diff options
Diffstat (limited to 'src/or/rephist.c')
| -rw-r--r-- | src/or/rephist.c | 1133 |
1 files changed, 588 insertions, 545 deletions
diff --git a/src/or/rephist.c b/src/or/rephist.c index dee74ed5c5..d85a8b7c3a 100644 --- a/src/or/rephist.c +++ b/src/or/rephist.c @@ -6,15 +6,21 @@ * \file rephist.c * \brief Basic history and "reputation" functionality to remember * which servers have worked in the past, how much bandwidth we've - * been using, which ports we tend to want, and so on. + * been using, which ports we tend to want, and so on; further, + * exit port statistics and cell statistics. **/ #include "or.h" +#include "circuitlist.h" +#include "circuituse.h" +#include "config.h" +#include "rephist.h" +#include "router.h" +#include "routerlist.h" #include "ht.h" static void bw_arrays_init(void); static void predicted_ports_init(void); -static void hs_usage_init(void); /** Total number of bytes currently allocated in fields used by rephist.c. */ uint64_t rephist_total_alloc=0; @@ -185,7 +191,6 @@ rep_hist_init(void) history_map = digestmap_new(); bw_arrays_init(); predicted_ports_init(); - hs_usage_init(); } /** Helper: note that we are no longer connected to the router with history @@ -795,12 +800,12 @@ rep_hist_record_mtbf_data(time_t now, int missing_means_down) static char * rep_hist_format_router_status(or_history_t *hist, time_t now) { - char buf[1024]; char sor_buf[ISO_TIME_LEN+1]; char sod_buf[ISO_TIME_LEN+1]; double wfu; double mtbf; int up = 0, down = 0; + char *cp = NULL; if (hist->start_of_run) { format_iso_time(sor_buf, hist->start_of_run); @@ -813,7 +818,7 @@ rep_hist_format_router_status(or_history_t *hist, time_t now) wfu = get_weighted_fractional_uptime(hist, now); mtbf = get_stability(hist, now); - tor_snprintf(buf, sizeof(buf), + tor_asprintf(&cp, "%s%s%s" "%s%s%s" "wfu %0.3lf\n" @@ -831,8 +836,7 @@ rep_hist_format_router_status(or_history_t *hist, time_t now) hist->weighted_run_length, hist->total_run_weights ); - - return tor_strdup(buf); + return cp; } /** The last stability analysis document that we created, or NULL if we never @@ -1280,13 +1284,21 @@ bw_array_new(void) static bw_array_t *read_array = NULL; /** Recent history of bandwidth observations for write operations. */ static bw_array_t *write_array = NULL; - -/** Set up read_array and write_array. */ +/** Recent history of bandwidth observations for read operations for the + directory protocol. */ +static bw_array_t *dir_read_array = NULL; +/** Recent history of bandwidth observations for write operations for the + directory protocol. */ +static bw_array_t *dir_write_array = NULL; + +/** Set up [dir-]read_array and [dir-]write_array. */ static void bw_arrays_init(void) { read_array = bw_array_new(); write_array = bw_array_new(); + dir_read_array = bw_array_new(); + dir_write_array = bw_array_new(); } /** We read <b>num_bytes</b> more bytes in second <b>when</b>. @@ -1320,6 +1332,24 @@ rep_hist_note_bytes_read(size_t num_bytes, time_t when) add_obs(read_array, when, num_bytes); } +/** We wrote <b>num_bytes</b> more directory bytes in second <b>when</b>. + * (like rep_hist_note_bytes_written() above) + */ +void +rep_hist_note_dir_bytes_written(size_t num_bytes, time_t when) +{ + add_obs(dir_write_array, when, num_bytes); +} + +/** We read <b>num_bytes</b> more directory bytes in second <b>when</b>. + * (like rep_hist_note_bytes_written() above) + */ +void +rep_hist_note_dir_bytes_read(size_t num_bytes, time_t when) +{ + add_obs(dir_read_array, when, num_bytes); +} + /** Helper: Return the largest value in b->maxima. (This is equal to the * most bandwidth used in any NUM_SECS_ROLLING_MEASURE period for the last * NUM_SECS_BW_SUM_IS_VALID seconds.) @@ -1355,9 +1385,9 @@ rep_hist_bandwidth_assess(void) return (int)(U64_TO_DBL(r)/NUM_SECS_ROLLING_MEASURE); } -/** Print the bandwidth history of b (either read_array or write_array) - * into the buffer pointed to by buf. The format is simply comma - * separated numbers, from oldest to newest. +/** Print the bandwidth history of b (either [dir-]read_array or + * [dir-]write_array) into the buffer pointed to by buf. The format is + * simply comma separated numbers, from oldest to newest. * * It returns the number of bytes written. */ @@ -1410,26 +1440,42 @@ rep_hist_fill_bandwidth_history(char *buf, size_t len, bw_array_t *b) * history in its descriptor. */ char * -rep_hist_get_bandwidth_lines(int for_extrainfo) +rep_hist_get_bandwidth_lines(void) { char *buf, *cp; char t[ISO_TIME_LEN+1]; int r; - bw_array_t *b; + bw_array_t *b = NULL; + const char *desc = NULL; size_t len; - /* opt (read|write)-history yyyy-mm-dd HH:MM:SS (n s) n,n,n,n,n... */ - len = (60+20*NUM_TOTALS)*2; + /* opt [dirreq-](read|write)-history yyyy-mm-dd HH:MM:SS (n s) n,n,n... */ + len = (67+21*NUM_TOTALS)*4; buf = tor_malloc_zero(len); cp = buf; - for (r=0;r<2;++r) { - b = r?read_array:write_array; + for (r=0;r<4;++r) { + switch (r) { + case 0: + b = write_array; + desc = "write-history"; + break; + case 1: + b = read_array; + desc = "read-history"; + break; + case 2: + b = dir_write_array; + desc = "dirreq-write-history"; + break; + case 3: + b = dir_read_array; + desc = "dirreq-read-history"; + break; + } tor_assert(b); format_iso_time(t, b->next_period-NUM_SECS_BW_SUM_INTERVAL); - tor_snprintf(cp, len-(cp-buf), "%s%s %s (%d s) ", - for_extrainfo ? "" : "opt ", - r ? "read-history" : "write-history", t, - NUM_SECS_BW_SUM_INTERVAL); + tor_snprintf(cp, len-(cp-buf), "%s %s (%d s) ", + desc, t, NUM_SECS_BW_SUM_INTERVAL); cp += strlen(cp); cp += rep_hist_fill_bandwidth_history(cp, len-(cp-buf), b); strlcat(cp, "\n", len-(cp-buf)); @@ -1444,20 +1490,41 @@ rep_hist_update_state(or_state_t *state) { int len, r; char *buf, *cp; - smartlist_t **s_values; - time_t *s_begins; - int *s_interval; - bw_array_t *b; + smartlist_t **s_values = NULL; + time_t *s_begins = NULL; + int *s_interval = NULL; + bw_array_t *b = NULL; len = 20*NUM_TOTALS+1; buf = tor_malloc_zero(len); - for (r=0;r<2;++r) { - b = r?read_array:write_array; - s_begins = r?&state->BWHistoryReadEnds :&state->BWHistoryWriteEnds; - s_interval= r?&state->BWHistoryReadInterval:&state->BWHistoryWriteInterval; - s_values = r?&state->BWHistoryReadValues :&state->BWHistoryWriteValues; - + for (r=0;r<4;++r) { + switch (r) { + case 0: + b = write_array; + s_begins = &state->BWHistoryWriteEnds; + s_interval = &state->BWHistoryWriteInterval; + s_values = &state->BWHistoryWriteValues; + break; + case 1: + b = read_array; + s_begins = &state->BWHistoryReadEnds; + s_interval = &state->BWHistoryReadInterval; + s_values = &state->BWHistoryReadValues; + break; + case 2: + b = dir_write_array; + s_begins = &state->BWHistoryDirWriteEnds; + s_interval = &state->BWHistoryDirWriteInterval; + s_values = &state->BWHistoryDirWriteValues; + break; + case 3: + b = dir_read_array; + s_begins = &state->BWHistoryDirReadEnds; + s_interval = &state->BWHistoryDirReadInterval; + s_values = &state->BWHistoryDirReadValues; + break; + } if (*s_values) { SMARTLIST_FOREACH(*s_values, char *, val, tor_free(val)); smartlist_free(*s_values); @@ -1497,23 +1564,45 @@ rep_hist_update_state(or_state_t *state) int rep_hist_load_state(or_state_t *state, char **err) { - time_t s_begins, start; + time_t s_begins = 0, start; time_t now = time(NULL); uint64_t v; int r,i,ok; int all_ok = 1; - int s_interval; - smartlist_t *s_values; - bw_array_t *b; + int s_interval = 0; + smartlist_t *s_values = NULL; + bw_array_t *b = NULL; /* Assert they already have been malloced */ tor_assert(read_array && write_array); - for (r=0;r<2;++r) { - b = r?read_array:write_array; - s_begins = r?state->BWHistoryReadEnds:state->BWHistoryWriteEnds; - s_interval = r?state->BWHistoryReadInterval:state->BWHistoryWriteInterval; - s_values = r?state->BWHistoryReadValues:state->BWHistoryWriteValues; + for (r=0;r<4;++r) { + switch (r) { + case 0: + b = write_array; + s_begins = state->BWHistoryWriteEnds; + s_interval = state->BWHistoryWriteInterval; + s_values = state->BWHistoryWriteValues; + break; + case 1: + b = read_array; + s_begins = state->BWHistoryReadEnds; + s_interval = state->BWHistoryReadInterval; + s_values = state->BWHistoryReadValues; + break; + case 2: + b = dir_write_array; + s_begins = state->BWHistoryDirWriteEnds; + s_interval = state->BWHistoryDirWriteInterval; + s_values = state->BWHistoryDirWriteValues; + break; + case 3: + b = dir_read_array; + s_begins = state->BWHistoryDirReadEnds; + s_interval = state->BWHistoryDirReadInterval; + s_values = state->BWHistoryDirReadValues; + break; + } if (s_values && s_begins >= now - NUM_SECS_BW_SUM_INTERVAL*NUM_TOTALS) { start = s_begins - s_interval*(smartlist_len(s_values)); if (start > now) @@ -1715,8 +1804,8 @@ rep_hist_get_predicted_internal(time_t now, int *need_uptime, return 0; /* too long ago */ if (predicted_internal_uptime_time + PREDICTED_CIRCS_RELEVANCE_TIME >= now) *need_uptime = 1; - if (predicted_internal_capacity_time + PREDICTED_CIRCS_RELEVANCE_TIME >= now) - *need_capacity = 1; + // Always predict that we need capacity. + *need_capacity = 1; return 1; } @@ -1845,564 +1934,518 @@ dump_pk_ops(int severity) pk_op_counts.n_rend_server_ops); } -/** Free all storage held by the OR/link history caches, by the - * bandwidth history arrays, or by the port history. */ +/*** Exit port statistics ***/ + +/* Some constants */ +/** To what multiple should byte numbers be rounded up? */ +#define EXIT_STATS_ROUND_UP_BYTES 1024 +/** To what multiple should stream counts be rounded up? */ +#define EXIT_STATS_ROUND_UP_STREAMS 4 +/** Number of TCP ports */ +#define EXIT_STATS_NUM_PORTS 65536 +/** Top n ports that will be included in exit stats. */ +#define EXIT_STATS_TOP_N_PORTS 10 + +/* The following data structures are arrays and no fancy smartlists or maps, + * so that all write operations can be done in constant time. This comes at + * the price of some memory (1.25 MB) and linear complexity when writing + * stats for measuring relays. */ +/** Number of bytes read in current period by exit port */ +static uint64_t *exit_bytes_read = NULL; +/** Number of bytes written in current period by exit port */ +static uint64_t *exit_bytes_written = NULL; +/** Number of streams opened in current period by exit port */ +static uint32_t *exit_streams = NULL; + +/** Start time of exit stats or 0 if we're not collecting exit stats. */ +static time_t start_of_exit_stats_interval; + +/** Initialize exit port stats. */ void -rep_hist_free_all(void) +rep_hist_exit_stats_init(time_t now) { - digestmap_free(history_map, free_or_history); - tor_free(read_array); - tor_free(write_array); - tor_free(last_stability_doc); - built_last_stability_doc_at = 0; - predicted_ports_free(); + start_of_exit_stats_interval = now; + exit_bytes_read = tor_malloc_zero(EXIT_STATS_NUM_PORTS * + sizeof(uint64_t)); + exit_bytes_written = tor_malloc_zero(EXIT_STATS_NUM_PORTS * + sizeof(uint64_t)); + exit_streams = tor_malloc_zero(EXIT_STATS_NUM_PORTS * + sizeof(uint32_t)); } -/****************** hidden service usage statistics ******************/ - -/** How large are the intervals for which we track and report hidden service - * use? */ -#define NUM_SECS_HS_USAGE_SUM_INTERVAL (15*60) -/** How far in the past do we remember and publish hidden service use? */ -#define