blob: f2eb8ba52aa2c127268e3f69fd3a974f34e00dbb [file] [log] [blame] [edit]
/*
* Copyright (c) 2014-2017 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.
*
* 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: Andreas Sandberg
*/
#include "dev/virtio/fs9p.hh"
#include <fcntl.h>
#include <netdb.h>
#include <netinet/in.h>
#include <sys/socket.h>
#include <sys/types.h>
#include <sys/un.h>
#include <sys/wait.h>
#include <unistd.h>
#include <csignal>
#include <fstream>
#include "base/callback.hh"
#include "base/output.hh"
#include "debug/VIO9P.hh"
#include "debug/VIO9PData.hh"
#include "params/VirtIO9PBase.hh"
#include "params/VirtIO9PDiod.hh"
#include "params/VirtIO9PProxy.hh"
#include "params/VirtIO9PSocket.hh"
#include "sim/system.hh"
struct P9MsgInfo {
P9MsgInfo(P9MsgType _type, std::string _name)
: type(_type), name(_name) {}
P9MsgType type;
std::string name;
};
typedef std::map<P9MsgType, P9MsgInfo> P9MsgInfoMap;
#define P9MSG(type, name) \
{ (type), P9MsgInfo((type), "T" # name ) }, \
{ (type + 1), P9MsgInfo((type + 1), "R" # name ) }
static const P9MsgInfoMap p9_msg_info {
P9MSG(6, LERROR),
P9MSG(8, STATFS),
P9MSG(12, LOPEN),
P9MSG(14, LCREATE),
P9MSG(16, SYMLINK),
P9MSG(18, MKNOD),
P9MSG(20, RENAME),
P9MSG(22, READLINK),
P9MSG(24, GETATTR),
P9MSG(26, SETATTR),
P9MSG(30, XATTRWALK),
P9MSG(32, XATTRCREATE),
P9MSG(40, READDIR),
P9MSG(50, FSYNC),
P9MSG(52, LOCK),
P9MSG(54, GETLOCK),
P9MSG(70, LINK),
P9MSG(72, MKDIR),
P9MSG(74, RENAMEAT),
P9MSG(76, UNLINKAT),
P9MSG(100, VERSION),
P9MSG(102, AUTH),
P9MSG(104, ATTACH),
P9MSG(106, ERROR),
P9MSG(108, FLUSH),
P9MSG(110, WALK),
P9MSG(112, OPEN),
P9MSG(114, CREATE),
P9MSG(116, READ),
P9MSG(118, WRITE),
P9MSG(120, CLUNK),
P9MSG(122, REMOVE),
P9MSG(124, STAT),
P9MSG(126, WSTAT),
};
#undef P9MSG
VirtIO9PBase::VirtIO9PBase(Params *params)
: VirtIODeviceBase(params, ID_9P,
sizeof(Config) + params->tag.size(),
F_MOUNT_TAG),
queue(params->system->physProxy, params->queueSize, *this)
{
config.reset((Config *)
operator new(configSize));
config->len = htov_legacy(params->tag.size());
memcpy(config->tag, params->tag.c_str(), params->tag.size());
registerQueue(queue);
}
VirtIO9PBase::~VirtIO9PBase()
{
}
void
VirtIO9PBase::readConfig(PacketPtr pkt, Addr cfgOffset)
{
readConfigBlob(pkt, cfgOffset, (uint8_t *)config.get());
}
void
VirtIO9PBase::FSQueue::onNotifyDescriptor(VirtDescriptor *desc)
{
DPRINTF(VIO9P, "Got input data descriptor (len: %i)\n", desc->size());
DPRINTF(VIO9P, "\tPending transactions: %i\n", parent.pendingTransactions.size());
P9MsgHeader header;
desc->chainRead(0, (uint8_t *)&header, sizeof(header));
header = p9toh(header);
uint8_t data[header.len - sizeof(header)];
