[OpenAFS-devel] Re: Change in openafs[master]: cmdebug -dcache

Jeffrey Hutzelman jhutz@cmu.edu
Thu, 03 Feb 2011 02:04:25 -0500


On Mon, 2011-01-24 at 20:17 +0000, Simon Wilkinson wrote:

> I'm in favour of Jeff's proposal that we stop treating cmdebug as a
>  "standard" interface. However, that means that we need to move the
>  RPCs used by cmdebug out of the callback service, and into their own
>  RX service. From memory, I think this has the added advantage that it
>  should be possible to build a cmdebug that returns results, even when
> the callback service thread is blocked.

In a user-mode server, it certainly would have that advantage.  In the
kernel, I'm not sure.  The threading model in the kernel is of course
completely different from any of those we use in userland.

Moving debugging into its own service is certainly feasible.  If the
goal is to avoid the need to maintain a registry, then I think it would
be desirable to establish a separate service per implementation (well,
per implementor, really).

-- Jeff