NUM_SECS_HS_USAGE_SUM_IS_VALID (24*60*60) -/** How many hidden service usage intervals do we remember? (derived) */ -#define NUM_TOTALS_HS_USAGE (NUM_SECS_HS_USAGE_SUM_IS_VALID/ \ - NUM_SECS_HS_USAGE_SUM_INTERVAL) - -/** List element containing a service id and the count. */ -typedef struct hs_usage_list_elem_t { - /** Service id of this elem. */ - char service_id[REND_SERVICE_ID_LEN_BASE32+1]; - /** Number of occurrences for the given service id. */ - uint32_t count; - /* Pointer to next list elem */ - struct hs_usage_list_elem_t *next; -} hs_usage_list_elem_t; - -/** Ordered list that stores service ids and the number of observations. It is - * ordered by the number of occurrences in descending order. Its purpose is to - * calculate the frequency distribution when the period is over. */ -typedef struct hs_usage_list_t { - /* Pointer to the first element in the list. */ - hs_usage_list_elem_t *start; - /* Number of total occurrences for all list elements. */ - uint32_t total_count; - /* Number of service ids, i.e. number of list elements. */ - uint32_t total_service_ids; -} hs_usage_list_t; - -/** Tracks service-related observations in the current period and their - * history. */ -typedef struct hs_usage_service_related_observation_t { - /** Ordered list that stores service ids and the number of observations in - * the current period. It is ordered by the number of occurrences in - * descending order. Its purpose is to calculate the frequency distribution - * when the period is over. */ - hs_usage_list_t *list; - /** Circular arrays that store the history of observations. totals stores all - * observations, twenty (ten, five) the number of observations related to a - * service id being accounted for the top 20 (10, 5) percent of all - * observations. */ - uint32_t totals[NUM_TOTALS_HS_USAGE]; - uint32_t five[NUM_TOTALS_HS_USAGE]; - uint32_t ten[NUM_TOTALS_HS_USAGE]; - uint32_t twenty[NUM_TOTALS_HS_USAGE]; -} hs_usage_service_related_observation_t; - -/** Tracks the history of general period-related observations, i.e. those that - * cannot be related to a specific service id. */ -typedef struct hs_usage_general_period_related_observations_t { - /** Circular array that stores the history of observations. */ - uint32_t totals[NUM_TOTALS_HS_USAGE]; -} hs_usage_general_period_related_observations_t; - -/** Keeps information about the current observation period and its relation to - * the histories of observations. */ -typedef struct hs_usage_current_observation_period_t { - /** Where do we write the next history entry? */ - int next_idx; - /** How many values in history have been set ever? (upper bound!) */ - int num_set; - /** When did this period begin? */ - time_t start_of_current_period; - /** When does the next period begin? */ - time_t start_of_next_period; -} hs_usage_current_observation_period_t; - -/** Usage statistics for the current observation period. */ -static hs_usage_current_observation_period_t *current_period = NULL; - -/** Total number of descriptor publish requests in the current observation - * period. */ -static hs_usage_service_related_observation_t *publish_total = NULL; - -/** Number of descriptor publish requests for services that have not been - * seen before in the current observation period. */ -static hs_usage_service_related_observation_t *publish_novel = NULL; - -/** Total number of descriptor fetch requests in the current observation - * period. */ -static hs_usage_service_related_observation_t *fetch_total = NULL; - -/** Number of successful descriptor fetch requests in the current - * observation period. */ -static hs_usage_service_related_observation_t *fetch_successful = NULL; - -/** Number of descriptors stored in the current observation period. */ -static hs_usage_general_period_related_observations_t *descs = NULL; - -/** Creates an empty ordered list element. */ -static hs_usage_list_elem_t * -hs_usage_list_elem_new(void) +/** Reset counters for exit port statistics. */ +void +rep_hist_reset_exit_stats(time_t now) { - hs_usage_list_elem_t *e; - e = tor_malloc_zero(sizeof(hs_usage_list_elem_t)); - rephist_total_alloc += sizeof(hs_usage_list_elem_t); - e->count = 1; - e->next = NULL; - return e; + start_of_exit_stats_interval = now; + memset(exit_bytes_read, 0, EXIT_STATS_NUM_PORTS * sizeof(uint64_t)); + memset(exit_bytes_written, 0, EXIT_STATS_NUM_PORTS * sizeof(uint64_t)); + memset(exit_streams, 0, EXIT_STATS_NUM_PORTS * sizeof(uint32_t)); } -/** Creates an empty ordered list. */ -static hs_usage_list_t * -hs_usage_list_new(void) +/** Stop collecting exit port stats in a way that we can re-start doing + * so in rep_hist_exit_stats_init(). */ +void +rep_hist_exit_stats_term(void) { - hs_usage_list_t *l; - l = tor_malloc_zero(sizeof(hs_usage_list_t)); - rephist_total_alloc += sizeof(hs_usage_list_t); - l->start = NULL; - l->total_count = 0; - l->total_service_ids = 0; - return l; + start_of_exit_stats_interval = 0; + tor_free(exit_bytes_read); + tor_free(exit_bytes_written); + tor_free(exit_streams); } -/** Creates an empty structure for storing service-related observations. */ -static hs_usage_service_related_observation_t * -hs_usage_service_related_observation_new(void) +/** Helper for qsort: compare two ints. */ +static int +_compare_int(const void *x, const void *y) { - hs_usage_service_related_observation_t *h; - h = tor_malloc_zero(sizeof(hs_usage_service_related_observation_t)); - rephist_total_alloc += sizeof(hs_usage_service_related_observation_t); - h->list = hs_usage_list_new(); - return h; + return (*(int*)x - *(int*)y); } -/** Creates an empty structure for storing general period-related - * observations. */ -static hs_usage_general_period_related_observations_t * -hs_usage_general_period_related_observations_new(void) -{ - hs_usage_general_period_related_observations_t *p; - p = tor_malloc_zero(sizeof(hs_usage_general_period_related_observations_t)); - rephist_total_alloc+= sizeof(hs_usage_general_period_related_observations_t); - return p; -} +/** Return a newly allocated string containing the exit port statistics + * until <b>now</b>, or NULL if we're not collecting exit stats. */ +char * +rep_hist_format_exit_stats(time_t now) +{ + int i, j, top_elements = 0, cur_min_idx = 0, cur_port; + uint64_t top_bytes[EXIT_STATS_TOP_N_PORTS]; + int top_ports[EXIT_STATS_TOP_N_PORTS]; + uint64_t cur_bytes = 0, other_read = 0, other_written = 0, + total_read = 0, total_written = 0; + uint32_t total_streams = 0, other_streams = 0; + char *buf; + smartlist_t *written_strings, *read_strings, *streams_strings; + char *written_string, *read_string, *streams_string; + char t[ISO_TIME_LEN+1]; + char *result; -/** Creates an empty structure for storing period-specific information. */ -static hs_usage_current_observation_period_t * -hs_usage_current_observation_period_new(void) -{ - hs_usage_current_observation_period_t *c; - time_t now; - c = tor_malloc_zero(sizeof(hs_usage_current_observation_period_t)); - rephist_total_alloc += sizeof(hs_usage_current_observation_period_t); - now = time(NULL); - c->start_of_current_period = now; - c->start_of_next_period = now + NUM_SECS_HS_USAGE_SUM_INTERVAL; - return c; -} + if (!start_of_exit_stats_interval) + return NULL; /* Not initialized. */ -/** Initializes the structures for collecting hidden service usage data. */ -static void -hs_usage_init(void) -{ - current_period = hs_usage_current_observation_period_new(); - publish_total = hs_usage_service_related_observation_new(); - publish_novel = hs_usage_service_related_observation_new(); - fetch_total = hs_usage_service_related_observation_new(); - fetch_successful = hs_usage_service_related_observation_new(); - descs = hs_usage_general_period_related_observations_new(); -} + /* Go through all ports to find the n ports that saw most written and + * read bytes. + * + * Invariant: at the end of the loop for iteration i, + * total_read is the sum of all exit_bytes_read[0..i] + * total_written is the sum of all exit_bytes_written[0..i] + * total_stream is the sum of all exit_streams[0..i] + * + * top_elements = MAX(EXIT_STATS_TOP_N_PORTS, + * #{j | 0 <= j <= i && volume(i) > 0}) + * + * For all 0 <= j < top_elements, + * top_bytes[j] > 0 + * 0 <= top_ports[j] <= 65535 + * top_bytes[j] = volume(top_ports[j]) + * + * There is no j in 0..i and k in 0..top_elements such that: + * volume(j) > top_bytes[k] AND j is not in top_ports[0..top_elements] + * + * There is no j!=cur_min_idx in 0..top_elements such that: + * top_bytes[j] < top_bytes[cur_min_idx] + * + * where volume(x) == exit_bytes_read[x]+exit_bytes_written[x] + * + * Worst case: O(EXIT_STATS_NUM_PORTS * EXIT_STATS_TOP_N_PORTS) + */ + for (i = 1; i < EXIT_STATS_NUM_PORTS; i++) { + total_read += exit_bytes_read[i]; + total_written += exit_bytes_written[i]; + total_streams += exit_streams[i]; + cur_bytes = exit_bytes_read[i] + exit_bytes_written[i]; + if (cur_bytes == 0) { + continue; + } + if (top_elements < EXIT_STATS_TOP_N_PORTS) { + top_bytes[top_elements] = cur_bytes; + top_ports[top_elements++] = i; + } else if (cur_bytes > top_bytes[cur_min_idx]) { + top_bytes[cur_min_idx] = cur_bytes; + top_ports[cur_min_idx] = i; + } else { + continue; + } + cur_min_idx = 0; + for (j = 1; j < top_elements; j++) { + if (top_bytes[j] < top_bytes[cur_min_idx]) { + cur_min_idx = j; + } + } + } -/** Clears the given ordered list by resetting its attributes and releasing - * the memory allocated by its elements. */ -static void -hs_usage_list_clear(hs_usage_list_t *lst) -{ - /* walk through elements and free memory */ - hs_usage_list_elem_t *current = lst->start; - hs_usage_list_elem_t *tmp; - while (current != NULL) { - tmp = current->next; - rephist_total_alloc -= sizeof(hs_usage_list_elem_t); - tor_free(current); - current = tmp; + /* Add observations of top ports to smartlists. */ + written_strings = smartlist_create(); + read_strings = smartlist_create(); + streams_strings = smartlist_create(); + other_read = total_read; + other_written = total_written; + other_streams = total_streams; + /* Sort the ports; this puts them out of sync with top_bytes, but we + * won't be using top_bytes again anyway */ + qsort(top_ports, top_elements, sizeof(int), _compare_int); + for (j = 0; j < top_elements; j++) { + cur_port = top_ports[j]; + if (exit_bytes_written[cur_port] > 0) { + uint64_t num = round_uint64_to_next_multiple_of( + exit_bytes_written[cur_port], + EXIT_STATS_ROUND_UP_BYTES); + num /= 1024; + buf = NULL; + tor_asprintf(&buf, "%d="U64_FORMAT, cur_port, U64_PRINTF_ARG(num)); + smartlist_add(written_strings, buf); + other_written -= exit_bytes_written[cur_port]; + } + if (exit_bytes_read[cur_port] > 0) { + uint64_t num = round_uint64_to_next_multiple_of( + exit_bytes_read[cur_port], + EXIT_STATS_ROUND_UP_BYTES); + num /= 1024; + buf = NULL; + tor_asprintf(&buf, "%d="U64_FORMAT, cur_port, U64_PRINTF_ARG(num)); + smartlist_add(read_strings, buf); + other_read -= exit_bytes_read[cur_port]; + } + if (exit_streams[cur_port] > 0) { + uint32_t num = round_uint32_to_next_multiple_of( + exit_streams[cur_port], + EXIT_STATS_ROUND_UP_STREAMS); + buf = NULL; + tor_asprintf(&buf, "%d=%u", cur_port, num); + smartlist_add(streams_strings, buf); + other_streams -= exit_streams[cur_port]; + } } - /* reset attributes */ - lst->start = NULL; - lst->total_count = 0; - lst->total_service_ids = 0; - return; -} -/** Frees the memory used by the given list. */ -static void -hs_usage_list_free(hs_usage_list_t *lst) -{ - if (!lst) - return; - hs_usage_list_clear(lst); - rephist_total_alloc -= sizeof(hs_usage_list_t); - tor_free(lst); + /* Add observations of other ports in a single element. */ + other_written = round_uint64_to_next_multiple_of(other_written, + EXIT_STATS_ROUND_UP_BYTES); + other_written /= 1024; + buf = NULL; + tor_asprintf(&buf, "other="U64_FORMAT, U64_PRINTF_ARG(other_written)); + smartlist_add(written_strings, buf); + other_read = round_uint64_to_next_multiple_of(other_read, + EXIT_STATS_ROUND_UP_BYTES); + other_read /= 1024; + buf = NULL; + tor_asprintf(&buf, "other="U64_FORMAT, U64_PRINTF_ARG(other_read)); + smartlist_add(read_strings, buf); + other_streams = round_uint32_to_next_multiple_of(other_streams, + EXIT_STATS_ROUND_UP_STREAMS); + buf = NULL; + tor_asprintf(&buf, "other=%u", other_streams); + smartlist_add(streams_strings, buf); + + /* Join all observations in single strings. */ + written_string = smartlist_join_strings(written_strings, ",", 0, NULL); + read_string = smartlist_join_strings(read_strings, ",", 0, NULL); + streams_string = smartlist_join_strings(streams_strings, ",", 0, NULL); + SMARTLIST_FOREACH(written_strings, char *, cp, tor_free(cp)); + SMARTLIST_FOREACH(read_strings, char *, cp, tor_free(cp)); + SMARTLIST_FOREACH(streams_strings, char *, cp, tor_free(cp)); + smartlist_free(written_strings); + smartlist_free(read_strings); + smartlist_free(streams_strings); + + /* Put everything together. */ + format_iso_time(t, now); + tor_asprintf(&result, "exit-stats-end %s (%d s)\n" + "exit-kibibytes-written %s\n" + "exit-kibibytes-read %s\n" + "exit-streams-opened %s\n", + t, (unsigned) (now - start_of_exit_stats_interval), + written_string, + read_string, + streams_string); + tor_free(written_string); + tor_free(read_string); + tor_free(streams_string); + return result; } -/** Frees the memory used by the given service-related observations. */ -static void -hs_usage_service_related_observation_free( - hs_usage_service_related_observation_t *s) +/** If 24 hours have passed since the beginning of the current exit port + * stats period, write exit stats to $DATADIR/stats/exit-stats (possibly + * overwriting an existing file) and reset counters. Return when we would + * next want to write exit stats or 0 if we never want to write. */ +time_t +rep_hist_exit_stats_write(time_t now) { - if (!s) - return; - hs_usage_list_free(s->list); - rephist_total_alloc -= sizeof(hs_usage_service_related_observation_t); - tor_free(s); -} + char *statsdir = NULL, *filename = NULL, *str = NULL; -/** Frees the memory used by the given period-specific observations. */ -static void -hs_usage_general_period_related_observations_free( - hs_usage_general_period_related_observations_t *s) -{ - rephist_total_alloc-=sizeof(hs_usage_general_period_related_observations_t); - tor_free(s); -} + if (!start_of_exit_stats_interval) + return 0; /* Not initialized. */ + if (start_of_exit_stats_interval + WRITE_STATS_INTERVAL > now) + goto done; /* Not ready to write. */ -/** Frees the memory used by period-specific information. */ -static void -hs_usage_current_observation_period_free( - hs_usage_current_observation_period_t *s) -{ - rephist_total_alloc -= sizeof(hs_usage_current_observation_period_t); - tor_free(s); -} + log_info(LD_HIST, "Writing exit port statistics to disk."); -/** Frees all memory that was used for collecting hidden service usage data. */ -void -hs_usage_free_all(void) -{ - hs_usage_general_period_related_observations_free(descs); - descs = NULL; - hs_usage_service_related_observation_free(fetch_successful); - hs_usage_service_related_observation_free(fetch_total); - hs_usage_service_related_observation_free(publish_novel); - hs_usage_service_related_observation_free(publish_total); - fetch_successful = fetch_total = publish_novel = publish_total = NULL; - hs_usage_current_observation_period_free(current_period); - current_period = NULL; -} + /* Generate history string. */ + str = rep_hist_format_exit_stats(now); -/** Inserts a new occurrence for the given service id to the given ordered - * list. */ -static void -hs_usage_insert_value(hs_usage_list_t *lst, const char *service_id) -{ - /* search if there is already an elem with same service_id in list */ - hs_usage_list_elem_t *current = lst->start; - hs_usage_list_elem_t *previous = NULL; - while (current != NULL && strcasecmp(current->service_id,service_id)) { - previous = current; - current = current->next; - } - /* found an element with same service_id? */ - if (current == NULL) { - /* not found! append to end (which could also be the end of a zero-length - * list), don't need to sort (1 is smallest value). */ - /* create elem */ - hs_usage_list_elem_t *e = hs_usage_list_elem_new(); - /* update list attributes (one new elem, one new occurrence) */ - lst->total_count++; - lst->total_service_ids++; - /* copy service id to elem */ - strlcpy(e->service_id,service_id,sizeof(e->service_id)); - /* let either l->start or previously last elem point to new elem */ - if (lst->start == NULL) { - /* this is the first elem */ - lst->start = e; - } else { - /* there were elems in the list before */ - previous->next = e; - } - } else { - /* found! add occurrence to elem and consider resorting */ - /* update list attributes (no new elem, but one new occurrence) */ - lst->total_count++; - /* add occurrence to elem */ - current->count++; - /* is it another than the first list elem? and has previous elem fewer - * count than current? then we need to resort */ - if (previous != NULL && previous->count < current->count) { - /* yes! we need to resort */ - /* remove current elem first */ - previous->next = current->next; - /* can we prepend elem to all other elements? */ - if (lst->start->count <= current->count) { - /* yes! prepend elem */ - current->next = lst->start; - lst->start = current; - } else { - /* no! walk through list a second time and insert at correct place */ - hs_usage_list_elem_t *insert_current = lst->start->next; - hs_usage_list_elem_t *insert_previous = lst->start; - while (insert_current != NULL && - insert_current->count > current->count) { - insert_previous = insert_current; - insert_current = insert_current->next; - } - /* insert here */ - current->next = insert_current; - insert_previous->next = current; - } - } - } -} + /* Reset counters. */ + rep_hist_reset_exit_stats(now); -/** Writes the current service-related observations to the history array and - * clears the observations of the current period. */ -static void -hs_usage_write_service_related_observations_to_history( - hs_usage_current_observation_period_t *p, - hs_usage_service_related_observation_t *h) -{ - /* walk through the first 20 % of list elements and calculate frequency - * distributions */ - /* maximum indices for the three frequencies */ - int five_percent_idx = h->list->total_service_ids/20; - int ten_percent_idx = h->list->total_service_ids/10; - int twenty_percent_idx = h->list->total_service_ids/5; - /* temp values */ - uint32_t five_percent = 0; - uint32_t ten_percent = 0; - uint32_t twenty_percent = 0; - /* walk through list */ - hs_usage_list_elem_t *current = h->list->start; - int i=0; - while (current != NULL && i <= twenty_percent_idx) { - twenty_percent += current->count; - if (i <= ten_percent_idx) - ten_percent += current->count; - if (i <= five_percent_idx) - five_percent += current->count; - current = current->next; - i++; + /* Try to write to disk. */ + statsdir = get_datadir_fname("stats"); + if (check_private_dir(statsdir, CPD_CREATE) < 0) { + log_warn(LD_HIST, "Unable to create stats/ directory!"); + goto done; } - /* copy frequencies */ - h->twenty[p->next_idx] = twenty_percent; - h->ten[p->next_idx] = ten_percent; - h->five[p->next_idx] = five_percent; - /* copy total number of observations */ - h->totals[p->next_idx] = h->list->total_count; - /* free memory of old list */ - hs_usage_list_clear(h->list); -} + filename = get_datadir_fname2("stats", "exit-stats"); + if (write_str_to_file(filename, str, 0) < 0) + log_warn(LD_HIST, "Unable to write exit port statistics to disk!"); -/** Advances to next observation period. */ -static void -hs_usage_advance_current_observation_period(void) -{ - /* aggregate observations to history, including frequency distribution - * arrays */ - hs_usage_write_service_related_observations_to_history( - current_period, publish_total); - hs_usage_write_service_related_observations_to_history( - current_period, publish_novel); - hs_usage_write_service_related_observations_to_history( - current_period, fetch_total); - hs_usage_write_service_related_observations_to_history( - current_period, fetch_successful); - /* write current number of descriptors to descs history */ - descs->totals[current_period->next_idx] = rend_cache_size(); - /* advance to next period */ - current_period->next_idx++; - if (current_period->next_idx == NUM_TOTALS_HS_USAGE) - current_period->next_idx = 0; - if (current_period->num_set < NUM_TOTALS_HS_USAGE) - ++current_period->num_set; - current_period->start_of_current_period=current_period->start_of_next_period; - current_period->start_of_next_period += NUM_SECS_HS_USAGE_SUM_INTERVAL; -} - -/** Checks if the current period is up to date, and if not, advances it. */ -static void -hs_usage_check_if_current_period_is_up_to_date(time_t now) -{ - while (now > current_period->start_of_next_period) { - hs_usage_advance_current_observation_period(); - } + done: + tor_free(str); + tor_free(statsdir); + tor_free(filename); + return start_of_exit_stats_interval + WRITE_STATS_INTERVAL; } -/** Adds a service-related observation, maybe after advancing to next - * observation period. */ -static void -hs_usage_add_service_related_observation( - hs_usage_service_related_observation_t *h, - time_t now, - const char *service_id) +/** Note that we wrote <b>num_written</b> bytes and read <b>num_read</b> + * bytes to/from an exit connection to <b>port</b>. */ +void +rep_hist_note_exit_bytes(uint16_t port, size_t num_written, + size_t num_read) { - if (now < current_period->start_of_current_period) { - /* don't record old data */ - return; - } - /* check if we are up-to-date */ - hs_usage_check_if_current_period_is_up_to_date(now); - /* add observation */ - hs_usage_insert_value(h->list, service_id); + if (!start_of_exit_stats_interval) + return; /* Not initialized. */ + exit_bytes_written[port] += num_written; + exit_bytes_read[port] += num_read; + log_debug(LD_HIST, "Written %lu bytes and read %lu bytes to/from an " + "exit connection to port %d.", + (unsigned long)num_written, (unsigned long)num_read, port); } -/** Adds the observation of storing a rendezvous service descriptor to our - * cache in our role as HS authoritative directory. */ +/** Note that we opened an exit stream to <b>port</b>. */ void -hs_usage_note_publish_total(const char *service_id, time_t now) +rep_hist_note_exit_stream_opened(uint16_t port) { - hs_usage_add_service_related_observation(publish_total, now, service_id); + if (!start_of_exit_stats_interval) + return; /* Not initialized. */ + exit_streams[port]++; + log_debug(LD_HIST, "Opened