Bug#998344: buster-pu: package llvm-toolchain-11/1:11.0.1-2~deb10u1
Hi Adam,
On Wed, Nov 03, 2021 at 02:20:35PM +0000, Adam D. Barratt wrote:
> On Tue, 2021-11-02 at 13:28 -0400, Roberto C. Sanchez wrote:
> > In order to support the update of rustc in buster, which in turn is
> > needed to support the updates of firefox-esr and thunderbird, I am
> > proposing an update of llvm-toolchain-11 in buster. The attached
> > diff represents the change from the current package in the buster-
> > backports repository.
>
> That diff appears to be between the git branches, rather than the
> generated packages. Would it be possible to have a source debdiff
> between your base and the package you're planning to upload?
>
I rebased my changes on 11.0.1-2 from buster. The debdiff attached to
this email represents the updated packge I am proposing for upload.
Note that I also updated the version of the proposed package to
11.0.1-2+deb10u1.
> Part of the reason that we request a debdiff rather than a VCS diff is
> that they can often reveal unexpected differences, for instance due to
> build system differences. In this case, the diff between the source
> package in bullseye and that uploaded to buster-backports includes 128
> generated files under debian/, which shouldn't really be in the source
> package.
>
> > As a result of mips build failures with the backport package, I am
> > running a test build on a mips porter box to verify that the mips
> > changes result in a successfully built package.
>
> How did that go?
>
It failed at the very end, but the failure seems to be spurious. That
is, I did a full build (dpkg-buildpackage with no options) rather than
an arch:any build. Some components are not built for mips (and other
architectures) and trying to build arch:all packages on those
architectures would actually fail because the components end up not
being built.
I started a new build with the --build=any option to dpkg-builpackage.
If you would like to wait for the result of that build, I am happy to
wait on uploading. However, I am confident that the changes I have in
the attached debdiff are completely ready for upload.
Regards,
-Roberto
--
Roberto C. Sánchez
diff -Nru llvm-toolchain-11-11.0.1/debian/changelog llvm-toolchain-11-11.0.1/debian/changelog
--- llvm-toolchain-11-11.0.1/debian/changelog 2021-01-06 14:16:26.000000000 -0500
+++ llvm-toolchain-11-11.0.1/debian/changelog 2021-10-30 13:14:49.000000000 -0400
@@ -1,3 +1,11 @@
+llvm-toolchain-11 (1:11.0.1-2+deb10u1) buster; urgency=medium
+
+ * Backport to buster.
+ - Disable tests on (big endian) mips due to timeout (i.e., test runtime
+ exceeds 10h).
+
+ -- Roberto C. Sánchez <roberto@debian.org> Sat, 30 Oct 2021 13:14:49 -0400
+
llvm-toolchain-11 (1:11.0.1-2) unstable; urgency=medium
* Fix the changelog
diff -Nru llvm-toolchain-11-11.0.1/debian/clang-tools-X.Y.install.in llvm-toolchain-11-11.0.1/debian/clang-tools-X.Y.install.in
--- llvm-toolchain-11-11.0.1/debian/clang-tools-X.Y.install.in 2020-11-01 04:19:28.000000000 -0500
+++ llvm-toolchain-11-11.0.1/debian/clang-tools-X.Y.install.in 2021-10-30 13:14:49.000000000 -0400
@@ -32,7 +32,7 @@
usr/lib/llvm-@LLVM_VERSION@/bin/clang-move
usr/lib/llvm-@LLVM_VERSION@/bin/clang-offload-wrapper
-[!armel !armhf !ppc64el !hurd-any !s390x !powerpc !ppc64 !mipsel !mips64el !sparc64 !riscv64] usr/lib/llvm-@LLVM_VERSION@/lib/clang/@LLVM_VERSION_FULL@/bin/hwasan_symbolize
+[!armel !armhf !ppc64el !hurd-any !s390x !powerpc !ppc64 !mips !mipsel !mips64el !sparc64 !riscv64] usr/lib/llvm-@LLVM_VERSION@/lib/clang/@LLVM_VERSION_FULL@/bin/hwasan_symbolize
clang/tools/scan-build-@LLVM_VERSION@ usr/share/clang/
clang/tools/scan-build-py-@LLVM_VERSION@ usr/share/clang/
diff -Nru llvm-toolchain-11-11.0.1/debian/rules llvm-toolchain-11-11.0.1/debian/rules
--- llvm-toolchain-11-11.0.1/debian/rules 2021-01-06 03:25:29.000000000 -0500
+++ llvm-toolchain-11-11.0.1/debian/rules 2021-10-30 13:14:49.000000000 -0400
@@ -196,7 +196,7 @@
endif
# llvm tests timeout, disable it on mipsel
-ifeq (mipsel,$(DEB_HOST_ARCH))
+ifneq (,$(filter $(DEB_HOST_ARCH), mips mipsel))
RUN_TEST=no
endif
Reply to: