[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Kernel compilation problem



I did finally get the 4.17.11 kernel to compile so that my large drives are visible now, but I had to use a 'shotgun' approach in including as many SCSI/SATA/etc variables as possible.


I do intend on comparing the non-functioning config for 4.17.11 against both the functioning config, as well as against the config for 4.9.0-7-amd.


Thanks.


Taren


On 08/02/2018 10:30 AM, David Wright wrote:
On Wed 01 Aug 2018 at 23:37:33 (-0600), Taren wrote:
I'm running Stretch, with kernel 4.9.0.7, and am trying to compile a
new kernel (preferably 4.17.11) into which I can boot.
The kernel builds successfully, but whenever I try booting into the
new kernel, I end up in emergency mode, with the error

     Unit dev-disk-by\x2duuid-<uuid>.device has failed.

     The result is timeout.

This device is anmd device, with two mirrors (each 2.7T in size).
The submirrors are present when I boot into 4.9.0.7 (installed when
the system was built).
However, they do not appear to be visible under any kernel which I
build and try to boot into.

I've tried setting LBDAF in the kernel configuration, but that
requires that a 32bit kernel be built (and x64 deselected), and I'm
running on an AMD 8350 chip, which is x86_64.
Kernel 4.9.0.7 does not have LBDAF set (and x64 is set), yet it's
able to see my 2.7T drives, and my raid device mounts with no
problem.
I think this is a red herring. 32bit kernels need LBDAF for large
disks because they have to be told to use 64bit addressing for
them. Obviously 64bit kernels don't need telling, so that option
is made unavailable.

Would someone point me in the correct direction for configuring a
new kernel, so that my 2T+ drives are visible?
In the absence of other replies, I can only suggest
(a) comparing /boot/config-4.9.0-7-amd64 with the one you've
     built to see whether something is missing,
(b) compare the output of lsinitramfs /boot/initrd.img-4.9.0-7-amd64
     with the one you've built to see likewise.
I'm assuming with (b) that the disks have to be found by using the
initramfs before the system can continue booting with the filesystem
contained on those disks.

Cheers,
David.




Reply to: