OpenCL GPU miner
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

1864 lines
42 KiB

/*
* Copyright 2011-2013 Con Kolivas
* Copyright 2010 Jeff Garzik
*
* This program is free software; you can redistribute it and/or modify it
* under the terms of the GNU General Public License as published by the Free
* Software Foundation; either version 3 of the License, or (at your option)
* any later version. See COPYING for more details.
*/
#include "config.h"
#include <stdio.h>
#include <stdlib.h>
#include <ctype.h>
#include <stdarg.h>
#include <string.h>
#include <jansson.h>
#include <curl/curl.h>
#include <time.h>
#include <errno.h>
#include <unistd.h>
#include <sys/types.h>
#ifndef WIN32
#include <fcntl.h>
# ifdef __linux
# include <sys/prctl.h>
# endif
# include <sys/socket.h>
# include <netinet/in.h>
# include <netinet/tcp.h>
# include <netdb.h>
#else
# include <windows.h>
# include <winsock2.h>
# include <ws2tcpip.h>
11 years ago
# include <mmsystem.h>
#endif
#include "miner.h"
#include "elist.h"
#include "compat.h"
#include "util.h"
bool successful_connect = false;
struct timeval nettime;
struct data_buffer {
void *buf;
size_t len;
};
struct upload_buffer {
const void *buf;
size_t len;
};
struct header_info {
char *lp_path;
int rolltime;
char *reason;
char *stratum_url;
bool hadrolltime;
bool canroll;
bool hadexpire;
};
struct tq_ent {
void *data;
struct list_head q_node;
};
static void databuf_free(struct data_buffer *db)
{
if (!db)
return;
free(db->buf);
memset(db, 0, sizeof(*db));
}
static size_t all_data_cb(const void *ptr, size_t size, size_t nmemb,
void *user_data)
{
struct data_buffer *db = user_data;
size_t len = size * nmemb;
size_t oldlen, newlen;
void *newmem;
static const unsigned char zero = 0;
oldlen = db->len;
newlen = oldlen + len;
newmem = realloc(db->buf, newlen + 1);
if (!newmem)
return 0;
db->buf = newmem;
db->len = newlen;
memcpy(db->buf + oldlen, ptr, len);
memcpy(db->buf + newlen, &zero, 1); /* null terminate */
return len;
}
static size_t upload_data_cb(void *ptr, size_t size, size_t nmemb,
void *user_data)
{
struct upload_buffer *ub = user_data;
unsigned int len = size * nmemb;
if (len > ub->len)
len = ub->len;
if (len) {
memcpy(ptr, ub->buf, len);
ub->buf += len;
ub->len -= len;
}
return len;
}
static size_t resp_hdr_cb(void *ptr, size_t size, size_t nmemb, void *user_data)
{
struct header_info *hi = user_data;
size_t remlen, slen, ptrlen = size * nmemb;
char *rem, *val = NULL, *key = NULL;
void *tmp;
val = calloc(1, ptrlen);
key = calloc(1, ptrlen);
if (!key || !val)
goto out;
tmp = memchr(ptr, ':', ptrlen);
if (!tmp || (tmp == ptr)) /* skip empty keys / blanks */
goto out;
slen = tmp - ptr;
if ((slen + 1) == ptrlen) /* skip key w/ no value */
goto out;
memcpy(key, ptr, slen); /* store & nul term key */
key[slen] = 0;
rem = ptr + slen + 1; /* trim value's leading whitespace */
remlen = ptrlen - slen - 1;
while ((remlen > 0) && (isspace(*rem))) {
remlen--;
rem++;
}
memcpy(val, rem, remlen); /* store value, trim trailing ws */
val[remlen] = 0;
while ((*val) && (isspace(val[strlen(val) - 1])))
val[strlen(val) - 1] = 0;
if (!*val) /* skip blank value */
goto out;
if (opt_protocol)
applog(LOG_DEBUG, "HTTP hdr(%s): %s", key, val);
if (!strcasecmp("X-Roll-Ntime", key)) {
hi->hadrolltime = true;
if (!strncasecmp("N", val, 1))
applog(LOG_DEBUG, "X-Roll-Ntime: N found");
else {
hi->canroll = true;
/* Check to see if expire= is supported and if not, set
* the rolltime to the default scantime */
if (strlen(val) > 7 && !strncasecmp("expire=", val, 7)) {
sscanf(val + 7, "%d", &hi->rolltime);
hi->hadexpire = true;
} else
hi->rolltime = opt_scantime;
applog(LOG_DEBUG, "X-Roll-Ntime expiry set to %d", hi->rolltime);
}
}
if (!strcasecmp("X-Long-Polling", key)) {
hi->lp_path = val; /* steal memory reference */
val = NULL;
}
if (!strcasecmp("X-Reject-Reason", key)) {
hi->reason = val; /* steal memory reference */
val = NULL;
}
if (!strcasecmp("X-Stratum", key)) {
hi->stratum_url = val;
val = NULL;
}
out:
free(key);
free(val);
return ptrlen;
}
static void keep_sockalive(SOCKETTYPE fd)
{
const int tcp_one = 1;
#ifndef WIN32
const int tcp_keepidle = 45;
const int tcp_keepintvl = 30;
int flags = fcntl(fd, F_GETFL, 0);
fcntl(fd, F_SETFL, O_NONBLOCK | flags);
#else
u_long flags = 1;
ioctlsocket(fd, FIONBIO, &flags);
#endif
setsockopt(fd, SOL_SOCKET, SO_KEEPALIVE, (const void *)&tcp_one, sizeof(tcp_one));
if (!opt_delaynet)
#ifndef __linux
setsockopt(fd, IPPROTO_TCP, TCP_NODELAY, (const void *)&tcp_one, sizeof(tcp_one));
#else /* __linux */
setsockopt(fd, SOL_TCP, TCP_NODELAY, (const void *)&tcp_one, sizeof(tcp_one));
setsockopt(fd, SOL_TCP, TCP_KEEPCNT, &tcp_one, sizeof(tcp_one));
setsockopt(fd, SOL_TCP, TCP_KEEPIDLE, &tcp_keepidle, sizeof(tcp_keepidle));
setsockopt(fd, SOL_TCP, TCP_KEEPINTVL, &tcp_keepintvl, sizeof(tcp_keepintvl));
#endif /* __linux */
#ifdef __APPLE_CC__
setsockopt(fd, IPPROTO_TCP, TCP_KEEPALIVE, &tcp_keepintvl, sizeof(tcp_keepintvl));
#endif /* __APPLE_CC__ */
}
#if CURL_HAS_KEEPALIVE
static void keep_curlalive(CURL *curl)
{
const int tcp_keepidle = 45;
const int tcp_keepintvl = 30;
const long int keepalive = 1;
curl_easy_setopt(curl, CURLOPT_TCP_KEEPALIVE, keepalive);
