aboutsummaryrefslogtreecommitdiff
path: root/pcap-rpcap.c
Commit message (Collapse)AuthorAgeFilesLines
...
* Check whether the compiler on Travis allows mixed declarations and code.Guy Harris2018-05-241-2/+1
| | | | It *should*, as we should be asking for at least C99, if not "gnu99".
* Rename SOCK_MESSAGE() to SOCK_DEBUG_MESSAGE().Guy Harris2018-04-011-4/+4
| | | | | | | This emphasizes that it's for *debugging* messages, not for errors sent to the peer or logged to a daemon log file. Any message that would be useful if printed/logged in a production environment should *not* be reported with SOCK_DEBUG_MESSAGE().
* Rename SOCK_ASSERT to SOCK_MESSAGEJoerg Mayer2018-04-011-4/+4
| | | | | SOCK_ASSERT was always called with a true expression ("1") thus only ever executing the message part. Rename the macro, remove the assert part and remove the "1" parameter.
* Note in a comment that no reply is sent to RPCAP_MSG_CLOSE.Guy Harris2018-03-311-1/+1
|
* Check whether the mode argument to rpcap_stats_rpcap() is valid.Guy Harris2018-03-311-0/+11
| | | | | | It's an internal routine, so it *shouldn't* ever be invalid, but this at least makes sure the mode argument is used, in case we ever have compilers check for unused arguments.
* Don't test ENABLE_REMOTE.Guy Harris2018-03-311-4/+4
| | | | | ENABLE_REMOTE is always defined when building this file. If you haven't enabled remote capture, you wouldn't even be compiling the file.
* Cast away warnings.Guy Harris2018-03-221-9/+9
| | | | | | | | | | | | | | On UN*X, a SOCKET is just an FD, hence an int, so this isn't an issue. On Windows, well, hopefully -1, -2, and -3 aren't valid values; if they are, this API can't be made to work (it's already documented as returning -1, -2, and -3 for various errors). So we might as well just cast the negative values to SOCKET, given that we're returning a SOCKET value. Once we support pcap_create()/pcap_activate() for remote capturing, we can provide a better API for active-mode capturing.
* Don't set auth_type until we know it's a type we support.Guy Harris2018-03-221-2/+2
| | | | | If it is, we know it fits into a uint16; cast the assignment to squelch a warning.
* Check for sampling method and value not fitting into a message.Guy Harris2018-03-221-2/+18
| | | | And once we know they fit, cast away warnings.
* Use for (;;) rather than while (1).Guy Harris2018-03-201-1/+1
| | | | That squelches "conditional expression is constant" warnings from MSVC.
* Free addrinfo in the error code path if necessary.Guy Harris2018-01-211-0/+3
| | | | This should fix Coverity CID 1418992.
* Fail if a given protocol version *isn't* supported.Guy Harris2017-11-281-1/+1
| | | | (Broken when RPCAP_VERSION_IS_SUPPORTED() was introduced.)
* More crap for HP-UX.Guy Harris2017-11-251-0/+2
| | | | What a mess.
* Suppress warnings if RPCAP_MIN_VERSION is 0.Guy Harris2017-11-251-2/+1
| | | | | | Version numbers are unsigned, so they can never be < 0, so checking whether a version is less than the minimum supported version will never fail.
* Add a routine to format error messages with an errno-based message at the end.Guy Harris2017-11-151-11/+20
| | | | | | | | | | | | | | That routine will use strerror_s() or strerror_r() if available, in a fashion that's thread-safe. Otherwise, it falls back on pcap_strerror(). Use it in both libpcap and rpcapd. Given that we check for errors in strerror_r(), hopefully this will squelch warnings with newer version of GCC and GNU libc; whilst the macOS (and other BSD-flavored?) strerror_r() always fills in a message, that's not required by the Single UNIX Specification, as far as I can tell, so we apparently really *do* need to check for errors.
* Plug memory leaks and clean up interface list creation.Guy Harris2017-11-131-13/+28
| | | | | | | | | | | | If we fail in the middle of constructing lists of savefiles or remote devices, free the entries we've already allocated. Use separate variables for "current device" and "last device in the list", to try to squelch some warnings. We already rely on the pointer to the place where the device list pointer should be put not being null; no need to handle the case where it is null.
* Clean up handing of signal interrupts in pcap_read_nocb_remote().Guy Harris2017-11-121-59/+146
| | | | | | | | | | | | | | | | | | | | | | | If the select() is interrupted by a signal, just return 0, don't treat it as an error. That'll let the read routine check whether we're supposed to break out of the loop and, if not, it'll let it keep processing packets, as we do for local capture modules. When reading packet messages from a TCP socket, use the bp and cc members of the pcap_t to keep track of where we're reading data and the number of bytes of data we have read so far. If a read is interrupted by a signal, just return 0, as per the above. Don't assume we're at the beginning of a message when we enter pcap_read_nocb_remote(); instead, if we've read less than the size of a packet message header, try to read the rest of the header, otherwise extract the packet payload length from the header, make sure it's not larger than the largest payload length we're expecting (if it is, return an error), and try to read the rest of the packet. That way, we can handle multiple calls to pcap_read_nocb_remote() reading multiple bits of the same packet, due to reads being interrupted. That obviates the need to discard the rest of the packet, as we read all of it into the buffer.
