diff options
author | Nick Mathewson <nickm@torproject.org> | 2008-12-18 16:11:24 +0000 |
---|---|---|
committer | Nick Mathewson <nickm@torproject.org> | 2008-12-18 16:11:24 +0000 |
commit | 122170c1d3333185bacf361c523c8e19f0a5e8ba (patch) | |
tree | 89abcc91294222de3353c64d0e00f98618e9f55e /src/or | |
parent | 6c6b0283cbc24f2d3d3bb7139e1cbeb5a35cfedd (diff) | |
download | tor-122170c1d3333185bacf361c523c8e19f0a5e8ba.tar.gz tor-122170c1d3333185bacf361c523c8e19f0a5e8ba.zip |
Downlgrade tweak, and answer lots of XXX021s. No actual code fixes in this patch.
svn:r17686
Diffstat (limited to 'src/or')
-rw-r--r-- | src/or/buffers.c | 2 | ||||
-rw-r--r-- | src/or/circuitbuild.c | 6 | ||||
-rw-r--r-- | src/or/circuitlist.c | 2 | ||||
-rw-r--r-- | src/or/circuituse.c | 5 | ||||
-rw-r--r-- | src/or/command.c | 4 | ||||
-rw-r--r-- | src/or/config.c | 14 | ||||
-rw-r--r-- | src/or/connection.c | 7 | ||||
-rw-r--r-- | src/or/connection_or.c | 6 | ||||
-rw-r--r-- | src/or/directory.c | 9 | ||||
-rw-r--r-- | src/or/dirserv.c | 7 | ||||
-rw-r--r-- | src/or/dns.c | 4 | ||||
-rw-r--r-- | src/or/dnsserv.c | 2 | ||||
-rw-r--r-- | src/or/networkstatus.c | 4 | ||||
-rw-r--r-- | src/or/or.h | 10 | ||||
-rw-r--r-- | src/or/policies.c | 12 | ||||
-rw-r--r-- | src/or/reasons.c | 2 | ||||
-rw-r--r-- | src/or/rendclient.c | 8 | ||||
-rw-r--r-- | src/or/rephist.c | 2 | ||||
-rw-r--r-- | src/or/routerlist.c | 16 |
19 files changed, 63 insertions, 59 deletions
diff --git a/src/or/buffers.c b/src/or/buffers.c index 17d5b3a9a6..141fe58b12 100644 --- a/src/or/buffers.c +++ b/src/or/buffers.c @@ -186,7 +186,7 @@ chunk_new_with_alloc_size(size_t alloc) freelist->lowest_length = freelist->cur_length; ++freelist->n_hit; } else { - /* XXXX021 take advantage of tor_malloc_roundup, once we know how that + /* XXXX take advantage of tor_malloc_roundup, once we know how that * affects freelists. */ if (freelist) ++freelist->n_alloc; diff --git a/src/or/circuitbuild.c b/src/or/circuitbuild.c index 521d347360..4a733bd125 100644 --- a/src/or/circuitbuild.c +++ b/src/or/circuitbuild.c @@ -1688,8 +1688,8 @@ choose_good_entry_server(uint8_t purpose, cpath_build_state_t *state) routerlist_add_family(excluded, r); } if (firewall_is_fascist_or()) { - /*XXXX021 This can slow things down a lot; use a smarter implementation */ - /* exclude all ORs that listen on the wrong port */ + /*XXXX This could slow things down a lot; use a smarter implementation */ + /* exclude all ORs that listen on the wrong port, if anybody notices. */ routerlist_t *rl = router_get_routerlist(); int i; @@ -1898,7 +1898,7 @@ build_state_get_exit_nickname(cpath_build_state_t *state) * * If it's not usable, set *<b>reason</b> to a static string explaining why. */ -/*XXXX021 take a routerstatus, not a routerinfo. */ +/*XXXX take a routerstatus, not a routerinfo. */ static int entry_guard_set_status(entry_guard_t *e, routerinfo_t *ri, time_t now, or_options_t *options, const char **reason) diff --git a/src/or/circuitlist.c b/src/or/circuitlist.c index 060dc73e73..229c84ea5f 100644 --- a/src/or/circuitlist.c +++ b/src/or/circuitlist.c @@ -872,7 +872,7 @@ origin_circuit_t * circuit_find_to_cannibalize(uint8_t purpose, extend_info_t *info, int flags) { - /*XXXX021 arma: The purpose argument is ignored. Can that possibly be + /* XXXX021 arma: The purpose argument is ignored. Can that possibly be * right? */ /* XXXX <arma> i don't know of any actual bugs that this causes. since i * think we only call