prac/mpcio.cpp

#include <sys/time.h>      // getrusage
#include <sys/resource.h>  // getrusage
#include "mpcio.hpp"
#include "rdpf.hpp"
#include "cdpf.hpp"
#include "bitutils.hpp"
#include "coroutine.hpp"

// T is the type being stored
// N is a type whose "name" static member is a string naming the type
//   so that we can report something useful to the user if they try
//   to read a type that we don't have any more values for
template<typename T, typename N>
PreCompStorage<T,N>::PreCompStorage(unsigned player, ProcessingMode mode,
        const char *filenameprefix, unsigned thread_num) :
        name(N::name), depth(0)
{
    init(player, mode, filenameprefix, thread_num);
}

template<typename T, typename N>
void PreCompStorage<T,N>::init(unsigned player, ProcessingMode mode,
        const char *filenameprefix, unsigned thread_num, nbits_t depth)
{
    if (mode != MODE_ONLINE) return;
    std::string filename(filenameprefix);
    char suffix[20];
    if (depth) {
        this->depth = depth;
        sprintf(suffix, "%02d.p%d.t%u", depth, player%10, thread_num);
    } else {
        sprintf(suffix, ".p%d.t%u", player%10, thread_num);
    }
    filename.append(suffix);
    storage.open(filename);
    // It's OK if not every file exists; so don't worry about checking
    // for errors here.  We'll report an error in get() if we actually
    // try to use a value for which we don't have a precomputed file.
    count = 0;
}

template<typename T, typename N>
void PreCompStorage<T,N>::get(T& nextval)
{
    storage >> nextval;
    if (!storage.good()) {
        std::cerr << "Failed to read precomputed value from " << name;
        if (depth) {
            std::cerr << (int)depth;
        }
        std::cerr << " storage\n";
        exit(1);
    }
    ++count;
}

void MPCSingleIO::async_send_from_msgqueue()
{
#ifdef SEND_LAMPORT_CLOCKS
    std::vector<boost::asio::const_buffer> tosend;
    tosend.push_back(boost::asio::buffer(messagequeue.front().header));
    tosend.push_back(boost::asio::buffer(messagequeue.front().message));
#endif
    boost::asio::async_write(sock,
#ifdef SEND_LAMPORT_CLOCKS
        tosend,
#else
        boost::asio::buffer(messagequeue.front()),
#endif
        [&](boost::system::error_code ec, std::size_t amt){
            messagequeuelock.lock();
            messagequeue.pop();
            if (messagequeue.size() > 0) {
                async_send_from_msgqueue();
            }
            messagequeuelock.unlock();
        });
}

size_t MPCSingleIO::queue(const void *data, size_t len, lamport_t lamport)
{
    // Is this a new message?
    size_t newmsg = 0;

    dataqueue.append((const char *)data, len);

    // If this is the first queue() since the last explicit send(),
    // which we'll know because message_lamport will be nullopt, set
    // message_lamport to the current Lamport clock.  Note that the
    // boolean test tests whether message_lamport is nullopt, not
    // whether its value is zero.
    if (!message_lamport) {
        message_lamport = lamport;
        newmsg = 1;
    }

#ifdef VERBOSE_COMMS
    printf("Queue %s.%d len=%lu lamp=%u: ", dest.c_str(), thread_num,
        len, message_lamport.value());
    for (size_t i=0;i<len;++i) {
        printf("%02x", ((const unsigned char*)data)[i]);
    }
    printf("\n");
#endif

    // If we already have some full packets worth of data, may as
    // well send it.
    if (dataqueue.size() > 28800) {
        send(true);
    }

    return newmsg;
}

void MPCSingleIO::send(bool implicit_send)
{
    size_t thissize = dataqueue.size();
    // Ignore spurious calls to send(), except for resetting
    // message_lamport if this was an explicit send().
    if (thissize == 0) {
#ifdef SEND_LAMPORT_CLOCKS
        // If this was an explicit send(), reset the message_lamport so
        // that it gets updated at the next queue().
        if (!implicit_send) {
            message_lamport.reset();
        }
#endif
        return;
    }

#ifdef RECORD_IOTRACE
    iotrace.push_back(thissize);
#endif

    messagequeuelock.lock();
    // Move the current message to send into the message queue (this
    // moves a pointer to the data, not copying the data itself)
#ifdef SEND_LAMPORT_CLOCKS
    messagequeue.emplace(std::move(dataqueue),
        message_lamport.value());
    // If this was an explicit send(), reset the message_lamport so
    // that it gets updated at the next queue().
    if (!implicit_send) {
        message_lamport.reset();
    }
#else
    messagequeue.emplace(std::move(dataqueue));
#endif
    // If this is now the first thing in the message queue, launch
    // an async_write to write it
    if (messagequeue.size() == 1) {
        async_send_from_msgqueue();
    }
    messagequeuelock.unlock();
}

size_t MPCSingleIO::recv(void *data, size_t len, lamport_t &lamport)
{
#ifdef VERBOSE_COMMS
    size_t orig_len = len;
    printf("Recv %s.%d len=%lu lamp=%u ", dest.c_str(), thread_num,
        len, lamport);
#endif

