Re: [PATCH spice-server 4/4] Provide and reuse default implementation for config_socket

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



> 
> On Wed, Feb 15, 2017 at 06:02:00AM -0500, Frediano Ziglio wrote:
> > > 
> > > On Tue, Feb 14, 2017 at 03:55:25PM +0100, Christophe Fergeau wrote:
> > > > I would not provide any default implementation, and just change
> > > > red_channel_config_socket to
> > > > 
> > > 
> > > Thinking more about this, I actually would provide a default
> > > implementation, and move the setsockopt/fnctl code from
> > > reds_init_client_connection() in the default implementation, and make
> > > sure the child classes properly chain up to it (or call it directly if
> > > chaining up is complicated)
> > > 
> > > Christophe
> > > 
> > 
> > This would introduce possible race conditions.
> > On reds.c the sockets are initialized before any usage.
> > Setting not blocking is quite mandatory to avoid some races using
> > some layers (like TLS). In this case before config_socket there
> > are the headers to tell which channel is this. A client could use
> > the race to cause the Qemu main thread to stop waiting for data
> > (potentially this could happen without authentication).
> 
> Does this include TCP_NODELAY as well?
> 
> Christophe
> 

No, just slows down connection a bit.

Frediano
_______________________________________________
Spice-devel mailing list
Spice-devel@xxxxxxxxxxxxxxxxxxxxx
https://lists.freedesktop.org/mailman/listinfo/spice-devel




[Index of Archives]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [IETF Annouce]     [Security]     [Bugtraq]     [Linux]     [Linux OMAP]     [Linux MIPS]     [ECOS]     [Asterisk Internet PBX]     [Linux API]     [Monitors]