the function for purposes where we want it to do what diff --git a/src/or/circuituse.c b/src/or/circuituse.c index 6e2c96392e..c6d8cde3c8 100644 --- a/src/or/circuituse.c +++ b/src/or/circuituse.c @@ -1084,8 +1084,7 @@ circuit_get_open_circ_or_launch(edge_connection_t *conn, return -1; } } else { - /* XXXX021 Duplicates checks in connection_ap_handshake_attach_circuit - * XXXX021 Fix this, then backport it? */ + /* XXXX021 Duplicates checks in connection_ap_handshake_attach_circuit */ routerinfo_t *router = router_get_by_nickname(conn->chosen_exit_name, 1); int opt = conn->chosen_exit_optional; if (router && !connection_ap_can_use_exit(conn, router)) { @@ -1363,7 +1362,7 @@ connection_ap_handshake_attach_chosen_circuit(edge_connection_t *conn, * Otherwise, associate conn with a safe live circuit, do the * right next step, and return 1. */ -/* XXXX021 this function should mark for close whenever it returns -1; +/* XXXX this function should mark for close whenever it returns -1; * its callers shouldn't have to worry about that. */ int connection_ap_handshake_attach_circuit(edge_connection_t *conn) diff --git a/src/or/command.c b/src/or/command.c index 30025acbf4..f7eb60f39d 100644 --- a/src/or/command.c +++ b/src/or/command.c @@ -597,7 +597,7 @@ command_process_netinfo_cell(cell_t *cell, or_connection_t *conn) router_get_by_digest(conn->identity_digest)) { char dbuf[64]; int severity; - /*XXXX021 be smarter about when everybody says we are skewed. */ + /*XXXX be smarter about when everybody says we are skewed. */ if (router_digest_is_trusted_dir(conn->identity_digest)) severity = LOG_WARN; else @@ -615,7 +615,7 @@ command_process_netinfo_cell(cell_t *cell, or_connection_t *conn) apparent_skew, conn->_base.address, conn->_base.port); } - /* XXX021 maybe act on my_apparent_addr, if the source is sufficiently + /* XXX maybe act on my_apparent_addr, if the source is sufficiently * trustworthy. */ if (connection_or_set_state_open(conn)<0) diff --git a/src/or/config.c b/src/or/config.c index 784e56f069..5972d54833 100644 --- a/src/or/config.c +++ b/src/or/config.c @@ -1349,9 +1349,9 @@ options_act(or_options_t *old_options) if (options->GeoIPFile && ((!old_options || !opt_streq(old_options->GeoIPFile, options->GeoIPFile)) || !geoip_is_loaded())) { - /** XXXX021 Don't use this "<default>" junk; make our filename options + /* XXXX Don't use this "<default>" junk; make our filename options * understand prefixes somehow. -NM */ - /** XXXX021 Reload GeoIPFile on SIGHUP. -NM */ + /* XXXX021 Reload GeoIPFile on SIGHUP. -NM */ char *actual_fname = tor_strdup(options->GeoIPFile); #ifdef WIN32 if (!strcmp(actual_fname, "<default>")) { @@ -2151,7 +2151,7 @@ options_trial_assign(config_line_t *list, int use_defaults, if (options_validate(get_options(), trial_options, 1, msg) < 0) { config_free(&options_format, trial_options); - return SETOPT_ERR_PARSE; /*XXX021 make this separate. */ + return SETOPT_ERR_PARSE; /*XXX make this a separate return value. */ } if (options_transition_allowed(get_options(), trial_options, msg) < 0) { @@ -3038,7 +3038,7 @@ options_validate(or_options_t *old_options, or_options_t *options, COMPLAIN("StrictEntryNodes set, but no EntryNodes listed."); if (options->EntryNodes && !routerset_is_list(options->EntryNodes)) { - /** XXXX021 fix this; see entry_guards_prepend_from_config(). */ + /* XXXX fix this; see entry_guards_prepend_from_config(). */ REJECT("IPs or countries are not yet supported in EntryNodes."); } @@ -3672,7 +3672,7 @@ static int options_transition_affects_descriptor(or_options_t *old_options, or_options_t *new_options) { - /* XXX021 We can be smarter here. If your DirPort isn't being + /* XXX We can be smarter here. If your DirPort isn't being * published and you just turned it off, no need to republish. If * you changed your bandwidthrate but maxadvertisedbandwidth still * trumps, no need to republish. Etc. */ @@ -4022,7 +4022,7 @@ options_init_from_string(const char *cf, * for a list of dependent config options, re-initialize newoptions * with the new defaults, and assign all options to it second time. */ if (newoptions->TestingTorNetwork) { - /* XXXX021 this is a bit of a kludge. perhaps there's a better way to do + /* XXXX this is a bit of a kludge. perhaps there's a better way to do * this? We could, for example, make the parsing algorithm do two passes * over the configuration. If it finds any "suite" options like * TestingTorNetwork, it could change the defaults before its second pass. @@ -4070,7 +4070,7 @@ options_init_from_string(const char *cf, /* Validate newoptions */ if (options_validate(oldoptions, newoptions, 0, msg) < 0) { - err = SETOPT_ERR_PARSE; /*XXX021 make this separate.*/ + err = SETOPT_ERR_PARSE; /*XXX make this a separate return value.*/ goto err; } diff --git a/src/or/connection.c b/src/or/connection.c index 7090e9135b..ca688052fc 100644 --- a/src/or/connection.c +++ b/src/or/connection.c @@ -563,7 +563,8 @@ connection_about_to_close_connection(connection_t *conn) int reason = tls_error_to_orconn_end_reason(or_conn->tls_error); control_event_or_conn_status(or_conn, OR_CONN_EVENT_FAILED, reason); - /* XXX021 come up with a better string for the first arg */ + /* XXX021 come up with a better string for the first arg -RD */ + /* What did you have in mind? -NM */ if (!authdir_mode_tests_reachability(options)) control_event_bootstrap_problem( orconn_end_reason_to_control_string(reason), reason); @@ -1800,7 +1801,7 @@ connection_bucket_refill_helper(int *bucket, int rate, int burst, if (*bucket > burst || *bucket < starting_bucket) { /* If we overflow the burst, or underflow our starting bucket, * cap the bucket value to burst. */ - /* XXXX021 this might be redundant now, but it doesn't show up + /* XXXX this might be redundant now, but it doesn't show up * in profiles. Remove it after analysis. */ *bucket = burst; } @@ -2033,7 +2034,7 @@ connection_read_to_buf(connection_t *conn, int *max_to_read, int *socket_error) if (at_most == -1) { /* we need to initialize it */ /* how many bytes are we allowed to read? */ - /* XXXX too many calls to time(). Do they hurt? */ + /* XXXX021 too many calls to time(). Do they hurt? */ at_most = connection_bucket_read_limit(conn, time(NULL)); } diff --git a/src/or/connection_or.c b/src/or/connection_or.c index 2cd96688f2..399ba1801c 100644 --- a/src/or/connection_or.c +++ b/src/or/connection_or.c @@ -398,14 +398,14 @@ connection_or_init_conn_from_address(or_connection_t *conn, tor_addr_copy(&conn->_base.addr, addr); tor_addr_copy(&conn->real_addr, addr); if (r) { - /* XXXX021 proposal 118 will make this more complex. */ + /* XXXX proposal 118 will make this more complex. */ if (tor_addr_eq_ipv4h(&conn->_base.addr, r->addr)) conn->is_canonical = 1; if (!started_here) { /* Override the addr/port, so our log messages will make sense. * This is dangerous, since if we ever try looking up a conn by * its actual addr/port, we won't remember. Careful! */ - /* XXXX021 arma: this is stupid, and it's the reason we need real_addr + /* XXXX arma: this is stupid, and it's the reason we need real_addr * to track is_canonical properly. What requires it? */ /* XXXX <arma> i believe the reason we did this, originally, is because * we wanted to log what OR a connection was to, and if we logged the @@ -929,7 +929,7 @@ connection_or_set_state_open(or_connection_t *conn) /* only report it to the geoip module if it's not a known router */ if (!router_get_by_digest(conn->identity_digest)) { if (tor_addr_family(&TO_CONN(conn)->addr) == AF_INET) { - /*XXXX021 IP6 support ipv6 geoip.*/ + /*XXXX IP6 support ipv6 geoip.*/ uint32_t a = tor_addr_to_ipv4h(&TO_CONN(conn)->addr); geoip_note_client_seen(GEOIP_CLIENT_CONNECT, a, now); } diff --git a/src/or/directory.c b/src/or/directory.c index 13c3f71ab8..926d1843f8 100644 --- a/src/or/directory.c +++ b/src/or/directory.c @@ -540,7 +540,7 @@ directory_conn_is_self_reachability_test(dir_connection_t *conn) routerinfo_t *me = router_get_my_routerinfo(); if (me && router_digest_is_me(conn->identity_digest) && - tor_addr_eq_ipv4h(&conn->_base.addr, me->addr) && /*XXXX021 prop 118*/ + tor_addr_eq_ipv4h(&conn->_base.addr, me->addr) && /*XXXX prop 118*/ me->dir_port == conn->_base.port) return 1; } @@ -1804,8 +1804,13 @@ connection_dir_client_reached_eof(dir_connection_t *conn) int rejected = 0; if (rejected_hdr) { if (!strcmp(rejected_hdr, "Yes")) { - /* XXXX021 use this information; be sure to upload next one + /* XXXX use this information; be sure to upload next one * sooner. -NM */ + /* XXXX021 On further thought, the task above implies that we're + * basing our regenerate-descriptor time on when we uploaded the + * last descriptor, not on the published time of the last + * descriptor. If those are different, that's a bad thing to + * do. -NM */ rejected = 1; } tor_free(rejected_hdr); diff --git a/src/or/dirserv.c b/src/or/dirserv.c index 336fa7c377..ff0e9663f6 100644 --- a/src/or/dirserv.c +++ b/src/or/dirserv.c @@ -887,7 +887,8 @@ directory_set_dirty(void) int set_v1_dirty=0; /* Regenerate stubs only every 8 hours. - * XXXX021 It would be nice to generate less often. */ + * XXXX It would be nice to generate less often, but these are just + * stubs: it doesn't matter. */ #define STUB_REGENERATE_INTERVAL (8*60*60) if (!the_directory || !the_runningrouters.dir) set_v1_dirty = 1; @@ -2727,8 +2728,8 @@ dirserv_get_routerdesc_fingerprints(smartlist_t *fps_out, const char *key, * If -1 is returned *<b>msg</b> will be set to an appropriate error * message. * - * XXXX021 rename this function. It's only called from the controller. - * XXXX021 in fact, refactor this function, mergeing as much as possible. + * XXXX rename this function. It's only called from the controller. + * XXXX in fact, refactor this function, mergeing as much as possible. */ int dirserv_get_routerdescs(smartlist_t *descs_out, const char *key, diff --git a/src/or/dns.c b/src/or/dns.c index 490444f752..a83c48981d 100644 --- a/src/or/dns.c +++ b/src/or/dns.c @@ -441,7 +441,7 @@ send_resolved_cell(edge_connection_t *conn, uint8_t answer_type) set_uint32(buf+6, htonl(ttl)); buflen = 10; break; - /*XXXX021 IP6 need ipv6 implementation */ + /*XXXX IP6 need ipv6 implementation */ case RESOLVED_TYPE_ERROR_TRANSIENT: case RESOLVED_TYPE_ERROR: { @@ -499,7 +499,7 @@ send_resolved_hostname_cell(edge_connection_t *conn, const char *hostname) * parse it and place the address in <b>in</b> if