April 2012 - tor-commits - lists.torproject.org

[tor/master] Add a check_no_tls_errors() to read_to_buf_tls
by nickm＠torproject.org 24 Apr '12

24 Apr '12

commit e0e4b8475708729a57461bb7336931351b874842 Author: Arturo Filastò <hellais(a)gmail.com> Date: Wed Feb 22 00:29:16 2012 -0800 Add a check_no_tls_errors() to read_to_buf_tls Fixes bug #4528 "read_to_buf_tls(): Inconsistency in code". This check was added back in 0.1.0.3-rc, but somehow we forgot to leave it in when we refactored read_to_buf_tls in 0.1.0.5-rc. (patch by Arturo; commit message and changes file by nickm) --- changes/bug4528 | 6 ++++++ src/or/buffers.c | 3 +++ 2 files changed, 9 insertions(+), 0 deletions(-) diff --git a/changes/bug4528 b/changes/bug4528 new file mode 100644 index 0000000..8703d6a --- /dev/null +++ b/changes/bug4528 @@ -0,0 +1,6 @@ + + o Minor bugfixes: + - Make sure that there are no unhandled pending TLS errors before + reading from a TLS stream. We had checks in 0.1.0.3-rc, but + lost them in 0.1.0.5-rc when we refactored read_to_buf_tls(). + Bugfix on 0.1.0.5-rc; fixes bug 4528. diff --git a/src/or/buffers.c b/src/or/buffers.c index d4fc44b..d60e54e 100644 --- a/src/or/buffers.c +++ b/src/or/buffers.c @@ -744,6 +744,9 @@ read_to_buf_tls(tor_tls_t *tls, size_t at_most, buf_t *buf) { int r = 0; size_t total_read = 0; + + check_no_tls_errors(); + check(); while (at_most > total_read) {

1 0

[tor/master] Don't exit on dirauths for some config transitions
by nickm＠torproject.org 24 Apr '12

24 Apr '12

commit 4ccc8d029266dc673d901c01a7a8f59b4f530542 Author: Sebastian Hahn <sebastian(a)torproject.org> Date: Tue Nov 8 12:44:12 2011 +0100 Don't exit on dirauths for some config transitions --- changes/bug4438 | 6 ++++++ src/or/router.c | 19 +++++++++++++++---- 2 files changed, 21 insertions(+), 4 deletions(-) diff --git a/changes/bug4438 b/changes/bug4438 new file mode 100644 index 0000000..f6d0c49 --- /dev/null +++ b/changes/bug4438 @@ -0,0 +1,6 @@ + o Minor bugfixes: + - Fix a dirauth-only exit on sighup that could happen during some + configuration state transitions. We now don't treat it as a fatal + error when the new descriptor we just generated in init_keys + isn't accepted. Fixes bug 4438; bugfix on FIXME. + diff --git a/src/or/router.c b/src/or/router.c index c9f141b..6199e9e 100644 --- a/src/or/router.c +++ b/src/or/router.c @@ -646,15 +646,26 @@ init_keys(void) return -1; } if (mydesc) { + was_router_added_t added; ri = router_parse_entry_from_string(mydesc, NULL, 1, 0, NULL); if (!ri) { log_err(LD_GENERAL,"Generated a routerinfo we couldn't parse."); return -1; } - if (!WRA_WAS_ADDED(dirserv_add_descriptor(ri, &m, "self"))) { - log_err(LD_GENERAL,"Unable to add own descriptor to directory: %s", - m?m:"<unknown error>"); - return -1; + added = dirserv_add_descriptor(ri, &m, "self"); + if (!WRA_WAS_ADDED(added)) { + if (WRA_WAS_REJECTED(added)) { + log_err(LD_GENERAL, "Unable to add own descriptor to directory: %s", + m?m:"<unknown error>"); + return -1; + } else { + /* If the descriptor wasn't rejected, that's ok. This can happen + * when some config options are toggled that affect workers, but + * we don't really need new keys yet so the descriptor doesn't + * change and the old one is still fresh. */ + log_info(LD_GENERAL, "Couldn't add own descriptor to directory: %s", + m?m:"unknown error>"); + } } } }

