Netjack dev: Squashed Interface selection plus IPv6 #27
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
I wonder what has happened to your previous commits. You sent four individual patches to the mailing list, but now, we only see two plus a bunch of merge commits. Could you rebase your feature branch against current master and update the pull request? I guess we simply merge in your work and wait for bug reports. ;) Last not least: please double-check whitespace, the patches you've sent to the ML contained some lines with wrong indentation. Just to be sure we don't end up with a mix of tabs/spaces. If need be, I could also create a whitespace /indentation patch for all the files your commits are going to touch. |
|
Hi Adrian, On Mon, Apr 15, 2013 at 08:10:53AM -0700, Adrian Knoth wrote:
Yes, I've reworked initial patches - squashed first commits into single commit (add interface selection) and added another commit with IPv6 support. Since then - since there were no comments I just kept pulling new upstream commits to spot on conflicts.
If there are no other comments - like to drop internal socket state tracker (or other tricks to preserve compatibility with behaviour of old socket API) at cost of deeper changes in jacknet2 (NetInterface/Driver/Adapter) - I think it should be ok to merge.
Thanks for feedback :) Regards, |
|
sorry for a shit^H^H^H^H push storm - was trying to clean up all the issues and keep tabs expanded... Now should be ready for test/merge |
|
I know it's been four years, but do you still care about this pull request? If so, please rebase/update, and I promise to take a timely look. Thanks for contributing, our lack of responsiveness is truly a shame. My sincerest apologies. |
* Multicast interface selection to UnixSocket - added methods Bind and JoinMCastGroup with <char *if_name> argument * Interface handling to NetInterface class tree - new field and its initialization/usage in Slave branch. * Multicast interface selection to NetSlaves (Adapter/Driver) and NetMaster, where master can now listen all interfaces.
|
Hi Adrian, Long time indeed, let me take a look. Regards, |
* Cleaned up direct references to sockaddr_in * Changed fRecvAddr & fSendAddr to sockaddr_storage * Added field fFamily indicating current probed AF. * Added protected method ProbeAF accepting IP to probe and addr to fill in. As well as final call to test which should be either connect or bind. * Reworked protocol specific methods to act on fFamily * IsLocal now checks assigned interface addresses * Introduced internal state tracker to avoid double bind * Workaround for GLIBC bug returning wrong order of AFs http://sourceware.org/bugzilla/show_bug.cgi?id=14967 * Corrected interface selection for Slaves - only one interface could be used to multicast the packet * Retab changes
|
Well, it appeared to be in the end better than I thought. And from the other side Unfortunately I've disassembled my daw setup so I cannot test it fully (i've used a docker on my router to simulate a slave) but devices with different arch connected and synced. Maybe later I'll recover some old gears and connect them all together. |
7890
added a commit
that referenced
this pull request
Feb 8, 2019