curl_easy_setopt(curl, CURLOPT_TCP_KEEPIDLE, tcp_keepidle);
curl_easy_setopt(curl, CURLOPT_TCP_KEEPINTVL, tcp_keepintvl);
}
#else
static void keep_curlalive(CURL *curl)
{
SOCKETTYPE sock;
curl_easy_getinfo(curl, CURLINFO_LASTSOCKET, (long *)&sock);
keep_sockalive(sock);
}
#endif
static void last_nettime(struct timeval *last)
{
rd_lock(&netacc_lock);
last->tv_sec = nettime.tv_sec;
last->tv_usec = nettime.tv_usec;
rd_unlock(&netacc_lock);
}
static void set_nettime(void)
{
wr_lock(&netacc_lock);
cgtime(&nettime);
wr_unlock(&netacc_lock);
}
static int curl_debug_cb(__maybe_unused CURL *handle, curl_infotype type,
__maybe_unused char *data, size_t size, void *userdata)
{
struct pool *pool = (struct pool *)userdata;
switch(type) {
case CURLINFO_HEADER_IN:
case CURLINFO_DATA_IN:
case CURLINFO_SSL_DATA_IN:
pool->cgminer_pool_stats.net_bytes_received += size;
break;
case CURLINFO_HEADER_OUT:
case CURLINFO_DATA_OUT:
case CURLINFO_SSL_DATA_OUT:
pool->cgminer_pool_stats.net_bytes_sent += size;
break;
case CURLINFO_TEXT:
default:
break;
}
return 0;
}
json_t *json_rpc_call(CURL *curl, const char *url,
const char *userpass, const char *rpc_req,
bool probe, bool longpoll, int *rolltime,
struct pool *pool, bool share)
{
long timeout = longpoll ? (60 * 60) : 60;
struct data_buffer all_data = {NULL, 0};
struct header_info hi = {NULL, 0, NULL, NULL, false, false, false};
char len_hdr[64], user_agent_hdr[128];
char curl_err_str[CURL_ERROR_SIZE];
struct curl_slist *headers = NULL;
struct upload_buffer upload_data;
json_t *val, *err_val, *res_val;
bool probing = false;
double byte_count;
json_error_t err;
int rc;
memset(&err, 0, sizeof(err));
/* it is assumed that 'curl' is freshly [re]initialized at this pt */
if (probe)
probing = !pool->probed;
curl_easy_setopt(curl, CURLOPT_TIMEOUT, timeout);
// CURLOPT_VERBOSE won't write to stderr if we use CURLOPT_DEBUGFUNCTION
curl_easy_setopt(curl, CURLOPT_DEBUGFUNCTION, curl_debug_cb);
curl_easy_setopt(curl, CURLOPT_DEBUGDATA, (void *)pool);
curl_easy_setopt(curl, CURLOPT_VERBOSE, 1);
curl_easy_setopt(curl, CURLOPT_NOSIGNAL, 1);
curl_easy_setopt(curl, CURLOPT_URL, url);
curl_easy_setopt(curl, CURLOPT_ENCODING, "");
curl_easy_setopt(curl, CURLOPT_FAILONERROR, 1);
/* Shares are staggered already and delays in submission can be costly
* so do not delay them */
if (!opt_delaynet || share)
curl_easy_setopt(curl, CURLOPT_TCP_NODELAY, 1);
curl_easy_setopt(curl, CURLOPT_WRITEFUNCTION, all_data_cb);
curl_easy_setopt(curl, CURLOPT_WRITEDATA, &all_data);
curl_easy_setopt(curl, CURLOPT_READFUNCTION, upload_data_cb);
curl_easy_setopt(curl, CURLOPT_READDATA, &upload_data);
curl_easy_setopt(curl, CURLOPT_ERRORBUFFER, curl_err_str);
curl_easy_setopt(curl, CURLOPT_FOLLOWLOCATION, 1);
curl_easy_setopt(curl, CURLOPT_HEADERFUNCTION, resp_hdr_cb);
curl_easy_setopt(curl, CURLOPT_HEADERDATA, &hi);
curl_easy_setopt(curl, CURLOPT_USE_SSL, CURLUSESSL_TRY);
if (pool->rpc_proxy) {
curl_easy_setopt(curl, CURLOPT_PROXY, pool->rpc_proxy);
curl_easy_setopt(curl, CURLOPT_PROXYTYPE, pool->rpc_proxytype);
} else if (opt_socks_proxy) {
curl_easy_setopt(curl, CURLOPT_PROXY, opt_socks_proxy);
curl_easy_setopt(curl, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS4);
}
if (userpass) {
curl_easy_setopt(curl, CURLOPT_USERPWD, userpass);
curl_easy_setopt(curl, CURLOPT_HTTPAUTH, CURLAUTH_BASIC);
}
if (longpoll)
keep_curlalive(curl);
curl_easy_setopt(curl, CURLOPT_POST, 1);
if (opt_protocol)
applog(LOG_DEBUG, "JSON protocol request:\n%s", rpc_req);
upload_data.buf = rpc_req;
upload_data.len = strlen(rpc_req);
sprintf(len_hdr, "Content-Length: %lu",
(unsigned long) upload_data.len);
sprintf(user_agent_hdr, "User-Agent: %s", PACKAGE_STRING);
headers = curl_slist_append(headers,
"Content-type: application/json");
headers = curl_slist_append(headers,
"X-Mining-Extensions: longpoll midstate rollntime submitold");
if (likely(global_hashrate)) {
char ghashrate[255];
sprintf(ghashrate, "X-Mining-Hashrate: %llu", global_hashrate);
headers = curl_slist_append(headers, ghashrate);
}
headers = curl_slist_append(headers, len_hdr);
headers = curl_slist_append(headers, user_agent_hdr);
headers = curl_slist_append(headers, "Expect:"); /* disable Expect hdr*/
curl_easy_setopt(curl, CURLOPT_HTTPHEADER, headers);
if (opt_delaynet) {
/* Don't delay share submission, but still track the nettime */
if (!share) {
long long now_msecs, last_msecs;
struct timeval now, last;
cgtime(&now);
last_nettime(&last);
now_msecs = (long long)now.tv_sec * 1000;
now_msecs += now.tv_usec / 1000;
last_msecs = (long long)last.tv_sec * 1000;
last_msecs += last.tv_usec / 1000;
if (now_msecs > last_msecs && now_msecs - last_msecs < 250) {
struct timespec rgtp;
rgtp.tv_sec = 0;
rgtp.tv_nsec = (250 - (now_msecs - last_msecs)) * 1000000;
nanosleep(&rgtp, NULL);
}
}
set_nettime();
}
rc = curl_easy_perform(curl);
if (rc) {
applog(LOG_INFO, "HTTP request failed: %s", curl_err_str);
goto err_out;
}
if (!all_data.buf) {
applog(LOG_DEBUG, "Empty data received in json_rpc_call.");
goto err_out;
}
pool->cgminer_pool_stats.times_sent++;
if (curl_easy_getinfo(curl, CURLINFO_SIZE_UPLOAD, &byte_count) == CURLE_OK)
pool->cgminer_pool_stats.bytes_sent += byte_count;
pool->cgminer_pool_stats.times_received++;
