Re: DomU crash during migration when suspending source domain

[Jeremy Fitzhardinge <jeremy@goop.org>]

Graham, Simon wrote:
> You have a better idea of that than I do - however, it seemed to me that
> a complete fix for the cases where allocation failed would be more
> likely to be accepted than something that only fixes one place.
>
> Another thing to consider -- hitting allocation failures is MUCH more
> likely in the Xen environment when live migration is in use; how often
> is a processor taken down and then restarted outside of live migration?
> However, since live migration can occur when the domain has been up for
> a long time, it's much more likely for the allocation to fail (although
> presumably still somewhat low)...
>
> Excuse my ignorance but what's the process for pushing something like
> this upstream?

In addition to all the things Keir mentioned, the other important thing
is to make sure that your comments at the top of the patch explain 1)
what the problem you're trying to solve is, and 2) how this patch solves
those problems.  And a good concise patch subject line helps as well. 
See http://www.zip.com.au/~akpm/linux/patches/stuff/tpp.txt for all the
details.

It seems to me that this patch should be pretty easy to get accepted,
since it is local, a pretty obvious bug, and not Xen-specific.  But it
may need some stylistic attention: for one, the kernel style tries to
avoid nesting things in deep if() statements, and an early return or
goto to a comment exit point might be more appropriate.

Andi Kleen <ak@muc.de> is probably the best person to send it to (as
well as cc:ing it linux-kernel).

    J