Splice vs memfd

0_benchmark.md

$ gcc -o main main.c
$ ./main
splice with discard took about 0.12762 seconds
splice with full discard took about 0.15904 seconds
normal splice took about 0.02920 seconds
sendfile from memfd took about 0.15539 seconds

main.c

#define _GNU_SOURCE
#include <unistd.h>
#include <sys/syscall.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <stdint.h>
#include <fcntl.h>
#include <stdio.h>
#include <time.h>
#include <sys/sendfile.h>
#include <sys/syscall.h>
#include <stdlib.h>

#  ifndef __NR_memfd_create
#    if defined __x86_64__
#      define __NR_memfd_create 319
#    elif defined __arm__
#      define __NR_memfd_create 385
#    elif defined __aarch64__
#      define __NR_memfd_create 279
#    elif defined __s390__
#      define __NR_memfd_create 350
#    elif defined _MIPS_SIM
#      if _MIPS_SIM == _MIPS_SIM_ABI32
#        define __NR_memfd_create 4354
#      endif
#      if _MIPS_SIM == _MIPS_SIM_NABI32
#        define __NR_memfd_create 6318
#      endif
#      if _MIPS_SIM == _MIPS_SIM_ABI64
#        define __NR_memfd_create 5314
#      endif
#    elif defined __i386__
#      define __NR_memfd_create 356
#    elif defined __arc__
#      define __NR_memfd_create 279
#    else
#      warning "__NR_memfd_create unknown for your architecture"
#    endif
#  endif

static inline int memfd_create(const char *name, unsigned int flags) {
#  ifdef __NR_memfd_create
            return syscall(__NR_memfd_create, name, flags);
#  else
                    errno = ENOSYS;
                            return -1;
#  endif
}

struct fuse_out_header {
  uint32_t  len;
  int32_t   error;
  uint64_t  unique;
};

struct fuse_write_out {
  uint32_t  size;
  uint32_t  padding;
};

int main() {
  int pipefd[2], pipefd2[2];
  pipe(&pipefd[0]);
  pipe(&pipefd2[0]);
  int fd = open("/run/user/1000/a", O_CREAT|O_TRUNC|O_RDWR, 0755);

  const int size = 1024 * 4;
  ftruncate(fd, size);
  fcntl(pipefd[0], F_SETPIPE_SZ, size + 4096);
  fcntl(pipefd2[0], F_SETPIPE_SZ, size + 4096);

  struct timespec tstart={0,0}, tend={0,0};
  clock_gettime(CLOCK_MONOTONIC, &tstart);
  uint8_t buf[sizeof(struct fuse_out_header) + sizeof(struct fuse_write_out)];
  loff_t offset = 0;
  for (int i = 0; i < 99999; i++) {
    offset = 0;
    splice(fd, &offset, pipefd[1], NULL, size, SPLICE_F_MOVE|SPLICE_F_NONBLOCK);

    write(pipefd2[1], buf, sizeof(buf));
    read(pipefd[0], buf, sizeof(buf));
    splice(pipefd[0], NULL, pipefd2[1], NULL, size - sizeof(buf), SPLICE_F_MOVE|SPLICE_F_NONBLOCK);

    offset = 0;
    splice(pipefd2[0], NULL, fd, &offset, size, SPLICE_F_MOVE|SPLICE_F_NONBLOCK);
  }

  clock_gettime(CLOCK_MONOTONIC, &tend);
  printf("splice with discard took about %.5f seconds\n",
      ((double)tend.tv_sec + 1.0e-9*tend.tv_nsec) -
      ((double)tstart.tv_sec + 1.0e-9*tstart.tv_nsec));

  clock_gettime(CLOCK_MONOTONIC, &tstart);
  uint8_t *buf2 = (uint8_t*)malloc(size);
  for (int i = 0; i < 99999; i++) {
    offset = 0;
    splice(fd, &offset, pipefd[1], NULL, size, SPLICE_F_MOVE|SPLICE_F_NONBLOCK);

    read(pipefd[0], buf2, size);
    write(fd, buf2, size);
  }
  free(buf2);
  clock_gettime(CLOCK_MONOTONIC, &tend);
  printf("splice with full discard took about %.5f seconds\n",
      ((double)tend.tv_sec + 1.0e-9*tend.tv_nsec) -
      ((double)tstart.tv_sec + 1.0e-9*tstart.tv_nsec));

  clock_gettime(CLOCK_MONOTONIC, &tstart);
  for (int i = 0; i < 99999; i++) {
    splice(fd, NULL, pipefd[1], NULL, size, SPLICE_F_MOVE|SPLICE_F_NONBLOCK);
    splice(pipefd[0], NULL, fd, NULL, size, SPLICE_F_MOVE|SPLICE_F_NONBLOCK);
  }
  clock_gettime(CLOCK_MONOTONIC, &tend);
  printf("normal splice took about %.5f seconds\n",
      ((double)tend.tv_sec + 1.0e-9*tend.tv_nsec) -
      ((double)tstart.tv_sec + 1.0e-9*tstart.tv_nsec));

  close(pipefd[0]);
  close(pipefd[1]);

  int tmpfd = memfd_create("test", 0);

  clock_gettime(CLOCK_MONOTONIC, &tstart);
  for (int i = 0; i < 99999; i++) {
    loff_t offset = 0;
    sendfile(fd, tmpfd, &offset, size);
    lseek(tmpfd, 0, SEEK_SET);
    sendfile(tmpfd, fd, &offset, size);
    lseek(fd, 0, SEEK_SET);
  }
  clock_gettime(CLOCK_MONOTONIC, &tend);
  printf("sendfile from memfd took about %.5f seconds\n",
      ((double)tend.tv_sec + 1.0e-9*tend.tv_nsec) -
      ((double)tstart.tv_sec + 1.0e-9*tstart.tv_nsec));
}