[ppc64el] LVM+crypto on ppc64el points to an invalid crypto_source

[Mathieu Trudel-Lapierre]

    4.582154] 8regs : 17120.000 MB/sec
[ 4.622155] 8regs_prefetch: 15152.000 MB/sec
[ 4.662154] 32regs : 17968.000 MB/sec
[ 4.702155] 32regs_prefetch: 16512.000 MB/sec
[ 4.742155] altivec : 26000.000 MB/sec
[ 4.742202] xor: using function: altivec (26000.000 MB/sec)
[ 4.743182] async_tx: api initialized (async)
done.
Spawning shell within the initramfs

BusyBox v1.27.2 (Ubuntu 1:1.27.2-2ubuntu4) built-in shell (ash)
Enter 'help' for a list of built-in commands.

(initramfs) grep -r mpatha3 *
cryptroot/crypttab:mpatha3_crypt /dev/mapper/mpatha3 none luks,discard

Boot reproducibly fails because it can't find /dev/mapper/mpatha3. This is expected, since after install the device is named mpatha-part3.

We really need to standardize on the naming...

[Mathieu Trudel-Lapierre (cyphermox)]

Filed in partman-crypto for now, I'm guessing that's what writes crypttab, and it might not be aware of multipath's special way of addressing devices?

This might well not be ppc64el-specific, given the symptoms, I'm pretty sure it will fail the same way everywhere. I will check on that.

[Mathieu Trudel-Lapierre (cyphermox)]

partman-crypto looks fine at first glance, but I'm starting to suspect multipath-tools might be addressing devices differently in d-i; possibly because of how we build the udeb, or what we have in multipath.conf?

[Ubuntu QA Website (ubuntuqa)]

This bug has been reported on the Ubuntu ISO testing tracker.

A list of all reports related to this bug can be found here:
http://iso.qa.ubuntu.com/qatracker/reports/bugs/1798512

[Launchpad Janitor (janitor)]

[Expired for multipath-tools (Ubuntu) because there has been no activity for 60 days.]

[Dimitri John Ledkov (xnox)]

initramfs hooks of cryptsetup got "smarter" and generate dev-mapper name as seen in d-i as the source device...
... which has wrong X/-partX naming for multipath devices
... thus partman-multipath, should rename multipath devices "for target" once partman is finally done, as otherwise at update-initramfs an invalid crypttab is generated and included in the initramfs irrespective of what /target/etc/fstab says.