Hoi
>
> On Sat, 2026-02-07 at 00:28 +0000, Johannes Schneider wrote:
> > Add 'expand=True' to directly resolve file checksums that are added
> > conditionally with inline python code.
> >
> > Otherwise it can happend that the license.bbclass still sees the
> > un-expanded values and either misses or even skipps adjacent licenses.
> >
> > E.g. an unvaforable combination of
> > LIC_FILES_CHKSUM +=
> > "file://docs/license.rst;md5=83b7626b8c7a37263c6a58af8d19bee1"
> > LIC_FILES_CHKSUM:append = "${@bb.utils.contains('TFA_MBEDTLS', '1', '
> > ${LIC_FILES_CHKSUM_MBEDTLS}', '', d)}"
> >
> > would lead to a
> > # pre-expansion value:
> > # "
> > file://docs/license.rst;md5=83b7626b8c7a37263c6a58af8d19bee1${@bb.utils.contains('TFA_MBEDTLS',
> > '1', ' ${LIC_FILES_CHKSUM_MBEDTLS}', '', d)}"
> > LIC_FILES_CHKSUM="
> > file://docs/license.rst;md5=83b7626b8c7a37263c6a58af8d19bee1"
> > and the license checksum verification on 'license.rst' being sometimes
> > skipped.
> >
> > Signed-off-by: Johannes Schneider <[email protected]>
> > ---
> > meta/classes-global/license.bbclass | 2 +-
> > 1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/meta/classes-global/license.bbclass
> > b/meta/classes-global/license.bbclass
> > index af5f1ed41d..62116ba31d 100644
> > --- a/meta/classes-global/license.bbclass
> > +++ b/meta/classes-global/license.bbclass
> > @@ -125,7 +125,7 @@ def find_license_files(d):
> > from collections import defaultdict, OrderedDict
> >
> > # All the license files for the package
> > - lic_files = d.getVar('LIC_FILES_CHKSUM') or ""
> > + lic_files = d.getVar('LIC_FILES_CHKSUM', expand=True) or ""
> > pn = d.getVar('PN')
> > # The license files are located in S/LIC_FILE_CHECKSUM.
> > srcdir = d.getVar('S')
>
> This is a little strange as expand has been the default value for a
> long time. On a modern code base, this should therefore be the default
> and the change would make no difference.
>
of the default i wasn't aware of (or checked ... my bad)
My train of thought was that bitbake expands variables lazily and
thought/remembered (wrongly) that this leads to edge cases with in-line python.
>
> Was this an issue you were seeing on an older release?
>
indeed, we ran into this with a 'scarthgap' setup
(oe-core and meta-oe on scarthgap; bitbake at 2.8)
so not that old :-S
More investigation later: it was neither variable expansion nor sstate - but a
change in our ci setup that i didn't get wind of; previously the do_populate_lic
task simply wasn't triggered - now it is, and it promptly steped on an old
landmine = a wrong checksum.
Please disregard this patch => sorry for the noise! :-S
>
> Cheers,
>
> Richard
>
gruß
Johannes
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#230672):
https://lists.openembedded.org/g/openembedded-core/message/230672
Mute This Topic: https://lists.openembedded.org/mt/117682470/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-