desc->chainRead(sizeof(header), data, sizeof(data));
// Keep track of pending transactions
parent.pendingTransactions[header.tag] = desc;
DPRINTF(VIO9P, "recvTMsg\n");
parent.dumpMsg(header, data, sizeof(data));
// Notify device of message
parent.recvTMsg(header, data, sizeof(data));
}
void
VirtIO9PBase::sendRMsg(const P9MsgHeader &header, const uint8_t *data, size_t size)
{
DPRINTF(VIO9P, "Sending RMsg\n");
dumpMsg(header, data, size);
DPRINTF(VIO9P, "\tPending transactions: %i\n", pendingTransactions.size());
assert(header.len >= sizeof(header));
VirtDescriptor *main_desc(pendingTransactions[header.tag]);
pendingTransactions.erase(header.tag);
// Find the first output descriptor
VirtDescriptor *out_desc(main_desc);
while (out_desc && !out_desc->isOutgoing())
out_desc = out_desc->next();
if (!out_desc)
panic("sendRMsg: Framing error, no output descriptor.\n");
P9MsgHeader header_out(htop9(header));
header_out.len = htop9(sizeof(P9MsgHeader) + size);
out_desc->chainWrite(0, (uint8_t *)&header_out, sizeof(header_out));
out_desc->chainWrite(sizeof(header_out), data, size);
queue.produceDescriptor(main_desc, sizeof(P9MsgHeader) + size);
kick();
}
void
VirtIO9PBase::dumpMsg(const P9MsgHeader &header, const uint8_t *data, size_t size)
{
#ifndef NDEBUG
if (!DTRACE(VIO9P))
return;
const P9MsgInfoMap::const_iterator it_msg(p9_msg_info.find(header.type));
if (it_msg != p9_msg_info.cend()) {
const P9MsgInfo &info(it_msg->second);
DPRINTF(VIO9P, "P9Msg[len = %i, type = %s (%i), tag = %i]\n",
header.len, info.name, header.type, header.tag);
} else {
DPRINTF(VIO9P, "P9Msg[len = %i, type = Unknown (%i), tag = %i]\n",
header.len, header.type, header.tag);
}
DDUMP(VIO9PData, data, size);
#endif
}
VirtIO9PProxy::VirtIO9PProxy(Params *params)
: VirtIO9PBase(params), deviceUsed(false)
{
}
VirtIO9PProxy::~VirtIO9PProxy()
{
}
void
VirtIO9PProxy::serialize(CheckpointOut &cp) const
{
if (deviceUsed) {
warn("Serializing VirtIO9Base device after device has been used. It is "
"likely that state will be lost, and that the device will cease "
"to work!");
}
SERIALIZE_SCALAR(deviceUsed);
VirtIO9PBase::serialize(cp);
}
void
VirtIO9PProxy::unserialize(CheckpointIn &cp)
{
UNSERIALIZE_SCALAR(deviceUsed);
if (deviceUsed) {
warn("Unserializing VirtIO9Base device after device has been used. It is "
"likely that state has been lost, and that the device will cease "
"to work!");
}
VirtIO9PBase::unserialize(cp);
}
void
VirtIO9PProxy::recvTMsg(const P9MsgHeader &header,
const uint8_t *data, size_t size)
{
deviceUsed = true;
assert(header.len == sizeof(header) + size);
// While technically not needed, we send the packet as one
// contiguous segment to make some packet dissectors happy.
