[mythtv] MythSocket class

Jim Westfall jwestfall at surrealistic.net
Thu Apr 20 01:22:32 UTC 2006


John P Poet <jppoet at gmail.com> wrote [04.19.06]:
> On 4/19/06, Isaac Richards <ijr at case.edu> wrote:
> > On Wednesday 19 April 2006 6:50 pm, John P Poet wrote:
> > > On 4/9/06, Janne Grunau <janne-mythtv at grunau.be> wrote:
> > > > On Sunday 09 April 2006 22:02, Jim Westfall wrote:
> > > > > Thanks for testing.  I didnt actually touch the http server part of
> > > > > the backend, so its still using QSocket.  I left it as is cause I
> > > > > need to figure out a workaround to handle canReadLine/readLine
> > > > > functions that QSocket provides.
> > > >
> > > > Had just a very short look at the patch. The deadlock occured in the
> > > > read/writestringlist functions and only if some other backend activity
> > > > (probably the scheduler) happens together with the http-server. So
> > > > changing the other part using QSocket is enough fix it.
> > > >
> > > > > > But if a slave backends connects, no further connection is
> > > > > > possible. Have right now no time to debug this, sorry.
> > > > >
> > > > > I will try and look into that, if I can get together a box for a
> > > > > slave backend.  It may just be the hack I had to do on the backend to
> > > > > force it to process an incoming request before returning from the
> > > > > readyRead callback.  Before it would hand the socket off to a worker
> > > > > thread and return right away.
> > > >
> > > > I will have time tomorrow to look into it.
> > >
> > > Is this still pending?  I looked over the commits, but did not see it.
> >
> > I haven't had time to commit it.
> 
> 
> No problem.  I was just wondering if the "slave backend" issue had
> been resolved.  I never saw a patch go by that addressed Janne
> concerns.
> 
> I would be willing to test a patch, but I have a slave backend and
> would not want to try something which has already been reported to
> cause a problem.

I wasnt able to reproduce the slave backend issue in testing.  I have an 
updated patch I will try and get submitted this weekend.  It addresses 
the backend worker thread hack and expands the conversion to include 
mythcontext and a few other classes that use QSocket[Device].

jim


More information about the mythtv-dev mailing list