From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mga03.intel.com ([143.182.124.21]) by linuxtogo.org with esmtp (Exim 4.72) (envelope-from ) id 1T2MPe-0005mr-Nc for openembedded-core@lists.openembedded.org; Fri, 17 Aug 2012 15:16:31 +0200 Received: from azsmga001.ch.intel.com ([10.2.17.19]) by azsmga101.ch.intel.com with ESMTP; 17 Aug 2012 06:04:31 -0700 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="4.77,784,1336374000"; d="scan'208,217";a="182211071" Received: from dell-desktop (HELO [10.237.105.32]) ([10.237.105.32]) by azsmga001.ch.intel.com with ESMTP; 17 Aug 2012 06:04:30 -0700 Message-ID: <502E41E3.4020903@intel.com> Date: Fri, 17 Aug 2012 16:06:43 +0300 From: Radu Moisan User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:14.0) Gecko/20120714 Thunderbird/14.0 MIME-Version: 1.0 To: Patches and discussions about the oe-core layer References: <1345208727-3029-1-git-send-email-radu.moisan@intel.com> In-Reply-To: <1345208727-3029-1-git-send-email-radu.moisan@intel.com> Subject: Re: [PATCH][RFC] coreutils: Upgrade to upstream version 8.17 X-BeenThere: openembedded-core@lists.openembedded.org X-Mailman-Version: 2.1.11 Precedence: list Reply-To: Patches and discussions about the oe-core layer List-Id: Patches and discussions about the oe-core layer List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Fri, 17 Aug 2012 13:16:31 -0000 Content-Type: multipart/alternative; boundary="------------000403090100000301050406" --------------000403090100000301050406 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit The problem: The new version of coreutils, adds use of canonicalize_file_name() function defined in canonicalize.h The problem I encountered is at compile time, the function is redefined to rpl_canonicalize_file_name by means of a macro, and I get an undefined reference error. //-- stdlib.in.h #if @GNULIB_CANONICALIZE_FILE_NAME@ # if @REPLACE_CANONICALIZE_FILE_NAME@ # if !(defined __cplusplus && defined GNULIB_NAMESPACE) # define canonicalize_file_name rpl_canonicalize_file_name # endif I've looked for REPLACE_CANONICALIZE_FILE_NAME and found out that is being set in configure, which is generated from canonicalize.m4 # canonicalize.m4 serial 24 dnl Copyright (C) 2003-2007, 2009-2012 Free Software Foundation, Inc. dnl This file is free software; the Free Software Foundation dnl gives unlimited permission to copy and/or distribute it, dnl with or without modifications, as long as this notice is preserved. # Provides canonicalize_file_name and canonicalize_filename_mode, but does # not provide or fix realpath. AC_DEFUN([gl_FUNC_CANONICALIZE_FILENAME_MODE], [ AC_REQUIRE([gl_USE_SYSTEM_EXTENSIONS]) AC_CHECK_FUNCS_ONCE([canonicalize_file_name]) AC_REQUIRE([gl_DOUBLE_SLASH_ROOT]) AC_REQUIRE([gl_FUNC_REALPATH_WORKS]) if test $ac_cv_func_canonicalize_file_name = no; then HAVE_CANONICALIZE_FILE_NAME=0 elif test "$gl_cv_func_realpath_works" != yes; then REPLACE_CANONICALIZE_FILE_NAME=1 fi ]) # Provides canonicalize_file_name and realpath. AC_DEFUN([gl_CANONICALIZE_LGPL], [ AC_REQUIRE([gl_STDLIB_H_DEFAULTS]) AC_REQUIRE([gl_CANONICALIZE_LGPL_SEPARATE]) if test $ac_cv_func_canonicalize_file_name = no; then HAVE_CANONICALIZE_FILE_NAME=0 if test $ac_cv_func_realpath = no; then HAVE_REALPATH=0 elif test "$gl_cv_func_realpath_works" != yes; then REPLACE_REALPATH=1 fi elif test "$gl_cv_func_realpath_works" != yes; then REPLACE_CANONICALIZE_FILE_NAME=1 REPLACE_REALPATH=1 fi ]) Then I followed gl_cv_func_realpath_works and found it to be set to "guessing yes" # canonicalize.m4 serial 24 [case "$host_os" in # Guess yes on glibc systems. *-gnu*) gl_cv_func_realpath_works="guessing yes" ;; # If we don't know, assume the worst. *) gl_cv_func_realpath_works="guessing no" ;; esac Being set this way it looked to me like it was always assumed to be "no" and thus assumed real path didn't work. In fact, I could not make a logic for "guessing yes" and "quessing no" because to me it looks like they are not verified against and thus I'm missing the reason of their usage. However, I'm assuming I missed something because they are used quite heavily. I don't have experience with autoconf so I may have missed the "right way" to fix this. Any comments well apreciated. radu On 08/17/2012 04:05 PM, Radu Moisan wrote: > Signed-off-by: Radu Moisan > --- > .../coreutils-8.17/realpath-works-yes.patch | 13 +++++++++++++ > .../remove-gets.patch | 20 +++++++++++--------- > .../remove-usr-local-lib-from-m4.patch | 0 > .../{coreutils_8.14.bb => coreutils_8.17.bb} | 9 +++++---- > 4 files changed, 29 insertions(+), 13 deletions(-) > create mode 100644 meta/recipes-core/coreutils/coreutils-8.17/realpath-works-yes.patch > rename meta/recipes-core/coreutils/{coreutils-8.14 => coreutils-8.17}/remove-gets.patch (47%) > rename meta/recipes-core/coreutils/{coreutils-8.14 => coreutils-8.17}/remove-usr-local-lib-from-m4.patch (100%) > rename meta/recipes-core/coreutils/{coreutils_8.14.bb => coreutils_8.17.bb} (91%) > > diff --git a/meta/recipes-core/coreutils/coreutils-8.17/realpath-works-yes.patch b/meta/recipes-core/coreutils/coreutils-8.17/realpath-works-yes.patch > new file mode 100644 > index 0000000..6bbe7ee > --- /dev/null > +++ b/meta/recipes-core/coreutils/coreutils-8.17/realpath-works-yes.patch > @@ -0,0 +1,13 @@ > +Index: coreutils-8.17/m4/canonicalize.m4 > +=================================================================== > +--- coreutils-8.17.orig/m4/canonicalize.m4 2012-05-08 12:05:23.000000000 +0300 > ++++ coreutils-8.17/m4/canonicalize.m4 2012-08-17 14:20:22.000000000 +0300 > +@@ -95,7 +95,7 @@ > + [gl_cv_func_realpath_works=no], > + [case "$host_os" in > + # Guess yes on glibc systems. > +- *-gnu*) gl_cv_func_realpath_works="guessing yes" ;; > ++ *-gnu*) gl_cv_func_realpath_works="yes" ;; > + # If we don't know, assume the worst. > + *) gl_cv_func_realpath_works="guessing no" ;; > + esac > diff --git a/meta/recipes-core/coreutils/coreutils-8.14/remove-gets.patch b/meta/recipes-core/coreutils/coreutils-8.17/remove-gets.patch > similarity index 47% > rename from meta/recipes-core/coreutils/coreutils-8.14/remove-gets.patch > rename to meta/recipes-core/coreutils/coreutils-8.17/remove-gets.patch > index 4f61c92..eaadf7e 100644 > --- a/meta/recipes-core/coreutils/coreutils-8.14/remove-gets.patch > +++ b/meta/recipes-core/coreutils/coreutils-8.17/remove-gets.patch > @@ -3,19 +3,21 @@ use gets iff its defined. eglibc 2.16 removed gets > Signed-off-by: Khem Raj > Upstream-Status: Pending > > -Index: coreutils-8.14/lib/stdio.in.h > +Index: coreutils-8.17/lib/stdio.in.h > =================================================================== > ---- coreutils-8.14.orig/lib/stdio.in.h 2011-09-24 04:20:48.000000000 -0700 > -+++ coreutils-8.14/lib/stdio.in.h 2012-07-03 10:36:19.886296576 -0700 > -@@ -713,11 +713,13 @@ > - _GL_CXXALIAS_SYS (gets, char *, (char *s)); > - # undef gets > +--- coreutils-8.17.orig/lib/stdio.in.h > ++++ coreutils-8.17/lib/stdio.in.h > +@@ -698,13 +698,14 @@ _GL_WARN_ON_USE (getline, "getline is un > + "use gnulib module getline for portability"); > # endif > + #endif > +- > +# if defined gets > - _GL_CXXALIASWARN (gets); > /* It is very rare that the developer ever has full control of stdin, > - so any use of gets warrants an unconditional warning. Assume it is > - always declared, since it is required by C89. */ > + so any use of gets warrants an unconditional warning; besides, C11 > + removed it. */ > + #undef gets > + #if HAVE_RAW_DECL_GETS > _GL_WARN_ON_USE (gets, "gets is a security hole - use fgets instead"); > +# endif > #endif > diff --git a/meta/recipes-core/coreutils/coreutils-8.14/remove-usr-local-lib-from-m4.patch b/meta/recipes-core/coreutils/coreutils-8.17/remove-usr-local-lib-from-m4.patch > similarity index 100% > rename from meta/recipes-core/coreutils/coreutils-8.14/remove-usr-local-lib-from-m4.patch > rename to meta/recipes-core/coreutils/coreutils-8.17/remove-usr-local-lib-from-m4.patch > diff --git a/meta/recipes-core/coreutils/coreutils_8.14.bb b/meta/recipes-core/coreutils/coreutils_8.17.bb > similarity index 91% > rename from meta/recipes-core/coreutils/coreutils_8.14.bb > rename to meta/recipes-core/coreutils/coreutils_8.17.bb > index 9a714a9..9d8170e 100644 > --- a/meta/recipes-core/coreutils/coreutils_8.14.bb > +++ b/meta/recipes-core/coreutils/coreutils_8.17.bb > @@ -6,8 +6,8 @@ HOMEPAGE = "http://www.gnu.org/software/coreutils/" > BUGTRACKER = "http://debbugs.gnu.org/coreutils" > LICENSE = "GPLv3+" > LIC_FILES_CHKSUM = "file://COPYING;md5=d32239bcb673463ab874e80d47fae504\ > - file://src/ls.c;startline=5;endline=16;md5=e1a509558876db58fb6667ba140137ad" > -PR = "r5" > + file://src/ls.c;startline=5;endline=16;md5=30c84fd2942cad91041e5e2dcd19ced6" > +PR = "r0" > DEPENDS = "gmp libcap" > DEPENDS_virtclass-native = "" > > @@ -16,9 +16,10 @@ inherit autotools gettext > SRC_URI = "${GNU_MIRROR}/coreutils/${BP}.tar.xz \ > file://remove-usr-local-lib-from-m4.patch \ > file://remove-gets.patch \ > + file://realpath-works-yes.patch \ > " > -SRC_URI[md5sum] = "bcb135ce553493a45aba01b39eb3920a" > -SRC_URI[sha256sum] = "0d120817c19292edb19e92ae6b8eac9020e03d51e0af9cb116cf82b65d18b02d" > +SRC_URI[md5sum] = "bbda656ce8ca2c6903948f9faa204ba3" > +SRC_URI[sha256sum] = "4e075a0d238072a5bd079046e1f024dc5e0d9133d43a39c73d0b86b0d1e2c5e5" > > EXTRA_OECONF_virtclass-native = "--without-gmp" > --------------000403090100000301050406 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit The problem:

The new version of coreutils, adds use of canonicalize_file_name() function defined in canonicalize.h The problem I encountered is at compile time, the function is redefined to rpl_canonicalize_file_name by means of a macro, and I get an undefined reference error.

//-- stdlib.in.h

#if @GNULIB_CANONICALIZE_FILE_NAME@
# if @REPLACE_CANONICALIZE_FILE_NAME@
#  if !(defined __cplusplus && defined GNULIB_NAMESPACE)
#   define canonicalize_file_name rpl_canonicalize_file_name
#  endif

I've looked for REPLACE_CANONICALIZE_FILE_NAME and found out that is being set in configure, which is generated from canonicalize.m4

# canonicalize.m4 serial 24

dnl Copyright (C) 2003-2007, 2009-2012 Free Software Foundation, Inc.

dnl This file is free software; the Free Software Foundation
dnl gives unlimited permission to copy and/or distribute it,
dnl with or without modifications, as long as this notice is preserved.

# Provides canonicalize_file_name and canonicalize_filename_mode, but does
# not provide or fix realpath.
AC_DEFUN([gl_FUNC_CANONICALIZE_FILENAME_MODE],
[
  AC_REQUIRE([gl_USE_SYSTEM_EXTENSIONS])
  AC_CHECK_FUNCS_ONCE([canonicalize_file_name])
  AC_REQUIRE([gl_DOUBLE_SLASH_ROOT])
  AC_REQUIRE([gl_FUNC_REALPATH_WORKS])
  if test $ac_cv_func_canonicalize_file_name = no; then
    HAVE_CANONICALIZE_FILE_NAME=0
  elif test "$gl_cv_func_realpath_works" != yes; then
    REPLACE_CANONICALIZE_FILE_NAME=1
  fi
])

# Provides canonicalize_file_name and realpath.
AC_DEFUN([gl_CANONICALIZE_LGPL],
[
  AC_REQUIRE([gl_STDLIB_H_DEFAULTS])
  AC_REQUIRE([gl_CANONICALIZE_LGPL_SEPARATE])
  if test $ac_cv_func_canonicalize_file_name = no; then
    HAVE_CANONICALIZE_FILE_NAME=0
    if test $ac_cv_func_realpath = no; then
      HAVE_REALPATH=0
    elif test "$gl_cv_func_realpath_works" != yes; then
      REPLACE_REALPATH=1
    fi
  elif test "$gl_cv_func_realpath_works" != yes; then
    REPLACE_CANONICALIZE_FILE_NAME=1
    REPLACE_REALPATH=1
  fi
])

Then I followed gl_cv_func_realpath_works and found it to be set to "guessing yes"

# canonicalize.m4 serial 24

     [case "$host_os" in
                # Guess yes on glibc systems.
        *-gnu*) gl_cv_func_realpath_works="guessing yes" ;;
                # If we don't know, assume the worst.
        *)      gl_cv_func_realpath_works="guessing no" ;;
      esac

Being set this way it looked to me like it was always assumed to be "no" and thus assumed real path didn't work. In fact, I could not make a logic for "guessing yes" and "quessing no" because to me it looks like they are not verified against and thus I'm missing the reason of their usage. However, I'm assuming I missed something because they are used quite heavily.

I don't have experience with autoconf so I may have missed the "right way" to fix this.
Any comments well apreciated.

radu



On 08/17/2012 04:05 PM, Radu Moisan wrote:
Signed-off-by: Radu Moisan <radu.moisan@intel.com>
---
 .../coreutils-8.17/realpath-works-yes.patch        |   13 +++++++++++++
 .../remove-gets.patch                              |   20 +++++++++++---------
 .../remove-usr-local-lib-from-m4.patch             |    0
 .../{coreutils_8.14.bb => coreutils_8.17.bb}       |    9 +++++----
 4 files changed, 29 insertions(+), 13 deletions(-)
 create mode 100644 meta/recipes-core/coreutils/coreutils-8.17/realpath-works-yes.patch
 rename meta/recipes-core/coreutils/{coreutils-8.14 => coreutils-8.17}/remove-gets.patch (47%)
 rename meta/recipes-core/coreutils/{coreutils-8.14 => coreutils-8.17}/remove-usr-local-lib-from-m4.patch (100%)
 rename meta/recipes-core/coreutils/{coreutils_8.14.bb => coreutils_8.17.bb} (91%)

diff --git a/meta/recipes-core/coreutils/coreutils-8.17/realpath-works-yes.patch b/meta/recipes-core/coreutils/coreutils-8.17/realpath-works-yes.patch
new file mode 100644
index 0000000..6bbe7ee
--- /dev/null
+++ b/meta/recipes-core/coreutils/coreutils-8.17/realpath-works-yes.patch
@@ -0,0 +1,13 @@
+Index: coreutils-8.17/m4/canonicalize.m4
+===================================================================
+--- coreutils-8.17.orig/m4/canonicalize.m4	2012-05-08 12:05:23.000000000 +0300
++++ coreutils-8.17/m4/canonicalize.m4	2012-08-17 14:20:22.000000000 +0300
+@@ -95,7 +95,7 @@
+      [gl_cv_func_realpath_works=no],
+      [case "$host_os" in
+                 # Guess yes on glibc systems.
+-        *-gnu*) gl_cv_func_realpath_works="guessing yes" ;;
++        *-gnu*) gl_cv_func_realpath_works="yes" ;;
+                 # If we don't know, assume the worst.
+         *)      gl_cv_func_realpath_works="guessing no" ;;
+       esac
diff --git a/meta/recipes-core/coreutils/coreutils-8.14/remove-gets.patch b/meta/recipes-core/coreutils/coreutils-8.17/remove-gets.patch
similarity index 47%
rename from meta/recipes-core/coreutils/coreutils-8.14/remove-gets.patch
rename to meta/recipes-core/coreutils/coreutils-8.17/remove-gets.patch
index 4f61c92..eaadf7e 100644
--- a/meta/recipes-core/coreutils/coreutils-8.14/remove-gets.patch
+++ b/meta/recipes-core/coreutils/coreutils-8.17/remove-gets.patch
@@ -3,19 +3,21 @@ use gets iff its defined. eglibc 2.16 removed gets
 Signed-off-by: Khem Raj <raj.khem@gmail.com>
 Upstream-Status: Pending
 
