This is the mail archive of the libc-alpha@sourceware.org mailing list for the glibc project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Use $(NM) not nm in tst-cancel-wrappers


On Thu, Oct 18, 2012 at 9:11 PM, Joseph S. Myers
<joseph@codesourcery.com> wrote:
> This is a first piece split out of the cross-testing changes: using $(NM)
> instead of hardcoded "nm" in tst-cancel-wrappers.  (The configure test for
> nm was introduced earlier this year, but previously the result was only
> used in configure and not substituted anywhere.)
>
> Tested x86_64.
>
> 2012-10-18  Jim Blandy  <jimb@codesourcery.com>
>             Joseph Myers  <joseph@codesourcery.com>
>
>         * config.make.in (NM): New variable.
>
> nptl:
> 2012-10-18  Jim Blandy  <jimb@codesourcery.com>
>             Joseph Myers  <joseph@codesourcery.com>
>
>         * Makefile ($(objpfx)tst-cancel-wrappers.out): Pass $(NM) to
>         tst-cancel-wrappers.sh.
>         * tst-cancel-wrappers.sh: Use nm program given as first argument,
>         not hardcoded "nm".

This looks good to me, and bordering on "so obviously good as not
needing review."

One minor nit.

> diff --git a/config.make.in b/config.make.in
> index b6d1c94..c22acf4 100644
> --- a/config.make.in
> +++ b/config.make.in
> @@ -111,6 +111,7 @@ CPPFLAGS-config = @CPPFLAGS@
>  CPPUNDEFS = @CPPUNDEFS@
>  ASFLAGS-config = @ASFLAGS_config@
>  AR = @AR@
> +NM = @NM@
>  MAKEINFO = @MAKEINFO@
>  AS = $(CC) -c
>  MIG = @MIG@
> diff --git a/nptl/Makefile b/nptl/Makefile
> index 184057b..bc563f8 100644
> --- a/nptl/Makefile
> +++ b/nptl/Makefile
> @@ -587,7 +587,8 @@ ifeq (no,$(cross-compiling))
>  ifeq (yes,$(build-shared))
>  tests: $(objpfx)tst-cancel-wrappers.out
>  $(objpfx)tst-cancel-wrappers.out: tst-cancel-wrappers.sh
> -       $(SHELL) $< $(common-objpfx)libc_pic.a \
> +       $(SHELL) $< '$(NM)' \
> +                   $(common-objpfx)libc_pic.a \
>                     $(common-objpfx)libc.a \
>                     $(objpfx)libpthread_pic.a \
>                     $(objpfx)libpthread.a > $@
> diff --git a/nptl/tst-cancel-wrappers.sh b/nptl/tst-cancel-wrappers.sh
> index dad2a67..a937efd 100644
> --- a/nptl/tst-cancel-wrappers.sh
> +++ b/nptl/tst-cancel-wrappers.sh
> @@ -1,6 +1,6 @@
>  #! /bin/sh
>  # Test whether all cancelable functions are cancelable.
> -# Copyright (C) 2002, 2003, 2007 Free Software Foundation, Inc.
> +# Copyright (C) 2002-2012 Free Software Foundation, Inc.
>  # This file is part of the GNU C Library.
>  # Contributed by Jakub Jelinek <jakub@redhat.com>, 2002.
>
> @@ -18,8 +18,9 @@
>  # License along with the GNU C Library; if not, see
>  # <http://www.gnu.org/licenses/>.
>
> +nm="$1"; shift
>  while [ $# -gt 0 ]; do
> -  ( nm -P $1; echo 'end[end]:' ) | gawk ' BEGIN {
> +  ( $nm -P $1; echo 'end[end]:' ) | gawk ' BEGIN {

I would rather this variable be called something other
than nm, such that it clearly fails if the "$" is left
off by accident.

I would accept NM, or NM_BIN or anything else like that.

>  C["accept"]=1
>  C["close"]=1
>  C["connect"]=1

Cheers,
Carlos.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]