present. Return 1 on success; * 0 if the address is not in in-addr.arpa format, and -1 if the address is * malformed. */ -/* XXXX021 move this to util.c. */ +/* XXXX021 move this to address.c; unify with logic in connection_edge.c */ int parse_inaddr_arpa_address(const char *address, struct in_addr *in) { diff --git a/src/or/dnsserv.c b/src/or/dnsserv.c index c8de359843..d9ec3f0fa0 100644 --- a/src/or/dnsserv.c +++ b/src/or/dnsserv.c @@ -258,7 +258,7 @@ dnsserv_resolved(edge_connection_t *conn, name = evdns_get_orig_address(req, answer_type, conn->socks_request->address); - /* XXXX021 Re-do; this is dumb. */ + /* XXXX Re-do; this is dumb. */ if (ttl < 60) ttl = 60; diff --git a/src/or/networkstatus.c b/src/or/networkstatus.c index b325e5620d..869d812a1b 100644 --- a/src/or/networkstatus.c +++ b/src/or/networkstatus.c @@ -908,7 +908,7 @@ router_get_consensus_status_by_nickname(const char *nickname, /* This name is not canonical for any server; go through the list and * see who it matches. */ - /*XXXX021 This is inefficient. */ + /*XXXX This is inefficient; optimize it if it matters. */ matches = smartlist_create(); SMARTLIST_FOREACH(current_consensus->routerstatus_list, routerstatus_t *, lrs, @@ -1746,7 +1746,7 @@ routers_update_status_from_consensus_networkstatus(smartlist_t *routers, } SMARTLIST_FOREACH_JOIN_END(rs, router); /* Now update last_listed_as_valid_until from v2 networkstatuses. */ - /* XXXX021 If this is slow, we need to rethink the code. */ + /* XXXX If this is slow, we need to rethink the code. */ SMARTLIST_FOREACH(networkstatus_v2_list, networkstatus_v2_t *, ns, { time_t live_until = ns->published_on + V2_NETWORKSTATUS_LIFETIME; SMARTLIST_FOREACH_JOIN(ns->entries, routerstatus_t *, rs, diff --git a/src/or/or.h b/src/or/or.h index 0e79b21430..b766083f44 100644 --- a/src/or/or.h +++ b/src/or/or.h @@ -1342,7 +1342,7 @@ typedef struct signed_descriptor_t { #ifdef TRACK_SERVED_TIME /** The last time we served anybody this descriptor. Used for internal * testing to see whether we're holding on to descriptors too long. */ - time_t last_served_at; /*XXXX021 remove if not useful. */ + time_t last_served_at; /*XXXX remove if not useful. */ #endif /* If true, we do not ever try to save this object in the cache. */ unsigned int do_not_cache : 1; @@ -4290,11 +4290,11 @@ smartlist_t *router_get_trusted_dir_servers(void); * * Passed to router_pick_directory_server (et al) * - * [XXXX021 NOTE: This option is only implemented for pick_trusteddirserver, + * [XXXX NOTE: This option is only implemented for pick_trusteddirserver, * not pick_directory_server. If we make it work on pick_directory_server * too, we could conservatively make it only prevent multiple fetches to * the same authority, or we could aggressively make it prevent multiple - * fetches to _any_ directory server.] + * fetches to _any_ single directory server.] */ #define PDS_NO_EXISTING_SERVERDESC_FETCH (1<<3) #define _PDS_PREFER_TUNNELED_DIR_CONNS (1<<16) @@ -4334,9 +4334,9 @@ typedef enum { CRN_NEED_CAPACITY = 1<<1, CRN_NEED_GUARD = 1<<2, CRN_ALLOW_INVALID = 1<<3, - /* XXXX021 not used, apparently. */ + /* XXXX not used, apparently. */ CRN_STRICT_PREFERRED = 1<<4, - /* XXXX021 not used, apparently. */ + /* XXXX not used, apparently. */ CRN_WEIGHT_AS_EXIT = 1<<5 } router_crn_flags_t; diff --git a/src/or/policies.c b/src/or/policies.c index 64d73375c8..610a34ed05 100644 --- a/src/or/policies.c +++ b/src/or/policies.c @@ -69,7 +69,7 @@ policy_expand_private(smartlist_t **policy) int i; smartlist_t *tmp; - if (!