On Sun, 2016-02-21 at 16:11 -0800, Vagrant Cascadian wrote: > On 2016-02-21, Ben Hutchings wrote: > > > --- a/debian/installer/armhf/modules/armhf-armmp/mmc-modules > > > +++ b/debian/installer/armhf/modules/armhf-armmp/mmc-modules > > > @@ -5,4 +5,5 @@ mmci > > > omap_hsmmc > > > sunxi-mmc > > > dw_mmc-exynos > > > +dw_mmc-rockchip > > > sdhci-bcm2835 > > > > Would it make sense to replace this list with a wildcard in the default > > mmc-modules file: > > > > drivers/mmc/host/** > > Overall, I like the idea of using wildcards to include enabled drivers > by default, so that drivers for new platforms get automatically > included. > > The only downside I see, is if a driver was accidentally disabled in the > config (or due to some new conflicting config), it will happily produce > a package without that driver...
I think we should add something to buildcheck.py to find unexpectedly
dropped drivers. But that is independent of whether drivers belong in
any particular udeb, or are needed at installation time at all.
> The obvious advantage is it will likely reduce the number of round-trip
> bug reports to enable drivers in both the kernel and the udebs, so I'd
> be supportive of using wildcards for that reason.
Please can you test whether that change would do the right thing for armhf.
Ben.
--
Ben Hutchings
The generation of random numbers is too important to be left to chance.
- Robert Coveyou
signature.asc
Description: This is a digitally signed message part
