Browse Source

Submitting an ntime offset nonce needs to be done on a copy of the work instead of the original so abstract out shared components as much as possible, minimising strdups in copy_work and make submit_work_async work take copied work, cleaning up code in the process.

nfactor-troky
Con Kolivas 11 years ago
parent
commit
8c0ea19e13
  1. 95
      cgminer.c
  2. 1
      miner.h

95
cgminer.c

@ -3511,9 +3511,23 @@ static void *submit_work_thread(void __maybe_unused *userdata)
} }
#endif /* HAVE_LIBCURL */ #endif /* HAVE_LIBCURL */
/* Return an adjusted ntime if we're submitting work that a device has
* internally offset the ntime. */
static char *offset_ntime(const char *ntime, int noffset)
{
unsigned char bin[4];
uint32_t h32, *be32 = (uint32_t *)bin;
hex2bin(bin, ntime, 4);
h32 = be32toh(*be32) + noffset;
*be32 = htobe32(h32);
return bin2hex(bin, 4);
}
/* Duplicates any dynamically allocated arrays within the work struct to /* Duplicates any dynamically allocated arrays within the work struct to
* prevent a copied work struct from freeing ram belonging to another struct */ * prevent a copied work struct from freeing ram belonging to another struct */
void __copy_work(struct work *work, struct work *base_work) static void _copy_work(struct work *work, const struct work *base_work, int noffset)
{ {
int id = work->id; int id = work->id;
@ -3526,8 +3540,12 @@ void __copy_work(struct work *work, struct work *base_work)
work->job_id = strdup(base_work->job_id); work->job_id = strdup(base_work->job_id);
if (base_work->nonce1) if (base_work->nonce1)
work->nonce1 = strdup(base_work->nonce1); work->nonce1 = strdup(base_work->nonce1);
if (base_work->ntime) if (base_work->ntime) {
work->ntime = strdup(base_work->ntime); if (noffset)
work->ntime = offset_ntime(base_work->ntime, noffset);
else
work->ntime = strdup(base_work->ntime);
}
if (base_work->coinbase) if (base_work->coinbase)
work->coinbase = strdup(base_work->coinbase); work->coinbase = strdup(base_work->coinbase);
} }
@ -3538,7 +3556,7 @@ struct work *copy_work(struct work *base_work)
{ {
struct work *work = make_work(); struct work *work = make_work();
__copy_work(work, base_work); _copy_work(work, base_work, 0);
return work; return work;
} }
@ -5972,14 +5990,13 @@ static struct work *get_work(struct thr_info *thr, const int thr_id)
return work; return work;
} }
static void submit_work_async(struct work *work_in, struct timeval *tv_work_found) /* Submit a copy of the tested, statistic recorded work item asynchronously */
static void submit_work_async(struct work *work)
{ {
struct work *work = copy_work(work_in);
struct pool *pool = work->pool; struct pool *pool = work->pool;
pthread_t submit_thread; pthread_t submit_thread;
if (tv_work_found) cgtime(&work->tv_work_found);
copy_time(&work->tv_work_found, tv_work_found);
if (stale_work(work, true)) { if (stale_work(work, true)) {
if (opt_submit_stale) if (opt_submit_stale)
@ -6018,6 +6035,9 @@ static void submit_work_async(struct work *work_in, struct timeval *tv_work_foun
void inc_hw_errors(struct thr_info *thr) void inc_hw_errors(struct thr_info *thr)
{ {
applog(LOG_INFO, "%s%d: invalid nonce - HW error", thr->cgpu->drv->name,
thr->cgpu->device_id);
mutex_lock(&stats_lock); mutex_lock(&stats_lock);
hw_errors++; hw_errors++;
thr->cgpu->hw_errors++; thr->cgpu->hw_errors++;
@ -6042,12 +6062,8 @@ bool test_nonce(struct work *work, uint32_t nonce)
return (be32toh(hash2_32[7]) <= diff1targ); return (be32toh(hash2_32[7]) <= diff1targ);
} }
/* To be used once the work has been tested to be meet diff1 and has had its static void update_work_stats(struct thr_info *thr, struct work *work)
* nonce adjusted. */
void submit_tested_work(struct thr_info *thr, struct work *work)
{ {
struct timeval tv_work_found;