if (curl_easy_getinfo(curl, CURLINFO_SIZE_DOWNLOAD, &byte_count) == CURLE_OK)
pool->cgminer_pool_stats.bytes_received += byte_count;
if (probing) {
pool->probed = true;
/* If X-Long-Polling was found, activate long polling */
if (hi.lp_path) {
if (pool->hdr_path != NULL)
free(pool->hdr_path);
pool->hdr_path = hi.lp_path;
} else
pool->hdr_path = NULL;
if (hi.stratum_url) {
pool->stratum_url = hi.stratum_url;
hi.stratum_url = NULL;
}
} else {
if (hi.lp_path) {
free(hi.lp_path);
hi.lp_path = NULL;
}
if (hi.stratum_url) {
free(hi.stratum_url);
hi.stratum_url = NULL;
}
}
*rolltime = hi.rolltime;
pool->cgminer_pool_stats.rolltime = hi.rolltime;
pool->cgminer_pool_stats.hadrolltime = hi.hadrolltime;
pool->cgminer_pool_stats.canroll = hi.canroll;
pool->cgminer_pool_stats.hadexpire = hi.hadexpire;
val = JSON_LOADS(all_data.buf, &err);
if (!val) {
applog(LOG_INFO, "JSON decode failed(%d): %s", err.line, err.text);
if (opt_protocol)
applog(LOG_DEBUG, "JSON protocol response:\n%s", (char *)(all_data.buf));
goto err_out;
}
if (opt_protocol) {
char *s = json_dumps(val, JSON_INDENT(3));
applog(LOG_DEBUG, "JSON protocol response:\n%s", s);
free(s);
}
/* JSON-RPC valid response returns a non-null 'result',
* and a null 'error'.
*/
res_val = json_object_get(val, "result");
err_val = json_object_get(val, "error");
if (!res_val ||(err_val && !json_is_null(err_val))) {
char *s;
if (err_val)
s = json_dumps(err_val, JSON_INDENT(3));
else
s = strdup("(unknown reason)");
applog(LOG_INFO, "JSON-RPC call failed: %s", s);
free(s);
goto err_out;
}
if (hi.reason) {
json_object_set_new(val, "reject-reason", json_string(hi.reason));
free(hi.reason);
hi.reason = NULL;
}
successful_connect = true;
databuf_free(&all_data);
curl_slist_free_all(headers);
curl_easy_reset(curl);
return val;
err_out:
databuf_free(&all_data);
curl_slist_free_all(headers);
curl_easy_reset(curl);
if (!successful_connect)
applog(LOG_DEBUG, "Failed to connect in json_rpc_call");
curl_easy_setopt(curl, CURLOPT_FRESH_CONNECT, 1);
return NULL;
}
#if (LIBCURL_VERSION_MAJOR == 7 && LIBCURL_VERSION_MINOR >= 10) || (LIBCURL_VERSION_MAJOR > 7)
static struct {
const char *name;
curl_proxytype proxytype;
} proxynames[] = {
{ "http:", CURLPROXY_HTTP },
#if (LIBCURL_VERSION_MAJOR > 7) || (LIBCURL_VERSION_MINOR > 19) || (LIBCURL_VERSION_MINOR == 19 && LIBCURL_VERSION_PATCH >= 4)
{ "http0:", CURLPROXY_HTTP_1_0 },
#endif
#if (LIBCURL_VERSION_MAJOR > 7) || (LIBCURL_VERSION_MINOR > 15) || (LIBCURL_VERSION_MINOR == 15 && LIBCURL_VERSION_PATCH >= 2)
{ "socks4:", CURLPROXY_SOCKS4 },
#endif
{ "socks5:", CURLPROXY_SOCKS5 },
#if (LIBCURL_VERSION_MAJOR > 7) || (LIBCURL_VERSION_MINOR >= 18)
{ "socks4a:", CURLPROXY_SOCKS4A },
{ "socks5h:", CURLPROXY_SOCKS5_HOSTNAME },
#endif
{ NULL, 0 }
};
#endif
const char *proxytype(curl_proxytype proxytype)
{
int i;
for (i = 0; proxynames[i].name; i++)
if (proxynames[i].proxytype == proxytype)
return proxynames[i].name;
return "invalid";
}
char *get_proxy(char *url, struct pool *pool)
{
pool->rpc_proxy = NULL;
#if (LIBCURL_VERSION_MAJOR == 7 && LIBCURL_VERSION_MINOR >= 10) || (LIBCURL_VERSION_MAJOR > 7)
char *split;
int plen, len, i;
for (i = 0; proxynames[i].name; i++) {
plen = strlen(proxynames[i].name);
if (strncmp(url, proxynames[i].name, plen) == 0) {
if (!(split = strchr(url, '|')))
return url;
*split = '\0';
len = split - url;
pool->rpc_proxy = malloc(1 + len - plen);
if (!(pool->rpc_proxy))
quit(1, "Failed to malloc rpc_proxy");
strcpy(pool->rpc_proxy, url + plen);
pool->rpc_proxytype = proxynames[i].proxytype;
url = split + 1;
break;
}
}
#endif
return url;
}
/* Returns a malloced array string of a binary value of arbitrary length. The
* array is rounded up to a 4 byte size to appease architectures that need
* aligned array sizes */
char *bin2hex(const unsigned char *p, size_t len)
{
unsigned int i;
ssize_t slen;
char *s;
slen = len * 2 + 1;
if (slen % 4)
slen += 4 - (slen % 4);
s = calloc(slen, 1);
if (unlikely(!s))
quit(1, "Failed to calloc in bin2hex");
for (i = 0; i < len; i++)
sprintf(s + (i * 2), "%02x", (unsigned int) p[i]);
return s;
}
/* Does the reverse of bin2hex but does not allocate any ram */
bool hex2bin(unsigned char *p, const char *hexstr, size_t len)
{
bool ret = false;
while (*hexstr && len) {
char hex_byte[4];
unsigned int v;
if (unlikely(!hexstr[1])) {
applog(LOG_ERR, "hex2bin str truncated");
return ret;
}
memset(hex_byte, 0, 4);
hex_byte[0] = hexstr[0];
hex_byte[1] = hexstr[1];
if (unlikely(sscanf(hex_byte, "%x", &v) != 1)) {
applog(LOG_ERR, "hex2bin sscanf '%s' failed", hex_byte);
return ret;
}
*p = (unsigned char) v;
p++;
hexstr += 2;
len--;
}
if (likely(len == 0 && *hexstr == 0))
ret = true;
return ret;
}
bool fulltest(const unsigned char *hash, const unsigned char *target)
{
unsigned char hash_swap[32], target_swap[32];
uint32_t *hash32 = (uint32_t *) hash_swap;
uint32_t *target32 = (uint32_t *) target_swap;
char *hash_str, *target_str;
bool rc = true;
int i;
swap256(hash_swap, hash);
swap256(target_swap, target);
for (i = 0; i < 32/4; i++) {
uint32_t h32tmp = htobe32(hash32[i]);
uint32_t t32tmp = htole32(target32[i]);
target32[i] = swab32(target32[i]); /* for printing */
if (h32tmp > t32tmp) {
rc = false;
break;
