Re: [PATCH v4 2/2] migration: check magic value for deciding the mapping of channels

[Peter Xu]

On Wed, Nov 23, 2022 at 09:28:14PM +0530, manish.mishra wrote:
> 
> On 23/11/22 9:22 pm, Peter Xu wrote:
> > On Wed, Nov 23, 2022 at 03:05:27PM +0000, manish.mishra wrote:
> > > +int migration_channel_read_peek(QIOChannel *ioc,
> > > +                                const char *buf,
> > > +                                const size_t buflen,
> > > +                                Error **errp)
> > > +{
> > > +   ssize_t len = 0;
> > > +   struct iovec iov = { .iov_base = (char *)buf, .iov_len = buflen };
> > > +
> > > +   while (len < buflen) {
> > > +       len = qio_channel_readv_full(ioc, &iov, 1, NULL,
> > > +                                    NULL, 
> > > QIO_CHANNEL_READ_FLAG_MSG_PEEK, errp);
> > > +
> > > +       if (len == QIO_CHANNEL_ERR_BLOCK) {
> > This needs to take care of partial len too?
> 
> 
> sorry Peter, I did not quite understand it. Can you please give some more 
> details.

As Daniel pointed out, I think if we peek partially it'll go into a loop.
Since we shouldn't read 0 anyway here, maybe you can directly write it as:

while (true) {
  len = read();
  if (len <= 0 && len != QIO_CHANNEL_ERR_BLOCK) {
      error_setg(...);
      return -1;
  }
  if (len == buflen) {
      break;
  }
  /* For either partial peek or QIO_CHANNEL_ERR_BLOCK, retry with timeout */
  sleep();
}

> 
> > 
> > > +            if (qemu_in_coroutine()) {
> > > +                /* 1ms sleep. */
> > > +                qemu_co_sleep_ns(QEMU_CLOCK_REALTIME, 1000000);
> > > +            } else {
> > > +                qio_channel_wait(ioc, G_IO_IN);
> > > +            }
> > > +            continue;
> > > +       }
> > > +       if (len == 0) {
> > > +           error_setg(errp,
> > > +                      "Unexpected end-of-file on channel");
> > > +           return -1;
> > > +       }
> > > +       if (len < 0) {
> > > +           return -1;
> > > +       }
> > > +   }
> > > +
> > > +   return 0;
> > > +}
> 
> Thanks
> 
> Manish Mishra
> 

-- 
Peter Xu