work->share_diff = share_diff(work); work->share_diff = share_diff(work);
mutex_lock(&stats_lock); mutex_lock(&stats_lock);
@ -6056,50 +6072,63 @@ void submit_tested_work(struct thr_info *thr, struct work *work)
work->pool->diff1 += work->device_diff; work->pool->diff1 += work->device_diff;
thr->cgpu->last_device_valid_work = time(NULL); thr->cgpu->last_device_valid_work = time(NULL);
mutex_unlock(&stats_lock); mutex_unlock(&stats_lock);
}
/* To be used once the work has been tested to be meet diff1 and has had its
* nonce adjusted. */
void submit_tested_work(struct thr_info *thr, struct work *work)
{
struct work *work_out;
update_work_stats(thr, work);
if (!fulltest(work->hash2, work->target)) { if (!fulltest(work->hash2, work->target)) {
applog(LOG_INFO, "Share below target"); applog(LOG_INFO, "Share below target");
return; return;
} }
work_out = copy_work(work);
cgtime(&tv_work_found); submit_work_async(work_out);
submit_work_async(work, &tv_work_found);
} }
/* Returns true if nonce for work was a valid share */ /* Returns true if nonce for work was a valid share */
bool submit_nonce(struct thr_info *thr, struct work *work, uint32_t nonce) bool submit_nonce(struct thr_info *thr, struct work *work, uint32_t nonce)
{ {
bool ret = true;
if (test_nonce(work, nonce)) if (test_nonce(work, nonce))
submit_tested_work(thr, work); submit_tested_work(thr, work);
else { else {
applog(LOG_INFO, "%s%d: invalid nonce - HW error",
thr->cgpu->drv->name, thr->cgpu->device_id);
inc_hw_errors(thr); inc_hw_errors(thr);
ret = false; return false;
} }
return ret; return true;
} }
/* Allows drivers to submit work items where the driver has changed the ntime /* Allows drivers to submit work items where the driver has changed the ntime
* value by noffset. Must be only used with a work protocol that does not ntime * value by noffset. Must be only used with a work protocol that does not ntime
* roll itself intrinsically to generate work (eg stratum). */ * roll itself intrinsically to generate work (eg stratum). We do not touch
bool submit_noffset_nonce(struct thr_info *thr, struct work *work, uint32_t nonce, * the original work struct, but the copy of it only. */
bool submit_noffset_nonce(struct thr_info *thr, struct work *work_in, uint32_t nonce,
int noffset) int noffset)
{ {
unsigned char bin[4]; struct work *work = make_work();
uint32_t h32, *be32 = (uint32_t *)bin; bool ret = false;
hex2bin(bin, work->ntime, 4); _copy_work(work, work_in, noffset);
h32 = be32toh(*be32) + noffset; if (!test_nonce(work, nonce)) {
*be32 = htobe32(h32); inc_hw_errors(thr);
free(work->ntime); goto out;
work->ntime = bin2hex(bin, 4); }
ret = true;
update_work_stats(thr, work);
if (!fulltest(work->hash2, work->target)) {
applog(LOG_INFO, "Share below target");
goto out;
}
submit_work_async(work);
return submit_nonce(thr, work, nonce); out:
if (!ret)
free_work(work);
return ret;
} }
static inline bool abandon_work(struct work *work, struct timeval *wdiff, uint64_t hashes) static inline bool abandon_work(struct work *work, struct timeval *wdiff, uint64_t hashes)

1
miner.h

@ -1419,7 +1419,6 @@ extern void adl(void);
extern void app_restart(void); extern void app_restart(void);
extern void clean_work(struct work *work); extern void clean_work(struct work *work);
extern void free_work(struct work *work); extern void free_work(struct work *work);
extern void __copy_work(struct work *work, struct work *base_work);
extern struct work *copy_work(struct work *base_work); extern struct work *copy_work(struct work *base_work);
extern struct thr_info *get_thread(int thr_id); extern struct thr_info *get_thread(int thr_id);
extern struct cgpu_info *get_devices(int id); extern struct cgpu_info *get_devices(int id);

Loading…
Cancel
Save