* You can't do pointer arithmetic on void * in standard C.Guy Harris2017-11-111-2/+2
| | | | | GCC and GCC-compatible compilers might allow it, but not all compilers do.
* Clean up packet message reception.Guy Harris2017-11-111-86/+152
| | | | | | | | | | | | | | | | | | | | | | | | | | | Have separate sockutils routines for reading from a stream socket and receiving messages from a datagram socket. The latter assumes you get the entire message in a single receive operation and checks for the datagram being bigger than the provided buffer. We don't need separate network and packet-to-provide-to-the-callback buffers; use the standard buffer for both purposes, making it big enough for the largest expected packet message. This eliminates a copy. Use the "receive messages from a datagram socket" routine for UDP data sockets. That way, we check for messages bigger than the largest expected packet message; add additional checks to make sure the packet is big enough to hold an rpcap header and, if so, that the packet length in the rpcap header isn't larger than the part of the message. Check for interrupted message receives in the "read a packet message" code path. If we get an invalid message from the server, discard the rest of it if it's sent over TCP. In the read routine, check for errors and, if we didn't get a packet and didn't get an error - which could be a timeout or an interrupted receive - check whether we were told to break out of the loop.
* Don't use fp->bufsize for the socket buffer size.Guy Harris2017-11-111-22/+36
| | | | | | Use a separate variable, to make it clearer that it's *not* the user-mode buffer attached to the pcap_t; it's more like the kernel buffer used by local capture mechanisms.
* Make the client-size buffer big enough for the largest packet message.Guy Harris2017-11-111-10/+29
| | | | | | It has to be large enough for the rpcap message header, the rpcap packet header, and the largest possible packet for the pcap_t, i.e. the snapshot length.
* In the server, don't complain if the client reset the connection.Guy Harris2017-11-101-8/+14
| | | | | If the client flakes out for whatever reason, including exiting, that's not a problem that the server needs to log.
* Handle the client closing the connection better.Guy Harris2017-11-101-6/+20
| | | | | | On the server side, don't treat an immediate EOF when trying to read a message header as an error, just silently terminate the session. Treat EOFs after reading *some* of the message as an error.
* Clean up error handling when closing an active connection.Guy Harris2017-11-011-11/+27
| | | | | | | | | | | If we get an error, return -1, and just report the first error; if we got an error sending RPCAP_MSG_CLOSE to the server, don't check for errors when closing the connection. For calls where we're not going to check for an error, don't pass an error message buffer. Addresses Coverity CID 1419000.
* Make sure rpcap_msg_err() returns a null-terminated string.Guy Harris2017-11-011-13/+12
| | | | | | | | | | Make sure we null-terminate the string before discarding the rest of a too-long error message fails. While we're at it, if we get a network error, we don't take any special shutdown action for the connection, so no need to have rpcap_msg_err() return a status; either it's reporting an error given to us by the server or it's reporting a network error.
* In pcap_findalldevs_ex_remote(), properly set the active-connection flag.Guy Harris2017-11-011-1/+3
| | | | | | It's used later to decide whether to close the socket or not. Addresses Coverity CID 1420505.
* Don't *report* errors when closing, but check for them in some cases.Guy Harris2017-11-011-22/+51
| | | | | | If you send a message to the server and wait for a reply, check whether the message was successfully sent before awaiting a reply; if it never made it onto the wire, you won't be getting a reply.
* Use rpcap_recv() and pcap_discard() for RPCAP_MSG_PACKET payloads over TCP.Guy Harris2017-11-011-24/+13
| | | | | That way, the appropriate packet length checks, and other checks, are done.
* Fix test.Guy Harris2017-10-311-1/+1
| | | | | It's a problem if we're trying to get *more* data from a message than is left in the message, not if we're trying to get *less* data.
* Get rid of sock_recv() call that should have been removed in a previous change.Guy Harris2017-10-311-3/+0
|
* Fix authentication/protocol version negotiation.Guy Harris2017-10-311-2/+19
| | | | | If the first authentication try returned 0, that's success; no need to try again.
* Fix up the negotiation, eliminating a failure mode with old clients.Guy Harris2017-10-311-14/+16
| | | | | | | | | | | | | | | If we send a "wrong version number" error message to an old client (before the changes to support version negotiation) with a version number other than 0, it'll treat it as a protocol error and return the wrong error message to the caller. Instead, if the client's version number is too old for the server, send back the version number they sent us; they'll try it again and get the same error, and give up. (This would happen only if there's a server that doesn't support version 0; let's not create so many protocol versions that there's a temptation to drop older ones.) Also, add details on protocol versioning and protocol negotiation (the construction of which pointed out the failure mode).