1 0

[tor/master] Say that bug4438 was a fix on 0.2.1.9-alpha.
by nickm＠torproject.org 24 Apr '12

24 Apr '12

commit 54520e49cd2a10778e7c58d797f22a768da6557a Author: Nick Mathewson <nickm(a)torproject.org> Date: Wed Mar 28 10:41:12 2012 -0400 Say that bug4438 was a fix on 0.2.1.9-alpha. Specifically, it was a fix on 33e2053ebca5d, where we introduced the WRA_* and ROUTER_* codes for dirserv_add_descriptor. Previously, we had checked for a _negative_ return from dirserv_add_descriptor, which meant "rejected". An insufficiently new descriptor would give a 0-valued return. But when we switched from numbers to enums, we got this check wrong and had init_keys() give an error whenever the descriptor wasn't accepted. --- changes/bug4438 | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-) diff --git a/changes/bug4438 b/changes/bug4438 index f6d0c49..94690e1 100644 --- a/changes/bug4438 +++ b/changes/bug4438 @@ -2,5 +2,5 @@ - Fix a dirauth-only exit on sighup that could happen during some configuration state transitions. We now don't treat it as a fatal error when the new descriptor we just generated in init_keys - isn't accepted. Fixes bug 4438; bugfix on FIXME. + isn't accepted. Fixes bug 4438; bugfix on 0.2.1.9-alpha.

1 0

[tor/master] Tweak the bug4438 fix a little: different check, better log
by nickm＠torproject.org 24 Apr '12

24 Apr '12

commit db81cdbb0f9a69bc7bb0a4f77ff0fe44875f7cda Author: Nick Mathewson <nickm(a)torproject.org> Date: Wed Mar 28 10:47:17 2012 -0400 Tweak the bug4438 fix a little: different check, better log Instead of checking for 'rejected' and calling everything else okay, let's check for 'outdated' and call everythign else a problem. This way we don't risk missing future errors so much. When logging a message that _looks_ like an error message at info, we should mention that it isn't really a problem. --- src/or/router.c | 9 +++++---- 1 files changed, 5 insertions(+), 4 deletions(-) diff --git a/src/or/router.c b/src/or/router.c index 6199e9e..451d95d 100644 --- a/src/or/router.c +++ b/src/or/router.c @@ -654,17 +654,18 @@ init_keys(void) } added = dirserv_add_descriptor(ri, &m, "self"); if (!WRA_WAS_ADDED(added)) { - if (WRA_WAS_REJECTED(added)) { + if (!WRA_WAS_OUTDATED(added)) { log_err(LD_GENERAL, "Unable to add own descriptor to directory: %s", m?m:"<unknown error>"); return -1; } else { - /* If the descriptor wasn't rejected, that's ok. This can happen + /* If the descriptor was outdated, that's ok. This can happen * when some config options are toggled that affect workers, but * we don't really need new keys yet so the descriptor doesn't * change and the old one is still fresh. */ - log_info(LD_GENERAL, "Couldn't add own descriptor to directory: %s", - m?m:"unknown error>"); + log_info(LD_GENERAL, "Couldn't add own descriptor to directory " + "after key init: %s. This is usually not a problem.", + m?m:"<unknown error>"); } } }

1 0

[tor/master] Merge branch 'bug4438-v2'
by nickm＠torproject.org 24 Apr '12

24 Apr '12

commit 461771ebbce3a86965eb2d696f9d42e9bcb777c9 Merge: 6f5a740 db81cdb Author: Nick Mathewson <nickm(a)torproject.org> Date: Tue Apr 24 11:18:41 2012 -0400 Merge branch 'bug4438-v2' changes/bug4438 | 6 ++++++ src/or/router.c | 20 ++++++++++++++++---- 2 files changed, 22 insertions(+), 4 deletions(-)

1 0

[tor/master] If SOCK_CLOEXEC and friends fail, fall back to regular socket() calls
by nickm＠torproject.org 24 Apr '12

