Re: [GIT PULL] ACPI & Power Management Patches for Linux-3.5-merge

[Len Brown <lenb@kernel.org>]

On 06/02/2012 07:51 PM, Linus Torvalds wrote:

> On Fri, Jun 1, 2012 at 11:32 PM, Len Brown <lenb@kernel.org> wrote:
>>
>> ps. Sorry for sending this request at the tails of the merge window --
>> I'll try to be earlier next time.
> 
> Christ, not only is it after I really wanted to do -rc1 (held up by
> the tty locking problems), but it doesn't even compile.
> 
> Find the bug (the compiler certainly did):
> 
> static inline int acpi_pm_device_sleep_state(struct device *d, int *p, int m)
> {
>         if (p)
>                 *p = ACPI_STATE_D0;
>         return (m >= ACPI_STATE_D0 && <= ACPI_STATE_D3) ? m : ACPI_STATE_D0;
> }
> 
> and no, it wasn't a merge error. That's what it looks like in your tree.

>

> The commit was done yesterday. It clearly had *zero* testing.

Hmm.

This hunk is in the CONFIG_PM=n case.

Of the several hundred x86_64 and i386 kernels I build
before sending you a pull request, only two do not have CONFIG_PM=y --
x86_64 allnoconfig and i386 allnoconfig.
Like the other kernels, those build fine.

I'm curious what config and compiler tripped on this for you.

> Looking more at the pull as a result of this, I notice that almost
> every commit in that tree is from yesterday, and thus cleary cannot
> have been in -next.


Yes, I did check in Ying's patch this week, and a few others.

But a bunch of the patches have been in linux-next for some time.

I know you'd prefer patches to live in the tree frozen at the
date that they were 1st checked in, but that doesn't work well
when patches change.  To update a patch in a series I need to re-base.
Yes, I could re-base in place -- in the context of an rc
that nobody anywhere (including me) will ever build or boot.
Or I could re-base up to the latest release boundary which
a lot of people (including me) will test.  In this case
I think I re-based everything up to 3.4.

> I was going to just fix up the obvious one-liner

> fixup, but looking at the bigger picture I'm going to say "3.6
> material" for this whole thing.


It would be sad for a simple, though embarrassing,
issue with this patch to delay the other patches.

-Len