On Mon, 2014-10-13 at 17:36 +0100, Ian Campbell wrote: > On Mon, 2014-10-13 at 16:30 +0100, Ben Hutchings wrote: > > On Sat, 11 Oct 2014 11:37:31 +0100 Ian Campbell <ijc@debian.org> wrote: > > > On Thu, 2014-09-18 at 00:17 +0100, Ian Campbell wrote: > > > > The xgene_ahci driver requires phy-xgene in order to be able to function (i.e. to mount rootfs). It was not included in the > > > > initramfs by default. Adding it to /etc/initramfs-tools/modules allowed the system to boot but ideally this would happen > > > > automatically. > > > > > > I also have a similar issue on the arndale platform which requires > > > i2c-s3c2410 and phy-exynos5250-sata to be listed > > > in /etc/initramfs-tools/modules in order to boot from the SATA disk. > > > (I've a feeling a different non-overlapping set will be required to boot > > > from MMC, but I'll have to perform that experiment later). > > > > Uh... so an *internal* PHY is controlled over I2C? That's a bit crazy > > but perhaps this IP block is also to be usable as an external PHY. > > Actually, I think the i2c here is used to talk to a power regulator of > some sort and is unrelated to the phy (it's just that both are needed). Looking at how the PHY driver uses the I2C device (just once), that does seem to make sense. Perhaps it's regulating power to the amplifier. > I made the i2c device built-in yesterday for just that reason. [...] OK. Ben. -- Ben Hutchings Any sufficiently advanced bug is indistinguishable from a feature.
Attachment:
signature.asc
Description: This is a digitally signed message part