#ifdef SEND_LAMPORT_CLOCKS
    char *cdata = (char *)data;
    size_t res = 0;

    while (len > 0) {
        while (recvdataremain == 0) {
            // Read a new header
            char hdr[sizeof(uint32_t) + sizeof(lamport_t)];
            uint32_t datalen;
            lamport_t recv_lamport;
            boost::asio::read(sock, boost::asio::buffer(hdr, sizeof(hdr)));
            memmove(&datalen, hdr, sizeof(datalen));
            memmove(&recv_lamport, hdr+sizeof(datalen), sizeof(lamport_t));
            lamport_t new_lamport = recv_lamport + 1;
            if (lamport < new_lamport) {
                lamport = new_lamport;
            }
            if (datalen > 0) {
                recvdata.resize(datalen, '\0');
                boost::asio::read(sock, boost::asio::buffer(recvdata));
                recvdataremain = datalen;
            }
        }
        size_t amttoread = len;
        if (amttoread > recvdataremain) {
            amttoread = recvdataremain;
        }
        memmove(cdata, recvdata.data()+recvdata.size()-recvdataremain,
            amttoread);
        cdata += amttoread;
        len -= amttoread;
        recvdataremain -= amttoread;
        res += amttoread;
    }
#else
    size_t res = boost::asio::read(sock, boost::asio::buffer(data, len));
#endif
#ifdef VERBOSE_COMMS
    printf("nlamp=%u: ", lamport);
    for (size_t i=0;i<orig_len;++i) {
        printf("%02x", ((const unsigned char*)data)[i]);
    }
    printf("\n");
#endif
#ifdef RECORD_IOTRACE
    iotrace.push_back(-(ssize_t(res)));
#endif
    return res;
}

#ifdef RECORD_IOTRACE
void MPCSingleIO::dumptrace(std::ostream &os, const char *label)
{
    if (label) {
        os << label << " ";
    }
    os << "IO trace:";
    for (auto& s: iotrace) {
        os << " " << s;
    }
    os << "\n";
}
#endif

void MPCIO::reset_stats()
{
    msgs_sent.clear();
    msg_bytes_sent.clear();
    aes_ops.clear();
    for (size_t i=0; i<num_threads; ++i) {
        msgs_sent.push_back(0);
        msg_bytes_sent.push_back(0);
        aes_ops.push_back(0);
    }
    steady_start = boost::chrono::steady_clock::now();
    cpu_start = boost::chrono::process_cpu_clock::now();
}

// Report the memory usage

void MPCIO::dump_memusage(std::ostream &os)
{
    struct rusage ru;
    getrusage(RUSAGE_SELF, &ru);
    os << "Mem: " << ru.ru_maxrss << " KiB\n";
}

void MPCIO::dump_stats(std::ostream &os)
{
    size_t tot_msgs_sent = 0;
    size_t tot_msg_bytes_sent = 0;
    size_t tot_aes_ops = 0;
    for (auto& n : msgs_sent) {
        tot_msgs_sent += n;
    }
    for (auto& n : msg_bytes_sent) {
        tot_msg_bytes_sent += n;
    }
    for (auto& n : aes_ops) {
        tot_aes_ops += n;
    }
    auto steady_elapsed =
        boost::chrono::steady_clock::now() - steady_start;
    auto cpu_elapsed =
        boost::chrono::process_cpu_clock::now() - cpu_start;

    os << tot_msgs_sent << " messages sent\n";
    os << tot_msg_bytes_sent << " message bytes sent\n";
    os << lamport << " Lamport clock (latencies)\n";
    os << tot_aes_ops << " local AES operations\n";
    os << boost::chrono::duration_cast
        <boost::chrono::milliseconds>(steady_elapsed) <<
        " wall clock time\n";
    os << cpu_elapsed << " {real;user;system}\n";
    dump_memusage(os);
}