24 Apr '12

commit 077b9f19a4a9486e5cd411a6fce0dec68fc2250c Author: Nick Mathewson <nickm(a)torproject.org> Date: Tue Feb 14 10:34:06 2012 -0500 If SOCK_CLOEXEC and friends fail, fall back to regular socket() calls Since 0.2.3.1-alpha, we've supported the Linux extensions to socket(), open(), socketpair(), and accept() that enable us to create an fd and make it close-on-exec with a single syscall. This not only saves us a syscall (big deal), but makes us less vulnerable to race conditions where we open a socket and then exec before we can make it close-on-exec. But these extensions are not supported on all Linuxes: They were added between 2.6.23 or so and 2.6.28 or so. If you were to build your Tor against a recent Linux's kernel headers, and then run it with a older kernel, you would find yourselve unable to open sockets. Ouch! The solution here is that, when one of these syscalls fails with EINVAL, we should try again in the portable way. This adds an extra syscall in the case where we built with new headers and are running with old ones, but it will at least allow Tor to work. Fixes bug 5112; bugfix on 0.2.3.1-alpha. --- changes/bug5112 | 5 ++ src/common/compat.c | 133 ++++++++++++++++++++++++++++++++++----------------- 2 files changed, 94 insertions(+), 44 deletions(-) diff --git a/changes/bug5112 b/changes/bug5112 new file mode 100644 index 0000000..9607a92 --- /dev/null +++ b/changes/bug5112 @@ -0,0 +1,5 @@ + o Minor bugfixes: + - When Tor is built with kernel headers from a recent (last few + years) Linux kernel, do not fail to run on older (pre-2.6.28 + Linux kernels). Fixes bug 5112; bugfix on 0.2.3.1-alpha. + diff --git a/src/common/compat.c b/src/common/compat.c index 64c0668..b648539 100644 --- a/src/common/compat.c +++ b/src/common/compat.c @@ -116,16 +116,24 @@ int tor_open_cloexec(const char *path, int flags, unsigned mode) { + int fd; #ifdef O_CLOEXEC - return open(path, flags|O_CLOEXEC, mode); -#else - int fd = open(path, flags, mode); + fd = open(path, flags|O_CLOEXEC, mode); + if (fd >= 0) + return fd; + /* If we got an error, see if it is EINVAL. EINVAL might indicate that, + * event though we were built on a system with O_CLOEXEC support, we + * are running on one without. */ + if (errno != EINVAL) + return -1; +#endif + + fd = open(path, flags, mode); #ifdef FD_CLOEXEC if (fd >= 0) fcntl(fd, F_SETFD, FD_CLOEXEC); #endif return fd; -#endif } /** DOCDOC */ @@ -961,19 +969,31 @@ tor_open_socket(int domain, int type, int protocol) { tor_socket_t s; #ifdef SOCK_CLOEXEC -#define LINUX_CLOEXEC_OPEN_SOCKET - type |= SOCK_CLOEXEC; -#endif + s = socket(domain, type|SOCK_CLOEXEC, protocol); + if (SOCKET_OK(s)) + goto socket_ok; + /* If we got an error, see if it is EINVAL. EINVAL might indicate that, + * event though we were built on a system with SOCK_CLOEXEC support, we + * are running on one without. */ + if (errno != EINVAL) + return s; +#endif /* SOCK_CLOEXEC */ + s = socket(domain, type, protocol); - if (SOCKET_OK(s)) { -#if !defined(LINUX_CLOEXEC_OPEN_SOCKET) && defined(FD_CLOEXEC) - fcntl(s, F_SETFD, FD_CLOEXEC); + if (! SOCKET_OK(s)) + return s; + +#if defined(FD_CLOEXEC) + fcntl(s, F_SETFD, FD_CLOEXEC); #endif - socket_accounting_lock(); - ++n_sockets_open; - mark_socket_open(s); - socket_accounting_unlock(); - } + + goto socket_ok; /* So that socket_ok will not be unused. */ + + socket_ok: + socket_accounting_lock(); + ++n_sockets_open; + mark_socket_open(s); + socket_accounting_unlock(); return s; } @@ -983,20 +1003,31 @@ tor_accept_socket(tor_socket_t sockfd, struct sockaddr *addr, socklen_t *len) { tor_socket_t s; #if defined(HAVE_ACCEPT4) && defined(SOCK_CLOEXEC) -#define LINUX_CLOEXEC_ACCEPT s = accept4(sockfd, addr, len, SOCK_CLOEXEC); -#else - s = accept(sockfd, addr, len); + if (SOCKET_OK(s)) + goto socket_ok; + /* If we got an error, see if it is EINVAL. EINVAL might indicate that, + * event though we were built on a system with accept4 support, we + * are running on one without. */ + if (errno != EINVAL) + return s; #endif - if (SOCKET_OK(s)) { -#if !defined(LINUX_CLOEXEC_ACCEPT) && defined(FD_CLOEXEC) - fcntl(s, F_SETFD, FD_CLOEXEC); + + s = accept(sockfd, addr, len); + if (!SOCKET_OK(s)) + return s; + +#if defined(FD_CLOEXEC) + fcntl(s, F_SETFD, FD_CLOEXEC); #endif - socket_accounting_lock(); - ++n_sockets_open; - mark_socket_open(s); - socket_accounting_unlock(); - } + + goto socket_ok; /* So that socket_ok will not be unused. */ + + socket_ok: + socket_accounting_lock(); + ++n_sockets_open; + mark_socket_open(s); + socket_accounting_unlock(); return s; } @@ -1047,29 +1078,43 @@ tor_socketpair(int family, int type, int protocol, tor_socket_t fd[2]) //don't use win32 socketpairs (they are always bad) #if defined(HAVE_SOCKETPAIR) && !defined(_WIN32) int r; + #ifdef SOCK_CLOEXEC - type |= SOCK_CLOEXEC; + r = socketpair(family, type|SOCK_CLOEXEC, protocol, fd); + if (r == 0) + goto sockets_ok; + /* If we got an error, see if it is EINVAL. EINVAL might indicate that, + * event though we were built on a system with SOCK_CLOEXEC support, we + * are running on one without. */ + if (errno != EINVAL) + return -errno; #endif + r = socketpair(family, type, protocol, fd); - if (r == 0) { -#if !defined(SOCK_CLOEXEC) && defined(FD_CLOEXEC) - if (SOCKET_OK(fd[0])) - fcntl(fd[0], F_SETFD, FD_CLOEXEC); - if (SOCKET_OK(fd[1])) - fcntl(fd[1], F_SETFD, FD_CLOEXEC); -#endif - socket_accounting_lock(); - if (SOCKET_OK(fd[0])) { - ++n_sockets_open; - mark_socket_open(fd[0]); - } - if (SOCKET_OK(fd[1])) { - ++n_sockets_open; - mark_socket_open(fd[1]); - } - socket_accounting_unlock(); + if (r < 0) + return -errno; + +#if defined(FD_CLOEXEC) + if (SOCKET_OK(fd[0])) + fcntl(fd[0], F_SETFD, FD_CLOEXEC); + if (SOCKET_OK(fd[1])) + fcntl(fd[1], F_SETFD, FD_CLOEXEC); +#endif + goto sockets_ok; /* So that sockets_ok will not be unused. */ + + sockets_ok: + socket_accounting_lock(); + if (SOCKET_OK(fd[0])) { + ++n_sockets_open; + mark_socket_open(fd[0]); } - return r < 0 ? -errno : r; + if (SOCKET_OK(fd[1])) { + ++n_sockets_open; + mark_socket_open(fd[1]); + } + socket_accounting_unlock(); + + return 0; #else /* This socketpair does not work when localhost is down. So * it's really not the same thing at all. But it's close enough

