[patch 2/2] KFAIL for mb-templates.exp PASS->FAIL regression [Re: [RFC stub-side break conditions 5/5] GDBServer-side changes]

[Luis Gustavo]

Hi Jan,

On 02/28/2012 01:05 PM, Jan Kratochvil wrote:
> Hi Luis,
>
> On Thu, 05 Jan 2012 15:59:25 +0100, Luis Machado wrote:
>> I ran the testsuite with GDBServer and only one regression showed up
>> in gdb.cp/mb-templates.exp. It may be a hidden bug that was exposed
>> by these changes since it gives me an internal error when trying to
>> parse the conditional expression.
>
> even despite it is not a bug in your patchset you have introduced PASS->FAIL
> regression in the results.  This is IMO not acceptable, KFAIl should be put
> there in such case.
>
> It is already difficult to find a cause of each PASS->FAIL, as it is typically
> by gcc/glibc but it may be also kernel or KVM and in some cases also CPU kind.
> If even GDB could voluntarily turn PASS->FAIL results on its own...
>
> I am aware PASS->FAILs happen for non-primary (=non-x86*) archs.
>
> I will check it in.
>

I agree we should keep it KFAIL for now, but i'll address this soon.

Regards,
Luis