[lfs-support] Booting LFS with systemd - SOLVED

Frans de Boer frans at fransdb.nl
Wed Jul 25 09:41:48 PDT 2018


On 07/24/2018 01:01 AM, Ken Moffat wrote:
> On Mon, Jul 23, 2018 at 05:20:41PM -0500, Douglas R. Reno wrote:
>> On Thu, Jul 19, 2018 at 8:47 AM Frans de Boer <frans at fransdb.nl> wrote:
>>
>>>>>>> This quite frustrating. After recompiling, following the book to the
>>>>>>> letter, I still get a frozen LFS system.
>>>>>>> One thing I do note however is that the freezing always occurs after
>>>>>>> systemd has detected that it is on a virtual machine. A number of
>>>>>>> error messages is send, but due to ratelimiting I can't see them
>>>>>>> because they are suppressed.
>>>>>>>
>>>>>>> I had even rebuild everything with systemd-232, and that worked as
>>>>>>> before. But after 232, things started to behave strange. Now way to
>>>>>>> debug systemd, whatever I do....
>>>>>>>
>>>>>>> Help?
>> Frans,
>>
>> Can you please try systemd-239? It should show up in the render tomorrow
>> morning (US Central time, I'm not sure what it is in UTC).
>>
>> I'll make sure it lands in BLFS here in the coming days, just extremely
>> busy outside of LFS.
> If that does't help, now that I've had to apply a workaround
> to two of my sysv systems to speed booting (lack of entropy on some
> machines with integrated video and only an SSD) I've got an
> alternative suggestion - if the kernel is 4.17 or later, or 4.16.4
> or later, or (perhaps) 4.14.36 or later, it contains a CVE fix which
> ensures that getrandom() will not return until the CRNG is properly
> initialised.
>
> That is reported to severely impact _some_ VMs' startup times.
>
> The easiest workaround is to install haveged in chroot, and its
> systemd file in your case.  If that is the problem, people differ
> about the quality of what haveged provides - if you need to generate
> long-lived security kees (e.g. for gpg) in the VM see
> https://lkml.org/lkml/2018/7/23/857 (Ted Ts'o's reply to me when I
> suggested that if I had to use haveged the boot was fast but didn't
> it weaken future entropy?)
>
> ĸen

Thanks to the availability of systemd-239 and the fact that finally the 
patch regarding the man-pages is available, I can now confirm that all 
is well again with systemd-239.
Thus, it was a systemd problem, I have tried the standard systemd-239 a 
long time ago, but since I have no idea how to make the specific patches 
to for LFS, I could not build that. Maybe make it clear what is done to 
make the specific systemd patch or is it a lot of hand work?

I remember that we only need doxygen as an additional package to build 
systemd without the LFS patch, right?

Regards, Frans.



More information about the lfs-support mailing list