}
if (h32tmp < t32tmp) {
rc = true;
break;
}
}
if (opt_debug) {
hash_str = bin2hex(hash_swap, 32);
target_str = bin2hex(target_swap, 32);
applog(LOG_DEBUG, " Proof: %s\nTarget: %s\nTrgVal? %s",
hash_str,
target_str,
rc ? "YES (hash <= target)" :
"no (false positive; hash > target)");
free(hash_str);
free(target_str);
}
return rc;
}
struct thread_q *tq_new(void)
{
struct thread_q *tq;
tq = calloc(1, sizeof(*tq));
if (!tq)
return NULL;
INIT_LIST_HEAD(&tq->q);
pthread_mutex_init(&tq->mutex, NULL);
pthread_cond_init(&tq->cond, NULL);
return tq;
}
void tq_free(struct thread_q *tq)
{
struct tq_ent *ent, *iter;
if (!tq)
return;
list_for_each_entry_safe(ent, iter, &tq->q, q_node) {
list_del(&ent->q_node);
free(ent);
}
pthread_cond_destroy(&tq->cond);
pthread_mutex_destroy(&tq->mutex);
memset(tq, 0, sizeof(*tq)); /* poison */
free(tq);
}
static void tq_freezethaw(struct thread_q *tq, bool frozen)
{
mutex_lock(&tq->mutex);
tq->frozen = frozen;
pthread_cond_signal(&tq->cond);
mutex_unlock(&tq->mutex);
}
void tq_freeze(struct thread_q *tq)
{
tq_freezethaw(tq, true);
}
void tq_thaw(struct thread_q *tq)
{
tq_freezethaw(tq, false);
}
bool tq_push(struct thread_q *tq, void *data)
{
struct tq_ent *ent;
bool rc = true;
ent = calloc(1, sizeof(*ent));
if (!ent)
return false;
ent->data = data;
INIT_LIST_HEAD(&ent->q_node);
mutex_lock(&tq->mutex);
if (!tq->frozen) {
list_add_tail(&ent->q_node, &tq->q);
} else {
free(ent);
rc = false;
}
pthread_cond_signal(&tq->cond);
mutex_unlock(&tq->mutex);
return rc;
}
void *tq_pop(struct thread_q *tq, const struct timespec *abstime)
{
struct tq_ent *ent;
void *rval = NULL;
int rc;
mutex_lock(&tq->mutex);
if (!list_empty(&tq->q))
goto pop;
if (abstime)
rc = pthread_cond_timedwait(&tq->cond, &tq->mutex, abstime);
else
rc = pthread_cond_wait(&tq->cond, &tq->mutex);
if (rc)
goto out;
if (list_empty(&tq->q))
goto out;
pop:
ent = list_entry(tq->q.next, struct tq_ent, q_node);
rval = ent->data;
list_del(&ent->q_node);
free(ent);
out:
mutex_unlock(&tq->mutex);
return rval;
}
int thr_info_create(struct thr_info *thr, pthread_attr_t *attr, void *(*start) (void *), void *arg)
{
return pthread_create(&thr->pth, attr, start, arg);
}
void thr_info_freeze(struct thr_info *thr)
{
struct tq_ent *ent, *iter;
struct thread_q *tq;
if (!thr)
return;
tq = thr->q;
if (!tq)
return;
mutex_lock(&tq->mutex);
tq->frozen = true;
list_for_each_entry_safe(ent, iter, &tq->q, q_node) {
list_del(&ent->q_node);
free(ent);
}
mutex_unlock(&tq->mutex);
}
void thr_info_cancel(struct thr_info *thr)
{
if (!thr)
return;
if (PTH(thr) != 0L) {
pthread_cancel(thr->pth);
PTH(thr) = 0L;
}
}
/* Provide a ms based sleep that uses nanosleep to avoid poor usleep accuracy
* on SMP machines */
void nmsleep(unsigned int msecs)
{
struct timespec twait, tleft;
int ret;
ldiv_t d;
#ifdef WIN32
timeBeginPeriod(1);
#endif
d = ldiv(msecs, 1000);
tleft.tv_sec = d.quot;
tleft.tv_nsec = d.rem * 1000000;
do {
twait.tv_sec = tleft.tv_sec;
twait.tv_nsec = tleft.tv_nsec;
ret = nanosleep(&twait, &tleft);
} while (ret == -1 && errno == EINTR);
#ifdef WIN32
timeEndPeriod(1);
#endif
}
/* This is a cgminer gettimeofday wrapper. Since we always call gettimeofday
* with tz set to NULL, and windows' default resolution is only 15ms, this
* gives us higher resolution times on windows. */
void cgtime(struct timeval *tv)
{
#ifdef WIN32
timeBeginPeriod(1);
#endif
gettimeofday(tv, NULL);
#ifdef WIN32
timeEndPeriod(1);
#endif
}
void subtime(struct timeval *a, struct timeval *b)
{
timersub(a, b, b);
}
void addtime(struct timeval *a, struct timeval *b)
{
timeradd(a, b, b);
}
bool time_more(struct timeval *a, struct timeval *b)
{
return timercmp(a, b, >);
}
bool time_less(struct timeval *a, struct timeval *b)
{
return timercmp(a, b, <);
}
void copy_time(struct timeval *dest, const struct timeval *src)
{
memcpy(dest, src, sizeof(struct timeval));
}
/* Returns the microseconds difference between end and start times as a double */
double us_tdiff(struct timeval *end, struct timeval *start)
{
return end->tv_sec * 1000000 + end->tv_usec - start->tv_sec * 1000000 - start->tv_usec;
}
/* Returns the seconds difference between end and start times as a double */
double tdiff(struct timeval *end, struct timeval *start)
{
return end->tv_sec - start->tv_sec + (end->tv_usec - start->tv_usec) / 1000000.0;
}
bool extract_sockaddr(struct pool *pool, char *url)
{
char *url_begin, *url_end, *ipv6_begin, *ipv6_end, *port_start = NULL;
char url_address[256], port[6];
12 years ago
int url_len, port_len = 0;
pool->sockaddr_url = url;
url_begin = strstr(url, "//");
if (!url_begin)
url_begin = url;
else
url_begin += 2;
/* Look for numeric ipv6 entries */
ipv6_begin = strstr(url_begin, "[");
ipv6_end = strstr(url_begin, "]");
if (ipv6_begin && ipv6_end && ipv6_end > ipv6_begin)
url_end = strstr(ipv6_end, ":");
else
url_end = strstr(url_begin, ":");
if (url_end) {
url_len = url_end - url_begin;
port_len = strlen(url_begin) - url_len - 1;
if (port_len < 1)
return false;
port_start = url_end + 1;
} else
url_len = strlen(url_begin);
if (url_len < 1)
return false;
sprintf(url_address, "%.*s", url_len, url_begin);
if (port_len)
12 years ago
snprintf(port, 6, "%.*s", port_len, port_start);
else
strcpy(port, "80");
pool->stratum_port = strdup(port);
pool->sockaddr_url = strdup(url_address);
return true;
}
enum send_ret {
SEND_OK,
SEND_SELECTFAIL,
SEND_SENDFAIL,
SEND_INACTIVE
};
/* Send a single command across a socket, appending \n to it. This should all
* be done under stratum lock except when first establishing the socket */
static enum send_ret __stratum_send(struct pool *pool, char *s, ssize_t len)
{
SOCKETTYPE sock = pool->sock;
ssize_t ssent = 0;
strcat(s, "\n");
len++;
while (len > 0 ) {
struct timeval timeout = {1, 0};
ssize_t sent;
fd_set wd;
FD_ZERO(&wd);
FD_SET(sock, &wd);
if (select(sock + 1, NULL, &wd, NULL, &timeout) < 1)
return SEND_SELECTFAIL;
#ifndef WIN32
sent = send(pool->sock, s + ssent, len, MSG_NOSIGNAL);
#else
sent = send(pool->sock, s + ssent, len, 0);
#endif
if (sent < 0) {
if (!sock_blocks())
return SEND_SENDFAIL;
sent = 0;
}
ssent += sent;
len -= sent;
}
pool->cgminer_pool_stats.times_sent++;
pool->cgminer_pool_stats.bytes_sent += ssent;
pool->cgminer_pool_stats.net_bytes_sent += ssent;
return SEND_OK;
}
bool stratum_send(struct pool *pool, char *s, ssize_t len)
{
enum send_ret ret = SEND_INACTIVE;
if (opt_protocol)
applog(LOG_DEBUG, "SEND: %s", s);
mutex_lock(&pool->stratum_lock);
if (pool->stratum_active)
ret = __stratum_send(pool, s, len);
mutex_unlock(&pool->stratum_lock);
/* This is to avoid doing applog under stratum_lock */
switch (ret) {
default:
case SEND_OK:
break;
case SEND_SELECTFAIL:
applog(LOG_DEBUG, "Write select failed on pool %d sock", pool->pool_no);
break;
case SEND_SENDFAIL:
applog(LOG_DEBUG, "Failed to curl_easy_send in stratum_send");
break;
case SEND_INACTIVE:
applog(LOG_DEBUG, "Stratum send failed due to no pool stratum_active");
break;
}
return (ret == SEND_OK);
}
static bool socket_full(struct pool *pool, bool wait)
{
SOCKETTYPE sock = pool->sock;
struct timeval timeout;
fd_set rd;
FD_ZERO(&rd);
FD_SET(sock, &rd);
timeout.tv_usec = 0;
if (wait)
timeout.tv_sec = 60;
else
timeout.tv_sec = 1;
if (select(sock + 1, &rd, NULL, NULL, &timeout) > 0)
return true;
return false;
}
/* Check to see if Santa's been good to you */
bool sock_full(struct pool *pool)
{
if (strlen(pool->sockbuf))
return true;
return (socket_full(pool, false));
}
static void clear_sockbuf(struct pool *pool)
{
strcpy(pool->sockbuf, "");
}
static void clear_sock(struct pool *pool)
{
ssize_t n;
mutex_lock(&pool->stratum_lock);
do {
n = recv(pool->sock, pool->sockbuf, RECVSIZE, 0);
} while (n > 0);
mutex_unlock(&pool->stratum_lock);
clear_sockbuf(pool);
}
/* Make sure the pool sockbuf is large enough to cope with any coinbase size
* by reallocing it to a large enough size rounded up to a multiple of RBUFSIZE
* and zeroing the new memory */
static void recalloc_sock(struct pool *pool, size_t len)
{
size_t old, new;
old = strlen(pool->sockbuf);
new = old + len + 1;
if (new < pool->sockbuf_size)
return;
new = new + (RBUFSIZE - (new % RBUFSIZE));
// Avoid potentially recursive locking
// applog(LOG_DEBUG, "Recallocing pool sockbuf to %d", new);
pool->sockbuf = realloc(pool->sockbuf, new);
if (!pool->sockbuf)
quit(1, "Failed to realloc pool sockbuf in recalloc_sock");
memset(pool->sockbuf + old, 0, new - old);
pool->sockbuf_size = new;
}
enum recv_ret {
RECV_OK,
RECV_CLOSED,
RECV_RECVFAIL
};
/* Peeks at a socket to find the first end of line and then reads just that
* from the socket and returns that as a malloced char */
char *recv_line(struct pool *pool)
{
ssize_t len, buflen;
char *tok, *sret = NULL;
if (!strstr(pool->sockbuf, "\n")) {
enum recv_ret ret = RECV_OK;
struct timeval rstart, now;
cgtime(&rstart);
if (!socket_full(pool, true)) {
applog(LOG_DEBUG, "Timed out waiting for data on socket_full");
goto out;
}
mutex_lock(&pool->stratum_lock);
do {
char s[RBUFSIZE];
size_t slen;
ssize_t n;
memset(s, 0, RBUFSIZE);
n = recv(pool->sock, s, RECVSIZE, 0);
if (!n) {
ret = RECV_CLOSED;
break;
}
if (n < 0) {
if (!sock_blocks() || !socket_full(pool, false)) {
ret = RECV_RECVFAIL;
break;
}
} else {
slen = strlen(s);
recalloc_sock(pool, slen);
strcat(pool->sockbuf, s);
}
cgtime(&now);
} while (tdiff(&now, &rstart) < 60 && !strstr(pool->sockbuf, "\n"));
mutex_unlock(&pool->stratum_lock);
switch (ret) {
default:
case RECV_OK:
break;
case RECV_CLOSED:
applog(LOG_DEBUG, "Socket closed waiting in recv_line");
goto out;
case RECV_RECVFAIL:
applog(LOG_DEBUG, "Failed to recv sock in recv_line");
goto out;
}
}
buflen = strlen(pool->sockbuf);
tok = strtok(pool->sockbuf, "\n");
if (!tok) {
applog(LOG_DEBUG, "Failed to parse a \\n terminated string in recv_line");
goto out;
}
sret = strdup(tok);
len = strlen(sret);
/* Copy what's left in the buffer after the \n, including the
* terminating \0 */
if (buflen > len + 1)
memmove(pool->sockbuf, pool->sockbuf + len + 1, buflen - len + 1);
else
strcpy(pool->sockbuf, "");
pool->cgminer_pool_stats.times_received++;
pool->cgminer_pool_stats.bytes_received += len;
pool->cgminer_pool_stats.net_bytes_received += len;
out:
if (!sret)
clear_sock(pool);
else if (opt_protocol)
applog(LOG_DEBUG, "RECVD: %s", sret);
return sret;
}
/* Extracts a string value from a json array with error checking. To be used
* when the value of the string returned is only examined and not to be stored.
