On Wed, Jan 29, 2020 at 10:11:26AM -0800, Linus Torvalds wrote:
> On Wed, Jan 29, 2020 at 2:14 AM Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx> wrote:
> >
> > Here is the big USB and Thunderbolt and PHY driver updates for 5.6-rc1.
>
> Hmm. This actually causes a new warning even before I start building it:
>
> WARNING: unmet direct dependencies detected for I2C_S3C2410
> Depends on [n]: I2C [=y] && HAS_IOMEM [=y] && HAVE_S3C2410_I2C [=n]
> Selected by [m]:
> - PHY_EXYNOS5250_SATA [=m] && (SOC_EXYNOS5250 || COMPILE_TEST
> [=y]) && HAS_IOMEM [=y] && OF [=y]
>
> and the cause seems to be
>
> 203b7ee14d3a ("phy: Enable compile testing for some of drivers")
>
> where PHY_EXYNOS5250_SATA now has a
>
> depends on SOC_EXYNOS5250 || COMPILE_TEST
> depends on HAS_IOMEM
> depends on OF
>
> and then blindly does a
>
> select I2C_S3C2410
>
> without having the dependencies that I2C_S3C2410 has.
>
> How did this ever pass any testing in linux-next without being
> noticed, when I noticed within five seconds of pulling it? It
> literally warns immediately on "make allmodconfig".
>
> The warnings happen during the build too, as it does the silentconfig.
> So I'm not sure how this was missed.
>
> Stephen, does linux-next perhaps miss these config-time warnings?
>
> I have partially reverted that commit in my merge (removing the "||
> COMPILE_TEST" for that PHY_EXYNOS5250_SATA entry) because warnings are
> not acceptable.
Hi Linus,
The I2C fix for this is in Wolfram's tree already:
https://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux.git/log/?h=i2c/for-next
Best regards,
Krzysztof