* Redo the message processing in the client, add protocol version negotiation.Guy Harris2017-10-311-372/+583
| | | | | | Currently, we only have one protocol version, but the code should be able to handle multiple versions in the future - and to be able to work with older code that doesn't do negotiation.
* Declare hints and addrinfo inside the if clause where we use it.Guy Harris2017-10-101-8/+3
| | | | | | | | The only uses of it outside there are invalid - there's no need to free the address list outside that if clause, we've already freed it by the time we leave it. Fixes Coverity CID 1419031.
* socket() and accept() return INVALID_SOCKET on errors.Guy Harris2017-10-071-3/+3
| | | | | | | | | | On Windows, you're supposed to check against INVALID_SOCKET. On UN*X, you're supposed to check against -1, but, on UN*X, we define INVALID_SOCKET to be -1 so you can compare against INVALID_SOCKET on both platforms. That means that sock_open() should return INVALID_SOCKET on errors as well.
* Don't try to close sockctrl() if the problem is that we couldn't open it.Guy Harris2017-10-071-1/+4
| | | | Fixes Coverity CID 1418987.
* Free the address info before we return.Guy Harris2017-10-071-0/+2
| | | | Fixes Coverity CID 1418989.
* Free the address list if we found an address in it.Guy Harris2017-10-071-0/+1
| | | | Fixes Coverity CID 1419004.
* Free an address list once we're done with it.Guy Harris2017-10-071-0/+2
| | | | This fixes Coverity CID 1419013.
* Only go to error if we have a control socket.Guy Harris2017-10-071-2/+6
| | | | | | | | | | | If we don't have a control socket, just free up the address list, close the pcap_t, and return. Fixes Coverity CID 1419021. Also, don't initialize the control socket; that lets compilers (and other tools) doing dataflow analysis catch cases where we haven't set the socket before using it.
* Call the flag to enable remote capture ENABLE_REMOTE.Guy Harris2017-10-031-4/+4
| | | | | | HAVE_REMOTE implies that there's some feature in our build environment that we have, but this is a flag that controls what the user specified at configuration time.
* Explicitly reject non-blocking mode on D-Bus and rpcap.Guy Harris2017-10-011-2/+19
| | | | | | | We currently don't support non-blocking mode on either of those, so have routines that return errors for pcap_getnonblock() and pcap_setnonblock(); you can't set those function pointers to null to say "we don't support it".
* Put rpcap protocol routines common to client and server in rpcap-protocol.cGuy Harris2017-09-301-257/+0
| | | | | That way, rpcapd doesn't depend on them being exported from libpcap (which they shouldn't be).
* Make sure sockutils.h is included before portability.h.Guy Harris2017-09-031-1/+1
| | | | | | | | sockutils.h may include <crtdbg.h> on Windows, and portability.h expects that, if <crtdbg.h> is going to be included, it'll be included before portability.h, so that, if __STDC__ is 0, strdup will have been defined by <crtdbg.h>, and portability.h won't define it only to have a later include of <crtdbg.h> redefine it and provoke warnings.
* Always include <config.h> rather than "config.h".Guy Harris2017-08-181-1/+1
| | | | | | | | This can prevent bizarre failures if, for example, you've done a configuration in the top-level source directory, leaving behind one config.h file, and then do an out-of-tree build in another directory, with different configuration options. This way, we always pick up the same config.h, in the build directory.
* Make the checks and adjustment of the snapshot length module-dependent.Guy Harris2017-06-011-0/+15
| | | | | | | | | | | | | Also, initialize the snapshot length to 0, meaning "not specified", so that the default snapshot length, if not specified, is also module-dependent. That way, D-Bus has a maximum and default of 128MB, as that's the maximum message size, but other capture devices have the current MAXIMUM_SNAPLEN, so we can handle full-size D-Bus messages without advertising an overly-large snapshot length for other devices, potentially causing libpcap and programs using it or reading libpcap files to allocate overly-large buffers for other capture devices.
* Move the RPCAP-specific functions to pcap-rpcap.h.Guy Harris2017-03-201-0/+1
| | | | | | | Move the declarations of the interfaces that are currently used for RPCAP by pcap_open_live(), pcap_open(), and pcap_findalldevs_ex() to pcap-rpcap.h; in the future, that will hold the routines used by pcap_create() and any future API for enumerating interfaces.
* Clean up some comments.Guy Harris2017-03-201-17/+18
|
* Get rid of unnecessary brackets and an unnecessary semicolon.Guy Harris2017-03-201-2/+0
|
* Make rpcap_sendauth() static.Guy Harris2017-03-141-1/+2
| | | | It's now used only in pcap-rpcap.c.
generated by cgit v1.2.3 (git 2.39.1) at 2025-05-11 06:54:42 +0000