-Index: coreutils-8.14/lib/stdio.in.h
+Index: coreutils-8.17/lib/stdio.in.h
 ===================================================================
---- coreutils-8.14.orig/lib/stdio.in.h	2011-09-24 04:20:48.000000000 -0700
-+++ coreutils-8.14/lib/stdio.in.h	2012-07-03 10:36:19.886296576 -0700
-@@ -713,11 +713,13 @@
- _GL_CXXALIAS_SYS (gets, char *, (char *s));
- #  undef gets
+--- coreutils-8.17.orig/lib/stdio.in.h
++++ coreutils-8.17/lib/stdio.in.h
+@@ -698,13 +698,14 @@ _GL_WARN_ON_USE (getline, "getline is un
+                  "use gnulib module getline for portability");
  # endif
+ #endif
+-
 +# if defined gets
- _GL_CXXALIASWARN (gets);
  /* It is very rare that the developer ever has full control of stdin,
-    so any use of gets warrants an unconditional warning.  Assume it is
-    always declared, since it is required by C89.  */
+    so any use of gets warrants an unconditional warning; besides, C11
+    removed it.  */
+ #undef gets
+ #if HAVE_RAW_DECL_GETS
  _GL_WARN_ON_USE (gets, "gets is a security hole - use fgets instead");
 +# endif
  #endif
diff --git a/meta/recipes-core/coreutils/coreutils-8.14/remove-usr-local-lib-from-m4.patch b/meta/recipes-core/coreutils/coreutils-8.17/remove-usr-local-lib-from-m4.patch
similarity index 100%
rename from meta/recipes-core/coreutils/coreutils-8.14/remove-usr-local-lib-from-m4.patch
rename to meta/recipes-core/coreutils/coreutils-8.17/remove-usr-local-lib-from-m4.patch
diff --git a/meta/recipes-core/coreutils/coreutils_8.14.bb b/meta/recipes-core/coreutils/coreutils_8.17.bb
similarity index 91%
rename from meta/recipes-core/coreutils/coreutils_8.14.bb
rename to meta/recipes-core/coreutils/coreutils_8.17.bb
index 9a714a9..9d8170e 100644
--- a/meta/recipes-core/coreutils/coreutils_8.14.bb
+++ b/meta/recipes-core/coreutils/coreutils_8.17.bb
@@ -6,8 +6,8 @@ HOMEPAGE = "http://www.gnu.org/software/coreutils/"
 BUGTRACKER = "http://debbugs.gnu.org/coreutils"
 LICENSE = "GPLv3+"
 LIC_FILES_CHKSUM = "file://COPYING;md5=d32239bcb673463ab874e80d47fae504\
-                    file://src/ls.c;startline=5;endline=16;md5=e1a509558876db58fb6667ba140137ad"
-PR = "r5"
+                    file://src/ls.c;startline=5;endline=16;md5=30c84fd2942cad91041e5e2dcd19ced6"
+PR = "r0"
 DEPENDS = "gmp libcap"
 DEPENDS_virtclass-native = ""
 
@@ -16,9 +16,10 @@ inherit autotools gettext
 SRC_URI = "${GNU_MIRROR}/coreutils/${BP}.tar.xz \
            file://remove-usr-local-lib-from-m4.patch \
            file://remove-gets.patch \
+           file://realpath-works-yes.patch \
           "
-SRC_URI[md5sum] = "bcb135ce553493a45aba01b39eb3920a"
-SRC_URI[sha256sum] = "0d120817c19292edb19e92ae6b8eac9020e03d51e0af9cb116cf82b65d18b02d"
+SRC_URI[md5sum] = "bbda656ce8ca2c6903948f9faa204ba3"
+SRC_URI[sha256sum] = "4e075a0d238072a5bd079046e1f024dc5e0d9133d43a39c73d0b86b0d1e2c5e5"
 
 EXTRA_OECONF_virtclass-native = "--without-gmp"

--------------000403090100000301050406--