Re: efivarfs: fix error on write to new variable leaving remnants

[James Bottomley <James.Bottomley@HansenPartnership.com>]

On Wed, 2025-02-26 at 12:30 +0100, Ard Biesheuvel wrote:
> On Tue, 25 Feb 2025 at 13:59, James Bottomley
> <James.Bottomley@hansenpartnership.com> wrote:
> > 
> > [added correct mailing list for bug report]
> > On Tue, 2025-02-25 at 12:10 +0000, Richard Hughes wrote:
> > > Hi,
> > > 
> > > I'm not sure what I'm supposed to do about:
> > > 
> > > commit 908af31f4896f2c0645031f8b74a89d3a8beb5b9
> > > Author: James Bottomley <James.Bottomley@HansenPartnership.com>
> > > Date:   Sun Jan 19 10:12:12 2025 -0500
> > > 
> > >     efivarfs: fix error on write to new variable leaving remnants
> > > 
> ...
> > > 
> > > It causes a regression in fwupd -- seen in
> > > https://github.com/fwupd/fwupd/issues/8495 and
> > > https://bugzilla.redhat.com/show_bug.cgi?id=2346831 so far -- and
> > > it seems broken for anyone (including me) updating to 6.14.
> > 
> > OK, so the problem with this as a bug report is that it doesn't
> > explain what you're doing.  However
> > 
> > > I can work around the behavior in
> > > https://github.com/fwupd/fwupd/pull/8500 (which is also the
> > > arguably correct thing to do) but it's going to cause a panic as
> > > I have to get an updated fwupd out on all distros so we'll need
> > > releases for multiple branches.
> 
> This code was introduced in the merge window for v6.14, which is due
> to be released end of March. How much time do you need?

The request in the original was for a few more months.

> Derailing LVFS is the last thing I want to do, but we all know how
> this works: once a workaround is put in, it is never going to be
> removed.

I think simply queueing the workaround now (If it works; I'd still like
to see a tested by since it's nowhere near a pure revert) in the fixes
branch for the v14-rc and immediately queueing its revert in the
updates for v15 would give an additional three months automatically and
no-one would have to do anything more (unless more time were needed).

Regards,

James