* See json_array_string below */
static char *__json_array_string(json_t *val, unsigned int entry)
{
json_t *arr_entry;
if (json_is_null(val))
return NULL;
if (!json_is_array(val))
return NULL;
if (entry > json_array_size(val))
return NULL;
arr_entry = json_array_get(val, entry);
if (!json_is_string(arr_entry))
return NULL;
return (char *)json_string_value(arr_entry);
}
/* Creates a freshly malloced dup of __json_array_string */
static char *json_array_string(json_t *val, unsigned int entry)
{
char *buf = __json_array_string(val, entry);
if (buf)
return strdup(buf);
return NULL;
}
static bool parse_notify(struct pool *pool, json_t *val)
{
char *job_id, *prev_hash, *coinbase1, *coinbase2, *bbversion, *nbit, *ntime;
bool clean, ret = false;
int merkles, i;
json_t *arr;
arr = json_array_get(val, 4);
if (!arr || !json_is_array(arr))
goto out;
merkles = json_array_size(arr);
job_id = json_array_string(val, 0);
prev_hash = json_array_string(val, 1);
coinbase1 = json_array_string(val, 2);
coinbase2 = json_array_string(val, 3);
bbversion = json_array_string(val, 5);
nbit = json_array_string(val, 6);
ntime = json_array_string(val, 7);
clean = json_is_true(json_array_get(val, 8));
if (!job_id || !prev_hash || !coinbase1 || !coinbase2 || !bbversion || !nbit || !ntime) {
/* Annoying but we must not leak memory */
if (job_id)
free(job_id);
if (prev_hash)
free(prev_hash);
if (coinbase1)
free(coinbase1);
if (coinbase2)
free(coinbase2);
if (bbversion)
free(bbversion);
if (nbit)
free(nbit);
if (ntime)
free(ntime);
goto out;
}
cg_wlock(&pool->data_lock);
free(pool->swork.job_id);
free(pool->swork.prev_hash);
free(pool->swork.coinbase1);
free(pool->swork.coinbase2);
free(pool->swork.bbversion);
free(pool->swork.nbit);
free(pool->swork.ntime);
pool->swork.job_id = job_id;
pool->swork.prev_hash = prev_hash;
pool->swork.coinbase1 = coinbase1;
pool->swork.cb1_len = strlen(coinbase1) / 2;
pool->swork.coinbase2 = coinbase2;
pool->swork.cb2_len = strlen(coinbase2) / 2;
pool->swork.bbversion = bbversion;
pool->swork.nbit = nbit;
pool->swork.ntime = ntime;
pool->swork.clean = clean;
pool->swork.cb_len = pool->swork.cb1_len + pool->n1_len + pool->n2size + pool->swork.cb2_len;
for (i = 0; i < pool->swork.merkles; i++)
free(pool->swork.merkle[i]);
if (merkles) {
pool->swork.merkle = realloc(pool->swork.merkle, sizeof(char *) * merkles + 1);
for (i = 0; i < merkles; i++)
pool->swork.merkle[i] = json_array_string(arr, i);
}
pool->swork.merkles = merkles;
if (clean)
pool->nonce2 = 0;
pool->swork.header_len = strlen(pool->swork.bbversion) +
strlen(pool->swork.prev_hash) +
strlen(pool->swork.ntime) +
strlen(pool->swork.nbit) +
/* merkle_hash */ 32 +
/* nonce */ 8 +
/* workpadding */ 96;
pool->swork.header_len = pool->swork.header_len * 2 + 1;
align_len(&pool->swork.header_len);
cg_wunlock(&pool->data_lock);
if (opt_protocol) {
applog(LOG_DEBUG, "job_id: %s", job_id);
applog(LOG_DEBUG, "prev_hash: %s", prev_hash);
applog(LOG_DEBUG, "coinbase1: %s", coinbase1);
applog(LOG_DEBUG, "coinbase2: %s", coinbase2);
for (i = 0; i < merkles; i++)
applog(LOG_DEBUG, "merkle%d: %s", i, pool->swork.merkle[i]);
applog(LOG_DEBUG, "bbversion: %s", bbversion);
applog(LOG_DEBUG, "nbit: %s", nbit);
applog(LOG_DEBUG, "ntime: %s", ntime);
applog(LOG_DEBUG, "clean: %s", clean ? "yes" : "no");
}
/* A notify message is the closest stratum gets to a getwork */
pool->getwork_requested++;
total_getworks++;
ret = true;
out:
return ret;
}
static bool parse_diff(struct pool *pool, json_t *val)
{
double diff;
diff = json_number_value(json_array_get(val, 0));
if (diff == 0)
return false;
cg_wlock(&pool->data_lock);
pool->swork.diff = diff;
cg_wunlock(&pool->data_lock);
applog(LOG_DEBUG, "Pool %d difficulty set to %f", pool->pool_no, diff);
return true;
}
static bool parse_reconnect(struct pool *pool, json_t *val)
{
char *url, *port, address[256];
memset(address, 0, 255);
url = (char *)json_string_value(json_array_get(val, 0));
if (!url)
url = pool->sockaddr_url;
port = (char *)json_string_value(json_array_get(val, 1));
if (!port)
port = pool->stratum_port;
sprintf(address, "%s:%s", url, port);
if (!extract_sockaddr(pool, address))
return false;
pool->stratum_url = pool->sockaddr_url;
applog(LOG_NOTICE, "Reconnect requested from pool %d to %s", pool->pool_no, address);
if (!restart_stratum(pool))
return false;
return true;
}
static bool send_version(struct pool *pool, json_t *val)
{
char s[RBUFSIZE];
int id = json_integer_value(json_object_get(val, "id"));
if (!id)
return false;
sprintf(s, "{\"id\": %d, \"result\": \""PACKAGE"/"VERSION"\", \"error\": null}", id);
if (!stratum_send(pool, s, strlen(s)))
return false;
return true;
}
static bool show_message(struct pool *pool, json_t *val)
{
char *msg;
if (!json_is_array(val))
return false;
msg = (char *)json_string_value(json_array_get(val, 0));
if (!msg)
return false;
applog(LOG_NOTICE, "Pool %d message: %s", pool->pool_no, msg);
return true;
}
bool parse_method(struct pool *pool, char *s)
{
json_t *val = NULL, *method, *err_val, *params;
json_error_t err;
bool ret = false;
char *buf;
if (!s)
goto out;
val = JSON_LOADS(s, &err);
