On Fri, Mar 17, 2017 at 16:10:35 +0000, Daniel Berrange wrote:
> On Fri, Mar 17, 2017 at 04:48:49PM +0100, Peter Krempa wrote:
> > In some cases a read error due to connection hangup is expected. This
> > patch adds a flag that removes the logging of a virError in such case.
> > ---
> > src/rpc/virnetsocket.c | 33 +++++++++++++++++++++++++++------
> > src/rpc/virnetsocket.h | 2 ++
> > 2 files changed, 29 insertions(+), 6 deletions(-)
[...]
> > + } else {
> > + if (errout)
> > + virReportSystemError(EIO,
> > + _("End of file while reading data: %s"),
> > + errout);
> > + else
> > + virReportSystemError(EIO, "%s",
> > + _("End of file while reading data"));
> > + }
> > ret = -1;
>
> I'm a little uncomfortable with the idea of returning '-1' without reporting
> an error message. I would suggest returning 0, but we used that to indicate
> EAGAIN condition. Can we at least make it return '-2' as a distinct code
> when we don't report errors.
That is fine by me. The callers that will ignore the error don't really
care why the read failed currently.
Attachment:
signature.asc
Description: PGP signature
-- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list