Branch netjack_squashed_interface_selection_plus_ipv6 Original PR: #27 by rufferson, first comment Mar 3, 2013 Commits on Mar 12, 2017 Add interface selection for NetJack components @rufferson rufferson committed on Mar 3, 2013 * Multicast interface selection to UnixSocket - added methods Bind and JoinMCastGroup with <char *if_name> argument * Interface handling to NetInterface class tree - new field and its initialization/usage in Slave branch. * Multicast interface selection to NetSlaves (Adapter/Driver) and NetMaster, where master can now listen all interfaces. Commits on Mar 15, 2017 Add IPv6 support to UnixSocket and NetInterfaces @rufferson rufferson committed on Feb 21, 2013 * Cleaned up direct references to sockaddr_in * Changed fRecvAddr & fSendAddr to sockaddr_storage * Added field fFamily indicating current probed AF. * Added protected method ProbeAF accepting IP to probe and addr to fill in. As well as final call to test which should be either connect or bind. * Reworked protocol specific methods to act on fFamily * IsLocal now checks assigned interface addresses * Introduced internal state tracker to avoid double bind * Workaround for GLIBC bug returning wrong order of AFs http://sourceware.org/bugzilla/show_bug.cgi?id=14967 * Corrected interface selection for Slaves - only one interface could be used to multicast the packet * Retab changes This branch has conflicts that must be resolved to resolve conflicts before continuing. Conflicting files common/JackNetDriver.h posix/JackNetUnixSocket.cpp wget https://github.com/jackaudio/jack2/pull/27.diff $ patch -p 1 < 27.diff patching file common/JackNetAdapter.cpp patching file common/JackNetDriver.cpp patching file common/JackNetDriver.h Hunk #1 FAILED at 79. 1 out of 1 hunk FAILED -- saving rejects to file common/JackNetDriver.h.rej patching file common/JackNetInterface.cpp patching file common/JackNetInterface.h patching file common/JackNetManager.cpp patching file common/JackNetManager.h patching file common/JackNetTool.h patching file posix/JackNetUnixSocket.cpp Hunk #4 FAILED at 229. Hunk #5 succeeded at 281 (offset -8 lines). Hunk #6 succeeded at 303 (offset -8 lines). Hunk #7 succeeded at 325 (offset -8 lines). Hunk #8 succeeded at 363 (offset -8 lines). Hunk #9 succeeded at 486 (offset -8 lines). Hunk #10 succeeded at 508 (offset -8 lines). Hunk #11 succeeded at 537 (offset -8 lines). Hunk #12 succeeded at 586 (offset -8 lines). Hunk #13 succeeded at 607 (offset -8 lines). Hunk #14 succeeded at 615 (offset -8 lines). Hunk #15 succeeded at 644 (offset -8 lines). Hunk #16 succeeded at 673 (offset -8 lines). 1 out of 16 hunks FAILED -- saving rejects to file posix/JackNetUnixSocket.cpp.rej find|grep rej ./posix/JackNetUnixSocket.cpp.rej ./common/JackNetDriver.h.rej --- posix/JackNetUnixSocket.cpp +++ posix/JackNetUnixSocket.cpp @@ -229,58 +311,117 @@ { if (strcmp(ip, "127.0.0.1") == 0) { return true; - } + } else if(!strcmp(ip,"::1")) + return true; - char host_name[32]; - gethostname(host_name, sizeof(host_name)); + struct ifaddrs *ifas, *ifa; + socklen_t len; - struct hostent* host = gethostbyname(host_name); - if (host) { - for (int i = 0; host->h_addr_list[i] != 0; ++i) { - struct in_addr addr; - memcpy(&addr, host->h_addr_list[i], sizeof(struct in_addr)); - if (strcmp(inet_ntoa(addr), ip) == 0) { - return true; - } - } - return false; - } else { - return false; + if (getifaddrs(&ifas) == -1) { + jack_error("JackNetUnixSocket::IsLocal error in getifaddrs"); + return false; } + for (ifa = ifas; ifa != NULL; ifa = ifa->ifa_next) { + if (ifa->ifa_addr == NULL) + continue; // Address is mandatory + len = (ifa->ifa_addr->sa_family==AF_INET)?sizeof(struct sockaddr_in):sizeof(struct sockaddr_in6); + if(!getnameinfo(ifa->ifa_addr, len, f_addr_buff, INET6_ADDRSTRLEN, NULL,0, NI_NUMERICSERV | NI_DGRAM | NI_NUMERICHOST)) + if(!strcmp(f_addr_buff,ip)) + break; + } + freeifaddrs(ifas); + return (ifa != NULL); } int JackNetUnixSocket::Bind() { - return bind(fSockfd, reinterpret_cast<socket_address_t*>(&fRecvAddr), sizeof(socket_address_t)); + int yes=1; + if(fState & JNS_BOUND) return 0; + // Multicast is incompatible with V4MAPPED or V4COMPAT addresses, if probe detected MC we need V6ONLY + if(fFamily == AF_INET6 && IN6_IS_ADDR_UNSPECIFIED(&_sock6(fRecvAddr).sin6_addr) && fState & JNS_MCAST) + if(SetOption(IPPROTO_IPV6, IPV6_V6ONLY, &yes, sizeof(yes))) return SOCKET_ERROR; + if(bind(fSockfd, reinterpret_cast<struct sockaddr*>(&fRecvAddr), sizeof(fRecvAddr))) return SOCKET_ERROR; + fState |= JNS_BOUND; + return 0; + } + int JackNetUnixSocket::Bind(const char *if_name) + { + int ret = Bind(); + if(!ret && strcmp(if_name,"any")) { + if(fFamily == AF_INET) { + // 'all' for this case will lead to 'last valid interface', which is not that one might expect + if(strcmp(if_name,"all")) + ret = BindMCastIface(if_name, IP_MULTICAST_IF, &_sock4(fSendAddr).sin_addr); + else + jack_error("Multicast Interface all not found, sending from default"); + } else if(fFamily == AF_INET6) { + struct if_nameindex *if_ni = if_nameindex(); // In V6 world we do everything differently. + if(if_ni) { + int i; + for (i=0; if_ni[i].if_index > 0; i++) { + if(if_ni[i].if_index == 1) + continue; // Skip loopback + if(!strcmp(if_ni[i].if_name,if_name)) { + ret = SetOption(IPPROTO_IPV6, IPV6_MULTICAST_IF, &if_ni[i].if_index, sizeof(if_ni[i].if_index)); + jack_log("JackNetUnixSocket::Bind Multicasting from %s",if_ni[i].if_name); + break; + } + } + if(if_ni[i].if_index == 0) jack_error("Multicast Interface %s not found, sending from default",if_name); + if_freenameindex(if_ni); + } + } + } + return ret; } int JackNetUnixSocket::BindWith(const char* ip) { - int addr_conv = inet_aton(ip, &fRecvAddr.sin_addr); - if (addr_conv < 0) { - return addr_conv; + if(fFamily == AF_UNSPEC) { + if(!fPort) return SOCKET_ERROR; + if(ProbeAF(ip,&fRecvAddr,&bind)<0) return SOCKET_ERROR; + fState |= JNS_BOUND; + return 0; + } else { + if(SetRecvIP(ip)==-1) return SOCKET_ERROR; + return Bind(); } - return Bind(); } int JackNetUnixSocket::BindWith(int port) { - fRecvAddr.sin_port = htons(port); - return Bind(); + if(fFamily == AF_UNSPEC) { + fPort = port; + if(ProbeAF(NULL,&fRecvAddr,&bind)<0) return SOCKET_ERROR; + fState |= JNS_BOUND; + return 0; + } else { + SetPort(port); + return Bind(); + } } int JackNetUnixSocket::Connect() { - return connect(fSockfd, reinterpret_cast<socket_address_t*>(&fSendAddr), sizeof(socket_address_t)); + if(fFamily != AF_UNSPEC) + return connect(fSockfd, (struct sockaddr*)&fSendAddr,sizeof(fSendAddr)); + jack_error("JackNetUnixSocket::Connect Family not initialized"); + return SOCKET_ERROR; } int JackNetUnixSocket::ConnectTo(const char* ip) { - int addr_conv = inet_aton(ip, &fSendAddr.sin_addr); - if (addr_conv < 0) { - return addr_conv; + socklen_t l=sizeof(fRecvAddr); + if(fPort==0) return SOCKET_ERROR; + if(fState & JNS_PROBED) { + Reset(); + fFamily=AF_UNSPEC; } - return Connect(); + if(fSockfd) + Close(); + if(ProbeAF(ip,&fSendAddr,&connect)<0) return SOCKET_ERROR; + fState |= JNS_CONNCD; + return getsockname(fSockfd, (struct sockaddr *)&fRecvAddr, &l); } void JackNetUnixSocket::Close() --- common/JackNetDriver.h +++ common/JackNetDriver.h @@ -79,7 +79,7 @@ public: JackNetDriver(const char* name, const char* alias, JackLockedEngine* engine, JackSynchro* table, - const char* ip, int port, int mtu, int midi_input_ports, int midi_output_ports, + const char* ip, int port, const char* mcif, int mtu, int midi_input_ports, int midi_output_ports, char* net_name, uint transport_sync, int network_latency, int celt_encoding, int opus_encoding, bool auto_save); virtual ~JackNetDriver();
|
Consolidated PR is here: #427 |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Hi,
Here's a new branch with rebased and squashed patches.
I've run some tests and ubuntu debs are available at https://launchpad.net/~rufferson/+archive/snd for further testing.
No windows support, mac should be ok since I've followed precisely POSIX's IEEE Std 1003.1, 2004 Edition.
Commits contain more descriptive logs.
Regards,
Ruslan