Re: [GIT PULL] platform-drivers-x86 for v7.4-1

[Linus Torvalds]

On Wed, 28 Oct 2026 at 09:26, Ilpo Järvinen
<ilpo.jarvinen@xxxxxxxxxxxxxxx> wrote:
>
> Okay, thanks for the feedback. I'll try to leave more out the next time
> and/or write it more directly to you specifically.

So yes, it should be to me in the sense that it should explain the
pull request, but I have actually suggested to people that what they
should think about as a target audience is something like lwn or
kernelnewbies or phoronix or similar.

IOW, if you write the pull request message to be of a form where "this
is what might interest a wider audience", I think you'll automatically
also end up writing it as a good explanation for a merge commit.

So a merge commit message should typically explain the "overall shape"
of what is brought in, not every detail. And it is typically meant for
an "outsider": not a developer of that subsystem, but somebody who
maybe bisected down to that merge (I find that particularly during the
merge window it's often useful to use "git bisect --first-parent" on
the mainline kernel to get to a "which big merge caused this problem"
before delving down into the weeds of whcih low-level commit it was).
Or, as mentioned, some outsider that is interested in what happens in
the kernel, but doesn't care about the exact details of every driver.

Because once you care about one particular driver, doing a

    git log drivers/xyz/abc.c

is typically *much* more informative than the alternatives. At that
point you get targeted information about one specific thing. That's
not what a mainline merge message is about (unless, of course, that
merge was only a fix for one single driver - at that point the merge
message obviously *is* just "Fix driver abc behavior")

        Linus