[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH] xen: credit2: fix spinlock irq-safety violation

On Wed, 2017-09-20 at 14:49 +0100, George Dunlap wrote:
> On 09/20/2017 12:59 PM, Jan Beulich wrote:
> > Hmm, killing the timer upfront is certainly fine, but is freeing
> > the
> > data before removing the element from the list safe not only
> > currently, but also going forward?
> I agree with Jan -- if you don't want to put the kill_timer() in the
> critical section, put it beforehand; but don't free the structure
> until
> after the sdom struct has been removed from the list.
In general, I agree, and so I'll do this.

In this case, considering what list we are talking about, and what it
is used for right now (i.e., only for debug dump!), there are no
dependencies between these two operations.

And in any scenario that I can anticipate, where such a dependency
would come into being, the level of restructuring of the code that is
necessary to use the list in any really useful way, would be
significant, and there may be a few other cases where a similar
dependency would also surface and become an issue, and that will
probably lead us to consider/remember this code here as well.

So, IOW, I wouldn't consider this a problem, in the specific case. But
since the net effect of this patch and what George suggests is the
same, and since I appreciate the value that it has, in principle, doing
changes like these with this approach, I'm fine with killing before and
freeing after the critical sections, and I'll send a patch for that

Thanks and Regards,
<<This happens because I choose it to happen!>> (Raistlin Majere)
Dario Faggioli, Ph.D, http://about.me/dario.faggioli
Senior Software Engineer, Citrix Systems R&D Ltd., Cambridge (UK)

Attachment: signature.asc
Description: This is a digitally signed message part

Xen-devel mailing list



Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.