MPCPeerIO::MPCPeerIO(unsigned player, ProcessingMode mode,
        std::deque<tcp::socket> &peersocks,
        std::deque<tcp::socket> &serversocks) :
    MPCIO(player, mode, peersocks.size())
{
    unsigned num_threads = unsigned(peersocks.size());
    for (unsigned i=0; i<num_threads; ++i) {
        multtriples.emplace_back(player, mode, "mults", i);
    }
    for (unsigned i=0; i<num_threads; ++i) {
        halftriples.emplace_back(player, mode, "halves", i);
    }
    for (unsigned i=0; i<num_threads; ++i) {
        andtriples.emplace_back(player, mode, "ands", i);
    }
    for (unsigned i=0; i<num_threads; ++i) {
        valselecttriples.emplace_back(player, mode, "selects", i);
    }
    rdpftriples.resize(num_threads);
    for (unsigned i=0; i<num_threads; ++i) {
        for (unsigned depth=1; depth<=ADDRESS_MAX_BITS; ++depth) {
            rdpftriples[i][depth-1].init(player, mode,
                "rdpf", i, depth);
        }
    }
    for (unsigned i=0; i<num_threads; ++i) {
        cdpfs.emplace_back(player, mode, "cdpf", i);
    }
    for (unsigned i=0; i<num_threads; ++i) {
        peerios.emplace_back(std::move(peersocks[i]), "peer", i);
    }
    for (unsigned i=0; i<num_threads; ++i) {
        serverios.emplace_back(std::move(serversocks[i]), "srv", i);
    }
}

void MPCPeerIO::dump_precomp_stats(std::ostream &os)
{
    for (size_t i=0; i<multtriples.size(); ++i) {
        size_t cnt;
        if (i > 0) {
            os << " ";
        }
        os << "T" << i;
        cnt = multtriples[i].get_stats();
        if (cnt > 0) {
            os << " m:" << cnt;
        }
        cnt = halftriples[i].get_stats();
        if (cnt > 0) {
            os << " h:" << cnt;
        }
        cnt = andtriples[i].get_stats();
        if (cnt > 0) {
            os << " a:" << cnt;
        }
        cnt = valselecttriples[i].get_stats();
        if (cnt > 0) {
            os << " s:" << cnt;
        }
        for (nbits_t depth=1; depth<=ADDRESS_MAX_BITS; ++depth) {
            cnt = rdpftriples[i][depth-1].get_stats();
            if (cnt > 0) {
                os << " r" << int(depth) << ":" << cnt;
            }
        }
        cnt = cdpfs[i].get_stats();
        if (cnt > 0) {
            os << " c:" << cnt;
        }
    }
    os << "\n";
}

void MPCPeerIO::reset_precomp_stats()
{
    for (size_t i=0; i<multtriples.size(); ++i) {
        multtriples[i].reset_stats();
        halftriples[i].reset_stats();
        andtriples[i].reset_stats();
        valselecttriples[i].reset_stats();
        for (nbits_t depth=1; depth<=ADDRESS_MAX_BITS; ++depth) {
            rdpftriples[i][depth-1].reset_stats();
        }
    }
}

void MPCPeerIO::dump_stats(std::ostream &os)
{
    MPCIO::dump_stats(os);
    os << "Precomputed values used: ";
    dump_precomp_stats(os);
}

MPCServerIO::MPCServerIO(ProcessingMode mode,
        std::deque<tcp::socket> &p0socks,
        std::deque<tcp::socket> &p1socks) :
    MPCIO(2, mode, p0socks.size())
{
    rdpfpairs.resize(num_threads);
    for (unsigned i=0; i<num_threads; ++i) {
        for (unsigned depth=1; depth<=ADDRESS_MAX_BITS; ++depth) {
            rdpfpairs[i][depth-1].init(player, mode,
                "rdpf", i, depth);
        }
    }
    for (unsigned i=0; i<num_threads; ++i) {
        p0ios.emplace_back(std::move(p0socks[i]), "p0", i);
    }
    for (unsigned i=0; i<num_threads; ++i) {
        p1ios.emplace_back(std::move(p1socks[i]), "p1", i);
    }
}

void MPCServerIO::dump_precomp_stats(std::ostream &os)
{
    for (size_t i=0; i<rdpfpairs.size(); ++i) {
        if (i > 0) {
            os << " ";
        }
        os << "T" << i;
        for (nbits_t depth=1; depth<=ADDRESS_MAX_BITS; ++depth) {
            size_t cnt = rdpfpairs[i][depth-1].get_stats();
            if (cnt > 0) {
                os << " r" << int(depth) << ":" << cnt;
            }
        }
    }
    os << "\n";
}

void MPCServerIO::reset_precomp_stats()
{
    for (size_t i=0; i<rdpfpairs.size(); ++i) {
        for (nbits_t depth=1; depth<=ADDRESS_MAX_BITS; ++depth) {
            rdpfpairs[i][depth-1].reset_stats();
        }
    }
}

void MPCServerIO::dump_stats(std::ostream &os)
{
    MPCIO::dump_stats(os);
    os << "Precomputed values used: ";
    dump_precomp_stats(os);
}

