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

Bug#944777: initramfs-tools-core: Resume fails when UUID= syntax used in conf.d/resume.



On Fri, 15 Nov 2019 22:20:53 +0000
Ben Hutchings <ben@decadent.org.uk> wrote:

> Control: tag -1 moreinfo
> 
> On Fri, 2019-11-15 at 01:30 -0600, Glenn Washburn wrote:
> > Package: initramfs-tools-core
> > Version: 0.130ubuntu3.9
> 
> This is an Ubuntu version, so maybe you should report this on
> Launchpad.

After further investigation, you're right its an ubuntu issue.  I've
reported it there.  I had thought the issue was in init where I pointed
to and since the code was the same between debian and ubuntu, I thought
debian would have the issue as well.  Now, I understand that the issue
is actually in scripts/local-premount/resume, which differs where the
bug is.

> > Severity: normal
> > Tags: newcomer
> > 
> > Dear Maintainer,
> > 
> > Resume from hibernation will fail when
> > /etc/initramfs-tools/conf.d/resume contains "RESUME=UUID=<UUID of
> > hibernation image>".
> 
> *Does* it fail, in practice?

I'm not sure and suspect it probably works in debian now that I
understand the bug better.

> > According to NEWS this
> > should work.  The reason is does not is because UUID= processing is
> > only done for the resume kernel parameter (see init lines 123-128).
> > No processing is done for the RESUME shell variable included from
> > conf.d.
> [...]
> 
> Whether the resume device is specified by a configuration file or from
> a kernel parameter, scripts/local-premount/resume will pass it to the
> local_device_setup function and that will pass it to resolve_device
> which processes all the "key=value" formats.
> 
> The early conversion of "UUID=" for the resume kernel parameter
> shouldn't be needed at all any more.

For the sake of clarity, I'd suggest removing the early conversion.  At
a minimum the conf.d/resume parameter should be treated the same as the
kernel parameter.

> Ben.
> 


Reply to: