On Sun, Aug 09, 2020 at 11:51:45AM -0600, Jens Axboe wrote:
> On 8/9/20 12:30 AM, Zorro Lang wrote:
> > @@ -2170,6 +2189,108 @@ do_aio_rw(int opno, long r, int flags)
> > }
> > #endif
> >
> > +#ifdef URING
> > +void
> > +do_uring_rw(int opno, long r, int flags)
> > +{
> > + char *buf;
> > + int e;
> > + pathname_t f;
> > + int fd;
> > + size_t len;
> > + int64_t lr;
> > + off64_t off;
> > + struct stat64 stb;
> > + int v;
> > + char st[1024];
> > + struct io_uring_sqe *sqe;
> > + struct io_uring_cqe *cqe;
> > + struct iovec iovec;
> > + int iswrite = (flags & (O_WRONLY | O_RDWR)) ? 1 : 0;
> > +
> > + init_pathname(&f);
> > + if (!get_fname(FT_REGFILE, r, &f, NULL, NULL, &v)) {
> > + if (v)
> > + printf("%d/%d: do_uring_rw - no filename\n", procid, opno);
> > + goto uring_out3;
> > + }
> > + fd = open_path(&f, flags);
> > + e = fd < 0 ? errno : 0;
> > + check_cwd();
> > + if (fd < 0) {
> > + if (v)
> > + printf("%d/%d: do_uring_rw - open %s failed %d\n",
> > + procid, opno, f.path, e);
> > + goto uring_out3;
> > + }
> > + if (fstat64(fd, &stb) < 0) {
> > + if (v)
> > + printf("%d/%d: do_uring_rw - fstat64 %s failed %d\n",
> > + procid, opno, f.path, errno);
> > + goto uring_out2;
> > + }
> > + inode_info(st, sizeof(st), &stb, v);
> > + if (!iswrite && stb.st_size == 0) {
> > + if (v)
> > + printf("%d/%d: do_uring_rw - %s%s zero size\n", procid, opno,
> > + f.path, st);
> > + goto uring_out2;
> > + }
> > + sqe = io_uring_get_sqe(&ring);
> > + if (!sqe) {
> > + if (v)
> > + printf("%d/%d: do_uring_rw - io_uring_get_sqe failed\n",
> > + procid, opno);
> > + goto uring_out2;
> > + }
> > + lr = ((int64_t)random() << 32) + random();
> > + len = (random() % FILELEN_MAX) + 1;
> > + buf = malloc(len);
> > + if (!buf) {
> > + if (v)
> > + printf("%d/%d: do_uring_rw - malloc failed\n",
> > + procid, opno);
> > + goto uring_out2;
> > + }
> > + iovec.iov_base = buf;
> > + iovec.iov_len = len;
> > + if (iswrite) {
> > + off = (off64_t)(lr % MIN(stb.st_size + (1024 * 1024), MAXFSIZE));
> > + off %= maxfsize;
> > + memset(buf, nameseq & 0xff, len);
> > + io_uring_prep_writev(sqe, fd, &iovec, 1, off);
> > + } else {
> > + off = (off64_t)(lr % stb.st_size);
> > + io_uring_prep_readv(sqe, fd, &iovec, 1, off);
> > + }
> > +
> > + if ((e = io_uring_submit(&ring)) != 1) {
> > + if (v)
> > + printf("%d/%d: %s - io_uring_submit failed %d\n", procid, opno,
> > + iswrite ? "uring_write" : "uring_read", e);
> > + goto uring_out1;
> > + }
> > + if ((e = io_uring_wait_cqe(&ring, &cqe)) < 0) {
> > + if (v)
> > + printf("%d/%d: %s - io_uring_wait_cqe failed %d\n", procid, opno,
> > + iswrite ? "uring_write" : "uring_read", e);
> > + goto uring_out1;
> > + }
>
> You could use io_uring_submit_and_wait() here, that'll save a system
> call for sync IO. Same comment goes for 4/4.
Hi Jens,
Sorry I think I haven't learned about io_uring enough, why the
io_uring_submit_and_wait can save a system call? Is it same with
io_uring_submit(), except a wait_nr ? The io_uring_wait_cqe() and
io_uring_cqe_seen() are still needed, right?
Thanks,
Zorro
>
> Apart from that, looks pretty straight forward to me.
>
> --
> Jens Axboe
>
