linux-kernel - Re: linux-next: Fixes tags need some work in the pm tree

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 16 Jan 2019 12:34:29 +0100
From:   "Rafael J. Wysocki" <rjw@...ysocki.net>
To:     Michael Ellerman <mpe@...erman.id.au>
Cc:     Stephen Rothwell <sfr@...b.auug.org.au>,
        Linux Next Mailing List <linux-next@...r.kernel.org>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        Sinan Kaya <okaya@...nel.org>,
        Paul Gortmaker <paul.gortmaker@...driver.com>
Subject: Re: linux-next: Fixes tags need some work in the pm tree

On Wednesday, January 16, 2019 12:43:31 AM CET Michael Ellerman wrote:
> "Rafael J. Wysocki" <rjw@...ysocki.net> writes:
> > On Tuesday, January 15, 2019 11:43:05 PM CET Stephen Rothwell wrote:
> >> Hi Rafael,
> >> 
> >> On Tue, 15 Jan 2019 23:13:16 +0100 "Rafael J. Wysocki" <rjw@...ysocki.net> wrote:
> >> >
> >> > On Tuesday, January 15, 2019 9:55:40 PM CET Stephen Rothwell wrote:
> >> > > [I am experimenting with checking the Fixes tags in commits in linux-next.
> >> > > Please let me know if you think I am being too strict.]
> >> > > 
> >> > > Hi Rafael,
> >> > > 
> >> > > Commits
> >> > > 
> >> > >   62b33d57c534 ("drivers: thermal: int340x_thermal: Make PCI dependency explicit")
> >> > >   cd793ab22a93 ("x86/intel/lpss: Make PCI dependency explicit")
> >> > >   42ac19e7b81e ("ACPI: EC: Look for ECDT EC after calling acpi_load_tables()")
> >> > >   6c29b81b5695 ("platform/x86: apple-gmux: Make PCI dependency explicit")
> >> > >   34783dc0182a ("platform/x86: intel_pmc: Make PCI dependency explicit")
> >> > >   704658d1d3ae ("platform/x86: intel_ips: make PCI dependency explicit")
> >> > >   5df37f3a1aa9 ("vga-switcheroo: make PCI dependency explicit")
> >> > >   da1df6ee4296 ("ata: pata_acpi: Make PCI dependency explicit")
> >> > >   ce97a22a596b ("ACPI / LPSS: Make PCI dependency explicit")
> >> > > 
> >> > > Have malformed Fixes tags:
> >> > > 
> >> > > There should be double quotes around the commit subject.  
> >> > 
> >> > Well, where does this requirement come from?
> >> > 
> >> > It hasn't been there before AFAICS.
> >> 
> >> Documentation/process/submitting-patches.rst has the following, but I
> >> am sure people are happy to discuss changes and it does say "For
> >> example", so maybe I am being to strict?
> >
> > If that's the source of it, then it's rather weak IMO.
> >
> > Formal requirements should be documented as such and I would expect that
> > to happen through the usual process: patch submission, review, acceptance etc.
> 
> It is documented, in submitting-patches.rst.
> 
> That was submitted to lkml:
> 
>   https://lore.kernel.org/lkml/1396949135-27122-1-git-send-email-jeffrey.t.kirsher@intel.com/
> 
> And committed by Linus:
> 
>   8401aa1f5997 ("Documentation/SubmittingPatches: describe the Fixes: tag")

Stephen has already quoted from that doc, but it only gives the format of the
summary line as an example.

> How would we make it more formal than that?

Say somewhere that this particular summary formatting is required?

Also, tags are more of a maintainers' thing and SubmittingPatches doesn't look
like the best place for documenting how the maintainers are expected to format
their commits.

> > Moreover, extending advice on to how submit paches to formatting requirements
> > for commits feels like a bit of a stretch to me.
> >
> >> The counter argument is that
> >> there are various (semi-)automated processes that use these tags and
> >> being consistent probably makes those processes (and life for those who
> >> run them) easier.
> >
> > And frankly I wouldn't expect any of these to even look at the summary
> > lines as they have not been consistent historically and the SHA-1 ID should
> > be sufficient to identify the commit in question.
> 
> It usually is, but it's still a good sanity check to have the subject in
> there, especially for cases where the SHA is wrong (though that should
> be less of a problem in future due to Stephen doing these checks).

A human can look at the summary after a script has not found the commit by
ID, but the human then doesn't care about the quoting characters.

Also it is rather straightforward to strip the quoting characters in a script
whatever they are.

My point basically is that in order to call something "malformed", you need
to provide a formal definition of what is expected.  An example in
SubmittingPatches doesn't seem quite sufficient to me for that role.

Cheers,
Rafael