Re: [Xen-devel] [PATCH 2/2] xen/livepatch: Don't crash on encountering STN_UNDEF relocations

[Andrew Cooper <andrew.cooper3@xxxxxxxxxx>]

On 19/06/2017 19:30, Konrad Rzeszutek Wilk wrote:
> On Wed, Jun 14, 2017 at 12:49:21PM -0600, Jan Beulich wrote:
>>>>> Andrew Cooper <andrew.cooper3@xxxxxxxxxx> 06/14/17 8:34 PM >>>
>>> Well - I've got a livepatch with such a relocation.  It is probably a
>>> livepatch build tools issue, but the question is whether Xen should ever
>>> accept such a livepatch or not (irrespective of whether this exact
>>> relocation is permitted within the ELF spec).
>> Since the spec explicitly mentions that case, I think we'd better support it.
>> But it wouldn't be the end of the world if we didn't, as presumably there
>> aren't that many use cases for it.
> OK. In that case:
>
> Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>
>
>
> Andrew,
>
> Do you think it would be possible to generate an test-case for this
> in arch/test/livepatch?

I can trivially cause this situation to occur with the current build
tools, but we are currently presuming a build tools bug to be the
underlying issue behind getting a STN_UNDEF relocation in the livepatch.

Given that a STN_UNDEF relocation (appears to) mean a NULL dereference
once the relocations are evaluated, I am not happy with supporting such
a case.

Therefore, I'm going to insist that we take a concrete decision as to
what to do in the hypervisor code, before adding a test case, and
advocate for excluding it outright rather than tolerating it in the
(certain?) knowledge that Xen will subsequently crash.

~Andrew

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
https://lists.xen.org/xen-devel