On Tue, 2022-07-05 at 06:36 +0200, Jacob Kroon wrote:
> On 7/4/22 15:25, Richard Purdie wrote:
> > Add a patch to avoid writing the full pathname to gperf into source
> > files which leads to reproducibility issues.
> >
> > This fixes issues with systemd reproducibility in particular.
> >
> > Signed-off-by: Richard Purdie <richard.pur...@linuxfoundation.org>
> > ---
> > .../gperf/gperf/reproducibility.patch | 26 +++++++++++++++++++
> > meta/recipes-extended/gperf/gperf_3.1.bb | 2 ++
> > 2 files changed, 28 insertions(+)
> > create mode 100644 meta/recipes-extended/gperf/gperf/reproducibility.patch
> >
> > diff --git a/meta/recipes-extended/gperf/gperf/reproducibility.patch
> > b/meta/recipes-extended/gperf/gperf/reproducibility.patch
> > new file mode 100644
> > index 00000000000..9f80828dbd7
> > --- /dev/null
> > +++ b/meta/recipes-extended/gperf/gperf/reproducibility.patch
> > @@ -0,0 +1,26 @@
> > +By default gperf puts a header into generated files with the full path to
> > +the tool along with the commandline used. This patch removes the path to
> > +the binary, allowing reproducible source files (which can be included in
> > +debug source packages).
> > +
> > +Upstream-Status: Pending
> > +Signed-off-by: Richard Purdie <richard.pur...@linuxfoundation.org>
> > +
> > +Index: gperf-3.1/src/options.cc
> > +===================================================================
> > +--- gperf-3.1.orig/src/options.cc
> > ++++ gperf-3.1/src/options.cc
> > +@@ -280,6 +280,13 @@ Options::print_options () const
> > + {
> > + const char *arg = _argument_vector[i];
> > +
> > ++ if (i == 0) {
> > ++ const char *shortarg = strrchr(arg, '/');
> > ++ if (shortarg) {
> > ++ arg = shortarg + 1;
> > ++ }
> > ++ }
> > ++
> > + /* Escape arg if it contains shell metacharacters. */
> > + if (*arg == '-')
> > + {
> > diff --git a/meta/recipes-extended/gperf/gperf_3.1.bb
> > b/meta/recipes-extended/gperf/gperf_3.1.bb
> > index 82750fca05c..3564ac0805b 100644
> > --- a/meta/recipes-extended/gperf/gperf_3.1.bb
> > +++ b/meta/recipes-extended/gperf/gperf_3.1.bb
> > @@ -9,6 +9,8 @@ SRC_URI = "${GNU_MIRROR}/${BPN}/${BP}.tar.gz"
> > SRC_URI[md5sum] = "9e251c0a618ad0824b51117d5d9db87e"
> > SRC_URI[sha256sum] =
> > "588546b945bba4b70b6a3a616e80b4ab466e3f33024a352fc2198112cdbb3ae2"
> >
> > +SRC_URI:append = " file://reproducibility.patch"
> > +
>
> Can I ask if :append was really necessary here, couldn't we just have
> used += ? I'm a little saddened with all these :append/:prepend/:remove
> creeping in to OE-Core.
I agree, I should and usually do better. I'll take a patch to fix it.
Cheers,
Richard
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#167649):
https://lists.openembedded.org/g/openembedded-core/message/167649
Mute This Topic: https://lists.openembedded.org/mt/92164384/21656
Group Owner: openembedded-core+ow...@lists.openembedded.org
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub
[arch...@mail-archive.com]
-=-=-=-=-=-=-=-=-=-=-=-
