Browse Source

cuda: reduce possible segfaults on exit

not perfect but helps...
master
Tanguy Pruvot 10 years ago
parent
commit
85394cf22b
  1. 15
      ccminer.cpp
  2. 19
      cuda.cpp
  3. 1
      miner.h
  4. 2
      nvml.cpp
  5. 1
      scrypt/salsa_kernel.h

15
ccminer.cpp

@ -58,6 +58,7 @@ BOOL WINAPI ConsoleHandler(DWORD);
// from cuda.cpp // from cuda.cpp
int cuda_num_devices(); int cuda_num_devices();
void cuda_devicenames(); void cuda_devicenames();
void cuda_reset_device(int thr_id, bool *init);
void cuda_shutdown(); void cuda_shutdown();
int cuda_finddevice(char *name); int cuda_finddevice(char *name);
void cuda_print_devices(); void cuda_print_devices();
@ -211,7 +212,6 @@ int device_lookup_gap[MAX_GPUS] = { 0 };
int device_interactive[MAX_GPUS] = { 0 }; int device_interactive[MAX_GPUS] = { 0 };
int opt_nfactor = 0; int opt_nfactor = 0;
bool opt_autotune = true; bool opt_autotune = true;
bool abort_flag = false;
char *jane_params = NULL; char *jane_params = NULL;
// pools (failover/getwork infos) // pools (failover/getwork infos)
@ -243,6 +243,7 @@ int longpoll_thr_id = -1;
int stratum_thr_id = -1; int stratum_thr_id = -1;
int api_thr_id = -1; int api_thr_id = -1;
bool stratum_need_reset = false; bool stratum_need_reset = false;
volatile bool abort_flag = false;
struct work_restart *work_restart = NULL; struct work_restart *work_restart = NULL;
static int app_exit_code = EXIT_CODE_OK; static int app_exit_code = EXIT_CODE_OK;
uint32_t zr5_pok = 0; uint32_t zr5_pok = 0;
@ -1947,7 +1948,7 @@ static void *miner_thread(void *userdata)
/* record scanhash elapsed time */ /* record scanhash elapsed time */
gettimeofday(&tv_end, NULL); gettimeofday(&tv_end, NULL);
if (rc && opt_debug) if (rc > 0 && opt_debug)
applog(LOG_NOTICE, CL_CYN "found => %08x" CL_GRN " %08x", nonceptr[0], swab32(nonceptr[0])); // data[19] applog(LOG_NOTICE, CL_CYN "found => %08x" CL_GRN " %08x", nonceptr[0], swab32(nonceptr[0])); // data[19]
if (rc > 1 && opt_debug) if (rc > 1 && opt_debug)
applog(LOG_NOTICE, CL_CYN "found => %08x" CL_GRN " %08x", nonceptr[2], swab32(nonceptr[2])); // data[21] applog(LOG_NOTICE, CL_CYN "found => %08x" CL_GRN " %08x", nonceptr[2], swab32(nonceptr[2])); // data[21]
@ -1979,7 +1980,7 @@ static void *miner_thread(void *userdata)
if (rc > 1) if (rc > 1)
work.scanned_to = nonceptr[2]; work.scanned_to = nonceptr[2];
else if (rc) else if (rc > 0)
work.scanned_to = nonceptr[0]; work.scanned_to = nonceptr[0];
else { else {
work.scanned_to = max_nonce; work.scanned_to = max_nonce;
@ -1990,6 +1991,9 @@ static void *miner_thread(void *userdata)
} }
} }
if (abort_flag)
break; // time to leave the mining loop...
if (check_dups) if (check_dups)
hashlog_remember_scan_range(&work); hashlog_remember_scan_range(&work);
@ -2023,7 +2027,7 @@ static void *miner_thread(void *userdata)
firstwork_time = time(NULL); firstwork_time = time(NULL);
/* if nonce found, submit work */ /* if nonce found, submit work */
if (rc && !opt_benchmark) { if (rc > 0 && !opt_benchmark) {
if (!submit_work(mythr, &work)) if (!submit_work(mythr, &work))
break; break;
@ -3486,7 +3490,8 @@ int main(int argc, char *argv[])
if (hnvml) { if (hnvml) {
applog(LOG_INFO, "NVML GPU monitoring enabled."); applog(LOG_INFO, "NVML GPU monitoring enabled.");
for (int n=0; n < opt_n_threads; n++) { for (int n=0; n < opt_n_threads; n++) {
nvml_set_clocks(hnvml, device_map[n]); if (nvml_set_clocks(hnvml, device_map[n]) == 1)
cuda_reset_device(n, NULL);
} }
} }
#else #else

19
cuda.cpp

@ -1,6 +1,7 @@
#include <stdio.h> #include <stdio.h>
#include <memory.h> #include <memory.h>
#include <string.h> #include <string.h>
#include <unistd.h>
#include <map> #include <map>
#ifndef _WIN32 #ifndef _WIN32
@ -60,6 +61,7 @@ void cuda_devicenames()
exit(1); exit(1);
} }
GPU_N = min(MAX_GPUS, GPU_N);
for (int i=0; i < GPU_N; i++) for (int i=0; i < GPU_N; i++)
{ {
cudaDeviceProp props; cudaDeviceProp props;
@ -177,13 +179,20 @@ int cuda_gpu_clocks(struct cgpu_info *gpu)
void cuda_reset_device(int thr_id, bool *init) void cuda_reset_device(int thr_id, bool *init)
{ {
int dev_id = device_map[thr_id]; int dev_id = device_map[thr_id];
for (int i=0; i < MAX_GPUS; i++) { cudaSetDevice(dev_id);
if (device_map[i] == dev_id) { if (init != NULL) {
init[i] = false; // with init array, its meant to be used in algo's scan code...
for (int i=0; i < MAX_GPUS; i++) {
if (device_map[i] == dev_id) {
init[i] = false;
}
} }
// force exit from algo's scan loops/function
restart_threads();
cudaDeviceSynchronize();
while (cudaStreamQuery(NULL) == cudaErrorNotReady)
usleep(1000);
} }
restart_threads();
cudaDeviceSynchronize();
cudaDeviceReset(); cudaDeviceReset();
} }

1
miner.h

@ -490,6 +490,7 @@ extern struct thr_info *thr_info;
extern int longpoll_thr_id; extern int longpoll_thr_id;
extern int stratum_thr_id; extern int stratum_thr_id;
extern int api_thr_id; extern int api_thr_id;
extern volatile bool abort_flag;
extern struct work_restart *work_restart; extern struct work_restart *work_restart;
extern bool opt_trust_pool; extern bool opt_trust_pool;
extern uint16_t opt_vote; extern uint16_t opt_vote;

2
nvml.cpp

@ -342,7 +342,7 @@ int nvml_set_clocks(nvml_handle *nvmlh, int dev_id)
} }
gpu_clocks_changed[dev_id] = 1; gpu_clocks_changed[dev_id] = 1;
return 0; return 1;
} }
/* reset default app clocks to an used device */ /* reset default app clocks to an used device */

1
scrypt/salsa_kernel.h

@ -23,7 +23,6 @@ extern char *device_name[MAX_GPUS];
extern bool opt_autotune; extern bool opt_autotune;
extern int opt_nfactor; extern int opt_nfactor;
extern char *jane_params; extern char *jane_params;
extern bool abort_flag;
extern int parallel; extern int parallel;
extern void get_currentalgo(char* buf, int sz); extern void get_currentalgo(char* buf, int sz);

Loading…
Cancel
Save