[PATCH] Make headers_check less chatty for success cases

[PATCH] Make headers_check less chatty for success cases

[Satyam Sharma]

Hi David, Sam,

"make headers_check" is too verbose. It likes to chat even if it has
just _successfully_ checked a header, and not just on seeing errors.
What is worse, even if you touch just one little header in some corner
(or possibly none) and type "make" on an "already-made" tree,
headers_check still feels obliged to check all the headers.

Both the above things combined mean that when you're working on your own
little subsystem, and simply "make" an already-made tree, and turn around
to speak to a colleague, then by the time you look back at the screen, it
does not contain any useful warnings from the actual compilation stage.
If you're unlucky, the entire scroll-back history of the xterm would have
also got polluted with headers_check's "CHECK include/linux/foo.h" spam.

As a result I've been totally put off by it and have CONFIG_HEADERS_CHECK=n
set in my .config -- but that could lead to breakages. Hence:


[PATCH] Make headers_check less chatty for success cases

We still echo verbosely (and fail, obviously) for any errors encountered.

Signed-off-by: Satyam Sharma <satyam@infradead.org>

---

[ just removing that line also works, but that loses symmetry with others ]

 scripts/Makefile.headersinst |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/scripts/Makefile.headersinst b/scripts/Makefile.headersinst
index 53dae3e..0a23dd9 100644
--- a/scripts/Makefile.headersinst
+++ b/scripts/Makefile.headersinst
@@ -96,7 +96,7 @@ quiet_cmd_unifdef	  = UNIFDEF $(patsubst $(INSTALL_HDR_PATH)/%,%,$@)
       cmd_unifdef	  = $(UNIFDEF) $(patsubst $(INSTALL_HDR_PATH)/$(_dst)/%,$(srctree)/$(obj)/%,$@) \
 				   | $(HDRSED) > $@ || :
 
-quiet_cmd_check		  = CHECK   $(patsubst $(INSTALL_HDR_PATH)/$(_dst)/.check.%,$(_dst)/%,$@)
+quiet_cmd_check		  =
       cmd_check		  = $(CONFIG_SHELL) $(srctree)/scripts/hdrcheck.sh \
                               $(INSTALL_HDR_PATH)/include $(subst /.check.,/,$@) $@
 

Re: [PATCH] Make headers_check less chatty for success cases

[David Woodhouse]

On Tue, 2007-07-31 at 22:10 +0530, Satyam Sharma wrote:
> "make headers_check" is too verbose. It likes to chat even if it has
> just _successfully_ checked a header, and not just on seeing errors.
> What is worse, even if you touch just one little header in some corner
> (or possibly none) and type "make" on an "already-made" tree,
> headers_check still feels obliged to check all the headers.

Didn't Sam have a better fix for that?

-- 
dwmw2

Re: [PATCH] Make headers_check less chatty for success cases

[Sam Ravnborg]

On Tue, Jul 31, 2007 at 05:30:09PM +0100, David Woodhouse wrote:
> On Tue, 2007-07-31 at 22:10 +0530, Satyam Sharma wrote:
> > "make headers_check" is too verbose. It likes to chat even if it has
> > just _successfully_ checked a header, and not just on seeing errors.
> > What is worse, even if you touch just one little header in some corner
> > (or possibly none) and type "make" on an "already-made" tree,
> > headers_check still feels obliged to check all the headers.
> 
> Didn't Sam have a better fix for that?
In another .git here but not yet quite ready.
Got sidetracked by even more section mismatch stuff. If I had knew
how much work that had generated I had not integrated these harmless
scripts in modpost I think ;-)

I will give it a spin soon and post an update.

	Sam

Re: [PATCH] Make headers_check less chatty for success cases

[Satyam Sharma]

On Tue, 31 Jul 2007, David Woodhouse wrote:

> On Tue, 2007-07-31 at 22:10 +0530, Satyam Sharma wrote:
> > "make headers_check" is too verbose. It likes to chat even if it has
> > just _successfully_ checked a header, and not just on seeing errors.
> > What is worse, even if you touch just one little header in some corner
> > (or possibly none) and type "make" on an "already-made" tree,
> > headers_check still feels obliged to check all the headers.
> 
> Didn't Sam have a better fix for that?

None that I see in kbuild.git, at least ... Anyway, now that I've silenced
headers_check, I keep seeing:

make[2]: `scripts/unifdef' is up to date.

which is because of the:

make -f scripts/Makefile.build obj=scripts scripts/unifdef
                       (passed separately) ^^^^^^^^^^^^^^^

from the top-level Makefile (scripts/Makefile says something about it
being built only on demand ...)