From: Petr Vorel <pvorel@suse.cz>
To: Andrea Cervesato <andrea.cervesato@suse.de>
Cc: Linux Test Project <ltp@lists.linux.it>
Subject: Re: [LTP] [PATCH] build: Parallelize subdirectory recursion in generic_trunk_target
Date: Tue, 20 Oct 2026 22:15:31 +0200 [thread overview]
Message-ID: <20261020201531.GA154402@pevik> (raw)
In-Reply-To: <20261020-build_system_optimize-v1-1-744f9fa88908@suse.com>
Hi Andrea,
thanks for the speedup.
Reviewed-by: Petr Vorel <pvorel@suse.cz>
...
> +# Recurse into subdirectories in parallel instead of using a serial shell loop.
> +# Each subdirectory is mapped to a phony target (<dir>-all, <dir>-install, etc.).
> +# GNU Make can then schedule sibling directories concurrently under -jN.
> +all: $(addsuffix -all,$(SUBDIRS))
> +install: $(addsuffix -install,$(SUBDIRS))
> +check: $(addsuffix -check,$(SUBDIRS))
> +
> +.PHONY: $(addsuffix -all,$(SUBDIRS)) $(addsuffix -install,$(SUBDIRS)) $(addsuffix -check,$(SUBDIRS))
> +
> +# For each target (e.g. 'open-all'), strip the suffix to get the directory name,
> +# build trunk-local targets first, and ensure the subdirectory exists.
> +$(addsuffix -all,$(SUBDIRS)): trunk-all | $(SUBDIRS) $(MAKE_DEPS)
> ifdef VERBOSE
> - @set -e; for dir in $(SUBDIRS); do \
> - $(MAKE) -C $$dir -f "$(abs_srcdir)/$$dir/Makefile" $@; \
> - done
Yeah, replacing shell for loop which is single process only to make processing
makes sense.
> + $(MAKE) -C "$(patsubst %-all,%,$@)" -f "$(abs_srcdir)/$(patsubst %-all,%,$@)/Makefile" all
> else
> - @set -e; for dir in $(SUBDIRS); do \
> - $(MAKE) --no-print-directory -C $$dir -f "$(abs_srcdir)/$$dir/Makefile" $@; \
> - done
> + @$(MAKE) --no-print-directory -C "$(patsubst %-all,%,$@)" -f "$(abs_srcdir)/$(patsubst %-all,%,$@)/Makefile" all
> +endif
> +
> +$(addsuffix -install,$(SUBDIRS)): trunk-install | $(SUBDIRS) $(MAKE_DEPS)
> +ifdef VERBOSE
> + $(MAKE) -C "$(patsubst %-install,%,$@)" -f "$(abs_srcdir)/$(patsubst %-install,%,$@)/Makefile" install
> +else
> + @$(MAKE) --no-print-directory -C "$(patsubst %-install,%,$@)" -f "$(abs_srcdir)/$(patsubst %-install,%,$@)/Makefile" install
> +endif
> +
> +$(addsuffix -check,$(SUBDIRS)): trunk-check | $(SUBDIRS) $(MAKE_DEPS)
> +ifdef VERBOSE
> + $(MAKE) -C "$(patsubst %-check,%,$@)" -f "$(abs_srcdir)/$(patsubst %-check,%,$@)/Makefile" check
> +else
> + @$(MAKE) --no-print-directory -C "$(patsubst %-check,%,$@)" -f "$(abs_srcdir)/$(patsubst %-check,%,$@)/Makefile" check
> endif
> endif
> diff --git a/testcases/kernel/Makefile b/testcases/kernel/Makefile
> index 29e26742d..3bf5e433e 100644
> --- a/testcases/kernel/Makefile
> +++ b/testcases/kernel/Makefile
> @@ -54,3 +54,6 @@ FILTER_OUT_DIRS += containers controllers device-drivers fs io ipc mem \
> endif
> include $(top_srcdir)/include/mk/generic_trunk_target.mk
> +
> +# libkerntest.a in lib/ must be built before syscalls and mem consumers
> +syscalls-all mem-all: lib-all
I wonder if this can be autogenerated, because this was not needed before.
I'm ok with this explicit definition. Probably in the future we should move it
to libs/ (using it as LTPLIBS), which will solve the dependency.
> diff --git a/testcases/network/rpc/Makefile b/testcases/network/rpc/Makefile
> index 6ecb1a665..8d12bdf1e 100644
> --- a/testcases/network/rpc/Makefile
> +++ b/testcases/network/rpc/Makefile
> @@ -9,3 +9,6 @@ include $(top_srcdir)/include/mk/env_pre.mk
> INSTALL_TARGETS := *.sh
> include $(top_srcdir)/include/mk/generic_trunk_target.mk
> +
> +# librpc-tirpc.a in rpc-tirpc/ must be built before basic_tests
> +basic_tests-all: rpc-tirpc-all
Also librpc-tirpc.a should be moved into libs/ (if we still keep this code and
not move it to libtirpc project where it'd make more sense).
Kind regards,
Petr
--
Mailing list info: https://lists.linux.it/listinfo/ltp
next prev parent reply other threads:[~2026-10-20 20:15 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-10-20 10:52 [LTP] [PATCH] build: Parallelize subdirectory recursion in generic_trunk_target Andrea Cervesato
2026-10-20 14:49 ` [LTP] " linuxtestproject.agent
2026-10-20 20:15 ` Petr Vorel [this message]
2026-10-21 6:11 ` [LTP] [PATCH] " Andrea Cervesato via ltp
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20261020201531.GA154402@pevik \
--to=pvorel@suse.cz \
--cc=andrea.cervesato@suse.de \
--cc=ltp@lists.linux.it \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.