llvm-project/lldb/source/Core/ConnectionFileDescriptor.cpp

1508 lines
51 KiB
C++
Raw Normal View History

//===-- ConnectionFileDescriptor.cpp ----------------------------*- C++ -*-===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
#if defined(__APPLE__)
// Enable this special support for Apple builds where we can have unlimited
// select bounds. We tried switching to poll() and kqueue and we were panicing
// the kernel, so we have to stick with select for now.
#define _DARWIN_UNLIMITED_SELECT
#endif
#include "lldb/Core/ConnectionFileDescriptor.h"
// C Includes
#include <errno.h>
#include <fcntl.h>
#include <arpa/inet.h>
#include <netdb.h>
#include <netinet/in.h>
#include <netinet/tcp.h>
#include <sys/socket.h>
#include <sys/un.h>
#include <sys/types.h>
#include <string.h>
#include <stdlib.h>
#include <unistd.h>
// C++ Includes
// Other libraries and framework includes
#if defined(__APPLE__)
#include "llvm/ADT/SmallVector.h"
#endif
// Project includes
#include "lldb/lldb-private-log.h"
#include "lldb/Interpreter/Args.h"
#include "lldb/Core/Communication.h"
#include "lldb/Core/Log.h"
#include "lldb/Core/RegularExpression.h"
#include "lldb/Core/Timer.h"
using namespace lldb;
using namespace lldb_private;
static bool
DecodeHostAndPort (const char *host_and_port,
std::string &host_str,
std::string &port_str,
int32_t& port,
Error *error_ptr)
{
static RegularExpression g_regex ("([^:]+):([0-9]+)");
RegularExpression::Match regex_match(2);
if (g_regex.Execute (host_and_port, &regex_match))
{
if (regex_match.GetMatchAtIndex (host_and_port, 1, host_str) &&
regex_match.GetMatchAtIndex (host_and_port, 2, port_str))
{
port = Args::StringToSInt32 (port_str.c_str(), INT32_MIN);
if (port != INT32_MIN)
{
if (error_ptr)
error_ptr->Clear();
return true;
}
}
}
host_str.clear();
port_str.clear();
port = INT32_MIN;
if (error_ptr)
error_ptr->SetErrorStringWithFormat("invalid host:port specification: '%s'", host_and_port);
return false;
}
ConnectionFileDescriptor::ConnectionFileDescriptor () :
Connection(),
m_fd_send (-1),
m_fd_recv (-1),
m_fd_send_type (eFDTypeFile),
m_fd_recv_type (eFDTypeFile),
m_udp_send_sockaddr (),
m_should_close_fd (false),
m_socket_timeout_usec(0),
m_pipe_read(-1),
m_pipe_write(-1),
m_mutex (Mutex::eMutexTypeRecursive),
m_shutting_down (false)
{
Log *log(lldb_private::GetLogIfAnyCategoriesSet (LIBLLDB_LOG_CONNECTION | LIBLLDB_LOG_OBJECT));
if (log)
log->Printf ("%p ConnectionFileDescriptor::ConnectionFileDescriptor ()", this);
}
ConnectionFileDescriptor::ConnectionFileDescriptor (int fd, bool owns_fd) :
Connection(),
m_fd_send (fd),
m_fd_recv (fd),
m_fd_send_type (eFDTypeFile),
m_fd_recv_type (eFDTypeFile),
m_udp_send_sockaddr (),
m_should_close_fd (owns_fd),
m_socket_timeout_usec(0),
m_pipe_read(-1),
m_pipe_write(-1),
m_mutex (Mutex::eMutexTypeRecursive),
m_shutting_down (false)
{
Log *log(lldb_private::GetLogIfAnyCategoriesSet (LIBLLDB_LOG_CONNECTION | LIBLLDB_LOG_OBJECT));
if (log)
log->Printf ("%p ConnectionFileDescriptor::ConnectionFileDescriptor (fd = %i, owns_fd = %i)", this, fd, owns_fd);
OpenCommandPipe ();
}
ConnectionFileDescriptor::~ConnectionFileDescriptor ()
{
Log *log(lldb_private::GetLogIfAnyCategoriesSet (LIBLLDB_LOG_CONNECTION | LIBLLDB_LOG_OBJECT));
if (log)
log->Printf ("%p ConnectionFileDescriptor::~ConnectionFileDescriptor ()", this);
Disconnect (NULL);
CloseCommandPipe ();
}
void
ConnectionFileDescriptor::OpenCommandPipe ()
{
CloseCommandPipe();
Log *log(lldb_private::GetLogIfAnyCategoriesSet (LIBLLDB_LOG_CONNECTION));
// Make the command file descriptor here:
int filedes[2];
int result = pipe (filedes);
if (result != 0)
{
if (log)
log->Printf ("%p ConnectionFileDescriptor::ConnectionFileDescriptor () - could not make pipe: %s",
this,
strerror(errno));
}
else
{
m_pipe_read = filedes[0];
m_pipe_write = filedes[1];
}
}
void
ConnectionFileDescriptor::CloseCommandPipe ()
{
if (m_pipe_read != -1)
{
close (m_pipe_read);
m_pipe_read = -1;
}
if (m_pipe_write != -1)
{
close (m_pipe_write);
m_pipe_write = -1;
}
}
bool
ConnectionFileDescriptor::IsConnected () const
{
return m_fd_send >= 0 || m_fd_recv >= 0;
}
ConnectionStatus
ConnectionFileDescriptor::Connect (const char *s, Error *error_ptr)
{
Mutex::Locker locker (m_mutex);
Log *log(lldb_private::GetLogIfAnyCategoriesSet (LIBLLDB_LOG_CONNECTION));
if (log)
log->Printf ("%p ConnectionFileDescriptor::Connect (url = '%s')", this, s);
OpenCommandPipe();
if (s && s[0])
{
char *end = NULL;
if (strstr(s, "listen://"))
{
// listen://HOST:PORT
unsigned long listen_port = ::strtoul(s + strlen("listen://"), &end, 0);
return SocketListen (listen_port, error_ptr);
}
else if (strstr(s, "unix-accept://"))
{
// unix://SOCKNAME
return NamedSocketAccept (s + strlen("unix-accept://"), error_ptr);
}
else if (strstr(s, "connect://"))
{
return ConnectTCP (s + strlen("connect://"), error_ptr);
}
else if (strstr(s, "tcp-connect://"))
{
return ConnectTCP (s + strlen("tcp-connect://"), error_ptr);
}
else if (strstr(s, "udp://"))
{
return ConnectUDP (s + strlen("udp://"), error_ptr);
}
else if (strstr(s, "fd://"))
{
// Just passing a native file descriptor within this current process
// that is already opened (possibly from a service or other source).
s += strlen ("fd://");
bool success = false;
m_fd_send = m_fd_recv = Args::StringToSInt32 (s, -1, 0, &success);
if (success)
{
// We have what looks to be a valid file descriptor, but we
// should make sure it is. We currently are doing this by trying to
// get the flags from the file descriptor and making sure it
// isn't a bad fd.
errno = 0;
int flags = ::fcntl (m_fd_send, F_GETFL, 0);
if (flags == -1 || errno == EBADF)
{
if (error_ptr)
error_ptr->SetErrorStringWithFormat ("stale file descriptor: %s", s);
m_fd_send = m_fd_recv = -1;
return eConnectionStatusError;
}
else
{
// Try and get a socket option from this file descriptor to
// see if this is a socket and set m_is_socket accordingly.
int resuse;
bool is_socket = GetSocketOption (m_fd_send, SOL_SOCKET, SO_REUSEADDR, resuse) == 0;
if (is_socket)
m_fd_send_type = m_fd_recv_type = eFDTypeSocket;
// Don't take ownership of a file descriptor that gets passed
// to us since someone else opened the file descriptor and
// handed it to us.
