[linux-uc20-efi][PATCH 1/1] UBUNTU: [Debian] Add megaraid_sas into initrd

Seth Forshee seth.forshee at canonical.com
Wed Feb 24 14:39:46 UTC 2021


On Wed, Feb 24, 2021 at 03:32:29PM +0800, Jesse Sung wrote:
> On Wed, Jan 27, 2021 at 4:00 AM Seth Forshee <seth.forshee at canonical.com> wrote:
> >
> > +Cc Dimitri
> >
> > On Tue, Jan 26, 2021 at 02:31:30PM -0500, Jeffrey Lane wrote:
> > > On Tue, Jan 26, 2021 at 12:57 PM Jesse Sung <jesse.sung at canonical.com>
> > > wrote:
> > >
> > > > On Wed, Jan 27, 2021 at 1:38 AM Seth Forshee <seth.forshee at canonical.com>
> > > > wrote:
> > > > >
> > > > > On Fri, Jan 22, 2021 at 06:56:00PM +0800, Wen-chien Jesse Sung wrote:
> > > > > > BugLink: https://launchpad.net/bugs/1912752
> > > > > >
> > > > > > Module megaraid_sas is required in order to some systems to boot with
> > > > > > the pc-kernel snap. Append megaraid_sas to ubuntu-core-initramfs.conf
> > > > > > before using ubuntu-core-initramfs to generate the initrd in order to
> > > > > > include the particular module.
> > > > > >
> > > > > > Signed-off-by: Wen-chien Jesse Sung <jesse.sung at canonical.com>
> > > > >
> > > > > Why not change this in ubuntu-core-initramfs instead? This is a kludgy
> > > > > way to add the module to the conf file, so we'd only want to do this if
> > > > > there's some reason it's unacceptable to add it to ubuntu-core-initramfs
> > > > > directly.
> > > >
> > > > Including megaraid_sas into evey initramfs (including the armhf/arm64
> > > > ones) may not be what we want, especially for those devices with
> > > > limited storage space. This driver most likely is required for the
> > > > pc-kernel only.
> > > >
> > >
> > > Relating to the project that triggers this bug, I agree, this only
> > > currently applies to one specific AMD64 machine.
> > >
> > > There is a much larger discussion needed, however, as in order to support
> > > servers,  the initrd/initramfs for core will need to grow a LOT of new
> > > drivers in order to support deploying core onto any supported server, and
> > > eventually doing so via MAAS.
> > >
> > > IMO not doing so by default today makes this only a band-aid until a point
> > > in the future where this will have to be done across architectures anyway
> > > and this will have to revisited, but that may belong to the larger
> > > discussion anyway.  Just my 2 cents worth of unsolicited opinion.
> >
> > Dimitri,
> >
> > Since the diff below is difficult to read with all the quoting at this
> > point, you can find the original here:
> >
> > https://lists.ubuntu.com/archives/kernel-team/2021-January/116703.html
> >
> > This strikes me as a pretty kludgy way to add this driver into the
> > initramfs, but it looks like ubuntu-core-initramfs doesn't really
> > support any way to add a module to the initramfs for only specific
> > architectures. Am I missing something, or are we stuck with this kludge
> > for now?
> 
> Could this be accepted as a temporary workaround until there's a
> solution? We really need the module in pc-kernel initramfs for this
> cycle...

Dimitri and I had talked off-list, and I thought he had decided that it
made sense to just add this module to the initramfs for all
architectures. Dimitri, am I misremembering?

Seth



More information about the kernel-team mailing list