DPDK patches and discussions
 help / color / mirror / Atom feed
From: Thomas Monjalon <thomas.monjalon@6wind.com>
To: Ferruh Yigit <ferruh.yigit@intel.com>
Cc: dev@dpdk.org
Subject: Re: [dpdk-dev] [PATCH] mk: parallelize make config
Date: Sun, 29 Jan 2017 16:29:35 +0100	[thread overview]
Message-ID: <1800797.YCEvINBUOG@xps13> (raw)
In-Reply-To: <20170122015034.19824-1-ferruh.yigit@intel.com>

2017-01-22 01:50, Ferruh Yigit:
> make config dependency resolving was always running serial,
> parallelize it for better performance.

It could be interesting to explain why it was not parallelized,
and how you made it possible.

The test script should be updated as below:

--- a/devtools/test-build.sh
+++ b/devtools/test-build.sh
-               make T=$2 O=$1 config
+               make -j$J T=$2 O=$1 config


> --- a/mk/internal/rte.depdirs-post.mk
> +++ b/mk/internal/rte.depdirs-post.mk
> @@ -29,11 +29,12 @@
>  #   (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
>  #   OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
>  
> -.PHONY: depdirs
> -depdirs:
> -	@for d in $(DEPDIRS-y); do \
> -		$(RTE_SDK)/buildtools/depdirs-rule.sh $(S) $$d ; \
> -	done
> +.PHONY: depdirs $(DEPDIRS-y)
> +depdirs: $(DEPDIRS-y)
> +	@echo ""

Why this echo "" ?

> +
> +$(DEPDIRS-y):
> +	@$(RTE_SDK)/buildtools/depdirs-rule.sh $(S) $@


> --- a/mk/rte.sdkdepdirs.mk
> +++ b/mk/rte.sdkdepdirs.mk
> @@ -36,19 +36,22 @@ ifeq (,$(wildcard $(RTE_OUTPUT)/Makefile))
>    $(error "need a make config first")
>  endif
>  
> +DEPDIRS = $(addsuffix /.depdirs, $(addprefix $(BUILDDIR)/,$(ROOTDIRS-y)))

These DEPDIRS are files, although DEPDIRS in other contexts are directories.
I think it should be renamed. DEPDIR_FILES?

>  # use a "for" in a shell to process dependencies: we don't want this
>  # task to be run in parallel.

You forgot to remove this obsolete comment.

>  .PHONY: depdirs
>  depdirs: $(RTE_OUTPUT)/.depdirs
> -$(RTE_OUTPUT)/.depdirs: $(RTE_OUTPUT)/.config
> -	@rm -f $(RTE_OUTPUT)/.depdirs ; \
> -	for d in $(ROOTDIRS-y); do \
> -		if [ -f $(RTE_SRCDIR)/$$d/Makefile ]; then \
> -			[ -d $(BUILDDIR)/$$d ] || mkdir -p $(BUILDDIR)/$$d ; \
> -			$(MAKE) S=$$d -f $(RTE_SRCDIR)/$$d/Makefile depdirs \
> -				>> $(RTE_OUTPUT)/.depdirs ; \
> -		fi ; \
> -	done
> +$(RTE_OUTPUT)/.depdirs: $(DEPDIRS)
> +	@rm -f $@
> +	@for f in $(DEPDIRS); do cat $$f >> $@; done
> +	@sort -u -o $@ $@
> +
> +$(DEPDIRS): $(RTE_OUTPUT)/.config
> +	@f=$(lastword $(subst /, ,$(dir $@))); \

Could you use $(notdir $(@D)) ?

> +	[ -d $(BUILDDIR)/$$f ] || mkdir -p $(BUILDDIR)/$$f; \
> +	rm -f $@; \

Why this removal?

> +	$(MAKE) S=$$f -f $(RTE_SRCDIR)/$$f/Makefile depdirs >> $@

This part is a bit complicated.
Could it be simplified by better naming $f?


