Re: [Xen-devel] Will hap_alloc fail?

[Keir Fraser <keir@xxxxxxxxxxxxx>]

On 13/7/07 09:23, "Tim Deegan" <Tim.Deegan@xxxxxxxxxxxxx> wrote:

> At 18:19 +0100 on 12 Jul (1184264340), Mats Petersson wrote:
>> I would have thought that domain_crash() is the right thing to do -
>> there's nothing "better" that can be done elsewhere, as far as I can
>> understand, and there's really no point in propagating an error
>> unless there's something that can be done about it (or it can be
>> ignored, which isn't the case in this instance), as this only leads
>> to potential misses of the propagated error, making it harder to debug.
> 
> You need to do both, unfortunately.  domain_crash() just marks the
> domain as crashed; we still need to survive the rest of the code path
> for the action we're taking without following a null pointer or similar.

Yes, we pretty much killed off usage of domain_crash_synchronous() because
it was being used as the lazy way out at the expense of correctness. These
low-level fallible routines often get called in spinlock contexts, for
example.

 -- Keir


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/xen-devel