents accidental state manipulation during serialization. Objects that rely on modifying state can use the serializeOld() call instead. The default implementation simply calls serialize(). Note: The old-style calls need to be explicitly called using the serializeOld()/serializeSectionOld() style APIs. These are used by default when serializing SimObjects. * Both the input and output checkpoints now use their own named types. This hides underlying checkpoint implementation from objects that need checkpointing and makes it easier to change the underlying checkpoint storage code.
2015-07-07 10:51:03 +02:00
PciDevice::unserialize(cp);
regs.unserialize(cp);
for (int x = 0; x < chan.size(); x++)
sim: Refactor the serialization base class Objects that are can be serialized are supposed to inherit from the Serializable class. This class is meant to provide a unified API for such objects. However, so far it has mainly been used by SimObjects due to some fundamental design limitations. This changeset redesigns to the serialization interface to make it more generic and hide the underlying checkpoint storage. Specifically: * Add a set of APIs to serialize into a subsection of the current object. Previously, objects that needed this functionality would use ad-hoc solutions using nameOut() and section name generation. In the new world, an object that implements the interface has the methods serializeSection() and unserializeSection() that serialize into a named /subsection/ of the current object. Calling serialize() serializes an object into the current section. * Move the name() method from Serializable to SimObject as it is no longer needed for serialization. The fully qualified section name is generated by the main serialization code on the fly as objects serialize sub-objects. * Add a scoped ScopedCheckpointSection helper class. Some objects need to serialize data structures, that are not deriving from Serializable, into subsections. Previously, this was done using nameOut() and manual section name generation. To simplify this, this changeset introduces a ScopedCheckpointSection() helper class. When this class is instantiated, it adds a new /subsection/ and subsequent serialization calls during the lifetime of this helper class happen inside this section (or a subsection in case of nested sections). * The serialize() call is now const which prevents accidental state manipulation during serialization. Objects that rely on modifying state can use the serializeOld() call instead. The default implementation simply calls serialize(). Note: The old-style calls need to be explicitly called using the serializeOld()/serializeSectionOld() style APIs. These are used by default when serializing SimObjects. * Both the input and output checkpoints now use their own named types. This hides underlying checkpoint implementation from objects that need checkpointing and makes it easier to change the underlying checkpoint storage code.
2015-07-07 10:51:03 +02:00
chan[x]->unserializeSection(cp, csprintf("channel%d", x));
}
void
sim: Refactor the serialization base class Objects that are can be serialized are supposed to inherit from the Serializable class. This class is meant to provide a unified API for such objects. However, so far it has mainly been used by SimObjects due to some fundamental design limitations. This changeset redesigns to the serialization interface to make it more generic and hide the underlying checkpoint storage. Specifically: * Add a set of APIs to serialize into a subsection of the current object. Previously, objects that needed this functionality would use ad-hoc solutions using nameOut() and section name generation. In the new world, an object that implements the interface has the methods serializeSection() and unserializeSection() that serialize into a named /subsection/ of the current object. Calling serialize() serializes an object into the current section. * Move the name() method from Serializable to SimObject as it is no longer needed for serialization. The fully qualified section name is generated by the main serialization code on the fly as objects serialize sub-objects. * Add a scoped ScopedCheckpointSection helper class. Some objects need to serialize data structures, that are not deriving from Serializable, into subsections. Previously, this was done using nameOut() and manual section name generation. To simplify this, this changeset introduces a ScopedCheckpointSection() helper class. When this class is instantiated, it adds a new /subsection/ and subsequent serialization calls during the lifetime of this helper class happen inside this section (or a subsection in case of nested sections). * The serialize() call is now const which prevents accidental state manipulation during serialization. Objects that rely on modifying state can use the serializeOld() call instead. The default implementation simply calls serialize(). Note: The old-style calls need to be explicitly called using the serializeOld()/serializeSectionOld() style APIs. These are used by default when serializing SimObjects. * Both the input and output checkpoints now use their own named types. This hides underlying checkpoint implementation from objects that need checkpointing and makes it easier to change the underlying checkpoint storage code.
2015-07-07 10:51:03 +02:00
CopyEngine::CopyEngineChannel::serialize(CheckpointOut &cp) const
{
SERIALIZE_SCALAR(channelId);
SERIALIZE_SCALAR(busy);
SERIALIZE_SCALAR(underReset);
SERIALIZE_SCALAR(refreshNext);
SERIALIZE_SCALAR(lastDescriptorAddr);
SERIALIZE_SCALAR(completionDataReg);
SERIALIZE_SCALAR(fetchAddress);
int nextState = this->nextState;
SERIALIZE_SCALAR(nextState);
sim: Refactor the serialization base class Objects that are can be serialized are supposed to inherit from the Serializable class. This class is meant to provide a unified API for such objects. However, so far it has mainly been used by SimObjects due to some fundamental design limitations. This changeset redesigns to the serialization interface to make it more generic and hide the underlying checkpoint storage. Specifically: * Add a set of APIs to serialize into a subsection of the current object. Previously, objects that needed this functionality would use ad-hoc solutions using nameOut() and section name generation. In the new world, an object that implements the interface has the methods serializeSection() and unserializeSection() that serialize into a named /subsection/ of the current object. Calling serialize() serializes an object into the current section. * Move the name() method from Serializable to SimObject as it is no longer needed for serialization. The fully qualified section name is generated by the main serialization code on the fly as objects serialize sub-objects. * Add a scoped ScopedCheckpointSection helper class. Some objects need to serialize data structures, that are not deriving from Serializable, into subsections. Previously, this was done using nameOut() and manual section name generation. To simplify this, this changeset introduces a ScopedCheckpointSection() helper class. When this class is instantiated, it adds a new /subsection/ and subsequent serialization calls during the lifetime of this helper class happen inside this section (or a subsection in case of nested sections). * The serialize() call is now const which prevents accidental state manipulation during serialization. Objects that rely on modifying state can use the serializeOld() call instead. The default implementation simply calls serialize(). Note: The old-style calls need to be explicitly called using the serializeOld()/serializeSectionOld() style APIs. These are used by default when serializing SimObjects. * Both the input and output checkpoints now use their own named types. This hides underlying checkpoint implementation from objects that need checkpointing and makes it easier to change the underlying checkpoint storage code.
2015-07-07 10:51:03 +02:00
arrayParamOut(cp, "curDmaDesc", (uint8_t*)curDmaDesc, sizeof(DmaDesc));
SERIALIZE_ARRAY(copyBuffer, ce->params()->XferCap);
sim: Refactor the serialization base class Objects that are can be serialized are supposed to inherit from the Serializable class. This class is meant to provide a unified API for such objects. However, so far it has mainly been used by SimObjects due to some fundamental design limitations. This changeset redesigns to the serialization interface to make it more generic and hide the underlying checkpoint storage. Specifically: * Add a set of APIs to serialize into a subsection of the current object. Previously, objects that needed this functionality would use ad-hoc solutions using nameOut() and section name generation. In the new world, an object that implements the interface has the methods serializeSection() and unserializeSection() that serialize into a named /subsection/ of the current object. Calling serialize() serializes an object into the current section. * Move the name() method from Serializable to SimObject as it is no longer needed for serialization. The fully qualified section name is generated by the main serialization code on the fly as objects serialize sub-objects. * Add a scoped ScopedCheckpointSection helper class. Some objects need to serialize data structures, that are not deriving from Serializable, into subsections. Previously, this was done using nameOut() and manual section name generation. To simplify this, this changeset introduces a ScopedCheckpointSection() helper class. When this class is instantiated, it adds a new /subsection/ and subsequent serialization calls during the lifetime of this helper class happen inside this section (or a subsection in case of nested sections). * The serialize() call is now const which prevents accidental state manipulation during serialization. Objects that rely on modifying state can use the serializeOld() call instead. The default implementation simply calls serialize(). Note: The old-style calls need to be explicitly called using the serializeOld()/serializeSectionOld() style APIs. These are used by default when serializing SimObjects. * Both the input and output checkpoints now use their own named types. This hides underlying checkpoint implementation from objects that need checkpointing and makes it easier to change the underlying checkpoint storage code.
2015-07-07 10:51:03 +02:00
cr.serialize(cp);
}
void
sim: Refactor the serialization base class Objects that are can be serialized are supposed to inherit from the Serializable class. This class is meant to provide a unified API for such objects. However, so far it has mainly been used by SimObjects due to some fundamental design limitations. This changeset redesigns to the serialization interface to make it more generic and hide the underlying checkpoint storage. Specifically: * Add a set of APIs to serialize into a subsection of the current object. Previously, objects that needed this functionality would use ad-hoc solutions using nameOut() and section name generation. In the new world, an object that implements the interface has the methods serializeSection() and unserializeSection() that serialize into a named /subsection/ of the current object. Calling serialize() serializes an object into the current section. * Move the name() method from Serializable to SimObject as it is no longer needed for serialization. The fully qualified section name is generated by the main serialization code on the fly as objects serialize sub-objects. * Add a scoped ScopedCheckpointSection helper class. Some objects need to serialize data structures, that are not deriving from Serializable, into subsections. Previously, this was done using nameOut() and manual section name generation. To simplify this, this changeset introduces a ScopedCheckpointSection() helper class. When this class is instantiated, it adds a new /subsection/ and subsequent serialization calls during the lifetime of this helper class happen inside this section (or a subsection in case of nested sections). * The serialize() call is now const which prevents accidental state manipulation during serialization. Objects that rely on modifying state can use the serializeOld() call instead. The default implementation simply calls serialize(). Note: The old-style calls need to be explicitly called using the serializeOld()/serializeSectionOld() style APIs. These are used by default when serializing SimObjects. * Both the input and output checkpoints now use their own named types. This hides underlying checkpoint implementation from objects that need checkpointing and makes it easier to change the underlying checkpoint storage code.
