On Wed, 23 Feb 2022 at 23:31, Peter Collingbourne <[email protected]> wrote:
[...]
> > > > Link: https://linux-review.googlesource.com/id/I20faa90126937bbee77d9d44709556c3dd4b40be
> > > > Signed-off-by: Peter Collingbourne <[email protected]>
> > > > Fixes: e5f4728767d2 ("kasan: test: add globals left-out-of-bounds test")
> > >
> > > This Fixes tag is unneeded.
> > >
> > > Except the above nit, this patch looks good to me. Thanks.
> > >
> > > Reviewed-by: Miaohe Lin <[email protected]>
> >
> > Reviewed-by: Marco Elver <[email protected]>
> >
> > And yes, the Fixes tag should be removed to not have stable teams do
> > unnecessary work.
>
> I thought that Cc: [email protected] controlled whether the patch
> is to be taken to the stable kernel and Fixes: was more of an
> informational tag. At least that's what this seems to say:
> https://www.kernel.org/doc/html/latest/process/submitting-patches.html#reviewer-s-statement-of-oversight
These days patches that just have a Fixes tag (and no Cc: stable) will
be auto-picked in many (most?) cases (by empirical observation).
I think there were also tree-specific variances of this policy, but am
not sure anymore. What is the latest policy?
> > +Cc'ing missing mailing lists (use get_maintainers.pl - in particular,
> > LKML is missing, which should always be Cc'd for archival purposes so
> > that things like b4 can work properly).
>
> get_maintainers.pl tends to list a lot of reviewers so I try to filter
> it to only the most important recipients or only use it for
> "important" patches (like the uaccess logging patch). It's also a bit
> broken in my workflow --
> https://lore.kernel.org/all/[email protected]/
> fixes one of the problems but there are others.
That's fair. It just seemed that something went wrong given
[email protected] wasn't Cc'd. FWIW, syzbot uses
'get_maintainer.pl --git-min-percent=20' which is a bit less
aggressive with Cc'ing folks not mentioned explicitly in MAINTAINERS.
> Doesn't b4 scan all the mailing lists? So I'd have imagined it
> wouldn't matter which one you send it to.
Those under lore.kernel.org or lists.linux.dev. Seems linux-mm does
get redirected to lore: https://lore.kernel.org/linux-mm/ -- It's not
entirely obvious which are lore managed and which aren't (obviously
things like [email protected] aren't).
On Thu, 24 Feb 2022 00:35:32 +0100 Marco Elver <[email protected]> wrote:
> > I thought that Cc: [email protected] controlled whether the patch
> > is to be taken to the stable kernel and Fixes: was more of an
> > informational tag. At least that's what this seems to say:
> > https://www.kernel.org/doc/html/latest/process/submitting-patches.html#reviewer-s-statement-of-oversight
>
> These days patches that just have a Fixes tag (and no Cc: stable) will
> be auto-picked in many (most?) cases (by empirical observation).
>
> I think there were also tree-specific variances of this policy, but am
> not sure anymore. What is the latest policy?
The -stable maintainers have been asked not to do that for MM patches -
to only take those which the developers (usually I) have explicitly tagged
for backporting.
I don't know how rigorously this is being followed. Probably OK for
patches to mm/* but if it's drivers/base/node.c then heaven knows.