// TODO: Since are using a URL to open connection we should
// eventually parse options using the web standard where we
// have "fd://123?opt1=value;opt2=value" and we can have an
// option be "owns=1" or "owns=0" or something like this to
// allow us to specify this. For now, we assume we must
// assume we don't own it.
m_should_close_fd = false;
return eConnectionStatusSuccess;
}
}
if (error_ptr)
error_ptr->SetErrorStringWithFormat ("invalid file descriptor: \"fd://%s\"", s);
m_fd_send = m_fd_recv = -1;
return eConnectionStatusError;
}
else if (strstr(s, "file://"))
{
// file:///PATH
const char *path = s + strlen("file://");
do
{
m_fd_send = m_fd_recv = ::open (path, O_RDWR);
} while (m_fd_send == -1 && errno == EINTR);
if (m_fd_send == -1)
{
if (error_ptr)
error_ptr->SetErrorToErrno();
return eConnectionStatusError;
}
int flags = ::fcntl (m_fd_send, F_GETFL, 0);
if (flags >= 0)
{
if ((flags & O_NONBLOCK) == 0)
{
flags |= O_NONBLOCK;
::fcntl (m_fd_send, F_SETFL, flags);
}
}
m_should_close_fd = true;
return eConnectionStatusSuccess;
}
if (error_ptr)
error_ptr->SetErrorStringWithFormat ("unsupported connection URL: '%s'", s);
return eConnectionStatusError;
}
if (error_ptr)
error_ptr->SetErrorString("invalid connect arguments");
return eConnectionStatusError;
}
ConnectionStatus
ConnectionFileDescriptor::Disconnect (Error *error_ptr)
{
Log *log(lldb_private::GetLogIfAnyCategoriesSet (LIBLLDB_LOG_CONNECTION));
if (log)
log->Printf ("%p ConnectionFileDescriptor::Disconnect ()", this);
ConnectionStatus status = eConnectionStatusSuccess;
if (m_fd_send < 0 && m_fd_recv < 0)
{
if (log)
log->Printf ("%p ConnectionFileDescriptor::Disconnect(): Nothing to disconnect", this);
return eConnectionStatusSuccess;
}
// Try to get the ConnectionFileDescriptor's mutex. If we fail, that is quite likely
// because somebody is doing a blocking read on our file descriptor. If that's the case,
// then send the "q" char to the command file channel so the read will wake up and the connection
// will then know to shut down.
m_shutting_down = true;
Mutex::Locker locker;
bool got_lock= locker.TryLock (m_mutex);
if (!got_lock)
{
if (m_pipe_write != -1 )
{
write (m_pipe_write, "q", 1);
close (m_pipe_write);
m_pipe_write = -1;
}
locker.Lock (m_mutex);
}
if (m_should_close_fd == true)
{
if (m_fd_send == m_fd_recv)
{
status = Close (m_fd_send, error_ptr);
}
else
{
// File descriptors are the different, close both if needed
if (m_fd_send >= 0)
status = Close (m_fd_send, error_ptr);
if (m_fd_recv >= 0)
{
ConnectionStatus recv_status = Close (m_fd_recv, error_ptr);
if (status == eConnectionStatusSuccess)
status = recv_status;
}
}
}
// Now set all our descriptors to invalid values.
m_fd_send = m_fd_recv = -1;
if (status != eConnectionStatusSuccess)
{
return status;
}
m_shutting_down = false;
return eConnectionStatusSuccess;
}
size_t
ConnectionFileDescriptor::Read (void *dst,
size_t dst_len,
uint32_t timeout_usec,
ConnectionStatus &status,
Error *error_ptr)
{
Log *log(lldb_private::GetLogIfAnyCategoriesSet (LIBLLDB_LOG_CONNECTION));
if (log)
log->Printf ("%p ConnectionFileDescriptor::Read () ::read (fd = %i, dst = %p, dst_len = %" PRIu64 ")...",
this, m_fd_recv, dst, (uint64_t)dst_len);
Mutex::Locker locker;
bool got_lock = locker.TryLock (m_mutex);
if (!got_lock)
{
if (log)
log->Printf ("%p ConnectionFileDescriptor::Read () failed to get the connection lock.",
this);
if (error_ptr)
error_ptr->SetErrorString ("failed to get the connection lock for read.");
status = eConnectionStatusTimedOut;
return 0;
}
else if (m_shutting_down)
return eConnectionStatusError;
ssize_t bytes_read = 0;
status = BytesAvailable (timeout_usec, error_ptr);
if (status == eConnectionStatusSuccess)
{
do
{
bytes_read = ::read (m_fd_recv, dst, dst_len);
} while (bytes_read < 0 && errno == EINTR);
}
if (status != eConnectionStatusSuccess)
return 0;
Error error;
if (bytes_read == 0)
{
error.Clear(); // End-of-file. Do not automatically close; pass along for the end-of-file handlers.
status = eConnectionStatusEndOfFile;
}
else if (bytes_read < 0)
{
error.SetErrorToErrno();
}
else
{
error.Clear();
}
if (log)
log->Printf ("%p ConnectionFileDescriptor::Read () ::read (fd = %i, dst = %p, dst_len = %" PRIu64 ") => %" PRIi64 ", error = %s",
this,
m_fd_recv,
dst,
(uint64_t)dst_len,
(int64_t)bytes_read,
error.AsCString());
if (error_ptr)
*error_ptr = error;
if (error.Fail())
{
uint32_t error_value = error.GetError();
switch (error_value)
{
case EAGAIN: // The file was marked for non-blocking I/O, and no data were ready to be read.
if (m_fd_recv_type == eFDTypeSocket || m_fd_recv_type == eFDTypeSocketUDP)
status = eConnectionStatusTimedOut;
else
status = eConnectionStatusSuccess;
return 0;
case EFAULT: // Buf points outside the allocated address space.
case EINTR: // A read from a slow device was interrupted before any data arrived by the delivery of a signal.
case EINVAL: // The pointer associated with fildes was negative.
case EIO: // An I/O error occurred while reading from the file system.
// The process group is orphaned.