2015-07-07 10:51:03 +02:00
CopyEngine::CopyEngineChannel::unserialize(CheckpointIn &cp)
{
UNSERIALIZE_SCALAR(channelId);
UNSERIALIZE_SCALAR(busy);
UNSERIALIZE_SCALAR(underReset);
UNSERIALIZE_SCALAR(refreshNext);
UNSERIALIZE_SCALAR(lastDescriptorAddr);
UNSERIALIZE_SCALAR(completionDataReg);
UNSERIALIZE_SCALAR(fetchAddress);
int nextState;
UNSERIALIZE_SCALAR(nextState);
this->nextState = (ChannelState)nextState;
sim: Refactor the serialization base class Objects that are can be serialized are supposed to inherit from the Serializable class. This class is meant to provide a unified API for such objects. However, so far it has mainly been used by SimObjects due to some fundamental design limitations. This changeset redesigns to the serialization interface to make it more generic and hide the underlying checkpoint storage. Specifically: * Add a set of APIs to serialize into a subsection of the current object. Previously, objects that needed this functionality would use ad-hoc solutions using nameOut() and section name generation. In the new world, an object that implements the interface has the methods serializeSection() and unserializeSection() that serialize into a named /subsection/ of the current object. Calling serialize() serializes an object into the current section. * Move the name() method from Serializable to SimObject as it is no longer needed for serialization. The fully qualified section name is generated by the main serialization code on the fly as objects serialize sub-objects. * Add a scoped ScopedCheckpointSection helper class. Some objects need to serialize data structures, that are not deriving from Serializable, into subsections. Previously, this was done using nameOut() and manual section name generation. To simplify this, this changeset introduces a ScopedCheckpointSection() helper class. When this class is instantiated, it adds a new /subsection/ and subsequent serialization calls during the lifetime of this helper class happen inside this section (or a subsection in case of nested sections). * The serialize() call is now const which prevents accidental state manipulation during serialization. Objects that rely on modifying state can use the serializeOld() call instead. The default implementation simply calls serialize(). Note: The old-style calls need to be explicitly called using the serializeOld()/serializeSectionOld() style APIs. These are used by default when serializing SimObjects. * Both the input and output checkpoints now use their own named types. This hides underlying checkpoint implementation from objects that need checkpointing and makes it easier to change the underlying checkpoint storage code.
2015-07-07 10:51:03 +02:00
arrayParamIn(cp, "curDmaDesc", (uint8_t*)curDmaDesc, sizeof(DmaDesc));
UNSERIALIZE_ARRAY(copyBuffer, ce->params()->XferCap);
sim: Refactor the serialization base class Objects that are can be serialized are supposed to inherit from the Serializable class. This class is meant to provide a unified API for such objects. However, so far it has mainly been used by SimObjects due to some fundamental design limitations. This changeset redesigns to the serialization interface to make it more generic and hide the underlying checkpoint storage. Specifically: * Add a set of APIs to serialize into a subsection of the current object. Previously, objects that needed this functionality would use ad-hoc solutions using nameOut() and section name generation. In the new world, an object that implements the interface has the methods serializeSection() and unserializeSection() that serialize into a named /subsection/ of the current object. Calling serialize() serializes an object into the current section. * Move the name() method from Serializable to SimObject as it is no longer needed for serialization. The fully qualified section name is generated by the main serialization code on the fly as objects serialize sub-objects. * Add a scoped ScopedCheckpointSection helper class. Some objects need to serialize data structures, that are not deriving from Serializable, into subsections. Previously, this was done using nameOut() and manual section name generation. To simplify this, this changeset introduces a ScopedCheckpointSection() helper class. When this class is instantiated, it adds a new /subsection/ and subsequent serialization calls during the lifetime of this helper class happen inside this section (or a subsection in case of nested sections). * The serialize() call is now const which prevents accidental state manipulation during serialization. Objects that rely on modifying state can use the serializeOld() call instead. The default implementation simply calls serialize(). Note: The old-style calls need to be explicitly called using the serializeOld()/serializeSectionOld() style APIs. These are used by default when serializing SimObjects. * Both the input and output checkpoints now use their own named types. This hides underlying checkpoint implementation from objects that need checkpointing and makes it easier to change the underlying checkpoint storage code.
2015-07-07 10:51:03 +02:00
cr.unserialize(cp);
}
void
CopyEngine::CopyEngineChannel::restartStateMachine()
{
switch(nextState) {
case AddressFetch:
fetchNextAddr(lastDescriptorAddr);
break;
case DescriptorFetch:
fetchDescriptor(fetchAddress);
break;
case DMARead:
readCopyBytes();
break;
case DMAWrite:
writeCopyBytes();
break;
case CompletionWrite:
writeCompletionStatus();
break;
case Idle:
break;
default:
panic("Unknown state for CopyEngineChannel\n");
}
}
void
CopyEngine::drainResume()
{
Drainable::drainResume();
for (int x = 0;x < chan.size(); x++)
chan[x]->drainResume();
}
void
CopyEngine::CopyEngineChannel::drainResume()
{
DPRINTF(DMACopyEngine, "Restarting state machine at state %d\n", nextState);
restartStateMachine();
}
CopyEngine *
CopyEngineParams::create()
{
return new CopyEngine(this);
}