On 10/18/2016 03:27 PM, Dr. David Alan Gilbert wrote:
> * Halil Pasic (pa...@linux.vnet.ibm.com) wrote:
>> > The handling of VMStateField.start is currently quite broken if
>> > VMS_ALLOC is present (that is for VMSTATE_VBUFFER_ALLOC_UINT32) but
>> > fortunately also quite underutilized -- nobody is using .start != 0.
>> > Let's prove with this patch that it's really broken (as a first
>> > step towards fixing things up).
> You can't play this trick - patch series have to work and pass
> tests after each test, otherwise it messes up people in the future
> trying to do a git bisect.
I think I understand the motivation. Does that mean
you are not supposed to expose a bug via a test? I might
be able to demonstrate that something is wrong but unable
to fix the problem myself (time constraints).
How was I supposed to do this?