From mboxrd@z Thu Jan 1 00:00:00 1970 Return-path: MIME-Version: 1.0 References: <592943eab0a32f9e1b892960e84e0524a6f56c14.camel@pengutronix.de> In-Reply-To: From: Brian Hutchinson Date: Wed, 16 Jun 2021 08:45:36 -0400 Message-ID: Content-Type: multipart/alternative; boundary="0000000000003db0d805c4e1788d" Subject: Re: [RAUC] Create bundles from command line outside of yocto? To: =?UTF-8?Q?Enrico_J=C3=B6rns?= Cc: rauc@pengutronix.de, =?UTF-8?Q?Jan_L=C3=BCbbe?= List-ID: --0000000000003db0d805c4e1788d Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable On Wed, Jun 16, 2021 at 1:06 AM Enrico J=C3=B6rns wrot= e: > Hi Brian, > > Am Dienstag, dem 15.06.2021 um 17:32 -0400 schrieb Brian Hutchinson: > > Ok, I think I figured out my problems. Appears that I was using busybo= x > > version of tar so I switched to the real thing (GNU Tar) and that fixed > one > > problem then I ran into a mkfs.vfat problem and I added dosfstools to m= y > image > > and that took care of that so now my bundle is updating correctly now > ... so > > sorry for the noise! > > glad you solved it! > > You could actually use busybox tar, too. But then you need to make sure > to have > enabled format autodetection and the required compression algorithm, see: > > https://rauc.readthedocs.io/en/latest/integration.html#required-target-to= ols > > About your original questions: > > > So I should use "parent" to tie both kernel and appfs slots to the > rootfs? > > Your system.conf (and parent relations) looked good from what I saw. > > > And for now, the rootfs is r/w on a ext4 filesystem, but in the future > it will > > be a squashfs. So once that happens would 'type=3Dext4' then change to > > 'type=3Draw'? > > Yes, 'raw' will be fine for squashfs as there is an explicit handler > available > for it. The 'raw' type ensures RAUC does not think it could mount the > bundle for > writing. > > If you have a separate shared data partition, then I would recommend to > configure RAUC to use this for its status file (which cannot be located > inside > the slot anymore when having "ro by design"): > > [system] > ... > statusfile=3D/path/to/datapart/rauc.status > > See https://rauc.readthedocs.io/en/latest/reference.html#statusfile I'll check that out. I'm using u-boot so I "thought" all the status stuff was being done with u-boot environment variables via fw_setenv/fw_printenv. Was it my imagination or did RAUC populate my rootfs during the update faster than when I just mount and manually untar my rootfs? How is it you guys are faster? Thanks again! Regards, Brian --0000000000003db0d805c4e1788d Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable

On Wed, Jun 16, 2021 at 1:06 AM Enrico J= =C3=B6rns <ejo@pengutronix.de&= gt; wrote:
Hi Br= ian,

Am Dienstag, dem 15.06.2021 um 17:32 -0400 schrieb Brian Hutchinson:
> Ok, I think I figured out my problems.=C2=A0 Appears that I was using = busybox
> version of tar so I switched to the real thing (GNU Tar) and that fixe= d one
> problem then I ran into a mkfs.vfat problem and I added dosfstools to = my image
> and that took care of that so now my bundle is updating correctly now = ... so
> sorry for the noise!

glad you solved it!

You could actually use busybox tar, too.=C2=A0 But then you need to make su= re to have
enabled format autodetection and the required compression algorithm, see: https://rauc.readthedocs= .io/en/latest/integration.html#required-target-tools

About your original questions:

> So I should use "parent" to tie both kernel and appfs slots = to the rootfs?

Your system.conf (and parent relations) looked good from what I saw.

> And for now, the rootfs is r/w on a ext4 filesystem, but in the future= it will
> be a squashfs.=C2=A0 So once that happens would 'type=3Dext4' = then change to
> 'type=3Draw'?

Yes, 'raw' will be fine for squashfs as there is an explicit handle= r available
for it. The 'raw' type ensures RAUC does not think it could mount t= he bundle for
writing.

If you have a separate shared data partition, then I would recommend to
configure RAUC to use this for its status file (which cannot be located ins= ide
the slot anymore when having "ro by design"):

=C2=A0 [system]
=C2=A0 ...
=C2=A0 statusfile=3D/path/to/datapart/rauc.status

See=C2=A0https://rauc.readthedocs.io/= en/latest/reference.html#statusfile

I&#= 39;ll check that out.=C2=A0 I'm using u-boot so I "thought" a= ll the status stuff was being done with u-boot environment variables via fw= _setenv/fw_printenv.

Was it my imagination or did= RAUC populate my rootfs during the update faster than when I just mount an= d manually untar my rootfs?=C2=A0 How is it you guys are faster?
=
Thanks again!

Regards,
Brian
--0000000000003db0d805c4e1788d--