// The file is a regular file, nbyte is greater than 0,
// the starting position is before the end-of-file, and
// the starting position is greater than or equal to the
// offset maximum established for the open file
// descriptor associated with fildes.
case EISDIR: // An attempt is made to read a directory.
case ENOBUFS: // An attempt to allocate a memory buffer fails.
case ENOMEM: // Insufficient memory is available.
status = eConnectionStatusError;
break; // Break to close....
case ENOENT: // no such file or directory
case EBADF: // fildes is not a valid file or socket descriptor open for reading.
case ENXIO: // An action is requested of a device that does not exist..
// A requested action cannot be performed by the device.
case ECONNRESET:// The connection is closed by the peer during a read attempt on a socket.
case ENOTCONN: // A read is attempted on an unconnected socket.
status = eConnectionStatusLostConnection;
break; // Break to close....
case ETIMEDOUT: // A transmission timeout occurs during a read attempt on a socket.
status = eConnectionStatusTimedOut;
return 0;
}
return 0;
}
return bytes_read;
}
size_t
ConnectionFileDescriptor::Write (const void *src, size_t src_len, ConnectionStatus &status, Error *error_ptr)
{
Log *log(lldb_private::GetLogIfAnyCategoriesSet (LIBLLDB_LOG_CONNECTION));
if (log)
log->Printf ("%p ConnectionFileDescriptor::Write (src = %p, src_len = %" PRIu64 ")", this, src, (uint64_t)src_len);
if (!IsConnected ())
{
if (error_ptr)
error_ptr->SetErrorString("not connected");
status = eConnectionStatusNoConnection;
return 0;
}
Error error;
ssize_t bytes_sent = 0;
switch (m_fd_send_type)
{
case eFDTypeFile: // Other FD requireing read/write
do
{
bytes_sent = ::write (m_fd_send, src, src_len);
} while (bytes_sent < 0 && errno == EINTR);
break;
case eFDTypeSocket: // Socket requiring send/recv
do
{
bytes_sent = ::send (m_fd_send, src, src_len, 0);
} while (bytes_sent < 0 && errno == EINTR);
break;
case eFDTypeSocketUDP: // Unconnected UDP socket requiring sendto/recvfrom
assert (m_udp_send_sockaddr.GetFamily() != 0);
do
{
bytes_sent = ::sendto (m_fd_send,
src,
src_len,
0,
m_udp_send_sockaddr,
m_udp_send_sockaddr.GetLength());
} while (bytes_sent < 0 && errno == EINTR);
break;
}
if (bytes_sent < 0)
error.SetErrorToErrno ();
else
error.Clear ();
if (log)
{
switch (m_fd_send_type)
{
case eFDTypeFile: // Other FD requireing read/write
log->Printf ("%p ConnectionFileDescriptor::Write() ::write (fd = %i, src = %p, src_len = %" PRIu64 ") => %" PRIi64 " (error = %s)",
this,
m_fd_send,
src,
(uint64_t)src_len,
(int64_t)bytes_sent,
error.AsCString());
break;
case eFDTypeSocket: // Socket requiring send/recv
log->Printf ("%p ConnectionFileDescriptor::Write() ::send (socket = %i, src = %p, src_len = %" PRIu64 ", flags = 0) => %" PRIi64 " (error = %s)",
this,
m_fd_send,
src,
(uint64_t)src_len,
(int64_t)bytes_sent,
error.AsCString());
break;
case eFDTypeSocketUDP: // Unconnected UDP socket requiring sendto/recvfrom
log->Printf ("%p ConnectionFileDescriptor::Write() ::sendto (socket = %i, src = %p, src_len = %" PRIu64 ", flags = 0) => %" PRIi64 " (error = %s)",
this,
m_fd_send,
src,
(uint64_t)src_len,
(int64_t)bytes_sent,
error.AsCString());
break;
}
}
if (error_ptr)
*error_ptr = error;
if (error.Fail())
{
switch (error.GetError())
{
case EAGAIN:
case EINTR:
status = eConnectionStatusSuccess;
return 0;
case ECONNRESET:// The connection is closed by the peer during a read attempt on a socket.
case ENOTCONN: // A read is attempted on an unconnected socket.
status = eConnectionStatusLostConnection;
break; // Break to close....
default:
status = eConnectionStatusError;
break; // Break to close....
}
return 0;
}
status = eConnectionStatusSuccess;
return bytes_sent;
}
#if defined(__APPLE__)
// This ConnectionFileDescriptor::BytesAvailable() uses select().
//
// PROS:
// - select is consistent across most unix platforms
// - this Apple specific version allows for unlimited fds in the fd_sets by
// setting the _DARWIN_UNLIMITED_SELECT define prior to including the
// required header files.
// CONS:
// - Darwin only
ConnectionStatus
ConnectionFileDescriptor::BytesAvailable (uint32_t timeout_usec, Error *error_ptr)
{
// Don't need to take the mutex here separately since we are only called from Read. If we
// ever get used more generally we will need to lock here as well.
Log *log(lldb_private::GetLogIfAnyCategoriesSet (LIBLLDB_LOG_CONNECTION));
if (log)
log->Printf("%p ConnectionFileDescriptor::BytesAvailable (timeout_usec = %u)", this, timeout_usec);
struct timeval *tv_ptr;
struct timeval tv;
if (timeout_usec == UINT32_MAX)
{
// Infinite wait...
tv_ptr = NULL;
}
else
{
TimeValue time_value;
time_value.OffsetWithMicroSeconds (timeout_usec);
tv = time_value.GetAsTimeVal();
tv_ptr = &tv;
}
// Make a copy of the file descriptors to make sure we don't
// have another thread change these values out from under us
// and cause problems in the loop below where like in FS_SET()
const int data_fd = m_fd_recv;
const int pipe_fd = m_pipe_read;
if (data_fd >= 0)
{
const bool have_pipe_fd = pipe_fd >= 0;
while (data_fd == m_fd_recv)
{
const int nfds = std::max<int>(data_fd, pipe_fd) + 1;
llvm::SmallVector<fd_set, 1> read_fds;
read_fds.resize((nfds/FD_SETSIZE) + 1);
for (size_t i=0; i<read_fds.size(); ++i)
FD_ZERO (&read_fds[i]);
// FD_SET doesn't bounds check, it just happily walks off the end
// but we have taken care of making the extra storage with our
// SmallVector of fd_set objects
FD_SET (data_fd, read_fds.data());
if (have_pipe_fd)
FD_SET (pipe_fd, read_fds.data());
Error error;
if (log)
{
if (have_pipe_fd)
log->Printf("%p ConnectionFileDescriptor::BytesAvailable() ::select (nfds=%i, fds={%i, %i}, NULL, NULL, timeout=%p)...",
this, nfds, data_fd, pipe_fd, tv_ptr);
else
log->Printf("%p ConnectionFileDescriptor::BytesAvailable() ::select (nfds=%i, fds={%i}, NULL, NULL, timeout=%p)...",
this, nfds, data_fd, tv_ptr);
}
const int num_set_fds = ::select (nfds, read_fds.data(), NULL, NULL, tv_ptr);
if (num_set_fds < 0)
error.SetErrorToErrno();
else
error.Clear();
if (log)
{
if (have_pipe_fd)
log->Printf("%p ConnectionFileDescriptor::BytesAvailable() ::select (nfds=%i, fds={%i, %i}, NULL, NULL, timeout=%p) => %d, error = %s",
this, nfds, data_fd, pipe_fd, tv_ptr, num_set_fds, error.AsCString());
else
log->Printf("%p ConnectionFileDescriptor::BytesAvailable() ::select (nfds=%i, fds={%i}, NULL, NULL, timeout=%p) => %d, error = %s",
this, nfds, data_fd, tv_ptr, num_set_fds, error.AsCString());
}
if (error_ptr)
*error_ptr = error;
if (error.Fail())
{
switch (error.GetError())
{
case EBADF: // One of the descriptor sets specified an invalid descriptor.
return eConnectionStatusLostConnection;
case EINVAL: // The specified time limit is invalid. One of its components is negative or too large.
default: // Other unknown error
return eConnectionStatusError;
case EAGAIN: // The kernel was (perhaps temporarily) unable to
// allocate the requested number of file descriptors,
// or we have non-blocking IO
case EINTR: // A signal was delivered before the time limit
// expired and before any of the selected events
// occurred.
break; // Lets keep reading to until we timeout
}
}
else if (num_set_fds == 0)
{
return eConnectionStatusTimedOut;
}
else if (num_set_fds > 0)
{
// FD_ISSET is happy to deal with a something larger than
// a single fd_set.
if (FD_ISSET(data_fd, read_fds.data()))
return eConnectionStatusSuccess;
if (have_pipe_fd && FD_ISSET(pipe_fd, read_fds.data()))
{
// We got a command to exit. Read the data from that pipe:
char buffer[16];
ssize_t bytes_read;
do
{
bytes_read = ::read (pipe_fd, buffer, sizeof(buffer));
} while (bytes_read < 0 && errno == EINTR);
assert (bytes_read == 1 && buffer[0] == 'q');
if (log)
log->Printf("%p ConnectionFileDescriptor::BytesAvailable() got data: %*s from the command channel.",
this, (int) bytes_read, buffer);
return eConnectionStatusEndOfFile;
}
}
}
}
if (error_ptr)
error_ptr->SetErrorString("not connected");
return eConnectionStatusLostConnection;
}
#else
// This ConnectionFileDescriptor::BytesAvailable() uses select().
//
// PROS:
// - select is consistent across most unix platforms
// CONS:
// - only supports file descriptors up to FD_SETSIZE. This implementation
// will assert if it runs into that hard limit to let users know that
// another ConnectionFileDescriptor::BytesAvailable() should be used
// or a new version of ConnectionFileDescriptor::BytesAvailable() should
// be written for the system that is running into the limitations. MacOSX
// uses kqueues, and there is a poll() based implementation below.
ConnectionStatus
ConnectionFileDescriptor::BytesAvailable (uint32_t timeout_usec, Error *error_ptr)
{
// Don't need to take the mutex here separately since we are only called from Read. If we
// ever get used more generally we will need to lock here as well.
Log *log(lldb_private::GetLogIfAnyCategoriesSet (LIBLLDB_LOG_CONNECTION));
if (log)
log->Printf("%p ConnectionFileDescriptor::BytesAvailable (timeout_usec = %u)", this, timeout_usec);
struct timeval *tv_ptr;
struct timeval tv;
if (timeout_usec == UINT32_MAX)
{
// Infinite wait...
tv_ptr = NULL;
}
else
{
TimeValue time_value;
time_value.OffsetWithMicroSeconds (timeout_usec);
tv = time_value.GetAsTimeVal();
tv_ptr = &tv;
}
// Make a copy of the file descriptors to make sure we don't
// have another thread change these values out from under us
// and cause problems in the loop below where like in FS_SET()
const int data_fd = m_fd_recv;
const int pipe_fd = m_pipe_read;
if (data_fd >= 0)
{
// If this assert fires off on MacOSX, we will need to switch to using
// libdispatch to read from file descriptors because poll() is causing
// kernel panics and if we exceed FD_SETSIZE we will have no choice...
assert (data_fd < FD_SETSIZE);
const bool have_pipe_fd = pipe_fd >= 0;
if (have_pipe_fd)
{
assert (pipe_fd < FD_SETSIZE);
}
while (data_fd == m_fd_recv)
{
fd_set read_fds;
FD_ZERO (&read_fds);
FD_SET (data_fd, &read_fds);
if (have_pipe_fd)
FD_SET (pipe_fd, &read_fds);
const int nfds = std::max<int>(data_fd, pipe_fd) + 1;
Error error;
if (log)
{
if (have_pipe_fd)
log->Printf("%p ConnectionFileDescriptor::BytesAvailable() ::select (nfds=%i, fds={%i, %i}, NULL, NULL, timeout=%p)...",
this, nfds, data_fd, pipe_fd, tv_ptr);
else
log->Printf("%p ConnectionFileDescriptor::BytesAvailable() ::select (nfds=%i, fds={%i}, NULL, NULL, timeout=%p)...",
this, nfds, data_fd, tv_ptr);
}
const int num_set_fds = ::select (nfds, &read_fds, NULL, NULL, tv_ptr);
if (num_set_fds < 0)
error.SetErrorToErrno();
else
error.Clear();
if (log)
{
if (have_pipe_fd)
log->Printf("%p ConnectionFileDescriptor::BytesAvailable() ::select (nfds=%i, fds={%i, %i}, NULL, NULL, timeout=%p) => %d, error = %s",
this, nfds, data_fd, pipe_fd, tv_ptr, num_set_fds, error.AsCString());
else
log->Printf("%p ConnectionFileDescriptor::BytesAvailable() ::select (nfds=%i, fds={%i}, NULL, NULL, timeout=%p) => %d, error = %s",
this, nfds, data_fd, tv_ptr, num_set_fds, error.AsCString());
}
if (error_ptr)
*error_ptr = error;
if (error.Fail())
{
switch (error.GetError())
{
case EBADF: // One of the descriptor sets specified an invalid descriptor.
return eConnectionStatusLostConnection;
case EINVAL: // The specified time limit is invalid. One of its components is negative or too large.
default: // Other unknown error
return eConnectionStatusError;
case EAGAIN: // The kernel was (perhaps temporarily) unable to
// allocate the requested number of file descriptors,
// or we have non-blocking IO
case EINTR: // A signal was delivered before the time limit
// expired and before any of the selected events
// occurred.
break; // Lets keep reading to until we timeout
}
}
else if (num_set_fds == 0)
{
return eConnectionStatusTimedOut;
}
else if (num_set_fds > 0)
{
if (FD_ISSET(data_fd, &read_fds))
return eConnectionStatusSuccess;
if (have_pipe_fd && FD_ISSET(pipe_fd, &read_fds))
{
// We got a command to exit. Read the data from that pipe:
char buffer[16];
ssize_t bytes_read;
do
{
bytes_read = ::read (pipe_fd, buffer, sizeof(buffer));
} while (bytes_read < 0 && errno == EINTR);
assert (bytes_read == 1 && buffer[0] == 'q');
if (log)
log->Printf("%p ConnectionFileDescriptor::BytesAvailable() got data: %*s from the command channel.",
this, (int) bytes_read, buffer);
return eConnectionStatusEndOfFile;
}
}
}
}
if (error_ptr)
error_ptr->SetErrorString("not connected");
return eConnectionStatusLostConnection;
}
#endif
#if 0
#include <poll.h>
// This ConnectionFileDescriptor::BytesAvailable() uses poll(). poll() should NOT
// be used on MacOSX as it has all sorts of restrictions on the types of file descriptors
// that it doesn't support.
//
// There may be some systems that properly support poll() that could use this
// implementation. I will let each system opt into this on their own.
//
// PROS:
// - no restrictions on the fd value that is used
// CONS:
// - varies wildly from platform to platform in its implementation restrictions
ConnectionStatus
ConnectionFileDescriptor::BytesAvailable (uint32_t timeout_usec, Error *error_ptr)
{
// Don't need to take the mutex here separately since we are only called from Read. If we
// ever get used more generally we will need to lock here as well.
Log *log(lldb_private::GetLogIfAnyCategoriesSet (LIBLLDB_LOG_CONNECTION));
if (log)
log->Printf("%p ConnectionFileDescriptor::BytesAvailable (timeout_usec = %u)", this, timeout_usec);
int timeout_msec = 0;
if (timeout_usec == UINT32_MAX)
{
// Infinite wait...
timeout_msec = -1;
}
else if (timeout_usec == 0)
{
// Return immediately, don't wait
timeout_msec = 0;
}
else
{
// Convert usec to msec
timeout_msec = (timeout_usec + 999) / 1000;
}
// Make a copy of the file descriptors to make sure we don't
// have another thread change these values out from under us
// and cause problems in the loop below where like in FS_SET()
const int data_fd = m_fd_recv;
const int pipe_fd = m_pipe_read;
// Make sure the file descriptor can be used with select as it
// must be in range
if (data_fd >= 0)
{
const bool have_pipe_fd = pipe_fd >= 0;
struct pollfd fds[2] =
{
{ data_fd, POLLIN, 0 },
{ pipe_fd, POLLIN, 0 }
};
const int nfds = have_pipe_fd ? 2 : 1;
Error error;
while (data_fd == m_fd_recv)
{
const int num_set_fds = ::poll (fds, nfds, timeout_msec);
if (num_set_fds < 0)
error.SetErrorToErrno();
else
error.Clear();
if (error_ptr)
*error_ptr = error;
if (log)
{
if (have_pipe_fd)
log->Printf("%p ConnectionFileDescriptor::BytesAvailable() ::poll (fds={{%i,POLLIN},{%i,POLLIN}}, nfds=%i, timeout_ms=%i) => %d, error = %s\n",
this,
data_fd,
pipe_fd,
nfds,
timeout_msec,
num_set_fds,
error.AsCString());
else
log->Printf("%p ConnectionFileDescriptor::BytesAvailable() ::poll (fds={{%i,POLLIN}}, nfds=%i, timeout_ms=%i) => %d, error = %s\n",
this,
data_fd,
nfds,
timeout_msec,
num_set_fds,
error.AsCString());
}
if (error.Fail())
{
switch (error.GetError())
{
case EBADF: // One of the descriptor sets specified an invalid descriptor.
return eConnectionStatusLostConnection;
case EINVAL: // The specified time limit is invalid. One of its components is negative or too large.
default: // Other unknown error
return eConnectionStatusError;
case EAGAIN: // The kernel was (perhaps temporarily) unable to
// allocate the requested number of file descriptors,
// or we have non-blocking IO
case EINTR: // A signal was delivered before the time limit
// expired and before any of the selected events
// occurred.
break; // Lets keep reading to until we timeout
}
}
else if (num_set_fds == 0)
{
return eConnectionStatusTimedOut;
}
else if (num_set_fds > 0)
{
if (fds[0].revents & POLLIN)
return eConnectionStatusSuccess;
if (fds[1].revents & POLLIN)
{
// We got a command to exit. Read the data from that pipe:
char buffer[16];
ssize_t bytes_read;
do
{
bytes_read = ::read (pipe_fd, buffer, sizeof(buffer));
} while (bytes_read < 0 && errno == EINTR);
assert (bytes_read == 1 && buffer[0] == 'q');
if (log)
log->Printf("%p ConnectionFileDescriptor::BytesAvailable() got data: %*s from the command channel.",
this, (int) bytes_read, buffer);
return eConnectionStatusEndOfFile;
}
}
}
}
if (error_ptr)
error_ptr->SetErrorString("not connected");
return eConnectionStatusLostConnection;
}
#endif
ConnectionStatus
ConnectionFileDescriptor::Close (int& fd, Error *error_ptr)
{
if (error_ptr)
error_ptr->Clear();
bool success = true;
// Avoid taking a lock if we can
if (fd >= 0)
{
Mutex::Locker locker (m_mutex);
// Check the FD after the lock is taken to ensure only one thread
// can get into the close scope below
if (fd >= 0)
{
Log *log(lldb_private::GetLogIfAnyCategoriesSet (LIBLLDB_LOG_CONNECTION));
if (log)
log->Printf ("%p ConnectionFileDescriptor::Close (fd = %i)", this,fd);
success = ::close (fd) == 0;
// A reference to a FD was passed in, set it to an invalid value
fd = -1;
if (!success && error_ptr)
{
// Only set the error if we have been asked to since something else
// might have caused us to try and shut down the connection and may
// have already set the error.
error_ptr->SetErrorToErrno();
}
}
}
if (success)
return eConnectionStatusSuccess;
else
return eConnectionStatusError;
}
ConnectionStatus
ConnectionFileDescriptor::NamedSocketAccept (const char *socket_name, Error *error_ptr)
{
ConnectionStatus result = eConnectionStatusError;
struct sockaddr_un saddr_un;
m_fd_send_type = m_fd_recv_type = eFDTypeSocket;
int listen_socket = ::socket (AF_UNIX, SOCK_STREAM, 0);
if (listen_socket == -1)
{
if (error_ptr)
error_ptr->SetErrorToErrno();
return eConnectionStatusError;
}
saddr_un.sun_family = AF_UNIX;
::strncpy(saddr_un.sun_path, socket_name, sizeof(saddr_un.sun_path) - 1);
saddr_un.sun_path[sizeof(saddr_un.sun_path) - 1] = '\0';
#if defined(__APPLE__) || defined(__FreeBSD__)
saddr_un.sun_len = SUN_LEN (&saddr_un);
#endif
if (::bind (listen_socket, (struct sockaddr *)&saddr_un, SUN_LEN (&saddr_un)) == 0)
{
if (::listen (listen_socket, 5) == 0)
{
m_fd_send = m_fd_recv = ::accept (listen_socket, NULL, 0);
if (m_fd_send > 0)
{
m_should_close_fd = true;
if (error_ptr)
error_ptr->Clear();
result = eConnectionStatusSuccess;
}
}
}
if (result != eConnectionStatusSuccess)
{
if (error_ptr)
error_ptr->SetErrorToErrno();
}
// We are done with the listen port
Close (listen_socket, NULL);
return result;
}
Moved the execution context that was in the Debugger into the CommandInterpreter where it was always being used. Make sure that Modules can track their object file offsets correctly to allow opening of sub object files (like the "__commpage" on darwin). Modified the Platforms to be able to launch processes. The first part of this move is the platform soon will become the entity that launches your program and when it does, it uses a new ProcessLaunchInfo class which encapsulates all process launching settings. This simplifies the internal APIs needed for launching. I want to slowly phase out process launching from the process classes, so for now we can still launch just as we used to, but eventually the platform is the object that should do the launching. Modified the Host::LaunchProcess in the MacOSX Host.mm to correctly be able to launch processes with all of the new eLaunchFlag settings. Modified any code that was manually launching processes to use the Host::LaunchProcess functions. Fixed an issue where lldb_private::Args had implicitly defined copy constructors that could do the wrong thing. This has now been fixed by adding an appropriate copy constructor and assignment operator. Make sure we don't add empty ModuleSP entries to a module list. Fixed the commpage module creation on MacOSX, but we still need to train the MacOSX dynamic loader to not get rid of it when it doesn't have an entry in the all image infos. Abstracted many more calls from in ProcessGDBRemote down into the GDBRemoteCommunicationClient subclass to make the classes cleaner and more efficient. Fixed the default iOS ARM register context to be correct and also added support for targets that don't support the qThreadStopInfo packet by selecting the current thread (only if needed) and then sending a stop reply packet. Debugserver can now start up with a --unix-socket (-u for short) and can then bind to port zero and send the port it bound to to a listening process on the other end. This allows the GDB remote platform to spawn new GDB server instances (debugserver) to allow platform debugging. llvm-svn: 129351
2011-04-12 13:54:46 +08:00
ConnectionStatus
ConnectionFileDescriptor::NamedSocketConnect (const char *socket_name, Error *error_ptr)
{
Disconnect (NULL);
m_fd_send_type = m_fd_recv_type = eFDTypeSocket;
Moved the execution context that was in the Debugger into the CommandInterpreter where it was always being used. Make sure that Modules can track their object file offsets correctly to allow opening of sub object files (like the "__commpage" on darwin). Modified the Platforms to be able to launch processes. The first part of this move is the platform soon will become the entity that launches your program and when it does, it uses a new ProcessLaunchInfo class which encapsulates all process launching settings. This simplifies the internal APIs needed for launching. I want to slowly phase out process launching from the process classes, so for now we can still launch just as we used to, but eventually the platform is the object that should do the launching. Modified the Host::LaunchProcess in the MacOSX Host.mm to correctly be able to launch processes with all of the new eLaunchFlag settings. Modified any code that was manually launching processes to use the Host::LaunchProcess functions. Fixed an issue where lldb_private::Args had implicitly defined copy constructors that could do the wrong thing. This has now been fixed by adding an appropriate copy constructor and assignment operator. Make sure we don't add empty ModuleSP entries to a module list. Fixed the commpage module creation on MacOSX, but we still need to train the MacOSX dynamic loader to not get rid of it when it doesn't have an entry in the all image infos. Abstracted many more calls from in ProcessGDBRemote down into the GDBRemoteCommunicationClient subclass to make the classes cleaner and more efficient. Fixed the default iOS ARM register context to be correct and also added support for targets that don't support the qThreadStopInfo packet by selecting the current thread (only if needed) and then sending a stop reply packet. Debugserver can now start up with a --unix-socket (-u for short) and can then bind to port zero and send the port it bound to to a listening process on the other end. This allows the GDB remote platform to spawn new GDB server instances (debugserver) to allow platform debugging. llvm-svn: 129351
2011-04-12 13:54:46 +08:00
// Open the socket that was passed in as an option
struct sockaddr_un saddr_un;
m_fd_send = m_fd_recv = ::socket (AF_UNIX, SOCK_STREAM, 0);
if (m_fd_send == -1)
Moved the execution context that was in the Debugger into the CommandInterpreter where it was always being used. Make sure that Modules can track their object file offsets correctly to allow opening of sub object files (like the "__commpage" on darwin). Modified the Platforms to be able to launch processes. The first part of this move is the platform soon will become the entity that launches your program and when it does, it uses a new ProcessLaunchInfo class which encapsulates all process launching settings. This simplifies the internal APIs needed for launching. I want to slowly phase out process launching from the process classes, so for now we can still launch just as we used to, but eventually the platform is the object that should do the launching. Modified the Host::LaunchProcess in the MacOSX Host.mm to correctly be able to launch processes with all of the new eLaunchFlag settings. Modified any code that was manually launching processes to use the Host::LaunchProcess functions. Fixed an issue where lldb_private::Args had implicitly defined copy constructors that could do the wrong thing. This has now been fixed by adding an appropriate copy constructor and assignment operator. Make sure we don't add empty ModuleSP entries to a module list. Fixed the commpage module creation on MacOSX, but we still need to train the MacOSX dynamic loader to not get rid of it when it doesn't have an entry in the all image infos. Abstracted many more calls from in ProcessGDBRemote down into the GDBRemoteCommunicationClient subclass to make the classes cleaner and more efficient. Fixed the default iOS ARM register context to be correct and also added support for targets that don't support the qThreadStopInfo packet by selecting the current thread (only if needed) and then sending a stop reply packet. Debugserver can now start up with a --unix-socket (-u for short) and can then bind to port zero and send the port it bound to to a listening process on the other end. This allows the GDB remote platform to spawn new GDB server instances (debugserver) to allow platform debugging. llvm-svn: 129351
2011-04-12 13:54:46 +08:00
{
if (error_ptr)
error_ptr->SetErrorToErrno();
return eConnectionStatusError;
}
saddr_un.sun_family = AF_UNIX;
::strncpy(saddr_un.sun_path, socket_name, sizeof(saddr_un.sun_path) - 1);
saddr_un.sun_path[sizeof(saddr_un.sun_path) - 1] = '\0';
#if defined(__APPLE__) || defined(__FreeBSD__)
saddr_un.sun_len = SUN_LEN (&saddr_un);
#endif
if (::connect (m_fd_send, (struct sockaddr *)&saddr_un, SUN_LEN (&saddr_un)) < 0)
Moved the execution context that was in the Debugger into the CommandInterpreter where it was always being used. Make sure that Modules can track their object file offsets correctly to allow opening of sub object files (like the "__commpage" on darwin). Modified the Platforms to be able to launch processes. The first part of this move is the platform soon will become the entity that launches your program and when it does, it uses a new ProcessLaunchInfo class which encapsulates all process launching settings. This simplifies the internal APIs needed for launching. I want to slowly phase out process launching from the process classes, so for now we can still launch just as we used to, but eventually the platform is the object that should do the launching. Modified the Host::LaunchProcess in the MacOSX Host.mm to correctly be able to launch processes with all of the new eLaunchFlag settings. Modified any code that was manually launching processes to use the Host::LaunchProcess functions. Fixed an issue where lldb_private::Args had implicitly defined copy constructors that could do the wrong thing. This has now been fixed by adding an appropriate copy constructor and assignment operator. Make sure we don't add empty ModuleSP entries to a module list. Fixed the commpage module creation on MacOSX, but we still need to train the MacOSX dynamic loader to not get rid of it when it doesn't have an entry in the all image infos. Abstracted many more calls from in ProcessGDBRemote down into the GDBRemoteCommunicationClient subclass to make the classes cleaner and more efficient. Fixed the default iOS ARM register context to be correct and also added support for targets that don't support the qThreadStopInfo packet by selecting the current thread (only if needed) and then sending a stop reply packet. Debugserver can now start up with a --unix-socket (-u for short) and can then bind to port zero and send the port it bound to to a listening process on the other end. This allows the GDB remote platform to spawn new GDB server instances (debugserver) to allow platform debugging. llvm-svn: 129351
2011-04-12 13:54:46 +08:00
{
if (error_ptr)
error_ptr->SetErrorToErrno();
Disconnect (NULL);
Moved the execution context that was in the Debugger into the CommandInterpreter where it was always being used. Make sure that Modules can track their object file offsets correctly to allow opening of sub object files (like the "__commpage" on darwin). Modified the Platforms to be able to launch processes. The first part of this move is the platform soon will become the entity that launches your program and when it does, it uses a new ProcessLaunchInfo class which encapsulates all process launching settings. This simplifies the internal APIs needed for launching. I want to slowly phase out process launching from the process classes, so for now we can still launch just as we used to, but eventually the platform is the object that should do the launching. Modified the Host::LaunchProcess in the MacOSX Host.mm to correctly be able to launch processes with all of the new eLaunchFlag settings. Modified any code that was manually launching processes to use the Host::LaunchProcess functions. Fixed an issue where lldb_private::Args had implicitly defined copy constructors that could do the wrong thing. This has now been fixed by adding an appropriate copy constructor and assignment operator. Make sure we don't add empty ModuleSP entries to a module list. Fixed the commpage module creation on MacOSX, but we still need to train the MacOSX dynamic loader to not get rid of it when it doesn't have an entry in the all image infos. Abstracted many more calls from in ProcessGDBRemote down into the GDBRemoteCommunicationClient subclass to make the classes cleaner and more efficient. Fixed the default iOS ARM register context to be correct and also added support for targets that don't support the qThreadStopInfo packet by selecting the current thread (only if needed) and then sending a stop reply packet. Debugserver can now start up with a --unix-socket (-u for short) and can then bind to port zero and send the port it bound to to a listening process on the other end. This allows the GDB remote platform to spawn new GDB server instances (debugserver) to allow platform debugging. llvm-svn: 129351
2011-04-12 13:54:46 +08:00
return eConnectionStatusError;
}
if (error_ptr)
error_ptr->Clear();
return eConnectionStatusSuccess;
}
ConnectionStatus
ConnectionFileDescriptor::SocketListen (uint16_t listen_port_num, Error *error_ptr)
{
Log *log(lldb_private::GetLogIfAnyCategoriesSet (LIBLLDB_LOG_CONNECTION));
if (log)
log->Printf ("%p ConnectionFileDescriptor::SocketListen (port = %i)", this, listen_port_num);
Disconnect (NULL);
m_fd_send_type = m_fd_recv_type = eFDTypeSocket;
int listen_port = ::socket (AF_INET, SOCK_STREAM, IPPROTO_TCP);
if (listen_port == -1)
{
if (error_ptr)
error_ptr->SetErrorToErrno();
return eConnectionStatusError;
}
// enable local address reuse
SetSocketOption (listen_port, SOL_SOCKET, SO_REUSEADDR, 1);
SocketAddress localhost;
if (localhost.SetToLocalhost (AF_INET, listen_port_num))
{
int err = ::bind (listen_port, localhost, localhost.GetLength());
if (err == -1)
{
if (error_ptr)
error_ptr->SetErrorToErrno();
Close (listen_port, NULL);
return eConnectionStatusError;
}
err = ::listen (listen_port, 1);
if (err == -1)
{
if (error_ptr)
error_ptr->SetErrorToErrno();
Close (listen_port, NULL);
return eConnectionStatusError;
}
m_fd_send = m_fd_recv = ::accept (listen_port, NULL, 0);
if (m_fd_send == -1)
{
if (error_ptr)
error_ptr->SetErrorToErrno();
Close (listen_port, NULL);
return eConnectionStatusError;
}
}
// We are done with the listen port
Close (listen_port, NULL);
m_should_close_fd = true;
// Keep our TCP packets coming without any delays.
SetSocketOption (m_fd_send, IPPROTO_TCP, TCP_NODELAY, 1);
if (error_ptr)
error_ptr->Clear();
return eConnectionStatusSuccess;
}
ConnectionStatus
ConnectionFileDescriptor::ConnectTCP (const char *host_and_port, Error *error_ptr)
{
Log *log(lldb_private::GetLogIfAnyCategoriesSet (LIBLLDB_LOG_CONNECTION));
if (log)
log->Printf ("%p ConnectionFileDescriptor::ConnectTCP (host/port = %s)", this, host_and_port);
Disconnect (NULL);
m_fd_send_type = m_fd_recv_type = eFDTypeSocket;
std::string host_str;
std::string port_str;
int32_t port = INT32_MIN;
if (!DecodeHostAndPort (host_and_port, host_str, port_str, port, error_ptr))
return eConnectionStatusError;
// Create the socket
m_fd_send = m_fd_recv = ::socket (AF_INET, SOCK_STREAM, IPPROTO_TCP);
if (m_fd_send == -1)
{
if (error_ptr)
error_ptr->SetErrorToErrno();
return eConnectionStatusError;
}
m_should_close_fd = true;
// Enable local address reuse
SetSocketOption (m_fd_send, SOL_SOCKET, SO_REUSEADDR, 1);
struct sockaddr_in sa;
::memset (&sa, 0, sizeof (sa));
sa.sin_family = AF_INET;
sa.sin_port = htons (port);
int inet_pton_result = ::inet_pton (AF_INET, host_str.c_str(), &sa.sin_addr);
if (inet_pton_result <= 0)
{
struct hostent *host_entry = gethostbyname (host_str.c_str());
if (host_entry)
host_str = ::inet_ntoa (*(struct in_addr *)*host_entry->h_addr_list);
inet_pton_result = ::inet_pton (AF_INET, host_str.c_str(), &sa.sin_addr);
if (inet_pton_result <= 0)
{
if (error_ptr)
{
if (inet_pton_result == -1)
error_ptr->SetErrorToErrno();
else
error_ptr->SetErrorStringWithFormat("invalid host string: '%s'", host_str.c_str());
}
Disconnect (NULL);
return eConnectionStatusError;
}
}
if (-1 == ::connect (m_fd_send, (const struct sockaddr *)&sa, sizeof(sa)))
{
if (error_ptr)
error_ptr->SetErrorToErrno();
Disconnect (NULL);
return eConnectionStatusError;
}
// Keep our TCP packets coming without any delays.
SetSocketOption (m_fd_send, IPPROTO_TCP, TCP_NODELAY, 1);
if (error_ptr)
error_ptr->Clear();
return eConnectionStatusSuccess;
}
ConnectionStatus
ConnectionFileDescriptor::ConnectUDP (const char *host_and_port, Error *error_ptr)
{
Log *log(lldb_private::GetLogIfAnyCategoriesSet (LIBLLDB_LOG_CONNECTION));
if (log)
log->Printf ("%p ConnectionFileDescriptor::ConnectUDP (host/port = %s)", this, host_and_port);
Disconnect (NULL);
m_fd_send_type = m_fd_recv_type = eFDTypeSocketUDP;
std::string host_str;
std::string port_str;
int32_t port = INT32_MIN;
if (!DecodeHostAndPort (host_and_port, host_str, port_str, port, error_ptr))
return eConnectionStatusError;
// Setup the receiving end of the UDP connection on this localhost
// on port zero. After we bind to port zero we can read the port.
m_fd_recv = ::socket (AF_INET, SOCK_DGRAM, 0);
if (m_fd_recv == -1)
{
// Socket creation failed...
if (error_ptr)
error_ptr->SetErrorToErrno();
}
else
{
// Socket was created, now lets bind to the requested port
SocketAddress addr;
addr.SetToLocalhost (AF_INET, 0);
if (::bind (m_fd_recv, addr, addr.GetLength()) == -1)
{
// Bind failed...
if (error_ptr)
error_ptr->SetErrorToErrno();
Disconnect (NULL);
}
}
if (m_fd_recv == -1)
return eConnectionStatusError;
// At this point we have setup the recieve port, now we need to
// setup the UDP send socket
struct addrinfo hints;
struct addrinfo *service_info_list = NULL;
::memset (&hints, 0, sizeof(hints));
hints.ai_family = AF_INET;
hints.ai_socktype = SOCK_DGRAM;
int err = ::getaddrinfo (host_str.c_str(), port_str.c_str(), &hints, &service_info_list);
if (err != 0)
{
if (error_ptr)
error_ptr->SetErrorStringWithFormat("getaddrinfo(%s, %s, &hints, &info) returned error %i (%s)",
host_str.c_str(),
port_str.c_str(),
err,
gai_strerror(err));
Disconnect (NULL);
return eConnectionStatusError;
}
for (struct addrinfo *service_info_ptr = service_info_list;
service_info_ptr != NULL;
service_info_ptr = service_info_ptr->ai_next)
{
m_fd_send = ::socket (service_info_ptr->ai_family,
service_info_ptr->ai_socktype,
service_info_ptr->ai_protocol);
if (m_fd_send != -1)
{
m_udp_send_sockaddr = service_info_ptr;
break;
}
else
continue;
}
:: freeaddrinfo (service_info_list);
if (m_fd_send == -1)
{
Disconnect (NULL);
return eConnectionStatusError;
}
if (error_ptr)
error_ptr->Clear();
m_should_close_fd = true;
return eConnectionStatusSuccess;
}
#if defined(__MINGW32__) || defined(__MINGW64__)
typedef const char * set_socket_option_arg_type;
typedef char * get_socket_option_arg_type;
#else // #if defined(__MINGW32__) || defined(__MINGW64__)
typedef const void * set_socket_option_arg_type;
typedef void * get_socket_option_arg_type;
#endif // #if defined(__MINGW32__) || defined(__MINGW64__)
int
ConnectionFileDescriptor::GetSocketOption(int fd, int level, int option_name, int &option_value)
{
get_socket_option_arg_type option_value_p = static_cast<get_socket_option_arg_type>(&option_value);
socklen_t option_value_size = sizeof(int);
return ::getsockopt(fd, level, option_name, option_value_p, &option_value_size);
}
int
ConnectionFileDescriptor::SetSocketOption(int fd, int level, int option_name, int option_value)
{
set_socket_option_arg_type option_value_p = static_cast<get_socket_option_arg_type>(&option_value);
return ::setsockopt(fd, level, option_name, option_value_p, sizeof(option_value));
}
bool
ConnectionFileDescriptor::SetSocketReceiveTimeout (uint32_t timeout_usec)
{
switch (m_fd_recv_type)
{
case eFDTypeFile: // Other FD requireing read/write
break;
case eFDTypeSocket: // Socket requiring send/recv
case eFDTypeSocketUDP: // Unconnected UDP socket requiring sendto/recvfrom
{
// Check in case timeout for m_fd has already been set to this value
if (timeout_usec == m_socket_timeout_usec)
return true;
//printf ("ConnectionFileDescriptor::SetSocketReceiveTimeout (timeout_usec = %u)\n", timeout_usec);
struct timeval timeout;
if (timeout_usec == UINT32_MAX)
{
timeout.tv_sec = 0;
timeout.tv_usec = 0;
}
else if (timeout_usec == 0)
{
// Sending in zero does an infinite timeout, so set this as low
// as we can go to get an effective zero timeout...
timeout.tv_sec = 0;
timeout.tv_usec = 1;
}
else
{
timeout.tv_sec = timeout_usec / TimeValue::MicroSecPerSec;
timeout.tv_usec = timeout_usec % TimeValue::MicroSecPerSec;
}
if (::setsockopt (m_fd_recv, SOL_SOCKET, SO_RCVTIMEO, &timeout, sizeof(timeout)) == 0)
{
m_socket_timeout_usec = timeout_usec;
return true;
}
}
}
return false;
}
in_port_t
ConnectionFileDescriptor::GetSocketPort (int fd)
{
// We bound to port zero, so we need to figure out which port we actually bound to
SocketAddress sock_addr;
socklen_t sock_addr_len = sock_addr.GetMaxLength ();
if (::getsockname (fd, sock_addr, &sock_addr_len) == 0)
return sock_addr.GetPort ();
return 0;
}
// If the read file descriptor is a socket, then return
// the port number that is being used by the socket.
in_port_t
ConnectionFileDescriptor::GetReadPort () const
{
return ConnectionFileDescriptor::GetSocketPort (m_fd_recv);
}
// If the write file descriptor is a socket, then return
// the port number that is being used by the socket.
in_port_t
ConnectionFileDescriptor::GetWritePort () const
{
return ConnectionFileDescriptor::GetSocketPort (m_fd_send);
}