MPCTIO::MPCTIO(MPCIO &mpcio, int thread_num, int num_threads) :
        thread_num(thread_num), local_cpu_nthreads(num_threads),
        communication_nthreads(num_threads),
        thread_lamport(mpcio.lamport), mpcio(mpcio),
#ifdef VERBOSE_COMMS
        round_num(0),
#endif
        last_andtriple_bits_remaining(0)
{
    if (mpcio.player < 2) {
        MPCPeerIO &mpcpio = static_cast<MPCPeerIO&>(mpcio);
        peer_iostream.emplace(mpcpio.peerios[thread_num],
            thread_lamport, mpcpio.msgs_sent[thread_num],
            mpcpio.msg_bytes_sent[thread_num]);
        server_iostream.emplace(mpcpio.serverios[thread_num],
            thread_lamport, mpcpio.msgs_sent[thread_num],
            mpcpio.msg_bytes_sent[thread_num]);
    } else {
        MPCServerIO &mpcsrvio = static_cast<MPCServerIO&>(mpcio);
        p0_iostream.emplace(mpcsrvio.p0ios[thread_num],
            thread_lamport, mpcsrvio.msgs_sent[thread_num],
            mpcsrvio.msg_bytes_sent[thread_num]);
        p1_iostream.emplace(mpcsrvio.p1ios[thread_num],
            thread_lamport, mpcsrvio.msgs_sent[thread_num],
            mpcsrvio.msg_bytes_sent[thread_num]);
    }
}

// Sync our per-thread lamport clock with the master one in the
// mpcio.  You only need to call this explicitly if your MPCTIO
// outlives your thread (in which case call it after the join), or
// if your threads do interthread communication amongst themselves
// (in which case call it in the sending thread before the send, and
// call it in the receiving thread after the receive).
void MPCTIO::sync_lamport()
{
    // Update the mpcio Lamport time to be max of the thread Lamport
    // time and what we thought it was before.  We use this
    // compare_exchange construction in order to atomically
    // do the comparison, computation, and replacement
    lamport_t old_lamport = mpcio.lamport;
    lamport_t new_lamport = thread_lamport;
    do {
        if (new_lamport < old_lamport) {
            new_lamport = old_lamport;
        }
    // The next line atomically checks if lamport still has
    // the value old_lamport; if so, it changes its value to
    // new_lamport and returns true (ending the loop).  If
    // not, it sets old_lamport to the current value of
    // lamport, and returns false (continuing the loop so
    // that new_lamport can be recomputed based on this new
    // value).
    } while (!mpcio.lamport.compare_exchange_weak(
        old_lamport, new_lamport));
    thread_lamport = new_lamport;
}

// Only call this if you can be sure that there are no outstanding
// messages in flight, you can call it on all existing MPCTIOs, and
// you really want to reset the Lamport clock in the midding of a
// run.
void MPCTIO::reset_lamport()
{
    // Reset both our own Lamport clock and the parent MPCIO's
    thread_lamport = 0;
    mpcio.lamport = 0;
}

// Queue up data to the peer or to the server

void MPCTIO::queue_peer(const void *data, size_t len)
{
    if (mpcio.player < 2) {
        MPCPeerIO &mpcpio = static_cast<MPCPeerIO&>(mpcio);
        size_t newmsg = mpcpio.peerios[thread_num].queue(data, len, thread_lamport);
        mpcpio.msgs_sent[thread_num] += newmsg;
        mpcpio.msg_bytes_sent[thread_num] += len;
    }
}

void MPCTIO::queue_server(const void *data, size_t len)
{
    if (mpcio.player < 2) {
        MPCPeerIO &mpcpio = static_cast<MPCPeerIO&>(mpcio);
        size_t newmsg = mpcpio.serverios[thread_num].queue(data, len, thread_lamport);
        mpcpio.msgs_sent[thread_num] += newmsg;
        mpcpio.msg_bytes_sent[thread_num] += len;
    }
}

// Receive data from the peer or to the server

size_t MPCTIO::recv_peer(void *data, size_t len)
{
    if (mpcio.player < 2) {
        MPCPeerIO &mpcpio = static_cast<MPCPeerIO&>(mpcio);
        return mpcpio.peerios[thread_num].recv(data, len, thread_lamport);
    }
    return 0;
}

size_t MPCTIO::recv_server(void *data, size_t len)
{
    if (mpcio.player < 2) {
        MPCPeerIO &mpcpio = static_cast<MPCPeerIO&>(mpcio);
        return mpcpio.serverios[thread_num].recv(data, len, thread_lamport);
    }
    return 0;
}

