barebox: readd lzop-native #133
Conversation
Emantor
force-pushed
the
topic/readd-lzop
branch
from
94f371b
to
ea4aae4
Compare
|
@Emantor Do we really still need it for newer barebox integrations? |
|
The upstream configs have been switched to LZ4 in https://lore.barebox.org/barebox/20221206112614.2612071-1-m.felsch@pengutronix.de/#t, but LZO remains a better compression choice if available. ZSTD would be a nice replacement for LZO, but the decompressor turned out to be quite heavy: https://lore.barebox.org/barebox/20220713100922.1880282-1-a.fatoum@pengutronix.de/ |
Regardless whether we have switched over integrations since the removal, LZO is currently supported within upstream barebox and as the poky commit notes still the best time/compression compromise on lower end ARM systems. I don't see a reason to not support LZO. |
It's not an option to add the dependency for cases where it's actually used? Also, note that the recipe is a bit deprecated anyway. nit: the referenced commit is a poky commit, not an oe-core commit |
In scarthgap, lzop has been readded to poky with commit 07006c486722
("Revert "lzop: remove recipe from oe-core""), readd it to the barebox
depends to reallow lzop compression.
This reverts commit 38ada68.
Signed-off-by: Rouven Czerwinski <r.czerwinski@pengutronix.de>
Emantor
force-pushed
the
topic/readd-lzop
branch
from
ea4aae4
to
c7dfa2e
Compare
Yes, I already did that. From a usability perspective I want to readd this to the upstream recipe since especially older kirkstone BSPs are prone to have this set and they will stumble upon this during a scarthgap bump. Since LZO is supported both in poky/oe-core and barebox, I don't see a reason to fix this for every BSP.
In that case we should add a deprecation note and have a removal plan.
Fixed this with a force push. |
In scarthgap, lzop has been readded to poky with commit 07006c486722 ("Revert "lzop: remove recipe from oe-core""), readd it to the barebox depends to reallow lzop compression.
This reverts commit 38ada68.