1 0

[tor/master] Merge remote-tracking branch 'public/bug5112'
by nickm＠torproject.org 24 Apr '12

24 Apr '12

commit 6f5a74002ac6406ebe602a3d8d36d251bd217df4 Merge: 3e4ccbc 077b9f1 Author: Nick Mathewson <nickm(a)torproject.org> Date: Tue Apr 24 11:14:22 2012 -0400 Merge remote-tracking branch 'public/bug5112' changes/bug5112 | 5 ++ src/common/compat.c | 133 ++++++++++++++++++++++++++++++++++----------------- 2 files changed, 94 insertions(+), 44 deletions(-)

1 0

[tor/master] We allow IPv6 connections, don't use sockaddr_in with getsockname
by nickm＠torproject.org 24 Apr '12

24 Apr '12

commit a74905cea45acf6d7278326360b179eda3d013ec Author: Nick Mathewson <nickm(a)torproject.org> Date: Fri Mar 30 16:43:52 2012 -0400 We allow IPv6 connections, don't use sockaddr_in with getsockname This fixes client_check_address_changed to work better with IPv6 Tor clients, and prevents them from spewing errors. Fix for ticket 5537. --- changes/bug5537 | 5 ++++ src/or/connection.c | 65 +++++++++++++++++++++++++++++++++------------------ 2 files changed, 47 insertions(+), 23 deletions(-) diff --git a/changes/bug5537 b/changes/bug5537 new file mode 100644 index 0000000..bcad2a2 --- /dev/null +++ b/changes/bug5537 @@ -0,0 +1,5 @@ + o Minor features: + - Make the code that clients use to detect an address change be + IPv6-aware, so that it won't fill clients' logs with error + messages when trying to get the IPv4 address of an IPv6 + connection. Implements ticket 5537. diff --git a/src/or/connection.c b/src/or/connection.c index c9093fe..d6b9adc 100644 --- a/src/or/connection.c +++ b/src/or/connection.c @@ -74,10 +74,14 @@ static void connection_send_socks5_connect(connection_t *conn); static const char *proxy_type_to_string(int proxy_type); static int get_proxy_type(void); -/** The last IPv4 address that our network interface seemed to have been - * binding to, in host order. We use this to detect when our IP changes. */ -static uint32_t last_interface_ip = 0; -/** A list of uint32_ts for addresses we've used in outgoing connections. +/** The last addresses that our network interface seemed to have been + * binding to. We use this as one way to detect when our IP changes. + * + * XXX024 We should really use the entire list of interfaces here. + **/ +static tor_addr_t *last_interface_ipv4 = NULL; +static tor_addr_t *last_interface_ipv6 = NULL; +/** A list of tor_addr_t for addresses we've used in outgoing connections. * Used to detect IP address changes. */ static smartlist_t *outgoing_addrs = NULL; @@ -3691,47 +3695,62 @@ alloc_http_authenticator(const char *authenticator) static void client_check_address_changed(tor_socket_t sock) { - uint32_t iface_ip, ip_out; /* host order */ - struct sockaddr_in out_addr; - socklen_t out_addr_len = (socklen_t) sizeof(out_addr); - uint32_t *ip; /* host order */ + struct sockaddr_storage out_sockaddr; + socklen_t out_addr_len = (socklen_t) sizeof(out_sockaddr); + tor_addr_t out_addr, iface_addr; + tor_addr_t **last_interface_ip_ptr; + sa_family_t family; - if (!last_interface_ip) - get_interface_address(LOG_INFO, &last_interface_ip); if (!outgoing_addrs) outgoing_addrs = smartlist_new(); - if (getsockname(sock, (struct sockaddr*)&out_addr, &out_addr_len)<0) { + if (getsockname(sock, (struct sockaddr*)&out_sockaddr, &out_addr_len)<0) { int e = tor_socket_errno(sock); log_warn(LD_NET, "getsockname() to check for address change failed: %s", tor_socket_strerror(e)); return; } + tor_addr_from_sockaddr(&out_addr, (struct sockaddr*)&out_sockaddr, NULL); + family = tor_addr_family(&out_addr); + + if (family == AF_INET) + last_interface_ip_ptr = &last_interface_ipv4; + else if (family == AF_INET6) + last_interface_ip_ptr = &last_interface_ipv6; + else + return; + + if (! *last_interface_ip_ptr) { + tor_addr_t *a = tor_malloc_zero(sizeof(tor_addr_t)); + if (get_interface_address6(LOG_INFO, family, a)==0) { + *last_interface_ip_ptr = a; + } else { + tor_free(a); + } + } /* If we've used this address previously, we're okay. */ - ip_out = ntohl(out_addr.sin_addr.s_addr); - SMARTLIST_FOREACH(outgoing_addrs, uint32_t*, ip_ptr, - if (*ip_ptr == ip_out) return; + SMARTLIST_FOREACH(outgoing_addrs, const tor_addr_t *, a_ptr, + if (tor_addr_eq(a_ptr, &out_addr)) + return; ); /* Uh-oh. We haven't connected from this address before. Has the interface * address changed? */ - if (get_interface_address(LOG_INFO, &iface_ip)<0) + if (get_interface_address6(LOG_INFO, family, &iface_addr)<0) return; - ip = tor_malloc(sizeof(uint32_t)); - *ip = ip_out; - if (iface_ip == last_interface_ip) { + if (tor_addr_eq(&iface_addr, *last_interface_ip_ptr)) { /* Nope, it hasn't changed. Add this address to the list. */ - smartlist_add(outgoing_addrs, ip); + smartlist_add(outgoing_addrs, tor_memdup(&out_addr, sizeof(tor_addr_t))); } else { /* The interface changed. We're a client, so we need to regenerate our * keys. First, reset the state. */ log(LOG_NOTICE, LD_NET, "Our IP address has changed. Rotating keys..."); - last_interface_ip = iface_ip; - SMARTLIST_FOREACH(outgoing_addrs, void*, ip_ptr, tor_free(ip_ptr)); + tor_addr_copy(*last_interface_ip_ptr, &iface_addr); + SMARTLIST_FOREACH(outgoing_addrs, tor_addr_t*, a_ptr, tor_free(a_ptr)); smartlist_clear(outgoing_addrs); - smartlist_add(outgoing_addrs, ip); + smartlist_add(outgoing_addrs, tor_memdup(&out_addr, sizeof(tor_addr_t))); /* Okay, now change our keys. */ ip_address_changed(1); } @@ -4231,7 +4250,7 @@ connection_free_all(void) SMARTLIST_FOREACH(conns, connection_t *, conn, _connection_free(conn)); if (outgoing_addrs) { - SMARTLIST_FOREACH(outgoing_addrs, void*, addr, tor_free(addr)); + SMARTLIST_FOREACH(outgoing_addrs, tor_addr_t *, addr, tor_free(addr)); smartlist_free(outgoing_addrs); outgoing_addrs = NULL; }

