[LTP] [PATCH v2 0/3] Speedup LTP compile time by ~20% on multi-machine

[Andrea Cervesato]

The following results show how this optimization moves compile times
really far from what we have now. With a single threaded build, LTP
would take approximately 12 minutes to compile. With a multi-threaded
build on 16 CPUs, this drops down to ~46 seconds.

time make -j16

Executed in   45.96 secs    fish           external
   usr time  519.11 secs    0.24 millis  519.11 secs
   sys time  167.13 secs    1.08 millis  167.13 secs

Signed-off-by: Andrea Cervesato <andrea.cervesato@suse.com>
---
Changes in v2:
- fix build race condition
- Link to v1: https://lore.kernel.org/20261020-build_system_optimize-v1-1-744f9fa88908@suse.com

---
Andrea Cervesato (3):
      build: Fix library prerequisites in testcases.mk
      build: Build shared libs once on subtree make
      build: Parallelize subdirectory recursion in generic_trunk_target

 Makefile                            |  2 ++
 include/mk/generic_trunk_target.inc | 58 ++++++++++++++++++++++++++++++++-----
 include/mk/testcases.mk             | 14 +++++----
 testcases/kernel/Makefile           |  3 ++
 testcases/kernel/include/lib.mk     |  4 +--
 testcases/network/rpc/Makefile      |  3 ++
 6 files changed, 69 insertions(+), 15 deletions(-)
---
base-commit: 121876b3da3ff59b0dd17b4798b2e4699ba431d1
change-id: 20261020-build_system_optimize-6cd0343f08e1

Best regards,
--  
Andrea Cervesato <andrea.cervesato@suse.com>