algo-gate-api.c

/////////////////////////////
////
////    NEW FEATURE: algo_gate
////
////    algos define targets for their common functions
////    and define a function for miner-thread to call to register
////    their targets. miner thread builds the gate, and array of structs
////    of function pointers, by calling each algo's register function.
//   Functions in this file are used simultaneously by myultiple
//   threads and must therefore be re-entrant.

#include "algo-gate-api.h"
#include <memory.h>
#include <stdbool.h>
#include <stdint.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>

// Define null and standard functions.
//
// Generic null functions do nothing except satisfy the syntax and
// can be used for optional safe gate functions.
//
// null gate functions are genarally used for mandatory and unsafe functions
// and will usually display an error massage and/or return a fail code.
// They are registered by default and are expected to be overwritten.
//
// std functions are non-null functions used by the most number of algos
// are are default.
//
// aux functions are functions used by many, but not most, algos and must
// be registered by eech algo using them. They usually have descriptive
// names.
//
// custom functions are algo spefic and are defined and registered in the
// algo's source file and are usually named [algo]_[function].
//
// In most cases the default is a null or std function. However in some
// cases, for convenience when the null function is not the most popular,
// the std function will be defined as default and the algo must register
// an appropriate null function.
//
// similar algos may share a gate function that may be defined here or
// in a source file common to the similar algos.
//
// gate functions may call other gate functions under the following
// restrictions. Any gate function defined here or used by more than one
// algo must call other functions using the gate: algo_gate.[function].
// custom functions may call other custom functions directly using
// [algo]_[function], howver it is recommended to alway use the gate.
//
// If, under rare circumstances, an algo with a custom gate function
// needs to call a function of another algo it must define and register
// a private gate from its rgistration function and use it to call
// forein functions: [private_gate].[function]. If the algo needs to call
// a utility function defined here it may do so directly.
//
// The algo's gate registration function is caled once from the main thread
// and can do other intialization in addition such as setting options or
// other global or local (to the algo) variables.

// A set of predefined generic null functions that can be used as any null
// gate function with the same signature.

void do_nothing() {}
bool return_true() { return true; }
bool return_false() { return false; }
void *return_null() { return NULL; }
void call_error() { printf("ERR: Uninitialized function pointer\n"); }

void algo_not_tested() {
  applog(LOG_WARNING, "Algo %s has not been tested live. It may not work",
         algo_names[opt_algo]);
  applog(LOG_WARNING, "and bad things may happen. Use at your own risk.");
}

void four_way_not_tested() {
  applog(LOG_WARNING, "Algo %s has not been tested using 4way. It may not",
         algo_names[opt_algo]);
  applog(LOG_WARNING, "work or may be slower. Please report your results.");
}

void algo_not_implemented() {
  applog(LOG_ERR, "Algo %s has not been Implemented.", algo_names[opt_algo]);
}

// default null functions
// deprecated, use generic as default
int null_scanhash() {
  applog(LOG_WARNING, "SWERR: undefined scanhash function in algo_gate");
  return 0;
}

// Default generic scanhash can be used in many cases.
int scanhash_generic(struct work *work, uint32_t max_nonce,
                     uint64_t *hashes_done, struct thr_info *mythr) {
  uint32_t edata[20] __attribute__((aligned(64)));
  uint32_t hash[8] __attribute__((aligned(64)));
  uint32_t *pdata = work->data;
  uint32_t *ptarget = work->target;
  const uint32_t first_nonce = pdata[19];
  const uint32_t last_nonce = max_nonce - 1;
  uint32_t n = first_nonce;
  const int thr_id = mythr->id;
  const bool bench = opt_benchmark;

  mm128_bswap32_80(edata, pdata);
  do {
    edata[19] = n;
    if (likely(algo_gate.hash(hash, edata, thr_id)))
      if (unlikely(valid_hash(hash, ptarget) && !bench)) {
        pdata[19] = bswap_32(n);
        submit_solution(work, hash, mythr);
      }
    n++;
  } while (n < last_nonce && !work_restart[thr_id].restart);
  *hashes_done = n - first_nonce;
  pdata[19] = n;
  return 0;
}

#if defined(__AVX2__)

// int scanhash_4way_64_64( struct work *work, uint32_t max_nonce,
//                      uint64_t *hashes_done, struct thr_info *mythr )

