v2:
 * Fix CI failures from newly-exposed logic
 * Drop -f's from $(RM)
 * Drop the 'run' rune patch.  Its clearly controvertial, but ignoring the
   problems isn't an available option in the longterm.

All other RFC questions still outstanding.

Andrew Cooper (4):
  tools/tests: Drop obsolete mce-test infrastructure
  tests/resource: Rework Makefile
  tests/cpu-policy: Rework Makefile
  tests/xenstore: Rework Makefile

 .gitignore                                         |   2 -
 tools/misc/.gitignore                              |   1 +
 tools/misc/Makefile                                |   4 +
 tools/{tests/mce-test/tools => misc}/xen-mceinj.c  |  32 +--
 tools/tests/Makefile                               |   1 -
 tools/tests/cpu-policy/Makefile                    |  31 ++-
 tools/tests/mce-test/Makefile                      |  12 -
 tools/tests/mce-test/README                        |  75 ------
 tools/tests/mce-test/cases/srao_llc/dom0/cases.sh  |  73 ------
 tools/tests/mce-test/cases/srao_llc/guest/cases.sh |  94 --------
 tools/tests/mce-test/cases/srao_llc/xen/cases.sh   |  69 ------
 tools/tests/mce-test/cases/srao_mem/dom0/cases.sh  |  73 ------
 tools/tests/mce-test/cases/srao_mem/guest/cases.sh |  94 --------
 tools/tests/mce-test/cases/srao_mem/xen/cases.sh   |  69 ------
 tools/tests/mce-test/cases/ucna_llc/dom0/cases.sh  |  72 ------
 tools/tests/mce-test/cases/ucna_llc/guest/cases.sh |  92 --------
 tools/tests/mce-test/cases/ucna_llc/xen/cases.sh   |  68 ------
 tools/tests/mce-test/config/setup.conf             |  24 --
 tools/tests/mce-test/lib/xen-mceinj-tool.sh        | 260 ---------------------
 tools/tests/mce-test/tools/Makefile                |  24 --
 tools/tests/mce-test/tools/README                  |  24 --
 tools/tests/resource/Makefile                      |  11 +-
 tools/tests/xenstore/.gitignore                    |   1 +
 tools/tests/xenstore/Makefile                      |  31 ++-
 .../tests/xenstore/{xs-test.c => test-xenstore.c}  |   8 +-
 25 files changed, 80 insertions(+), 1165 deletions(-)
 rename tools/{tests/mce-test/tools => misc}/xen-mceinj.c (97%)
 delete mode 100644 tools/tests/mce-test/Makefile
 delete mode 100644 tools/tests/mce-test/README
 delete mode 100644 tools/tests/mce-test/cases/srao_llc/dom0/cases.sh
 delete mode 100644 tools/tests/mce-test/cases/srao_llc/guest/cases.sh
 delete mode 100644 tools/tests/mce-test/cases/srao_llc/xen/cases.sh
 delete mode 100644 tools/tests/mce-test/cases/srao_mem/dom0/cases.sh
 delete mode 100644 tools/tests/mce-test/cases/srao_mem/guest/cases.sh
 delete mode 100644 tools/tests/mce-test/cases/srao_mem/xen/cases.sh
 delete mode 100644 tools/tests/mce-test/cases/ucna_llc/dom0/cases.sh
 delete mode 100644 tools/tests/mce-test/cases/ucna_llc/guest/cases.sh
 delete mode 100644 tools/tests/mce-test/cases/ucna_llc/xen/cases.sh
 delete mode 100644 tools/tests/mce-test/config/setup.conf
 delete mode 100644 tools/tests/mce-test/lib/xen-mceinj-tool.sh
 delete mode 100644 tools/tests/mce-test/tools/Makefile
 delete mode 100644 tools/tests/mce-test/tools/README
 create mode 100644 tools/tests/xenstore/.gitignore
 rename tools/tests/xenstore/{xs-test.c => test-xenstore.c} (98%)

-- 
2.11.0

On 22.06.2021 20:21, Andrew Cooper wrote:
> v2:
>  * Fix CI failures from newly-exposed logic
>  * Drop -f's from $(RM)
>  * Drop the 'run' rune patch.  Its clearly controvertial, but ignoring the
>    problems isn't an available option in the longterm.

What is "the problem" here? The presence of the run targets in
the first place (and their wiring up from the top level
Makefile, allowing direct invocation)? If so, I'm afraid so far
I haven't seen replacement proposals by you (nor why exactly
this would be a problem).

> All other RFC questions still outstanding.

I didn't find any here or in the individual patches.

Also a remark on patches 2 ... 4 each saying "fill in the
install/uninstall rules so this test can be packaged to be
automated sensibly": Why is running (or at least picking) tests
from the build area not an option in an automated environment?
And why is installing tests unconditionally a generally good
idea? I'd view this as unnecessary bloat for the majority of
downstreams.

Jan