uint8_t out[header.len];
P9MsgHeader header_out(htop9(header));
memcpy(out, (uint8_t *)&header_out, sizeof(header_out));
memcpy(out + sizeof(header_out), data, size);
writeAll(out, sizeof(header_out) + size);
}
void
VirtIO9PProxy::serverDataReady()
{
P9MsgHeader header;
readAll((uint8_t *)&header, sizeof(header));
header = p9toh(header);
const ssize_t payload_len(header.len - sizeof(header));
if (payload_len < 0)
panic("Payload length is negative!\n");
uint8_t data[payload_len];
readAll(data, payload_len);
sendRMsg(header, data, payload_len);
}
void
VirtIO9PProxy::readAll(uint8_t *data, size_t len)
{
while (len) {
ssize_t ret;
while ((ret = read(data, len)) == -EAGAIN)
;
if (ret < 0)
panic("readAll: Read failed: %i\n", -ret);
len -= ret;
data += ret;
}
}
void
VirtIO9PProxy::writeAll(const uint8_t *data, size_t len)
{
while (len) {
ssize_t ret;
while ((ret = write(data, len)) == -EAGAIN)
;
if (ret < 0)
panic("writeAll: write failed: %i\n", -ret);
len -= ret;
data += ret;
}
}
VirtIO9PDiod::VirtIO9PDiod(Params *params)
: VirtIO9PProxy(params),
fd_to_diod(-1), fd_from_diod(-1), diod_pid(-1)
{
// Register an exit callback so we can kill the diod process
Callback* cb = new MakeCallback<VirtIO9PDiod,
&VirtIO9PDiod::terminateDiod>(this);
registerExitCallback(cb);
}
VirtIO9PDiod::~VirtIO9PDiod()
{
}
void
VirtIO9PDiod::startup()
{
startDiod();
dataEvent.reset(new DiodDataEvent(*this, fd_from_diod, POLLIN));
pollQueue.schedule(dataEvent.get());
}
void
VirtIO9PDiod::startDiod()
{
const Params *p(dynamic_cast<const Params *>(params()));
int pipe_rfd[2];
int pipe_wfd[2];
const int DIOD_RFD = 3;
const int DIOD_WFD = 4;
const char *diod(p->diod.c_str());
DPRINTF(VIO9P, "Using diod at %s \n", p->diod.c_str());
if (pipe(pipe_rfd) == -1 || pipe(pipe_wfd) == -1)
panic("Failed to create DIOD pipes: %i\n", errno);
fd_to_diod = pipe_rfd[1];
fd_from_diod = pipe_wfd[0];
// Create Unix domain socket
int socket_id = socket(AF_UNIX, SOCK_STREAM, 0);
if (socket_id == -1) {
panic("Socket creation failed %i \n", errno);
}
// Bind the socket to a path which will not be read
struct sockaddr_un socket_address;
memset(&socket_address, 0, sizeof(struct sockaddr_un));
socket_address.sun_family = AF_UNIX;
const std::string socket_path = simout.resolve(p->socketPath);
fatal_if(!OutputDirectory::isAbsolute(socket_path), "Please make the" \
" output directory an absolute path, else diod will fail!\n");
// Prevent overflow in strcpy
fatal_if(sizeof(socket_address.sun_path) <= socket_path.length(),
"Incorrect length of socket path");
strncpy(socket_address.sun_path, socket_path.c_str(),
sizeof(socket_address.sun_path));
if (bind(socket_id, (struct sockaddr*) &socket_address,
sizeof(struct sockaddr_un)) == -1){
perror("Socket binding");
panic("Socket binding to %i failed - most likely the output dir" \
" and hence unused socket already exists \n", socket_id);
}
diod_pid = fork();
if (diod_pid == -1) {
panic("Fork failed: %i\n", errno);
} else if (diod_pid == 0) {
// Create the socket which will later by used by the diod process
close(STDIN_FILENO);
if (dup2(pipe_rfd[0], DIOD_RFD) == -1 ||
dup2(pipe_wfd[1], DIOD_WFD) == -1) {
panic("Failed to setup read/write pipes: %i\n",
errno);
}
// Start diod
execlp(diod, diod,
"-f", // start in foreground
"-r", "3", // setup read FD
"-w", "4", // setup write FD
"-e", p->root.c_str(), // path to export
"-n", // disable security
"-S", // squash all users
"-l", socket_path.c_str(), // pass the socket
