[0/3] Add DWARF self-comparison tests to libabigail-tests

Message ID 20240823093150.791835-1-dodji@redhat.com
Headers
Series Add DWARF self-comparison tests to libabigail-tests |

Message

Dodji Seketeli Aug. 23, 2024, 9:31 a.m. UTC
  From: Dodji Seketeli <dodji@seketeli.org>

The libabigail-tests repository [1] contains big integration tests for
the libabigail project.  One set of tests is the self-comparison of
Linux kernels (vmlinux & modules).  For now, those self-comparison
tests have been essentially using BTF and CTF format because of their
speed.

This test adds a DWARF self-comparison test to that set.  The test
uses a stablelist to restrict the size of the result KMI (aka, Kernel
Module Interface, really what people call KABI) and speed-up the
processing.  Without such a stablelist, analyzing vmlinux and its
3000+ modules takes ~ 35 minutes.  With a stablelist of ~1200 ELF
symbols, the analysis takes less than one minute.


The first patch adds the test.  The second patch fixes typo in the
build system of the test.  The third patch adds some information to
the README.

[1]: https://sourceware.org/git/?p=libabigail-tests.git;a=shortlog;h=refs/heads/master

Dodji Seketeli (3):
  test-self-compare-linux: Add a test that reads DWARF
  configure.ac: Fix the default path to the include dir
  README: Add that built kernel is 5.15

 configure.ac                                  |    2 +-
 .../test-minimal-ctf-1/README                 |    2 +-
 .../kmi-el94-stablelist-x86_6                 | 1429 +++++++++++++++++
 test-self-compare-linux.cc                    |   90 +-
 4 files changed, 1508 insertions(+), 15 deletions(-)
 create mode 100644 data/test-self-compare-linux/test-minimal-ctf-1/kmi-el94-stablelist-x86_6