Re: [PATCH 07/13] RFC migration: icp/server is a mess

[Juan Quintela]

"Nicholas Piggin" <npiggin@gmail.com> wrote:
> On Fri Oct 20, 2023 at 7:39 AM AEST, Greg Kurz wrote:
>> On Thu, 19 Oct 2023 21:08:25 +0200
>> Juan Quintela <quintela@redhat.com> wrote:


> So the reason we can't have duplicate names registered, aside from it
> surely going bad if we actually send or receive a stream at the point
> they are registered, is the duplcate check introduced in patch 9? But
> before that, this hack does seem to actually work because the duplicate
> is unregistered right away.

You are creating a new general case that has only a single use that you
agree it is "hacky" O:-)

The problem here is that you haven't made your mind what "ipc/server"
means.  You want sometimes to mean pre_2_10, sometimes to mean other
thing.  That is not how this is supposed to work.  See my proposed
change, it is one line change, and just do the right thing.

I know, it breaks backwards compatibility.  But for one machine type
that people are proposing to deprecate/remove.

> If I understand the workaround, there is an asymmetry in the migration
> sequence in that receiving an unexpected object would cause a failure,
> but going from newer to older would just skip some "expected" objects
> and that didn't cause a problem. So you only have to deal with ignoring
> the unexpected ones going form older to newer.


Ok, found a different workaround.
Sending a new version of the series with a different hack that maintains
backwards compatibility.

Later, Juan.