On Fri, 2020-12-04 at 15:55 +0100, Andrea Bolognani wrote:
> On Thu, 2020-12-03 at 22:05 -0300, Daniel Henrique Barboza wrote:
> > This reverts commit d3f3c2c97f9b92c982ff809479495f44614edb88
> > (domain_conf.c: auto-align pSeries NVDIMM in virDomainMemoryDefPostParse()).
> >
> > This reverts commit ace5931553c87beebb6b3cd994061742b3f88238
> > (conf, qemu: move qemuDomainNVDimmAlignSizePseries to domain_conf.c).
>
> Please revert the two commits separately.
Actually, looking at the rest of the series, I think it's better if
you do *not* go ahead with a straight revert, since that results in
1) returning the function to its old place and signature;
2) having to change the signature back;
3) having to move the function because you need it to be usable
earlier in the file.
That's a whole lot of churn for arguably very little benefit.
I suggest you instead do a pseudo-revert, where you rename the
function (without otherwise altering its signature) and move it to
the part of qemu_domain.c where you are ultimately going to need it;
in the commit message, you can still mention the commit that such a
change is a "spiritual revert" of, but this way we avoid muddying the
waters more than necessary.
Sorry for not realizing earlier that my initial suggestion would have
these consequences.
--
Andrea Bolognani / Red Hat / Virtualization