*policy) /*XXXX021 disallow NULL policies */ + if (!*policy) /*XXXX disallow NULL policies? */ return; tmp = smartlist_create(); @@ -235,7 +235,7 @@ addr_policy_permits_tor_addr(const tor_addr_t *addr, uint16_t port, } } -/* DOCDOC XXXX021 deprecate? */ +/* DOCDOC XXXX deprecate when possible. */ static int addr_policy_permits_address(uint32_t addr, uint16_t port, smartlist_t *policy) @@ -258,7 +258,7 @@ fascist_firewall_allows_address_or(const tor_addr_t *addr, uint16_t port) int fascist_firewall_allows_or(routerinfo_t *ri) { - /* XXXX021 proposal 118 */ + /* XXXX proposal 118 */ tor_addr_t addr; tor_addr_from_ipv4h(&addr, ri->addr); return fascist_firewall_allows_address_or(&addr, ri->or_port); @@ -556,7 +556,7 @@ addr_policy_get_canonical_entry(addr_policy_t *e) addr_policy_result_t compare_addr_to_addr_policy(uint32_t addr, uint16_t port, smartlist_t *policy) { - /*XXXX021 deprecate this function? */ + /*XXXX deprecate this function when possible. */ tor_addr_t a; tor_addr_from_ipv4h(&a, addr); return compare_tor_addr_to_addr_policy(&a, port, policy); @@ -825,7 +825,7 @@ exit_policy_is_general_exit(smartlist_t *policy) static const int ports[] = { 80, 443, 6667 }; int n_allowed = 0; int i; - if (!policy) /*XXXX021 disallow NULL policies */ + if (!policy) /*XXXX disallow NULL policies? */ return 0; for (i = 0; i < 3; ++i) { @@ -851,7 +851,7 @@ exit_policy_is_general_exit(smartlist_t *policy) int policy_is_reject_star(smartlist_t *policy) { - if (!policy) /*XXXX021 disallow NULL policies */ + if (!policy) /*XXXX disallow NULL policies? */ return 1; SMARTLIST_FOREACH(policy, addr_policy_t *, p, { if (p->policy_type == ADDR_POLICY_ACCEPT) diff --git a/src/or/reasons.c b/src/or/reasons.c index f261a87ad1..dd989950b4 100644 --- a/src/or/reasons.c +++ b/src/or/reasons.c @@ -248,8 +248,6 @@ tls_error_to_orconn_end_reason(int e) /** Given an errno from a failed ORConn connection, return a reason code * appropriate for use in the controller orconn events. */ -/* XXX021 somebody should think about whether the assignments I've made - * are accurate or useful. -RD */ int errno_to_orconn_end_reason(int e) { diff --git a/src/or/rendclient.c b/src/or/rendclient.c index 050397532a..0c091d4d40 100644 --- a/src/or/rendclient.c +++ b/src/or/rendclient.c @@ -601,9 +601,9 @@ rend_client_rendezvous_acked(origin_circuit_t *circ, const char *request, log_info(LD_REND,"Got rendezvous ack. This circuit is now ready for " "rendezvous."); circ->_base.purpose = CIRCUIT_PURPOSE_C_REND_READY; - /*XXXX021 This is a pretty brute approach. It'd be better to + /*XXXX This is a pretty brute approach. It'd be better to * attach only the connections that are waiting on this circuit, rather - * than trying to attach them all. */ + * than trying to attach them all. See bug 743. */ /* If we already have the introduction circuit built, make sure we send * the INTRODUCE cell _now_ */ connection_ap_attach_pending(); @@ -669,9 +669,9 @@ rend_client_receive_rendezvous(origin_circuit_t *circ, const char *request, onion_append_to_cpath(&circ->cpath, hop); circ->build_state->pending_final_cpath = NULL; /* prevent double-free */ - /*XXXX021 This is a pretty brute approach. It'd be better to + /*XXXX This is a pretty brute approach. It'd be better to * attach only the connections that are waiting on this circuit, rather - * than trying to attach them all. */ + * than trying to attach them all. See bug 743.