if (!val) {
applog(LOG_INFO, "JSON decode failed(%d): %s", err.line, err.text);
goto out;
}
method = json_object_get(val, "method");
if (!method)
goto out;
err_val = json_object_get(val, "error");
params = json_object_get(val, "params");
if (err_val && !json_is_null(err_val)) {
char *ss;
if (err_val)
ss = json_dumps(err_val, JSON_INDENT(3));
else
ss = strdup("(unknown reason)");
applog(LOG_INFO, "JSON-RPC method decode failed: %s", ss);
free(ss);
goto out;
}
buf = (char *)json_string_value(method);
if (!buf)
goto out;
if (!strncasecmp(buf, "mining.notify", 13)) {
if (parse_notify(pool, params))
pool->stratum_notify = ret = true;
else
pool->stratum_notify = ret = false;
goto out;
}
if (!strncasecmp(buf, "mining.set_difficulty", 21) && parse_diff(pool, params)) {
ret = true;
goto out;
}
if (!strncasecmp(buf, "client.reconnect", 16) && parse_reconnect(pool, params)) {
ret = true;
goto out;
}
if (!strncasecmp(buf, "client.get_version", 18) && send_version(pool, val)) {
ret = true;
goto out;
}
if (!strncasecmp(buf, "client.show_message", 19) && show_message(pool, params)) {
ret = true;
goto out;
}
out:
if (val)
json_decref(val);
return ret;
}
bool auth_stratum(struct pool *pool)
{
json_t *val = NULL, *res_val, *err_val;
char s[RBUFSIZE], *sret = NULL;
json_error_t err;
bool ret = false;
sprintf(s, "{\"id\": %d, \"method\": \"mining.authorize\", \"params\": [\"%s\", \"%s\"]}",
swork_id++, pool->rpc_user, pool->rpc_pass);
if (!stratum_send(pool, s, strlen(s)))
goto out;
/* Parse all data in the queue and anything left should be auth */
while (42) {
sret = recv_line(pool);
if (!sret)
goto out;
if (parse_method(pool, sret))
free(sret);
else
break;
}
val = JSON_LOADS(sret, &err);
free(sret);
res_val = json_object_get(val, "result");
err_val = json_object_get(val, "error");
if (!res_val || json_is_false(res_val) || (err_val && !json_is_null(err_val))) {
char *ss;
if (err_val)
ss = json_dumps(err_val, JSON_INDENT(3));
else
ss = strdup("(unknown reason)");
applog(LOG_WARNING, "pool %d JSON stratum auth failed: %s", pool->pool_no, ss);
free(ss);
goto out;
}
ret = true;
applog(LOG_INFO, "Stratum authorisation success for pool %d", pool->pool_no);
pool->probed = true;
successful_connect = true;
out:
if (val)
json_decref(val);
return ret;
}
static bool setup_stratum_socket(struct pool *pool)
{
struct addrinfo *servinfo, *hints, *p;
int sockd;
mutex_lock(&pool->stratum_lock);
pool->stratum_active = false;
if (pool->sock)
CLOSESOCKET(pool->sock);
pool->sock = 0;
mutex_unlock(&pool->stratum_lock);
hints = &pool->stratum_hints;
memset(hints, 0, sizeof(struct addrinfo));
hints->ai_family = AF_UNSPEC;
hints->ai_socktype = SOCK_STREAM;
if (getaddrinfo(pool->sockaddr_url, pool->stratum_port, hints, &servinfo) != 0) {
if (!pool->probed) {
applog(LOG_WARNING, "Failed to resolve (?wrong URL) %s:%s",
pool->sockaddr_url, pool->stratum_port);
} else {
applog(LOG_INFO, "Failed to getaddrinfo for %s:%s",
pool->sockaddr_url, pool->stratum_port);
}
return false;
}
for (p = servinfo; p != NULL; p = p->ai_next) {
sockd = socket(p->ai_family, p->ai_socktype, p->ai_protocol);
if (sockd == -1) {
applog(LOG_DEBUG, "Failed socket");
continue;
}
if (connect(sockd, p->ai_addr, p->ai_addrlen) == -1) {
CLOSESOCKET(sockd);
applog(LOG_DEBUG, "Failed connect");
continue;
}
break;
}
if (p == NULL) {
applog(LOG_INFO, "Failed to find a stratum servinfo on %s:%s",
pool->sockaddr_url, pool->stratum_port);
freeaddrinfo(servinfo);
return false;
}
freeaddrinfo(servinfo);
if (!pool->sockbuf) {
pool->sockbuf = calloc(RBUFSIZE, 1);
if (!pool->sockbuf)
quit(1, "Failed to calloc pool sockbuf in initiate_stratum");
pool->sockbuf_size = RBUFSIZE;
}
pool->sock = sockd;
keep_sockalive(sockd);
return true;
}
static char *get_sessionid(json_t *val)
{
char *ret = NULL;
json_t *arr_val;
int arrsize, i;
arr_val = json_array_get(val, 0);
if (!arr_val || !json_is_array(arr_val))
goto out;
arrsize = json_array_size(arr_val);
for (i = 0; i < arrsize; i++) {
json_t *arr = json_array_get(arr_val, i);
char *notify;
if (!arr | !json_is_array(arr))
break;
notify = __json_array_string(arr, 0);
if (!notify)
continue;
if (!strncasecmp(notify, "mining.notify", 13)) {
ret = json_array_string(arr, 1);
break;
}
}
out:
return ret;
}
void suspend_stratum(struct pool *pool)
{
clear_sockbuf(pool);
applog(LOG_INFO, "Closing socket for stratum pool %d", pool->pool_no);
mutex_lock(&pool->stratum_lock);
pool->stratum_active = pool->stratum_notify = false;
if (pool->sock)
CLOSESOCKET(pool->sock);
pool->sock = 0;
mutex_unlock(&pool->stratum_lock);
}
bool initiate_stratum(struct pool *pool)
{
bool ret = false, recvd = false, noresume = false, sockd = false;
char s[RBUFSIZE], *sret = NULL, *nonce1, *sessionid;
json_t *val = NULL, *res_val, *err_val;
json_error_t err;
int n2size;
resend:
if (!setup_stratum_socket(pool)) {
sockd = false;
goto out;
}
sockd = true;
if (recvd) {
/* Get rid of any crap lying around if we're resending */
clear_sock(pool);
sprintf(s, "{\"id\": %d, \"method\": \"mining.subscribe\", \"params\": []}", swork_id++);
} else {
if (pool->sessionid)
