public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
From: Arnaldo Carvalho de Melo <acme@redhat.com>
To: Bernd Petrovitsch <bernd@sysprog.at>
Cc: Peter Zijlstra <peterz@infradead.org>,
	Kusanagi Kouichi <slash@ac.auone-net.jp>,
	Paul Mackerras <paulus@samba.org>, Ingo Molnar <mingo@elte.hu>,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH] perf tools: Don't use brace expansion.
Date: Tue, 17 Aug 2010 15:16:54 -0300	[thread overview]
Message-ID: <20100817181654.GH13095@ghostprotocols.net> (raw)
In-Reply-To: <1282061389.5822.22.camel@thorin>

Em Tue, Aug 17, 2010 at 06:09:49PM +0200, Bernd Petrovitsch escreveu:
> On Die, 2010-08-17 at 12:42 -0300, Arnaldo Carvalho de Melo wrote:
> > +# we compile into subdirectories. if the target directory is not the source directory, they might not exists. So
> > +# we depend the various files onto their directories.
> > +$(LIB_OBJS) $(BUILTIN_OBJS) $(OUTPUT)PERF-VERSION-FILE $(OUTPUT)common-cmds.h: $(sort $(dir $(LIB_OBJS) $(BUILTIN_OBJS)))
 
> Hmm, that adds that $(OUTPUT)PERF-VERSION-FILE and
> $(OUTPUT)common-cmds.h (also) depends on the subdirectories of the other
> objects (and thus it works always because at least one of them is a
> subdirectory of $(OUTPUT)).
> To be 110% anal, it should look like e.g.
> +$(LIB_OBJS) $(BUILTIN_OBJS) $(OUTPUT)PERF-VERSION-FILE $(OUTPUT)common-cmds.h: $(sort $(dir $(LIB_OBJS) $(BUILTIN_OBJS)) $(OUTPUT))

> or actually all targets.

> > +# In the second step, we make a rule to actually create these directories
> > +$(sort $(dir $(LIB_OBJS) $(BUILTIN_OBJS))):
> > +       mkdir -p $@ 2>/dev/null

> BTW there is no automatic variable or other make-construct to refer in
> the dependencies on the own target. Therefore the copy-paste-the-
> variables solution.

> Perhaps an additional variable reduces clutter (and eases maintenance)?

> +# we compile into subdirectories. if the target directory is not the source directory, they might not exists. So
> +# we depend the various files onto their directories.
> +DIRECTORY_DEPS = $(LIB_OBJS) $(BUILTIN_OBJS) $(OUTPUT)PERF-VERSION-FILE $(OUTPUT)common-cmds.h
> +$(DIRECTORY_DEPS): $(sort $(dir $(DIRECTORY_DEPS)))
> +# In the second step, we make a rule to actually create these directories
> +$(sort $(dir $(DIRECTORY_DEPS))):
> +       mkdir -p $@ 2>/dev/null

> That should guarantee that all directories from the targets are created.

> > As it was failing when I did:

> > rm -rf ~/build/perf
> > make -C tools/perf O=~/build/perf

> > With that it retains the existing functionality,

> Ah, I `mkdir`ed the output directory explicitly before the `make` (and
> after the `rm -rf`).

> BTW which is the preferred tree to base patches on (for the "perf"
> subsystem)?

git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip

perf/urgent for 2.6.36
perf/core for 2.6.37

I'm targeting perf/urgent for this fix.

Or git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux-2.6

Sometimes, when Ingo is busy and patches accumulate, perf/{core,urgent}
are where I put things to go to the branches with the same names on his
tree.

Please look if patch is OK with you.

- Arnaldo

commit 033a273f9836b592dd568abd0f655be469d66704
Author: Bernd Petrovitsch <bernd@sysprog.at>
Date:   Tue Aug 17 12:22:08 2010 -0300

    perf tools: Fix build on POSIX shells
    
    POSIX sh does not specify the brace expansion, so fix it by replacing the
    global $(shell ...) lines quite at the top creating the output directories with
    real rules.
    
    Cc: Ingo Molnar <mingo@elte.hu>
    Cc: Kusanagi Kouichi <slash@ac.auone-net.jp>
    Cc: Peter Zijlstra <peterz@infradead.org>
    Cc: Paul Mackerras <paulus@samba.org>
    LKML-Reference: <1282046280.5822.4.camel@thorin>
    Signed-off-by: Bernd Petrovitsch <bernd@sysprog.at>
    Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

diff --git a/tools/perf/Makefile b/tools/perf/Makefile
index 41abb90..dcb9700 100644
--- a/tools/perf/Makefile
+++ b/tools/perf/Makefile
@@ -157,10 +157,6 @@ all::
 #
 # Define NO_DWARF if you do not want debug-info analysis feature at all.
 
