glib-2.0: Don't check function "qsort_r" in configure.ac

From: Xiaofeng Yan <[email protected]>

The function "g_qsort_with_data" has two kinds of realization. One calls qsort_r
from libglib, the other realize itself. the realization method is controlled by macro "HAVE_QSORT_R". qsort_r from libglib cause sort error. Add no-check-qsort_r.patch for fixing this problem. This patch is to fix bug 1959. After this patch is merged into OE-core I will open a new bug to track the reason why qsort_r report normally. For finishing lsb test in YOCTO 1.2 M3 this patch is applied temporarily.

Pull URL: git://git.pokylinux.org/poky-contrib.git
Branch: xiaofeng/1959
Browse: http://git.pokylinux.org/cgit.cgi/poky-contrib/log/?h=xiaofeng/1959

Thanks,
Xiaofeng Yan <[email protected]>
---

Xiaofeng Yan (2):
glib-2.0: Don't check function qsort_r
glib-2.0: Add patch "no-check-qsort_r.patch"

 .../glib-2.0/glib-2.0/no-check-qsort_r.patch       |   61 ++++++++++++++++++++
 meta/recipes-core/glib-2.0/glib-2.0_2.30.2.bb      |    3 +-
 2 files changed, 63 insertions(+), 1 deletions(-)
 create mode 100644 meta/recipes-core/glib-2.0/glib-2.0/no-check-qsort_r.patch

From: Xiaofeng Yan <[email protected]>

function "g_qsort_with_data" has two kinds of realization. One calls qsort_r from libglib, the other realize itself.
qsort_r from libglib cause sort error.
For fixing this problem no checking "qsort_r" in configure.ac.

[YOCTO #1959]

---
 .../glib-2.0/glib-2.0/no-check-qsort_r.patch       |   61 ++++++++++++++++++++
 1 files changed, 61 insertions(+), 0 deletions(-)
 create mode 100644 meta/recipes-core/glib-2.0/glib-2.0/no-check-qsort_r.patch

new file mode 100644
index 0000000..99e229c
@@ -0,0 +1,61 @@
+Upstream-Status: Pending
+
+function "g_qsort_with_data" has two kinds of realization. One calls qsort_r
+from libglib, the other realize itself.
+qsort_r from libglib cause sort error. Don't check qsort_r for fixing this problem.
+
+Xiaofeng Yan <[email protected]>
+
+Index: glib-2.30.0/configure.ac
+===================================================================
+--- a/configure.ac	2012-02-06 11:15:37.885295264 +0800
++++ b/configure.ac.new	2012-02-06 11:16:04.239795382 +0800
+@@ -584,47 +584,7 @@
+ dnl don't use AC_CHECK_FUNCS here, otherwise HAVE_QSORT_R will
+ dnl be automatically defined, which we don't want to do
+ dnl until we have checked this function is actually usable
+-AC_CHECK_FUNC([qsort_r])
+-
+-# BSD has a qsort_r with wrong argument order
+-if test x$ac_cv_func_qsort_r = xyes ; then
+-  AC_CACHE_CHECK([if qsort_r uses glibc compatible argument order], glib_cv_have_qsort_r, [
+-  AC_RUN_IFELSE([AC_LANG_SOURCE([[
+-  #define _GNU_SOURCE
+-  #include <stdlib.h>
+-
+-  static int
+-  cmp (const void *a, const void *b, void *c)
+-  {
+-    const int *ia = a;
+-    const int *ib = b;
+-
+-    if (*ia < *ib)
+-      return -1;
+-    else if (*ia > *ib)
+-      return 1;
+-    else
+-      return 0;
+-  }
+-
+-  int
+-  main (int argc, char **argv)
+-  {
+-    int arr[3] = { 1, 2, 0 };
+-    int d = 3;
+-
+-    qsort_r (arr, 3, sizeof (int), cmp, &d);
+-
+-    if (arr[0] == 0 && arr[1] == 1 && arr[2] == 2)
+-      return 0;
+-    else
+-      return 1;
+-  }]])],[glib_cv_have_qsort_r=yes],[glib_cv_have_qsort_r=no])])
+-fi
+-
+-if test x$glib_cv_have_qsort_r = xyes ; then
+-  AC_DEFINE(HAVE_QSORT_R, 1, [Define to 1 if you have the 'qsort_r' function])
+-fi
++#AC_CHECK_FUNC([qsort_r])
+ 
+ AC_CHECK_SIZEOF(char)
+ AC_CHECK_SIZEOF(short)

1.7.0.4

This description doesn't make much sense. libglib is part of glib-2.0 itself; I guess you meant to say that qsort_r is from libc6. But, even leaving that aside, if qsort_r is broken then the appropriate patch would be to fix it, not just to hack glib to use a standalone implementation. So this patch seems like the wrong thing by any measure.

Does glibc's own testsuite also have failures for qsort_r?

p.

glibc one works ok last time I checked however uclibc does not implement it

Yes, it does. It was implemented somewhere in between 0.9.32 and 0.9.33 [1]. Last time I've checked it even worked fine with glib-2.0. You just need to change glib_cv_have_qsort to yes in meta/site/common-uclibc.

/ptw

[1] http://git.uclibc.org/uClibc/commit/?id=515d54433138596e81267237542bd9168b8cc787

yes I was merely pointing out the need since we still dont have 0.9.33 in metadata

From: Xiaofeng Yan <[email protected]>

Apply this patch for avoiding sort error. Please check no-check-qsort_r.patch for detailed information.

