Re: [PATCH] xenstored: close socket connections on error

[Ian Jackson <iwj@xxxxxxxxxxxxxx>]

Ian Jackson writes ("Re: [PATCH] xenstored: close socket connections on error"):
> Manuel Bouyer writes ("[PATCH] xenstored: close socket connections on error"):
> > On error, don't keep socket connection in ignored state but close them.
> > When the remote end of a socket is closed, xenstored will flag it as an
> > error and switch the connection to ignored. But on some OSes (e.g.
> > NetBSD), poll(2) will return only POLLIN in this case, so sockets in ignored
> > state will stay open forever in xenstored (and it will loop with CPU 100%
> > busy).
> 
> Juergen, I think you probably know this code the best.  Would you be
> able to review this ?
> 
> I'm not sure I understand what the specific behaviour on NetBSD is
> that is upsetting xenstored.  Or rather, what it is that xenstored is
> using to tell when the socket should be closed.
> 
> I grepped for POLLERR and nothing came up.
> 
> Or to put it another way, is this commit
> 
> > Fixes: d2fa370d3ef9cbe22d7256c608671cdcdf6e0083
> 
> broken on Linux too ?

Andy pointed me to the recent thread "xenstored file descriptor leak"
which answers all these questions.  I think it would have been nice if
some tools maintainer(s) had been CC'd on that :-).

Juergen, I guess I will get a formal R-b from you ?

Release-Acked-by: Ian Jackson <iwj@xxxxxxxxxxxxxx>


Manuel, in response to this:

> When I started, I looked at the wiki for instructions about
> patches, but didn't find any ...

Earlier I offered you help with git, in private email.  I agree that
git is confusing and sometimes impenetrable.  But it seems that what
you are doing now is worse!  Please take me up on my offer of help.

Our wiki doesn't give instructions on how to use git to maintain a
patch series.  Those instructions would not be Xen-specific.  Perhaps
we could have a pointer or two, but everyone has their own pet methods
and tooling so the result would perhaps be more confusing than
helpful.

Regards,
Ian.