Re: initramfs-tools - forgotten change?
On Sat 10 Jun 2017 at 15:46:14 (+0200), Hans wrote:
> Hi Ben,
>
> I believe, this is what happened:
I don't know what the patch (in the OP) is or who Ben is (but I assume
they post from 4ax.com or decadent.org.uk).
I can only give a data point, with no particular insight into
its generation.
> Most people will have
>
> /etc/initramfs-tools/conf.d/resume
>
> file, which has a "RESUME=" content. So it will be the same, as set in
> initramfs.conf. This "resume"-file (I believe, but I am not sure!) was some day
> created by package uswsusp (If I am wrong. please correct me!)
My Dell laptop hibernates and wakes up when you close and open the
lid. None of my computers has knowingly installed uswsusp (no logs,
no archived dpkg -l listing), and there's no trace of it in my
apt-cacher-ng cache.
On the Dell, the resume file has been edited to RESUME=LABEL=john04
where john04 is a (presently unused) swap partition, and that is
reflected in its initrd.img.
> So, no people do see a difference, because a variable RESUME= exists, either
> with /dev/sdaX or UUID=somewhatever.
>
> I checked a fresh installation of initramfs-tools (0.130) and latest uswsusp:
> None of these installation creates
>
> /etc/initramfs-tools/conf.d/resume
>
> file. So, maybe this should be checked and confirmed somehow.
>
> I also found out, that it is recommended, to use UUID entries in /etc/fstab,
> otherwise you get some warnings. However, they do not harm. They are like:
>
> cryptsetup: can not find canonical name for /dev/sda3
>
> I suppose, this is well known and not important in this case.
I haven't used cryptsetup, nor seen warnings. My /etc/fstab files
use LABELs for ext partitions as a matter of course.
> As far as I tested, adding RESUME=auto in initramfs.conf did find my swap
> device. Hibernation (=suspen-to-disk) is working but Resume from swap is still
> not working. This is also another story, and might not be involved in this one
> (especially, because at boot the resume device IS FOUND, and the saved image
> IS loaded, but THEN crashes). So, do not look at this bug, just wanted to
> mention it.
>
> Back to my state: People should make sure. if there is a "resume" file, because
I believe, a fresh installation does not create one. Instead, the debian
> installer would do, but neither initramfs-tools-package, nor uswsusp-package
> don't. The latter one previous did create one and edited it, too. Now this
> configuration option is no more called again by dpkg-reconfigure (as it did
> before!)
My data point is concerned with this last paragraph. I recently
installed stretch RC3 on this Acer 386 laptop. The installation
is just about the minimum possible, an expert install with nothing
selected in the Software selection screen.
/etc/initramfs-tools/conf.d/resume was installed and contains the
correct UUID for the laptop's swap partition. I don't see what
creates the file at first sight, but don't intend trying to follow
the calls in the installation scripts. /etc/initramfs-tools/conf.d/
is created when unpacking the initramfs-tools or initramfs-tools-core
version 0.130 deb, I assume. IIUC mkinitramfs creates the resume file
in the initrd, but I don't know about the /etc file.
Installation idiosyncrasies disclaimer: I didn't select a swap
space at installation time. Between the Partition disks step and
the Install the base system step, I executed # swapon /dev/sda4
in a shell because this laptop only has ½GB memory. I do observe
that its /etc/fstab doesn't mention swap, only the root filesystem
and the optical drive. The system is virgin, not having been booted
since installation; just the 248 installed packages.
Cheers,
David.
Reply to: