rpi0: journal corrupted / boot_id does not change

[Gosi]

Hi,

The journal of my raspberry pi zero w is corrupted.

On every boot i have a reproducible error:
$this->bbcode_second_pass_code('', 'Entry timestamp out of synchronization
File corruption detected at /var/log/journal/57ca52580e1c409ca7e07fd9b806fa55/user-1001.journal:1d0b90
')

Also the the entries are not in order.

I think I have found the reason for this, but I am not sure and I can not fix it:

Because my rpi has no RTC journald relies on the boot_id /proc/sys/kernel/random/boot_id to order/assign log entires.

But it seems that the boot_id of my rpi does not change in between boots, which trips up journald (I assume).

So for the last boots my boot_id has been
$this->bbcode_second_pass_code('', '$ cat /proc/sys/kernel/random/boot_id
0cf98fd5-b98d-49f1-92b5-727a28c25c76
')

Here is a list of the last boots (it should be longer):
$this->bbcode_second_pass_code('', '
$ journalctl --list-boots
-3 5a6abaa768814db3bfb95c1c16c0df28 Tue 2018-03-06 19:56:27 CET—Tue 2018-03-06 20:19:40 CET
-2 5e32ee5e58254f51943c14ad0b99a3b1 Tue 2018-03-06 20:20:24 CET—Tue 2018-03-06 20:56:07 CET
-1 60ab17631c234f2faedbfab156aeb6fa Thu 2018-03-22 03:14:57 CET—Thu 2018-03-22 20:39:28 CET
0 0cf98fd5b98d49f192b5727a28c25c76 Sat 2018-03-24 13:34:59 CET—Sat 2018-03-24 01:50:25 CET')

It looks as if boot_id does change eventually but stays the same most of the time.

Can someone reproduce this behaviour?
Is the unchanged boot_id and corrupt journal even related?
Does anyone has information about how the boot_id is generated?

Thanks for reading this long post, I can provide more information if necessary.

[Gosi]

Solved by adding the rng modules to mkinicpio.conf.

$this->bbcode_second_pass_code('', 'MODULES=(rng_core bcm2835_rng)
')

[tracernz]

I was seeing this too but never got around to looking into it. Thanks.

[Gosi]

Except that it is not solved, because the boot_id does change if you change the initramfs by adding modules, but only to another constant value.

I think it it is a kernel bug and there is not much you can do about it.

[tracernz]

Damn. Oh well, it was close, thanks for trying anyway.