Re: [PATCH 2/5] Provide acpi_check_{mem_}region.

[Andrew Morton <akpm@linux-foundation.org>]

On Sun, 10 Feb 2008 13:25:36 +0100
Jean Delvare <khali@linux-fr.org> wrote:

> Hi Andrew,
> 
> On Thu, 7 Feb 2008 23:40:02 -0800, Andrew Morton wrote:
> > On Wed, 24 Oct 2007 16:32:59 +0200 Thomas Renninger <trenn@suse.de> wrote:
> > 
> > > Provide acpi_check_{mem_}region.
> > > 
> > > Drivers can additionally check against possible ACPI interference by also
> > > invoking this shortly before they call request_region.
> > > If -EBUSY is returned, the driver must not load.
> > > Use acpi_enforce_resources=strict/lax/no options to:
> > >   - strict: let conflicting drivers fail to load with an error message
> > >   - lax:    let conflicting driver work normal with a warning message
> > >   - no:     no functional change at all
> > > 
> > > 
> > 
> > OK, so Len has merged these into the acpi test tree.  My understanding is
> > that once this work hits mainline, we can then merge
> > check-for-acpi-resource-conflicts-in-hwmon-drivers.patch.
> 
> Correct. Same applies to a second patch:
> check-for-acpi-resource-conflicts-in-i2c-bus-drivers.patch
> Both patches should be merged upstream at the same time.
> 
> > My normal approach would be to send
> > check-for-acpi-resource-conflicts-in-hwmon-drivers.patch to Mark for
> > inclusion in git-hwmon one Len has merged the prerequisites into mainline. 
> > 
> > Problem is, if Len merges late in the 2.6.26 merge window, Mark might not
> > have time to gets these changes into mainline before 2.6.27.  Which is all
> > getting a bit dumb considering I first merged everything in October. 
> > Fortunately things aren't mormally _this_ inefficient when one follows the
> > rules - this was an unusual patchset.
> > 
> > But still, I think we could afford to speed things up a bit more than that.
> >  We could ask Len to consider merging this work into 2.6.25 and then if
> > Mark can ack check-for-acpi-resource-conflicts-in-hwmon-drivers.patch
> > (below) for an akpm-merge, we're good to go.  But I do recall that people
> > were a bit uncertain about it all back in October.
> > 
> > Please share your thoughts with us.
> 
> Len already merged all the acpi bits for 2.6.25 as far as I can see, so
> all that is missing now is these two patches:
> check-for-acpi-resource-conflicts-in-hwmon-drivers.patch
> check-for-acpi-resource-conflicts-in-i2c-bus-drivers.patch
> Both have been in -mm for quite some time.

Yup, the prerequisites appear to be in mainline now.

> In the default mode (acpi_enforce_resources=lax) these patches simply
> print warnings but still let the drivers load, so they are safe to
> merge, and the sooner, the better. The idea is to get feedback on how
> many systems out there have ACPI resource conflicts. Then we'll see how
> we can address them (if at all.)
> 
> I don't remember anyone objecting to these patches, and anyway the
> problem has been there for years and nobody took care, so if anyone
> really isn't happy with the solution designed by Thomas, that person
> will have to do the work and submit something better later. That
> shouldn't delay the merge of what we have now.
> 
> Andrew, both patches are
> 
> Acked-by: Jean Delvare <khali@linux-fr.org>

We already have Signed-off-by:you, which I figure outranks acked-by: ;)

> and I am totally fine with you pushing them to Linus now. But of course
> having Mark's ack would be good too.

That would be nice.  But I'll merge them mid-week anyway unless Mark actually
nacks them:

http://userweb.kernel.org/~akpm/mmotm/broken-out/check-for-acpi-resource-conflicts-in-hwmon-drivers.patch

http://userweb.kernel.org/~akpm/mmotm/broken-out/check-for-acpi-resource-conflicts-in-i2c-bus-drivers.patch

Thanks.