// Queue up data to p0 or p1

void MPCTIO::queue_p0(const void *data, size_t len)
{
    if (mpcio.player == 2) {
        MPCServerIO &mpcsrvio = static_cast<MPCServerIO&>(mpcio);
        size_t newmsg = mpcsrvio.p0ios[thread_num].queue(data, len, thread_lamport);
        mpcsrvio.msgs_sent[thread_num] += newmsg;
        mpcsrvio.msg_bytes_sent[thread_num] += len;
    }
}

void MPCTIO::queue_p1(const void *data, size_t len)
{
    if (mpcio.player == 2) {
        MPCServerIO &mpcsrvio = static_cast<MPCServerIO&>(mpcio);
        size_t newmsg = mpcsrvio.p1ios[thread_num].queue(data, len, thread_lamport);
        mpcsrvio.msgs_sent[thread_num] += newmsg;
        mpcsrvio.msg_bytes_sent[thread_num] += len;
    }
}

// Receive data from p0 or p1

size_t MPCTIO::recv_p0(void *data, size_t len)
{
    if (mpcio.player == 2) {
        MPCServerIO &mpcsrvio = static_cast<MPCServerIO&>(mpcio);
        return mpcsrvio.p0ios[thread_num].recv(data, len, thread_lamport);
    }
    return 0;
}

size_t MPCTIO::recv_p1(void *data, size_t len)
{
    if (mpcio.player == 2) {
        MPCServerIO &mpcsrvio = static_cast<MPCServerIO&>(mpcio);
        return mpcsrvio.p1ios[thread_num].recv(data, len, thread_lamport);
    }
    return 0;
}

// Send all queued data for this thread

void MPCTIO::send()
{
#ifdef VERBOSE_COMMS
    printf("Thread %u sending round %lu\n", thread_num, ++round_num);
#endif
    if (mpcio.player < 2) {
        MPCPeerIO &mpcpio = static_cast<MPCPeerIO&>(mpcio);
        mpcpio.peerios[thread_num].send();
        mpcpio.serverios[thread_num].send();
    } else {
        MPCServerIO &mpcsrvio = static_cast<MPCServerIO&>(mpcio);
        mpcsrvio.p0ios[thread_num].send();
        mpcsrvio.p1ios[thread_num].send();
    }
}

// Functions to get precomputed values.  If we're in the online
// phase, get them from PreCompStorage.  If we're in the
// preprocessing or online-only phase, read them from the server.
MultTriple MPCTIO::multtriple(yield_t &yield)
{
    MultTriple val;
    if (mpcio.player < 2) {
        MPCPeerIO &mpcpio = static_cast<MPCPeerIO&>(mpcio);
        if (mpcpio.mode != MODE_ONLINE) {
            yield();
            recv_server(&val, sizeof(val));
            mpcpio.multtriples[thread_num].inc();
        } else {
            mpcpio.multtriples[thread_num].get(val);
        }
    } else if (mpcio.mode != MODE_ONLINE) {
        // Create multiplication triples (X0,Y0,Z0),(X1,Y1,Z1) such that
        // (X0*Y1 + Y0*X1) = (Z0+Z1)
        value_t X0, Y0, Z0, X1, Y1, Z1;
        arc4random_buf(&X0, sizeof(X0));
        arc4random_buf(&Y0, sizeof(Y0));
        arc4random_buf(&Z0, sizeof(Z0));
        arc4random_buf(&X1, sizeof(X1));
        arc4random_buf(&Y1, sizeof(Y1));
        Z1 = X0 * Y1 + X1 * Y0 - Z0;
        MultTriple T0, T1;
        T0 = std::make_tuple(X0, Y0, Z0);
        T1 = std::make_tuple(X1, Y1, Z1);
        queue_p0(&T0, sizeof(T0));
        queue_p1(&T1, sizeof(T1));
        yield();
    }
    return val;
}

// When halftriple() is used internally to another preprocessing
// operation, don't tally it, so that it doesn't appear sepearately in
// the stats from the preprocessing operation that invoked it
HalfTriple MPCTIO::halftriple(yield_t &yield, bool tally)
{
    HalfTriple val;
    if (mpcio.player < 2) {
        MPCPeerIO &mpcpio = static_cast<MPCPeerIO&>(mpcio);
        if (mpcpio.mode != MODE_ONLINE) {
            yield();
            recv_server(&val, sizeof(val));
            if (tally) {
                mpcpio.halftriples[thread_num].inc();
            }
        } else {
            mpcpio.halftriples[thread_num].get(val);
        }
    } else if (mpcio.mode != MODE_ONLINE) {
        // Create half-triples (X0,Z0),(Y1,Z1) such that
        // X0*Y1 = Z0 + Z1
        value_t X0, Z0, Y1, Z1;
        arc4random_buf(&X0, sizeof(X0));
        arc4random_buf(&Z0, sizeof(Z0));
        arc4random_buf(&Y1, sizeof(Y1));
        Z1 = X0 * Y1 - Z0;
        HalfTriple H0, H1;
        H0 = std::make_tuple(X0, Z0);
        H1 = std::make_tuple(Y1, Z1);
        queue_p0(&H0, sizeof(H0));
        queue_p1(&H1, sizeof(H1));
        yield();
    }
    return val;
}