sprintf(s, "{\"id\": %d, \"method\": \"mining.subscribe\", \"params\": [\""PACKAGE"/"VERSION"\", \"%s\"]}", swork_id++, pool->sessionid);
else
sprintf(s, "{\"id\": %d, \"method\": \"mining.subscribe\", \"params\": [\""PACKAGE"/"VERSION"\"]}", swork_id++);
}
if (__stratum_send(pool, s, strlen(s)) != SEND_OK) {
applog(LOG_DEBUG, "Failed to send s in initiate_stratum");
goto out;
}
if (!socket_full(pool, true)) {
applog(LOG_DEBUG, "Timed out waiting for response in initiate_stratum");
goto out;
}
sret = recv_line(pool);
if (!sret)
goto out;
recvd = true;
val = JSON_LOADS(sret, &err);
free(sret);
if (!val) {
applog(LOG_INFO, "JSON decode failed(%d): %s", err.line, err.text);
goto out;
}
res_val = json_object_get(val, "result");
err_val = json_object_get(val, "error");
if (!res_val || json_is_null(res_val) ||
(err_val && !json_is_null(err_val))) {
char *ss;
if (err_val)
ss = json_dumps(err_val, JSON_INDENT(3));
else
ss = strdup("(unknown reason)");
applog(LOG_INFO, "JSON-RPC decode failed: %s", ss);
free(ss);
goto out;
}
sessionid = get_sessionid(res_val);
if (!sessionid)
applog(LOG_DEBUG, "Failed to get sessionid in initiate_stratum");
nonce1 = json_array_string(res_val, 1);
if (!nonce1) {
applog(LOG_INFO, "Failed to get nonce1 in initiate_stratum");
free(sessionid);
goto out;
}
n2size = json_integer_value(json_array_get(res_val, 2));
if (!n2size) {
applog(LOG_INFO, "Failed to get n2size in initiate_stratum");
free(sessionid);
free(nonce1);
goto out;
}
cg_wlock(&pool->data_lock);
pool->sessionid = sessionid;
pool->nonce1 = nonce1;
pool->n1_len = strlen(nonce1) / 2;
pool->n2size = n2size;
cg_wunlock(&pool->data_lock);
if (sessionid)
applog(LOG_DEBUG, "Pool %d stratum session id: %s", pool->pool_no, pool->sessionid);
ret = true;
out:
if (val)
json_decref(val);
if (ret) {
if (!pool->stratum_url)
pool->stratum_url = pool->sockaddr_url;
pool->stratum_active = true;
pool->swork.diff = 1;
if (opt_protocol) {
applog(LOG_DEBUG, "Pool %d confirmed mining.subscribe with extranonce1 %s extran2size %d",
pool->pool_no, pool->nonce1, pool->n2size);
}
} else {
if (recvd && !noresume) {
/* Reset the sessionid used for stratum resuming in case the pool
* does not support it, or does not know how to respond to the
* presence of the sessionid parameter. */
cg_wlock(&pool->data_lock);
free(pool->sessionid);
free(pool->nonce1);
pool->sessionid = pool->nonce1 = NULL;
cg_wunlock(&pool->data_lock);
applog(LOG_DEBUG, "Failed to resume stratum, trying afresh");
noresume = true;
goto resend;
}
applog(LOG_DEBUG, "Initiate stratum failed");
if (sockd)
suspend_stratum(pool);
}
return ret;
}
bool restart_stratum(struct pool *pool)
{
if (pool->stratum_active)
suspend_stratum(pool);
if (!initiate_stratum(pool))
return false;
if (!auth_stratum(pool))
return false;
return true;
}
void dev_error(struct cgpu_info *dev, enum dev_reason reason)
{
dev->device_last_not_well = time(NULL);
dev->device_not_well_reason = reason;
12 years ago
switch (reason) {
case REASON_THREAD_FAIL_INIT:
dev->thread_fail_init_count++;
12 years ago
break;
case REASON_THREAD_ZERO_HASH:
dev->thread_zero_hash_count++;
12 years ago
break;
case REASON_THREAD_FAIL_QUEUE:
dev->thread_fail_queue_count++;
12 years ago
break;
case REASON_DEV_SICK_IDLE_60:
dev->dev_sick_idle_60_count++;
12 years ago
break;
case REASON_DEV_DEAD_IDLE_600:
dev->dev_dead_idle_600_count++;
12 years ago
break;
case REASON_DEV_NOSTART:
dev->dev_nostart_count++;
12 years ago
break;
case REASON_DEV_OVER_HEAT:
dev->dev_over_heat_count++;
12 years ago
break;
case REASON_DEV_THERMAL_CUTOFF:
dev->dev_thermal_cutoff_count++;
12 years ago
break;
case REASON_DEV_COMMS_ERROR:
dev->dev_comms_error_count++;
12 years ago
break;
case REASON_DEV_THROTTLE:
dev->dev_throttle_count++;
12 years ago
break;
}
}
/* Realloc an existing string to fit an extra string s, appending s to it. */
void *realloc_strcat(char *ptr, char *s)
{
size_t old = strlen(ptr), len = strlen(s);
char *ret;
if (!len)
return ptr;
len += old + 1;
align_len(&len);
ret = malloc(len);
if (unlikely(!ret))
quit(1, "Failed to malloc in realloc_strcat");
sprintf(ret, "%s%s", ptr, s);
free(ptr);
return ret;
}
/* Make a text readable version of a string using 0xNN for < ' ' or > '~'
* Including 0x00 at the end
* You must free the result yourself */
void *str_text(char *ptr)
{
unsigned char *uptr;
char *ret, *txt;
if (ptr == NULL) {
ret = strdup("(null)");
if (unlikely(!ret))
quit(1, "Failed to malloc in text_str null");
}
uptr = (unsigned char *)ptr;
ret = txt = malloc(strlen(ptr)*4+5); // Guaranteed >= needed
if (unlikely(!txt))
quit(1, "Failed to malloc in text_str txt");
do {
if (*uptr < ' ' || *uptr > '~') {
sprintf(txt, "0x%02x", *uptr);
txt += 4;
} else
*(txt++) = *uptr;
} while (*(uptr++));
*txt = '\0';
return ret;
}
void RenameThread(const char* name)
{
#if defined(PR_SET_NAME)
// Only the first 15 characters are used (16 - NUL terminator)
prctl(PR_SET_NAME, name, 0, 0, 0);
#elif (defined(__FreeBSD__) || defined(__OpenBSD__))
pthread_set_name_np(pthread_self(), name);
#elif defined(MAC_OSX)
pthread_setname_np(name);
#else
// Prevent warnings for unused parameters...
(void)name;
#endif
}