// int scanhash_4way_64_640( struct work *work, uint32_t max_nonce,
//                      uint64_t *hashes_done, struct thr_info *mythr )

int scanhash_4way_64in_32out(struct work *work, uint32_t max_nonce,
                             uint64_t *hashes_done, struct thr_info *mythr) {
  uint32_t hash32[8 * 4] __attribute__((aligned(64)));
  uint32_t vdata[20 * 4] __attribute__((aligned(64)));
  uint32_t lane_hash[8] __attribute__((aligned(64)));
  uint32_t *hash32_d7 = &(hash32[7 * 4]);
  uint32_t *pdata = work->data;
  const uint32_t *ptarget = work->target;
  const uint32_t first_nonce = pdata[19];
  const uint32_t last_nonce = max_nonce - 4;
  __m256i *noncev = (__m256i *)vdata + 9;
  uint32_t n = first_nonce;
  const int thr_id = mythr->id;
  const uint32_t targ32_d7 = ptarget[7];
  const bool bench = opt_benchmark;

  mm256_bswap32_intrlv80_4x64(vdata, pdata);
  *noncev = mm256_intrlv_blend_32(
      _mm256_set_epi32(n + 3, 0, n + 2, 0, n + 1, 0, n, 0), *noncev);
  do {
    if (likely(algo_gate.hash(hash32, vdata, thr_id)))
      for (int lane = 0; lane < 4; lane++)
        if (unlikely(hash32_d7[lane] <= targ32_d7 && !bench)) {
          extr_lane_4x32(lane_hash, hash32, lane, 256);
          if (valid_hash(lane_hash, ptarget)) {
            pdata[19] = bswap_32(n + lane);
            submit_solution(work, lane_hash, mythr);
          }
        }
    *noncev = _mm256_add_epi32(*noncev, m256_const1_64(0x0000000400000000));
    n += 4;
  } while (likely((n <= last_nonce) && !work_restart[thr_id].restart));
  pdata[19] = n;
  *hashes_done = n - first_nonce;
  return 0;
}

// int scanhash_8way_32_32( struct work *work, uint32_t max_nonce,
//                      uint64_t *hashes_done, struct thr_info *mythr )

#endif

#if defined(__AVX512F__) && defined(__AVX512VL__) && defined(__AVX512DQ__) &&  \
    defined(__AVX512BW__)

// int scanhash_8way_64_64( struct work *work, uint32_t max_nonce,
//                      uint64_t *hashes_done, struct thr_info *mythr )

// int scanhash_8way_64_640( struct work *work, uint32_t max_nonce,
//                      uint64_t *hashes_done, struct thr_info *mythr )

int scanhash_8way_64in_32out(struct work *work, uint32_t max_nonce,
                             uint64_t *hashes_done, struct thr_info *mythr) {
  uint32_t hash32[8 * 8] __attribute__((aligned(128)));
  uint32_t vdata[20 * 8] __attribute__((aligned(64)));
  uint32_t lane_hash[8] __attribute__((aligned(64)));
  uint32_t *hash32_d7 = &(hash32[7 * 8]);
  uint32_t *pdata = work->data;
  const uint32_t *ptarget = work->target;
  const uint32_t first_nonce = pdata[19];
  const uint32_t last_nonce = max_nonce - 8;
  __m512i *noncev = (__m512i *)vdata + 9;
  uint32_t n = first_nonce;
  const int thr_id = mythr->id;
  const uint32_t targ32_d7 = ptarget[7];
  const bool bench = opt_benchmark;

  mm512_bswap32_intrlv80_8x64(vdata, pdata);
  *noncev = mm512_intrlv_blend_32(_mm512_set_epi32(n + 7, 0, n + 6, 0, n + 5, 0,
                                                   n + 4, 0, n + 3, 0, n + 2, 0,
                                                   n + 1, 0, n, 0),
                                  *noncev);
  do {
    if (likely(algo_gate.hash(hash32, vdata, thr_id)))
      for (int lane = 0; lane < 8; lane++)
        if (unlikely((hash32_d7[lane] <= targ32_d7) && !bench)) {
          extr_lane_8x32(lane_hash, hash32, lane, 256);
          if (likely(valid_hash(lane_hash, ptarget))) {
            pdata[19] = bswap_32(n + lane);
            submit_solution(work, lane_hash, mythr);
          }
        }
    *noncev = _mm512_add_epi32(*noncev, m512_const1_64(0x0000000800000000));
    n += 8;
  } while (likely((n < last_nonce) && !work_restart[thr_id].restart));
  pdata[19] = n;
  *hashes_done = n - first_nonce;
  return 0;
}