1 0

[tor/master] Merge remote-tracking branch 'public/bug5537'
by nickm＠torproject.org 24 Apr '12

24 Apr '12

commit 3e4ccbc4ba748b76c2e8d3c0d99b48191c01d1e4 Merge: da820bb a74905c Author: Nick Mathewson <nickm(a)torproject.org> Date: Tue Apr 24 11:05:50 2012 -0400 Merge remote-tracking branch 'public/bug5537' changes/bug5537 | 5 ++++ src/or/connection.c | 65 +++++++++++++++++++++++++++++++++------------------ 2 files changed, 47 insertions(+), 23 deletions(-)

1 0

[torspec/master] Expanding or-address as per trac feedback
by nickm＠torproject.org 24 Apr '12

24 Apr '12

commit 6285decabe0cfa088e471e070531f5bf1823de7d Author: Damian Johnson <atagar(a)torproject.org> Date: Wed Apr 11 08:51:29 2012 -0700 Expanding or-address as per trac feedback Applying feedback from... https://trac.torproject.org/projects/tor/ticket/5594#comment:2 --- dir-spec.txt | 12 ++++++++++-- 1 files changed, 10 insertions(+), 2 deletions(-) diff --git a/dir-spec.txt b/dir-spec.txt index b1e80aa..315114f 100644 --- a/dir-spec.txt +++ b/dir-spec.txt @@ -606,11 +606,19 @@ PORTSPEC = PORT PORT = a number between 1 and 65535 inclusive. - An alternative address that can be used instead of ORListenAddress. + An alternative for the address and ORPort of the "router" line, but with + two added capabilities: - A descriptor should not include an or-address line that does nothing but + * or-address can be either an IPv4 or IPv6 address + * or-address allows for multiple ORPorts and addresses + + A descriptor SHOULD NOT include an or-address line that does nothing but duplicate the address:port pair from its "router" line. + The ordering of or-address lines and their PORT entries matter because + Tor MAY accept a limited number of addresses or ports. As of Tor 0.2.3.x + only the first address and the first port are used. + 2.2. Extra-info documents Extra-info documents consist of the following items:

1 0