MultTriple MPCTIO::andtriple(yield_t &yield)
{
    AndTriple val;
    if (mpcio.player < 2) {
        MPCPeerIO &mpcpio = static_cast<MPCPeerIO&>(mpcio);
        if (mpcpio.mode != MODE_ONLINE) {
            yield();
            recv_server(&val, sizeof(val));
            mpcpio.andtriples[thread_num].inc();
        } else {
            mpcpio.andtriples[thread_num].get(val);
        }
    } else if (mpcio.mode != MODE_ONLINE) {
        // Create AND triples (X0,Y0,Z0),(X1,Y1,Z1) such that
        // (X0&Y1 ^ Y0&X1) = (Z0^Z1)
        value_t X0, Y0, Z0, X1, Y1, Z1;
        arc4random_buf(&X0, sizeof(X0));
        arc4random_buf(&Y0, sizeof(Y0));
        arc4random_buf(&Z0, sizeof(Z0));
        arc4random_buf(&X1, sizeof(X1));
        arc4random_buf(&Y1, sizeof(Y1));
        Z1 = (X0 & Y1) ^ (X1 & Y0) ^ Z0;
        AndTriple T0, T1;
        T0 = std::make_tuple(X0, Y0, Z0);
        T1 = std::make_tuple(X1, Y1, Z1);
        queue_p0(&T0, sizeof(T0));
        queue_p1(&T1, sizeof(T1));
        yield();
    }
    return val;
}

SelectTriple<DPFnode> MPCTIO::nodeselecttriple(yield_t &yield)
{
    SelectTriple<DPFnode> val;
    if (mpcio.player < 2) {
        MPCPeerIO &mpcpio = static_cast<MPCPeerIO&>(mpcio);
        if (mpcpio.mode != MODE_ONLINE) {
            uint8_t Xbyte;
            yield();
            recv_server(&Xbyte, sizeof(Xbyte));
            val.X = Xbyte & 1;
            recv_server(&val.Y, sizeof(val.Y));
            recv_server(&val.Z, sizeof(val.Z));
        } else {
            std::cerr << "Attempted to read SelectTriple<DPFnode> in online phase\n";
        }
    } else if (mpcio.mode != MODE_ONLINE) {
        // Create triples (X0,Y0,Z0),(X1,Y1,Z1) such that
        // (X0*Y1 ^ Y0*X1) = (Z0^Z1)
        bit_t X0, X1;
        DPFnode Y0, Z0, Y1, Z1;
        X0 = arc4random() & 1;
        arc4random_buf(&Y0, sizeof(Y0));
        arc4random_buf(&Z0, sizeof(Z0));
        X1 = arc4random() & 1;
        arc4random_buf(&Y1, sizeof(Y1));
        DPFnode X0ext, X1ext;
        // Sign-extend X0 and X1 (so that 0 -> 0000...0 and
        // 1 -> 1111...1)
        X0ext = if128_mask[X0];
        X1ext = if128_mask[X1];
        Z1 = ((X0ext & Y1) ^ (X1ext & Y0)) ^ Z0;
        queue_p0(&X0, sizeof(X0));
        queue_p0(&Y0, sizeof(Y0));
        queue_p0(&Z0, sizeof(Z0));
        queue_p1(&X1, sizeof(X1));
        queue_p1(&Y1, sizeof(Y1));
        queue_p1(&Z1, sizeof(Z1));
        yield();
    }
    return val;
}

SelectTriple<value_t> MPCTIO::valselecttriple(yield_t &yield)
{
    SelectTriple<value_t> val;
    if (mpcio.player < 2) {
        MPCPeerIO &mpcpio = static_cast<MPCPeerIO&>(mpcio);
        if (mpcpio.mode != MODE_ONLINE) {
            uint8_t Xbyte;
            yield();
            recv_server(&Xbyte, sizeof(Xbyte));
            val.X = Xbyte & 1;
            recv_server(&val.Y, sizeof(val.Y));
            recv_server(&val.Z, sizeof(val.Z));
            mpcpio.valselecttriples[thread_num].inc();
        } else {
            mpcpio.valselecttriples[thread_num].get(val);
        }
    } else if (mpcio.mode != MODE_ONLINE) {
        // Create triples (X0,Y0,Z0),(X1,Y1,Z1) such that
        // (X0*Y1 ^ Y0*X1) = (Z0^Z1)
        bit_t X0, X1;
        value_t Y0, Z0, Y1, Z1;
        X0 = arc4random() & 1;
        arc4random_buf(&Y0, sizeof(Y0));
        arc4random_buf(&Z0, sizeof(Z0));
        X1 = arc4random() & 1;
        arc4random_buf(&Y1, sizeof(Y1));
        value_t X0ext, X1ext;
        // Sign-extend X0 and X1 (so that 0 -> 0000...0 and
        // 1 -> 1111...1)
        X0ext = -value_t(X0);
        X1ext = -value_t(X1);
        Z1 = ((X0ext & Y1) ^ (X1ext & Y0)) ^ Z0;
        queue_p0(&X0, sizeof(X0));
        queue_p0(&Y0, sizeof(Y0));
        queue_p0(&Z0, sizeof(Z0));
        queue_p1(&X1, sizeof(X1));
        queue_p1(&Y1, sizeof(Y1));
        queue_p1(&Z1, sizeof(Z1));
        yield();
    }
    return val;
}

SelectTriple<bit_t> MPCTIO::bitselecttriple(yield_t &yield)
{
    // Do we need to fetch a new AND triple?
    if (last_andtriple_bits_remaining == 0) {
        last_andtriple = andtriple(yield);
        last_andtriple_bits_remaining = 8*sizeof(value_t);
    }
    --last_andtriple_bits_remaining;
    value_t mask = value_t(1) << last_andtriple_bits_remaining;
    SelectTriple<bit_t> val;
    val.X = !!(std::get<0>(last_andtriple) & mask);
    val.Y = !!(std::get<1>(last_andtriple) & mask);
    val.Z = !!(std::get<2>(last_andtriple) & mask);
    return val;
}

// Only computational peers call this; the server should be calling
// rdpfpair() at the same time
RDPFTriple<1> MPCTIO::rdpftriple(yield_t &yield, nbits_t depth,
    bool keep_expansion)
{
    assert(mpcio.player < 2);
    RDPFTriple<1> val;

    MPCPeerIO &mpcpio = static_cast<MPCPeerIO&>(mpcio);
    if (mpcio.mode == MODE_ONLINE) {
        mpcpio.rdpftriples[thread_num][depth-1].get(val);
    } else {
        val = RDPFTriple<1>(*this, yield, depth,
            keep_expansion);
        iostream_server() <<
            val.dpf[(mpcio.player == 0) ? 1 : 2];
        mpcpio.rdpftriples[thread_num][depth-1].inc();
        yield();
    }
    return val;
}

// Only the server calls this; the computational peers should be calling
// rdpftriple() at the same time
RDPFPair<1> MPCTIO::rdpfpair(yield_t &yield, nbits_t depth)
{
    assert(mpcio.player == 2);
    RDPFPair<1> val;

    MPCServerIO &mpcsrvio = static_cast<MPCServerIO&>(mpcio);
    if (mpcio.mode == MODE_ONLINE) {
        mpcsrvio.rdpfpairs[thread_num][depth-1].get(val);
    } else {
        RDPFTriple<1> trip(*this, yield, depth, true);
        yield();
        iostream_p0() >> val.dpf[0];
        iostream_p1() >> val.dpf[1];
        mpcsrvio.rdpfpairs[thread_num][depth-1].inc();
    }
    return val;
}

CDPF MPCTIO::cdpf(yield_t &yield)
{
    CDPF val;
    if (mpcio.player < 2) {
        MPCPeerIO &mpcpio = static_cast<MPCPeerIO&>(mpcio);
        if (mpcpio.mode != MODE_ONLINE) {
            yield();
            iostream_server() >> val;
            mpcpio.cdpfs[thread_num].inc();
        } else {
            mpcpio.cdpfs[thread_num].get(val);
        }
    } else if (mpcio.mode != MODE_ONLINE) {
        auto [ cdpf0, cdpf1 ] = CDPF::generate(aes_ops());
        iostream_p0() << cdpf0;
        iostream_p1() << cdpf1;
        yield();
    }
    return val;
}

// The port number for the P1 -> P0 connection
static const unsigned short port_p1_p0 = 2115;

// The port number for the P2 -> P0 connection
static const unsigned short port_p2_p0 = 2116;

