Re: [Qemu-devel] [PATCH v1 4/7] io: pass a struct iovec into qio_channel_websock_encode

[Daniel P. Berrange]

On Tue, Oct 10, 2017 at 12:18:26PM -0500, Eric Blake wrote:
> On 10/10/2017 10:43 AM, Daniel P. Berrange wrote:
> > Instead of requiring use of another Buffer, pass a struct iovec
> > into qio_channel_websock_encode, which gives callers more
> > flexibility in how they process data.
> > 
> > Signed-off-by: Daniel P. Berrange <address@hidden>
> > ---
> >  io/channel-websock.c | 69 
> > ++++++++++++++++++++++++++++++++--------------------
> >  1 file changed, 42 insertions(+), 27 deletions(-)
> > 
> 
> > +static void qio_channel_websock_encode(QIOChannelWebsock *ioc,
> > +                                       uint8_t opcode,
> > +                                       const struct iovec *iov,
> > +                                       size_t niov,
> > +                                       size_t size)
> >  {
> 
> Is size redundant with iov_size(iov, niov)?  Or is a caller allowed to
> pass a smaller size (tail end of iov is not encoded) or larger size
> (encoding stops at end of iov, even if size is not exhausted)?  I'd lean
> towards the former (one less parameter), especially since all callers in
> this patch could have passed iov_size(&iov, 1) for the same effect.

You've already noticed the code in later patch which passes a smaller
size. This lets me avoid having to duplicate the iovec array and
set smaller iov_len

> 
> > -    trace_qio_channel_websock_encode(ioc, opcode, header_size, 
> > buffer->offset);
> > -    buffer_reserve(output, header_size + buffer->offset);
> > -    buffer_append(output, header.buf, header_size);
> > -    buffer_append(output, buffer->buffer, buffer->offset);
> > +    trace_qio_channel_websock_encode(ioc, opcode, header_size, size);
> > +    buffer_reserve(&ioc->encoutput, header_size + size);
> > +    buffer_append(&ioc->encoutput, header.buf, header_size);
> > +    for (i = 0; i < niov && size != 0; i++) {
> > +        size_t want = iov->iov_len;
> > +        if (want > size) {
> > +            want = size;
> > +        }
> > +        buffer_append(&ioc->encoutput, iov->iov_base, want);
> > +        size -= want;
> > +    }
> 
> Umm, where are you incrementing iov? It appears you only tested with
> niov == 1.

Opps. Yeah, we need to cope with niov > 0 to satisfy qio_channel_writev
API contract, but the VNC server only ever sends a single iov element
so I didn't hit the bug.

Regards,
Daniel
-- 
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|