// int scanhash_16way_32_32( struct work *work, uint32_t max_nonce,
//                      uint64_t *hashes_done, struct thr_info *mythr )

#endif

int null_hash() {
  applog(LOG_WARNING, "SWERR: null_hash unsafe null function");
  return 0;
};

void init_algo_gate(algo_gate_t *gate) {
  gate->miner_thread_init = (void *)&return_true;
  gate->scanhash = (void *)&scanhash_generic;
  gate->hash = (void *)&null_hash;
  gate->get_new_work = (void *)&std_get_new_work;
  gate->work_decode = (void *)&std_le_work_decode;
  gate->decode_extra_data = (void *)&do_nothing;
  gate->gen_merkle_root = (void *)&sha256d_gen_merkle_root;
  gate->build_stratum_request = (void *)&std_le_build_stratum_request;
  gate->malloc_txs_request = (void *)&std_malloc_txs_request;
  gate->submit_getwork_result = (void *)&std_le_submit_getwork_result;
  gate->build_block_header = (void *)&std_build_block_header;
  gate->build_extraheader = (void *)&std_build_extraheader;
  gate->set_work_data_endian = (void *)&do_nothing;
  gate->calc_network_diff = (void *)&std_calc_network_diff;
  gate->ready_to_mine = (void *)&std_ready_to_mine;
  gate->resync_threads = (void *)&do_nothing;
  gate->do_this_thread = (void *)&return_true;
  gate->longpoll_rpc_call = (void *)&std_longpoll_rpc_call;
  gate->get_work_data_size = (void *)&std_get_work_data_size;
  gate->optimizations = EMPTY_SET;
  gate->ntime_index = STD_NTIME_INDEX;
  gate->nbits_index = STD_NBITS_INDEX;
  gate->nonce_index = STD_NONCE_INDEX;
  gate->work_cmp_size = STD_WORK_CMP_SIZE;
}

// Ignore warnings for not yet defined register functions
#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wimplicit-function-declaration"

// Called once by main
bool register_algo_gate(int algo, algo_gate_t *gate) {
  bool rc = false;

  if (NULL == gate) {
    applog(LOG_ERR, "FAIL: algo_gate registration failed, NULL gate\n");
    return false;
  }

  init_algo_gate(gate);

  switch (algo) {
  case ALGO_GR:
    rc = register_gr_algo(gate);
    break;
  default:
    applog(LOG_ERR, "BUG: unregistered algorithm %s.\n", algo_names[opt_algo]);
    return false;
  } // switch

  if (!rc) {
    applog(LOG_ERR, "FAIL: %s algorithm failed to initialize\n",
           algo_names[opt_algo]);
    return false;
  }
  return true;
}

// restore warnings
#pragma GCC diagnostic pop

void exec_hash_function(int algo, void *output, const void *pdata) {
  algo_gate_t gate;
  gate.hash = (void *)&null_hash;
  register_algo_gate(algo, &gate);
  gate.hash(output, pdata, 0);
}

#define PROPER (1)
#define ALIAS (0)

// The only difference between the alias and the proper algo name is the
// proper name is the one that is defined in ALGO_NAMES. There may be
// multiple aliases that map to the same proper name.
// New aliases can be added anywhere in the array as long as NULL is last.
// Alphabetic order of alias is recommended.
const char *const algo_alias_map[][2] = {
    //   alias                proper
    {"grhash", "gr"},
    {NULL, NULL}};

// if arg is a valid alias for a known algo it is updated with the proper
// name. No validation of the algo or alias is done, It is the responsinility
// of the calling function to validate the algo after return.
void get_algo_alias(char **algo_or_alias) {
  int i;
  for (i = 0; algo_alias_map[i][ALIAS]; i++)
    if (!strcasecmp(*algo_or_alias, algo_alias_map[i][ALIAS])) {
      // found valid alias, return proper name
      *algo_or_alias = (char *)(algo_alias_map[i][PROPER]);
      return;
    }
}

#undef ALIAS
#undef PROPER