exit stream to port %d", port); } -/** Adds the observation of storing a novel rendezvous service descriptor to - * our cache in our role as HS authoritative directory. */ +/*** cell statistics ***/ + +/** Start of the current buffer stats interval or 0 if we're not + * collecting buffer statistics. */ +static time_t start_of_buffer_stats_interval; + +/** Initialize buffer stats. */ void -hs_usage_note_publish_novel(const char *service_id, time_t now) +rep_hist_buffer_stats_init(time_t now) { - hs_usage_add_service_related_observation(publish_novel, now, service_id); + start_of_buffer_stats_interval = now; } -/** Adds the observation of being requested for a rendezvous service descriptor - * in our role as HS authoritative directory. */ +typedef struct circ_buffer_stats_t { + uint32_t processed_cells; + double mean_num_cells_in_queue; + double mean_time_cells_in_queue; + uint32_t local_circ_id; +} circ_buffer_stats_t; + +/** Holds stats. */ +smartlist_t *circuits_for_buffer_stats = NULL; + +/** Remember cell statistics for circuit <b>circ</b> at time + * <b>end_of_interval</b> and reset cell counters in case the circuit + * remains open in the next measurement interval. */ void -hs_usage_note_fetch_total(const char *service_id, time_t now) +rep_hist_buffer_stats_add_circ(circuit_t *circ, time_t end_of_interval) +{ + circ_buffer_stats_t *stat; + time_t start_of_interval; + int interval_length; + or_circuit_t *orcirc; + if (CIRCUIT_IS_ORIGIN(circ)) + return; + orcirc = TO_OR_CIRCUIT(circ); + if (!orcirc->processed_cells) + return; + if (!circuits_for_buffer_stats) + circuits_for_buffer_stats = smartlist_create(); + start_of_interval = circ->timestamp_created > + start_of_buffer_stats_interval ? + circ->timestamp_created : + start_of_buffer_stats_interval; + interval_length = (int) (end_of_interval - start_of_interval); + stat = tor_malloc_zero(sizeof(circ_buffer_stats_t)); + stat->processed_cells = orcirc->processed_cells; + /* 1000.0 for s -> ms; 2.0 because of app-ward and exit-ward queues */ + stat->mean_num_cells_in_queue = interval_length == 0 ? 0.0 : + (double) orcirc->total_cell_waiting_time / + (double) interval_length / 1000.0 / 2.0; + stat->mean_time_cells_in_queue = + (double) orcirc->total_cell_waiting_time / + (double) orcirc->processed_cells; + smartlist_add(circuits_for_buffer_stats, stat); + orcirc->total_cell_waiting_time = 0; + orcirc->processed_cells = 0; +} + +/** Sorting helper: return -1, 1, or 0 based on comparison of two + * circ_buffer_stats_t */ +static int +_buffer_stats_compare_entries(const void **_a, const void **_b) { - hs_usage_add_service_related_observation(fetch_total, now, service_id); + const circ_buffer_stats_t *a = *_a, *b = *_b; + if (a->processed_cells < b->processed_cells) + return 1; + else if (a->processed_cells > b->processed_cells) + return -1; + else + return 0; } -/** Adds the observation of being requested for a rendezvous service descriptor - * in our role as HS authoritative directory and being able to answer that - * request successfully. */ +/** Stop collecting cell stats in a way that we can re-start doing so in + * rep_hist_buffer_stats_init(). */ void -hs_usage_note_fetch_successful(const char *service_id, time_t now) +rep_hist_buffer_stats_term(void) { - hs_usage_add_service_related_observation(fetch_successful, now, service_id); + start_of_buffer_stats_interval = 0; + if (!circuits_for_buffer_stats) + circuits_for_buffer_stats = smartlist_create(); + SMARTLIST_FOREACH(circuits_for_buffer_stats, circ_buffer_stats_t *, + stat, tor_free(stat)); + smartlist_clear(circuits_for_buffer_stats); } -/** Writes the given circular array to a string. */ -static size_t -hs_usage_format_history(char *buf, size_t len, uint32_t *data) +/** Write buffer statistics to $DATADIR/stats/buffer-stats and return when + * we would next want to write exit stats. */ +time_t +rep_hist_buffer_stats_write(time_t now) { - char *cp = buf; /* pointer where we are in the buffer */ - int i, n; - if (current_period->num_set <= current_period->next_idx) { - i = 0; /* not been through circular array */ - } else { - i = current_period->next_idx; + char *statsdir = NULL, *filename = NULL; + char written[ISO_TIME_LEN+1]; + open_file_t *open_file = NULL; + FILE *out; +#define SHARES 10 + int processed_cells[SHARES], circs_in_share[SHARES], + number_of_circuits, i; + double queued_cells[SHARES], time_in_queue[SHARES]; + smartlist_t *str_build = smartlist_create(); + char *str = NULL, *buf=NULL; + circuit_t *circ; + + if (!start_of_buffer_stats_interval) + return 0; /* Not initialized. */ + if (start_of_buffer_stats_interval + WRITE_STATS_INTERVAL > now) + goto done; /* Not ready to write */ + + /* add current circuits to stats */ + for (circ = _circuit_get_global_list(); circ; circ = circ->next) + rep_hist_buffer_stats_add_circ(circ, now); + /* calculate deciles */ + memset(processed_cells, 0, SHARES * sizeof(int)); + memset(circs_in_share, 0, SHARES * sizeof(int)); + memset(queued_cells, 0, SHARES * sizeof(double)); + memset(time_in_queue, 0, SHARES * sizeof(double)); + if (!circuits_for_buffer_stats) + circuits_for_buffer_stats = smartlist_create(); + smartlist_sort(circuits_for_buffer_stats, + _buffer_stats_compare_entries); + number_of_circuits = smartlist_len(circuits_for_buffer_stats); + if (number_of_circuits < 1) { + log_info(LD_HIST, "Attempt to write cell statistics to disk failed. " + "We haven't seen a single circuit to report about."); + goto done; } - for (n = 0; n < current_period->num_set; ++n,++i) { - if (i >= NUM_TOTALS_HS_USAGE) - i -= NUM_TOTALS_HS_USAGE; - tor_assert(i < NUM_TOTALS_HS_USAGE); - if (n == (current_period->num_set-1)) - tor_snprintf(cp, len-(cp-buf), "%d", data[i]); - else - tor_snprintf(cp, len-(cp-buf), "%d,", data[i]); - cp += strlen(cp); + i = 0; + SMARTLIST_FOREACH_BEGIN(circuits_for_buffer_stats, + circ_buffer_stats_t *, stat) + { + int share = i++ * SHARES / number_of_circuits; + processed_cells[share] += stat->processed_cells; + queued_cells[share] += stat->mean_num_cells_in_queue; + time_in_queue[share] += stat->mean_time_cells_in_queue; + circs_in_share[share]++; } - return cp-buf; -} - -/** Writes the complete usage history as hidden service authoritative directory - * to a string. */ -static char * -hs_usage_format_statistics(void) -{ - char *buf, *cp, *s = NULL; - char t[ISO_TIME_LEN+1]; - int r; - uint32_t *data = NULL; - size_t len; - len = (70+20*NUM_TOTALS_HS_USAGE)*11; - buf = tor_malloc_zero(len); - cp = buf; - for (r = 0; r < 11; ++r) { - switch (r) { - case 0: - s = (char*) "publish-total-history"; - data = publish_total->totals; - break; - case 1: - s = (char*) "publish-novel-history"; - data = publish_novel->totals; - break; - case 2: - s = (char*) "publish-top-5-percent-history"; - data = publish_total->five; - break; - case 3: - s = (char*) "publish-top-10-percent-history"; - data = publish_total->ten; - break; - case 4: - s = (char*) "publish-top-20-percent-history"; - data = publish_total->twenty; - break; - case 5: - s = (char*) "fetch-total-history"; - data = fetch_total->totals; - break; - case 6: - s = (char*) "fetch-successful-history"; - data = fetch_successful->totals; - break; - case 7: - s = (char*) "fetch-top-5-percent-history"; - data = fetch_total->five; - break; - case 8: - s = (char*) "fetch-top-10-percent-history"; - data = fetch_total->ten; - break; - case 9: - s = (char*) "fetch-top-20-percent-history"; - data = fetch_total->twenty; - break; - case 10: - s = (char*) "desc-total-history"; - data = descs->totals; - break; - } - format_iso_time(t, current_period->start_of_current_period); - tor_snprintf(cp, len-(cp-buf), "%s %s (%d s) ", s, t, - NUM_SECS_HS_USAGE_SUM_INTERVAL); - cp += strlen(cp); - cp += hs_usage_format_history(cp, len-(cp-buf), data); - strlcat(cp, "\n", len-(cp-buf)); - ++cp; + SMARTLIST_FOREACH_END(stat); + /* clear buffer stats history */ + SMARTLIST_FOREACH(circuits_for_buffer_stats, circ_buffer_stats_t *, + stat, tor_free(stat)); + smartlist_clear(circuits_for_buffer_stats); + /* write to file */ + statsdir = get_datadir_fname("stats"); + if (check_private_dir(statsdir, CPD_CREATE) < 0) + goto done; + filename = get_datadir_fname2("stats", "buffer-stats"); + out = start_writing_to_stdio_file(filename, OPEN_FLAGS_APPEND, + 0600, &open_file); + if (!out) + goto done; + format_iso_time(written, now); + if (fprintf(out, "cell-stats-end %s (%d s)\n", written, + (unsigned) (now - start_of_buffer_stats_interval)) < 0) + goto done; + for (i = 0; i < SHARES; i++) { + tor_asprintf(&buf,"%d", !circs_in_share[i] ? 0 : + processed_cells[i] / circs_in_share[i]); + smartlist_add(str_build, buf); } - return buf; + str = smartlist_join_strings(str_build, ",", 0, NULL); + if (fprintf(out, "cell-processed-cells %s\n", str) < 0) + goto done; + tor_free(str); + SMARTLIST_FOREACH(str_build, char *, c, tor_free(c)); + smartlist_clear(str_build); + for (i = 0; i < SHARES; i++) { + tor_asprintf(&buf, "%.2f", circs_in_share[i] == 0 ? 0.0 : + queued_cells[i] / (double) circs_in_share[i]); + smartlist_add(str_build, buf); + } + str = smartlist_join_strings(str_build, ",", 0, NULL); + if (fprintf(out, "cell-queued-cells %s\n", str) < 0) + goto done; + tor_free(str); + SMARTLIST_FOREACH(str_build, char *, c, tor_free(c)); + smartlist_clear(str_build); + for (i = 0; i < SHARES; i++) { + tor_asprintf(&buf, "%.0f", circs_in_share[i] == 0 ? 0.0 : + time_in_queue[i] / (double) circs_in_share[i]); + smartlist_add(str_build, buf); + } + str = smartlist_join_strings(str_build, ",", 0, NULL); + if (fprintf(out, "cell-time-in-queue %s\n", str) < 0) + goto done; + tor_free(str); + SMARTLIST_FOREACH(str_build, char *, c, tor_free(c)); + smartlist_free(str_build); + str_build = NULL; + if (fprintf(out, "cell-circuits-per-decile %d\n", + (number_of_circuits + SHARES - 1) / SHARES) < 0) + goto done; + finish_writing_to_file(open_file); + open_file = NULL; + start_of_buffer_stats_interval = now; + done: + if (open_file) + abort_writing_to_file(open_file); + tor_free(filename); + tor_free(statsdir); + if (str_build) { + SMARTLIST_FOREACH(str_build, char *, c, tor_free(c)); + smartlist_free(str_build); + } + tor_free(str); +#undef SHARES + return start_of_buffer_stats_interval + WRITE_STATS_INTERVAL; } -/** Write current statistics about hidden service usage to file. */ +/** Free all storage held by the OR/link history caches, by the + * bandwidth history arrays, by the port history, or by statistics . */ void -hs_usage_write_statistics_to_file(time_t now) +rep_hist_free_all(void) { - char *buf; - size_t len; - char *fname; - or_options_t *options = get_options(); - /* check if we are up-to-date */ - hs_usage_check_if_current_period_is_up_to_date(now); - buf = hs_usage_format_statistics(); - len = strlen(options->DataDirectory) + 16; - fname = tor_malloc(len); - tor_snprintf(fname, len, "%s"PATH_SEPARATOR"hsusage", - options->DataDirectory); - write_str_to_file(fname,buf,0); - tor_free(buf); - tor_free(fname); + digestmap_free(history_map, free_or_history); + tor_free(read_array); + tor_free(write_array); + tor_free(last_stability_doc); + tor_free(exit_bytes_read); + tor_free(exit_bytes_written); + tor_free(exit_streams); + built_last_stability_doc_at = 0; + predicted_ports_free(); } |