On Mon, 23 Oct 2017 12:31:53 +0200 Daniel Stone <[email protected]> wrote:
> Hi all, > > On 12 October 2017 at 13:13, Emre Ucan <[email protected]> wrote: > > + <interface name="weston_debug_v1" version="1"> > > + <description summary="weston internal debugging"> > > + This is a generic debugging interface for Weston internals, the > > global > > + object advertized through wl_registry. > > + > > + WARNING: This interface by design allows a denial-of-service attack. > > It > > + should not be offered in production, or proper authorization > > mechnisms > > + must be enforced. > > + > > + The idea is for a client to provide a file descriptor that the server > > + uses for printing debug information. The server uses the file > > + descriptor in blocking writes mode, which exposes the > > denial-of-service > > + risk. The blocking mode is necessary to ensure all debug messages can > > + be easily printed in place. It also ensures message ordering if a > > + client subcribes to more than one debug stream. > > This is pretty unpleasant indeed. My worry is that enabling the debug > interface could substantially change timing characteristics with > blocking writes, as the compositor waits on the clients. Wrapping > weston_debug_scope_write() or weston_debug_stream_write() with time > queries would at least allow us to emit a 'your client is making the > compositor slower' message. I fixed all my email addresses that had got mangled: https://gitlab.collabora.com/pq/weston/commits/weston-debug-v4.3 Thanks, pq
pgp_URhPK8UV1.pgp
Description: OpenPGP digital signature
_______________________________________________ wayland-devel mailing list [email protected] https://lists.freedesktop.org/mailman/listinfo/wayland-devel