*/ /* */ connection_ap_attach_pending(); return 0; diff --git a/src/or/rephist.c b/src/or/rephist.c index 45dbd6c4fc..4a77bad2b1 100644 --- a/src/or/rephist.c +++ b/src/or/rephist.c @@ -736,7 +736,7 @@ rep_hist_record_mtbf_data(void) PUT("data\n"); - /* XXX021 Nick: now bridge auths record this for all routers too. + /* XXX Nick: now bridge auths record this for all routers too. * Should we make them record it only for bridge routers? -RD * Not for 0.2.0. -NM */ for (orhist_it = digestmap_iter_init(history_map); diff --git a/src/or/routerlist.c b/src/or/routerlist.c index ec325b8361..e26809d675 100644 --- a/src/or/routerlist.c +++ b/src/or/routerlist.c @@ -80,7 +80,7 @@ static smartlist_t *warned_nicknames = NULL; * download is low. */ static time_t last_routerdesc_download_attempted = 0; -/* DOCDOC This is a massive massive kludge XXXX021 */ +/* DOCDOC This is a massive massive kludge XXXX */ static uint64_t sl_last_total_weighted_bw = 0; static uint64_t sl_last_weighted_bw_of_me = 0; @@ -1058,7 +1058,7 @@ router_pick_directory_server_impl(authority_type_t type, int flags) !router_supports_extrainfo(status->identity_digest, 0)) continue; - /* XXXX021 IP6 proposal 118 */ + /* XXXX IP6 proposal 118 */ tor_addr_from_ipv4h(&addr, status->addr); if (prefer_tunnel && @@ -1139,7 +1139,7 @@ router_pick_trusteddirserver_impl(authority_type_t type, int flags, if (requireother && me && router_digest_is_me(d->digest)) continue; - /* XXXX021 IP6 proposal 118 */ + /* XXXX IP6 proposal 118 */ tor_addr_from_ipv4h(&addr, d->addr); if (no_serverdesc_fetching) { @@ -3238,7 +3238,7 @@ routerlist_remove_old_routers(void) networkstatus_v2_list && smartlist_len(networkstatus_v2_list)) { SMARTLIST_FOREACH(networkstatus_v2_list, networkstatus_v2_t *, ns, n_expected_retain += smartlist_len(ns->entries)); - /*XXXX021 too much magic. */ + /* DOCDOC XXX021 too much magic. */ n_expected_retain /= (smartlist_len(networkstatus_v2_list)/2+1); } //log_notice(LD_DIR,"n_expected_retain=%d",n_expected_retain); @@ -4089,7 +4089,7 @@ update_router_descriptor_cache_downloads_v2(time_t now) smartlist_t *dl = download_from[i]; int pds_flags = PDS_RETRY_IF_NO_SERVERS; if (! authdir_mode_any_nonhidserv(options)) - pds_flags |= PDS_NO_EXISTING_SERVERDESC_FETCH; /* XXXX021 ignored*/ + pds_flags |= PDS_NO_EXISTING_SERVERDESC_FETCH; /* XXXX ignored*/ if (!ds) { log_warn(LD_BUG, "Networkstatus with no corresponding authority!"); @@ -4673,7 +4673,7 @@ routerlist_assert_ok(routerlist_t *rl) r->cache_info.signed_descriptor_digest); tor_assert(&(r->cache_info) == sd2); tor_assert(r->cache_info.routerlist_index == r_sl_idx); - /* XXXX021. + /* XXXX * * Hoo boy. We need to fix this one, and the fix is a bit tricky, so * commenting this out is just a band-aid. @@ -4704,7 +4704,7 @@ routerlist_assert_ok(routerlist_t *rl) sd2 = sdmap_get(rl->desc_digest_map, sd->signed_descriptor_digest); tor_assert(sd == sd2); tor_assert(sd->routerlist_index == sd_sl_idx); - /* XXXX021 see above. + /* XXXX see above. if (!tor_digest_is_zero(sd->extra_info_digest)) { signed_descriptor_t *sd3 = sdmap_get(rl->desc_by_eid_map, sd->extra_info_digest); @@ -4730,7 +4730,7 @@ routerlist_assert_ok(routerlist_t *rl) d, DIGEST_LEN)); sd = sdmap_get(rl->desc_by_eid_map, ei->cache_info.signed_descriptor_digest); - // tor_assert(sd); // XXXX021 see above + // tor_assert(sd); // XXXX see above if (sd) { tor_assert(!memcmp(ei->cache_info.signed_descriptor_digest, sd->extra_info_digest, DIGEST_LEN)); |