-$(shell sh -c 'mkdir -p $(OUTPUT)scripts/{perl,python}/Perf-Trace-Util/' 2> /dev/null)
-$(shell sh -c 'mkdir -p $(OUTPUT)util/{ui/browsers,scripting-engines}/' 2> /dev/null)
-$(shell sh -c 'mkdir $(OUTPUT)bench' 2> /dev/null)
-
 $(OUTPUT)PERF-VERSION-FILE: .FORCE-PERF-VERSION-FILE
 	@$(SHELL_PATH) util/PERF-VERSION-GEN $(OUTPUT)
 -include $(OUTPUT)PERF-VERSION-FILE
@@ -186,8 +182,6 @@ ifeq ($(ARCH),x86_64)
         ARCH := x86
 endif
 
-$(shell sh -c 'mkdir -p $(OUTPUT)arch/$(ARCH)/util/' 2> /dev/null)
-
 # CFLAGS and LDFLAGS are for the users to override from the command line.
 
 #
@@ -268,6 +262,7 @@ export prefix bindir sharedir sysconfdir
 CC = $(CROSS_COMPILE)gcc
 AR = $(CROSS_COMPILE)ar
 RM = rm -f
+MKDIR = mkdir
 TAR = tar
 FIND = find
 INSTALL = install
@@ -838,6 +833,7 @@ ifndef V
 	QUIET_CC       = @echo '   ' CC $@;
 	QUIET_AR       = @echo '   ' AR $@;
 	QUIET_LINK     = @echo '   ' LINK $@;
+	QUIET_MKDIR    = @echo '   ' MKDIR $@;
 	QUIET_BUILT_IN = @echo '   ' BUILTIN $@;
 	QUIET_GEN      = @echo '   ' GEN $@;
 	QUIET_SUBDIR0  = +@subdir=
@@ -1012,6 +1008,14 @@ $(LIB_OBJS) $(BUILTIN_OBJS): $(LIB_H)
 $(patsubst perf-%$X,%.o,$(PROGRAMS)): $(LIB_H) $(wildcard */*.h)
 builtin-revert.o wt-status.o: wt-status.h
 
+# we compile into subdirectories. if the target directory is not the source directory, they might not exists. So
+# we depend the various files onto their directories.
+DIRECTORY_DEPS = $(LIB_OBJS) $(BUILTIN_OBJS) $(OUTPUT)PERF-VERSION-FILE $(OUTPUT)common-cmds.h
+$(DIRECTORY_DEPS): $(sort $(dir $(DIRECTORY_DEPS)))
+# In the second step, we make a rule to actually create these directories
+$(sort $(dir $(DIRECTORY_DEPS))):
+	$(QUIET_MKDIR)$(MKDIR) -p $@ 2>/dev/null
+
 $(LIB_FILE): $(LIB_OBJS)
 	$(QUIET_AR)$(RM) $@ && $(AR) rcs $@ $(LIB_OBJS)
 

  reply	other threads:[~2010-08-17 18:17 UTC|newest]

Thread overview: 15+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-08-16 12:41 [PATCH] perf tools: Don't use brace expansion Kusanagi Kouichi
2010-08-16 13:30 ` Peter Zijlstra
2010-08-16 14:09 ` Bernd Petrovitsch
2010-08-16 14:30   ` Peter Zijlstra
2010-08-16 14:54     ` Bernd Petrovitsch
2010-08-16 15:29       ` Arnaldo Carvalho de Melo
2010-08-16 15:43         ` Bernd Petrovitsch
2010-08-16 15:50           ` Arnaldo Carvalho de Melo
2010-08-17 11:58             ` Bernd Petrovitsch
2010-08-17 15:42               ` Arnaldo Carvalho de Melo
2010-08-17 16:09                 ` Bernd Petrovitsch
2010-08-17 18:16                   ` Arnaldo Carvalho de Melo [this message]
2010-08-18  9:47                     ` Bernd Petrovitsch
2010-08-18  8:18               ` [tip:perf/urgent] perf tools: Fix build on POSIX shells tip-bot for Bernd Petrovitsch
2010-08-16 15:24 ` [PATCH] perf tools: Don't use brace expansion Arnaldo Carvalho de Melo

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=20100817181654.GH13095@ghostprotocols.net \
    --to=acme@redhat.com \
    --cc=bernd@sysprog.at \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mingo@elte.hu \
    --cc=paulus@samba.org \
    --cc=peterz@infradead.org \
    --cc=slash@ac.auone-net.jp \
    /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