From: Wei Liu <wei.liu2@citrix.com>
To: Venu Busireddy <venu.busireddy@oracle.com>
Cc: Wei Liu <wei.liu2@citrix.com>,
Andrew Cooper <andrew.cooper3@citrix.com>,
Ian Jackson <ian.jackson@eu.citrix.com>,
xen-devel@lists.xen.org, Jan Beulich <jbeulich@suse.com>
Subject: Re: [PATCH v2 1/2] libxl: Implement the handler to handle unrecoverable AER errors.
Date: Mon, 31 Jul 2017 16:03:36 +0100 [thread overview]
Message-ID: <20170731150336.f25ksvarhsoionzz@citrix.com> (raw)
In-Reply-To: <20170728235656.GA49069@fett.us.oracle.com>
On Fri, Jul 28, 2017 at 04:56:56PM -0700, Venu Busireddy wrote:
> On 2017-07-28 17:39:52 +0100, Ian Jackson wrote:
> > Venu Busireddy writes ("[PATCH v2 1/2] libxl: Implement the handler to handle unrecoverable AER errors."):
> > > Implement the callback function to handle unrecoverable AER errors, and
> > > also the public APIs that can be used to register/unregister the handler.
> > > When an AER error occurs, the handler will forcibly remove the erring
> > > PCIe device from the guest.
> >
> > Why is this only sometimes the right thing to do ? On what basis
> > might a user choose ?
>
> This is not an "only sometimes" thing. User doesn't choose it. We always
> want to watch for AER errors.
>
> > If this is always the right thing to do then maybe we need to recast
> > this as a general "please run monitoring for this domain" call ?
>
> What does "recast" imply? Which "call" are you referring to?
>
> > > +int libxl_reg_aer_events_handler(libxl_ctx *, uint32_t) LIBXL_EXTERNAL_CALLERS_ONLY;
> > > +void libxl_unreg_aer_events_handler(libxl_ctx *, uint32_t) LIBXL_EXTERNAL_CALLERS_ONLY;
> >
> > I think these names are very unintuitive. They describe the
> > implementation, not the effect.
>
> These names can be changed to anything you want. Please suggest any names
> of your choice, and I will change them. That ensures that we don't spend
> more review revisions in fine tuning those names.
>
> > The API seems awkward. Inside libxl, events are only processed while
> > the application is inside libxl. So for these functions to be
> > effective, the calling application must arrange to be running the
> > libxl event loop. This should be documented, at least.
>
> I am afraid I don't follow you. This scheme is tested and it works. So, I
> do not follow you when you say, "...for these functions to be effective,..."
Libxl has an internal event loop. The code as-is registers a watch which
runs on the internal event loop. The event loop's event is only
processed when the process enters libxl (calls libxl functions).
Imagine a toolstack which doesn't use libxl's internal event loop -- for
example libvirt has its own event loop, or a toolstack which only calls
the register function then nothing else. Your API would not work for
those cases.
Ian, please correct me if I'm wrong.
>
> > What happens if more than one process calls this at once ?
>
> Each call is handled within a separate 'xl' process's context. I don't
> see a problem there. Do you have anything specific in mind?
>
It is possible that both xl processes see its watch fires and try to
write to the same node at the same time.
_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel
next prev parent reply other threads:[~2017-07-31 15:03 UTC|newest]
Thread overview: 12+ messages / expand[flat|nested] mbox.gz Atom feed top
2017-07-27 0:16 [PATCH v2 0/2] Containing AER unrecoverable errors Venu Busireddy
2017-07-27 0:16 ` [PATCH v2 1/2] libxl: Implement the handler to handle unrecoverable AER errors Venu Busireddy
2017-07-28 15:58 ` Wei Liu
2017-07-28 17:15 ` Venu Busireddy
2017-07-31 14:37 ` Wei Liu
2017-07-28 16:39 ` Ian Jackson
2017-07-28 23:56 ` Venu Busireddy
2017-07-31 15:03 ` Wei Liu [this message]
2017-07-27 0:16 ` [PATCH v2 2/2] xl: Register the AER event handler to handle " Venu Busireddy
2017-07-28 15:58 ` Wei Liu
2017-07-28 16:06 ` Venu Busireddy
2017-07-28 16:14 ` Wei Liu
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20170731150336.f25ksvarhsoionzz@citrix.com \
--to=wei.liu2@citrix.com \
--cc=andrew.cooper3@citrix.com \
--cc=ian.jackson@eu.citrix.com \
--cc=jbeulich@suse.com \
--cc=venu.busireddy@oracle.com \
--cc=xen-devel@lists.xen.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.