> --- a/mk/rte.subdir.mk
> +++ b/mk/rte.subdir.mk
> @@ -76,7 +76,7 @@ clean: _postclean
>  # include .depdirs and define rules to order priorities between build
>  # of directories.
>  #
> -include $(RTE_OUTPUT)/.depdirs
> +-include $(RTE_OUTPUT)/.depdirs
>  
>  define depdirs_rule
>  $(1): $(sort $(patsubst $(S)/%,%,$(LOCAL_DEPDIRS-$(S)/$(1))))
> @@ -84,16 +84,15 @@ endef
>  
>  $(foreach d,$(DIRS-y),$(eval $(call depdirs_rule,$(d))))
>  
> +DEPDIRS = $(wildcard $(addprefix $(S)/,$(DIRS-y)))
>  
>  # use a "for" in a shell to process dependencies: we don't want this
>  # task to be run in parallel.

You forgot to remove this obsolete comment.

> -.PHONY: depdirs
> -depdirs:
> -	@for d in $(DIRS-y); do \
> -		if [ -f $(SRCDIR)/$$d/Makefile ]; then \
> -			$(MAKE) S=$S/$$d -f $(SRCDIR)/$$d/Makefile depdirs ; \
> -		fi ; \
> -	done
> +.PHONY: depdirs $(DEPDIRS)
> +depdirs: $(DEPDIRS)
> +
> +$(DEPDIRS):
> +	@$(MAKE) S=$@ -f $(RTE_SRCDIR)/$@/Makefile depdirs

  parent reply	other threads:[~2017-01-29 15:29 UTC|newest]

Thread overview: 38+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2017-01-22  1:50 Ferruh Yigit
2017-01-23 17:18 ` Olivier Matz
2017-01-23 17:19   ` [dpdk-dev] [PATCH] mk: optimize directory dependencies Olivier Matz
2017-01-24 11:19     ` Robin Jarry
2017-01-24 11:26       ` Bruce Richardson
2017-01-24 12:31         ` Robin Jarry
2017-01-24 11:40     ` Jerin Jacob
2017-01-24 12:15       ` Bruce Richardson
2017-01-24 12:56         ` Jerin Jacob
2017-01-24 13:26           ` Richardson, Bruce
2017-01-24 14:50             ` Olivier MATZ
2017-01-24 14:55               ` Wiles, Keith
2017-03-01 11:25                 ` Thomas Monjalon
2017-03-01 12:10                   ` Bruce Richardson
2017-03-01 12:30                   ` Olivier Matz
2017-01-24 13:05     ` Ferruh Yigit
2017-03-17 17:13     ` Olivier Matz
2017-03-17 17:47       ` Robin Jarry
2017-03-20  8:31         ` Olivier Matz
2017-03-24 13:21       ` [dpdk-dev] [PATCH v2] " Olivier Matz
2017-03-27 21:33         ` Thomas Monjalon
2017-03-28 10:34         ` Ferruh Yigit
2017-03-30  8:51           ` Olivier Matz
2017-03-30  9:27             ` Ferruh Yigit
2017-03-30 12:11               ` Olivier Matz
2017-03-30 12:32                 ` [dpdk-dev] [PATCH] mk: fix dependencies to optional configs Olivier Matz
2017-03-30 12:37                   ` Ferruh Yigit
2017-03-30 13:37                     ` Thomas Monjalon
2017-01-23 17:50   ` [dpdk-dev] [PATCH] mk: parallelize make config Wiles, Keith
2017-01-24  8:42     ` Olivier MATZ
2017-01-24 10:02       ` Bruce Richardson
2017-01-23 19:03 ` Michał Mirosław
2017-01-30  9:41   ` Ferruh Yigit
2017-01-24 10:52 ` Bruce Richardson
2017-01-29 15:29 ` Thomas Monjalon [this message]
2017-01-30  9:46   ` Ferruh Yigit
2017-01-30 10:21 ` [dpdk-dev] [PATCH v2] " Ferruh Yigit
2017-01-30 18:13   ` Thomas Monjalon

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=1800797.YCEvINBUOG@xps13 \
    --to=thomas.monjalon@6wind.com \
    --cc=dev@dpdk.org \
    --cc=ferruh.yigit@intel.com \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).