On Thursday, July 14, 2016 10:02:36 PM CEST Felix Fietkau wrote:
> On 2016-07-14 21:39, Christian Lamparter wrote:
> > This is mostly because with subtargets, it's possible to have different
> > kernel configs.
> >
> > MyBook Live is somewhat unique compared to the MR24 or the WNDR4700. It has
> > to
> > do without any flash (the 512Kb flash on the MBL only contains the u-boot).
> > The target's boot and rootfs has to be located on the SATA drive, so the MBL
> > needs to have the ext2/4 built-in to boot. (WD's u-boot requires the bootfs
> > to be ext2. We could do without it, if we didn't care about sysupgrade).
> >
> > The problem with having a shared target for MR24 and WNDR4700 is that the
> > they
> > have different hard-coded addresses (CONFIG_PPC_EARLY_DEBUG_44x_PHYSLOW)
> > for
> > the early console. Other than that, there is AFICT no reason.
> >
> > So, what's the plan? Have a NAS and ROUTER subtarget? And forget about the
> > early
> > console?
> I don't think early console is worth having subtargets for.
Having individual kernel-configs for targets would solve this.
> And regarding router vs nas: how much flash does the router have? Can it
> handle the extra stuff that NAS devices need in their kernel config?
The WNDR4700's kernel's uimage can be max 2M. Currently a lzma'd uimage
is at 1.5M. ext4 built-in will add another 150k. dm and md
> Having unnecessary extra subtargets wastes resources on snapshot builds
> and releases, so I want as few of them as possible.
I've looked into a shared subtarget. One problem that came up was with the
"ext4" [0], "ramdisk" [1], squashfs, nand features in the FEATURES list.
[0] MR24 will now select the CONFIG_TARGET_ROOTFS_EXT4FS option by
default and generate a useless ext4 rootfs. (Default 48M)
Note: The WNDR4700 will also generate a ext4 image. However, since it
has the option for a HDD, a rootfs on the HDD is do-able with
some additional work (e.g.: cmdline).
[1] MyBook Live will now generate useless ramdisk and squashfs images.
What should I do about those?
Regards,
Christian
_______________________________________________
Lede-dev mailing list
Lede-dev@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/lede-dev
