[OpenWrt-Devel] [PATCH v2 1/2] base-files: improve lib/upgrade/common.sh

Klaus Kudielka klaus.kudielka at gmail.com
Fri May 3 14:16:39 EDT 2019

On 03.05.19 19:05, Petr Štetiar wrote:
> Klaus Kudielka <klaus.kudielka at gmail.com> [2019-04-24 21:14:38]:
> Hi Klaus,
>> Remove workarounds for the old, inconsistent behaviour from the following
>> targets: apm821xx, brcm2708, omap, sunxi.
> I'm willing to push the fix for mvebu (which the patch series is about), so
> please don't touch other targets.  This should go into separate patch(es) and
> in order to push it, I would need some Tested-by, Reviewed-by or Acked-by for
> every platform separately.

Petr, do I understand correctly, I should split this single patch into
the common.sh one, and four additional ones (one per target)?


Let me remind you that the common one *alone* breaks sysupgrade for those
four targets, as Tomasz already pointed out earlier.

In more detail:

The root of the problem is that the *existing* export_bootdevice in
/lib/upgrade/common.sh behaves differently, if the kernel is booted with
root=/dev/..., or if it is booted with root=PARTUUID=...

In the former case, it reports back major/minor of the root partition,
in the latter case it reports back major/minor of the complete boot disk.

The targets mentioned above have added workarounds to this behaviour, by
specifying *negative* increments to the export_partdevice function.

And then came the mvebu target to use export_bootdevice /
export_partdevice as well. Now, different subtargets boot differently,
and the workaround would be even more complex.

I think now is the time to make export_bootdevice behave consistently,
and to report major/minor of the boot disk, period.

Consequently, those targets, which boot with root=/dev/... *and* use
export_bootdevice / export_partdevice, have to be adapted to use
positive increments, otherwise they are broken by the change
to export_bootdevice.

The targets affected were easy to spot with find & grep.

>> Fixes: 4e8345ff68 ("mvebu: base-files: autodetect upgrade device")
>> Signed-off-by: Klaus Kudielka <klaus.kudielka at gmail.com>
>> Acked-by: Tomasz Maciej Nowak <tomek_n at o2.pl>
> Tomasz, you seem pretty confident about this changes, how is that? :-)
> -- ynezz

openwrt-devel mailing list
openwrt-devel at lists.openwrt.org

More information about the openwrt-devel mailing list