(char *)NULL);
perror("Starting DIOD");
panic("Failed to execute diod to %s: %i\n",socket_path, errno);
} else {
close(pipe_rfd[0]);
close(pipe_wfd[1]);
inform("Started diod with PID %u, you might need to manually kill " \
" diod if gem5 crashes \n", diod_pid);
}
#undef DIOD_RFD
#undef DIOD_WFD
}
ssize_t
VirtIO9PDiod::read(uint8_t *data, size_t len)
{
assert(fd_from_diod != -1);
const int ret(::read(fd_from_diod, (void *)data, len));
return ret < 0 ? -errno : ret;
}
ssize_t
VirtIO9PDiod::write(const uint8_t *data, size_t len)
{
assert(fd_to_diod != -1);
const int ret(::write(fd_to_diod, (const void *)data, len));
return ret < 0 ? -errno : ret;
}
void
VirtIO9PDiod::DiodDataEvent::process(int revent)
{
parent.serverDataReady();
}
void
VirtIO9PDiod::terminateDiod()
{
assert(diod_pid != -1);
DPRINTF(VIO9P, "Trying to kill diod at pid %u \n", diod_pid);
if (kill(diod_pid, SIGTERM) != 0) {
perror("Killing diod process");
warn("Failed to kill diod using SIGTERM");
return;
}
// Check if kill worked
for (unsigned i = 0; i < 5; i++) {
int wait_return = waitpid(diod_pid, NULL, WNOHANG);
if (wait_return == diod_pid) {
// Managed to kill diod
return;
} else if (wait_return == 0) {
// Diod is not killed so sleep and try again
usleep(500);
} else {
// Failed in waitpid
perror("Waitpid");
warn("Failed in waitpid");
}
}
// Try again to kill diod with sigkill
inform("Trying to kill diod with SIGKILL as SIGTERM failed \n");
if (kill(diod_pid, SIGKILL) != 0) {
perror("Killing diod process");
warn("Failed to kill diod using SIGKILL");
} else {
// Managed to kill diod
return;
}
}
VirtIO9PDiod *
VirtIO9PDiodParams::create()
{
return new VirtIO9PDiod(this);
}
VirtIO9PSocket::VirtIO9PSocket(Params *params)
: VirtIO9PProxy(params), fdSocket(-1)
{
}
VirtIO9PSocket::~VirtIO9PSocket()
{
}
void
VirtIO9PSocket::startup()
{
connectSocket();
dataEvent.reset(new SocketDataEvent(*this, fdSocket, POLLIN));
pollQueue.schedule(dataEvent.get());
}
void
VirtIO9PSocket::connectSocket()
{
const Params &p(dynamic_cast<const Params &>(*params()));
int ret;
struct addrinfo hints, *result;
memset(&hints, 0, sizeof(hints));
hints.ai_family = AF_UNSPEC;
hints.ai_socktype = SOCK_STREAM;
hints.ai_flags = 0;
hints.ai_protocol = 0;
if ((ret = getaddrinfo(p.server.c_str(), p.port.c_str(),
&hints, &result)) != 0)
panic("getaddrinfo: %s\n", gai_strerror(ret));
DPRINTF(VIO9P, "Connecting to 9p server '%s'.\n", p.server);
for (struct addrinfo *rp = result; rp; rp = rp->ai_next) {
fdSocket = socket(rp->ai_family, rp->ai_socktype,
rp->ai_protocol);
if (fdSocket == -1) {
continue;
} else if (connect(fdSocket, rp->ai_addr, rp->ai_addrlen) != -1) {
break;
} else {
close(fdSocket);
fdSocket = -1;
}
}
freeaddrinfo(result);
if (fdSocket == -1)
panic("Failed to connect to 9p server (%s:%s)", p.server, p.port);
}
void
VirtIO9PSocket::socketDisconnect()
{
panic("9P Socket disconnected!\n");
}
ssize_t
VirtIO9PSocket::read(uint8_t *data, size_t len)
{
assert(fdSocket != -1);
int ret;
ret = ::recv(fdSocket, (void *)data, len, 0);
if (ret == 0)
socketDisconnect();
return ret < 0 ? -errno : ret;
}
ssize_t
VirtIO9PSocket::write(const uint8_t *data, size_t len)
{
assert(fdSocket != -1);
int ret(::send(fdSocket, (const void *)data, len, 0));
return ret < 0 ? -errno : ret;
}
void
VirtIO9PSocket::SocketDataEvent::process(int revent)
{
parent.serverDataReady();
}
VirtIO9PSocket *
VirtIO9PSocketParams::create()
{
return new VirtIO9PSocket(this);
}