// The port number for the P2 -> P1 connection
static const unsigned short port_p2_p1 = 2117;

void mpcio_setup_computational(unsigned player,
    boost::asio::io_context &io_context,
    const char *p0addr,  // can be NULL when player=0
    int num_threads,
    std::deque<tcp::socket> &peersocks,
    std::deque<tcp::socket> &serversocks)
{
    if (player == 0) {
        // Listen for connections from P1 and from P2
        tcp::acceptor acceptor_p1(io_context,
            tcp::endpoint(tcp::v4(), port_p1_p0));
        tcp::acceptor acceptor_p2(io_context,
            tcp::endpoint(tcp::v4(), port_p2_p0));

        peersocks.clear();
        serversocks.clear();
        for (int i=0;i<num_threads;++i) {
            peersocks.emplace_back(io_context);
            serversocks.emplace_back(io_context);
        }
        for (int i=0;i<num_threads;++i) {
            tcp::socket peersock = acceptor_p1.accept();
            // Read 2 bytes from the socket, which will be the thread
            // number
            unsigned short thread_num;
            boost::asio::read(peersock,
                boost::asio::buffer(&thread_num, sizeof(thread_num)));
            if (thread_num >= num_threads) {
                std::cerr << "Received bad thread number from peer\n";
            } else {
                peersocks[thread_num] = std::move(peersock);
            }
        }
        for (int i=0;i<num_threads;++i) {
            tcp::socket serversock = acceptor_p2.accept();
            // Read 2 bytes from the socket, which will be the thread
            // number
            unsigned short thread_num;
            boost::asio::read(serversock,
                boost::asio::buffer(&thread_num, sizeof(thread_num)));
            if (thread_num >= num_threads) {
                std::cerr << "Received bad thread number from server\n";
            } else {
                serversocks[thread_num] = std::move(serversock);
            }
        }
    } else if (player == 1) {
        // Listen for connections from P2, make num_threads connections to P0
        tcp::acceptor acceptor_p2(io_context,
            tcp::endpoint(tcp::v4(), port_p2_p1));

        tcp::resolver resolver(io_context);
        boost::system::error_code err;
        peersocks.clear();
        serversocks.clear();
        for (int i=0;i<num_threads;++i) {
            serversocks.emplace_back(io_context);
        }
        for (unsigned short thread_num = 0; thread_num < num_threads; ++thread_num) {
            tcp::socket peersock(io_context);
            while(1) {
                boost::asio::connect(peersock,
                    resolver.resolve(p0addr, std::to_string(port_p1_p0)), err);
                if (!err) break;
                std::cerr << "Connection to p0 refused, will retry.\n";
                sleep(1);
            }
            // Write 2 bytes to the socket indicating which thread
            // number this socket is for
            boost::asio::write(peersock,
                boost::asio::buffer(&thread_num, sizeof(thread_num)));
            peersocks.push_back(std::move(peersock));
        }
        for (int i=0;i<num_threads;++i) {
            tcp::socket serversock = acceptor_p2.accept();
            // Read 2 bytes from the socket, which will be the thread
            // number
            unsigned short thread_num;
            boost::asio::read(serversock,
                boost::asio::buffer(&thread_num, sizeof(thread_num)));
            if (thread_num >= num_threads) {
                std::cerr << "Received bad thread number from server\n";
            } else {
                serversocks[thread_num] = std::move(serversock);
            }
        }
    } else {
        std::cerr << "Invalid player number passed to mpcio_setup_computational\n";
    }
}

void mpcio_setup_server(boost::asio::io_context &io_context,
    const char *p0addr, const char *p1addr, int num_threads,
    std::deque<tcp::socket> &p0socks,
    std::deque<tcp::socket> &p1socks)
{
    // Make connections to P0 and P1
    tcp::resolver resolver(io_context);
    boost::system::error_code err;
    p0socks.clear();
    p1socks.clear();
    for (unsigned short thread_num = 0; thread_num < num_threads; ++thread_num) {
        tcp::socket p0sock(io_context);
        while(1) {
            boost::asio::connect(p0sock,
                resolver.resolve(p0addr, std::to_string(port_p2_p0)), err);
            if (!err) break;
            std::cerr << "Connection to p0 refused, will retry.\n";
            sleep(1);
        }
        // Write 2 bytes to the socket indicating which thread
        // number this socket is for
        boost::asio::write(p0sock,
            boost::asio::buffer(&thread_num, sizeof(thread_num)));
        p0socks.push_back(std::move(p0sock));
    }
    for (unsigned short thread_num = 0; thread_num < num_threads; ++thread_num) {
        tcp::socket p1sock(io_context);
        while(1) {
            boost::asio::connect(p1sock,
                resolver.resolve(p1addr, std::to_string(port_p2_p1)), err);
            if (!err) break;
            std::cerr << "Connection to p1 refused, will retry.\n";
            sleep(1);
        }
        // Write 2 bytes to the socket indicating which thread
        // number this socket is for
        boost::asio::write(p1sock,
            boost::asio::buffer(&thread_num, sizeof(thread_num)));
        p1socks.push_back(std::move(p1sock));
    }
}