Comment 0 for bug 1930125

Revision history for this message
Matthias Klose (doko) wrote : SRU: update gcc-9 and gcc-10 packages for focal and groovy

as done before in LP: #1879481 with GCC 9.3.0 and GCC 10.2.0, updating again to GCC 9.4.0 and GCC 10.3.0 in 20.04 LTS and 20.10.

 - gcc-10, update to the gcc-10.3.0 release
 - gcc-9, update to the gcc-9.4.0 release
 - corresponding gcc-cross packages

The package builds are prepared in
built with only the security pocket enabled.

Provide an upstream release for GCC 10.3, and an upstream release for GCC 9.4.
gcc-9 is the default compiler for focal, gcc-10 the default compiler for groovy.

[Regression Potential]
There is regression potential, however with a test rebuild of the main portion of the archive for both focal and groovy, we make sure that we don't see any regressions in package builds.

[Test Case]
No regressions in the GCC test suite, and no regressions in the test
rebuild of the main component of the archive (all architectures).

Also check that the rebuilt kernel boots on all architectures.

TODO: As a test, a complete archive rebuild was performed, and no regressions were found with this new package. The archive rebuild also contained updated versions of gcc-9, gcc-10, python3.8, and python3-stdlib-extensions. The GCC and Python packages should not infer with each other.

Analyze the GCC build logs for regressions.
For the cross builds, just make sure that the packages build using
the updated sources.

TODO: Summary of the test rebuilds:

The first one is a reference build, the second one the test rebuild with the updated components.

There are no additional regressions except for pyxdg, which is unrelated to the GCC update, and which was fixed by a separate SRU. Note that the riscv64 builds are purely informational, and are not checked for any regressions.