Re: [RFC][PATCH -mm] PM: add /sys/power documentation toDocumentation/ABI

From: Andrew Morton
Date: Wed Aug 30 2006 - 18:30:02 EST


On Wed, 30 Aug 2006 15:03:58 -0700
"Randy.Dunlap" <rdunlap@xxxxxxxxxxxx> wrote:

> On Wed, 30 Aug 2006 14:43:50 -0700 Andrew Morton wrote:
>
> > On Wed, 30 Aug 2006 23:38:06 +0200
> > "Rafael J. Wysocki" <rjw@xxxxxxx> wrote:
> >
> > > Documentation/ABI/testing
> >
> > psst, Greg, they're still sending too much stuff: need more paperwork!
> >
> > Some words about this in Documentation/SubmitChecklist would be nice.
>
> and people actually using it would be nice(r).

c'mon, it must be taped to 1,000 monitors and bathroom doors by now?

> Is this close to what you are looking for?

I guess so, thanks. Pointing at the README was sneaky.

>
> From: Randy Dunlap <rdunlap@xxxxxxxxxxxx>
>
> Mention Documenation/ABI/ requirements in Documentation/SubmitChecklist.
>
> Signed-off-by: Randy Dunlap <rdunlap@xxxxxxxxxxxx>
> ---
> Documentation/SubmitChecklist | 3 +++
> 1 files changed, 3 insertions(+)
>
> --- linux-2618-rc5all.orig/Documentation/SubmitChecklist
> +++ linux-2618-rc5all/Documentation/SubmitChecklist
> @@ -61,3 +61,6 @@ kernel patches.
> Documentation/kernel-parameters.txt.
>
> 18: All new module parameters are documented with MODULE_PARM_DESC()
> +
> +19: All new userspace interfaces are documented in Documentation/ABI/.
> + See Documentation/ABI/README for more information.

This ABI/ thing rather snuck under my radar (I saw it go past, but a lot of
things go past).

It'll be good if it works, but it is going to take quite a lot of thought,
effort and maintainer vigilance to be successful and to avoid becoming
rotware.

I wonder how hard it would be to write a script which parses a diff, works
out if it touches ABI things, complain if it doesn't alter
Documentation/ABI/*? Not very - it's just a matter of defining a suitable
regexp.

What _should_ be documented in there, anyway?

- syscalls, obviously.

- /proc? If so, everything, or are there exceptions?

- /sys? If so, everything, or are there exceptions?

- ioctl numbers and payloads?

- netlink messages?

- ethtool thingies? netdev interface names? /proc/iomem identifiers?
module names? kernel-thread comm[] contents? The ABI is pretty fat.

scary.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/