---
 meta/recipes-core/glib-2.0/glib-2.0_2.30.2.bb |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

index bdf92aa..a5ec28a 100644
@@ -1,6 +1,6 @@
 require glib.inc
 
-PR = "r2"
+PR = "r3"
 PE = "1"
 
 DEPENDS += "libffi python-argparse-native zlib"

@@ -14,6 +14,7 @@ SRC_URI = "${GNOME_MIRROR}/glib/${SHRT_VER}/glib-${PV}.tar.bz2 \
            file://60_wait-longer-for-threads-to-die.patch \
            file://g_once_init_enter.patch \
            file://glib-2.0_fix_for_x32.patch \
+           file://file://no-check-qsort_r.patch \
           "
 SRC_URI[md5sum] = "b40f3889e8d24e1b367763673ca6deb5"
 SRC_URI[sha256sum] = "94b1f1a1456c67060ca868d299bef3f7268a2c1c5c360aabb7149d4d9b2fdcd3"

1.7.0.4

really ?

I run qsort-test.c in core-image-lsb-sdk, The result is ok. I didn't know how to resolve this bug. I think I need to apply this patch only in core-image-lsb-qt3 and other images don't apply this patch.

no-check-qsort_r.patch is the same as remove.test.for.qsort_r.patch from
http://git.openembedded.org/openembedded-core/commit/?id=74f43e3b269c01dba7cd5215a8e825229bd5ecff

which was later removed from glib-2.0-native in
http://git.openembedded.org/openembedded-core/commit/?id=37579d7d74d127c90c1e078d05c5bf4ba0b3f755

and later removed completely in
http://git.openembedded.org/openembedded-core/commit/?id=812342e44e9e361a0fa93cfcfe65cf4dd6f2fb57

Are you sure this patch is right fix? And no need to split adding patch itself and adding it to SRC_URI to 2 patches.

Cheers,

Martin 'JaMa' Jansa     jabber: [email protected]

Hi Martin Jansa,
I reviewed the above link, no-check-qsort_r.patch is the same as remove.test.for.qsort_r.patch really. Please check attachment for the test program from LSB 4.1. I think we need this patch because it is an error really. LSB report the next information:

After sorting array, the 0-th element of array should be 1, but is 2 After sorting array, the 1-th element of array should be 2, but is 1 After sorting array, the 2-th element of array should be 3, but is 5 After sorting array, the 4-th element of array should be 5, but is 3

I can reuse this patch which was removed before if you think this is a bug. Do you any suggestion?

Compile this test program.

$ gcc -g `pkg-config --cflags glib-2.0` gobject_value_arrays.c -o gobject -lgobject-2.0

///////////////////////////////////////////////////////////////////////////// // Copyright (C) 2010 The Linux Foundation. All rights reserved. //
// This program is free software; you can redistribute it and/or // modify it under the terms of the GNU General Public License
// as published by the Free Software Foundation; either version 2 // of the License, or (at your option) any later version.
//
// This program is distributed in the hope that it will be useful, // but WITHOUT ANY WARRANTY; without even the implied warranty of // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the // GNU General Public License for more details.
//
// You should have received a copy of the GNU General Public License // along with this program; if not, write to the Free Software
// Foundation, Inc.,
// 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA. /////////////////////////////////////////////////////////////////////////////

///////////////////////////////////////////////////////////////////////////// // This file contains tests for the following library:
// libgobject-2.0,
// section: "GVALUE_ARRAYS".
//
// This C file was generated by the T2C system developed in ISPRAS // for The Linux Foundation. /////////////////////////////////////////////////////////////////////////////

///////////////////////////////////////////////////////////////////////////// // To build a standalone version of the tests, i.e. the one that does not // depend on TETware Lite test harness, please follow the instructions below. // (A standalone executable like this can prove useful for debugging.)
//
// 1. Make sure the T2C_ROOT and T2C_SUITE_ROOT environment variables are set properly. // T2C_ROOT should contain a path to the main directory of T2C Framework. // For example, the executable of the T2C C code generator is $T2C_ROOT/t2c/bin/t2c. // T2C_SUITE_ROOT is main test suite directory, the one where the tet execution // configuration file (tetexec.cfg) resides as well as tet_code file etc. //
// 2. If the standalone version of the tests is to be built with lsbcc or lsbc++, // make sure that lsbcc (or lsbc++, respectively) is somewhere in your PATH. // You should also add a path to the LSB pkg-config data
// (usually /opt/lsb/lib/pkgconfig or /opt/lsb/lib64/pkgconfig) to the beginning of // the string contained in PKG_CONFIG_PATH environment variable.
//
// 3. Before building the standalone version of a test, check if the following // files and directories exist:

//  $T2C_SUITE_ROOT/gobject-t2c/tests 
//  $T2C_SUITE_ROOT/gobject-t2c/tests/common.mk
//  $T2C_SUITE_ROOT/gobject-t2c/tests/gobject_value_arrays/
//  $T2C_SUITE_ROOT/gobject-t2c/tests/gobject_value_arrays/gobject_value_arrays.c
//  $T2C_SUITE_ROOT/gobject-t2c/tests/gobject_value_arrays/Makefile

//
// If some of these are missing, just run ./gen_code.sh script from the $T2C_SUITE_ROOT // directory. After that all necessary files and directories should exist. //
// 4. Execute the following commands:

//      cd $T2C_SUITE_ROOT/gobject-t2c/tests/gobject_value_arrays/
//      make clean
//      make debug

//
// The C-code of the tests is the same both for release and debug (standalone) // versions.
// main() function for standalone execution is defined in
// "$T2C_ROOT/t2c/debug/src/dbg_main.c" along with some utility stuff. // The TET API stubs are declared in "$T2C_ROOT/t2c/debug/include/tet_api.h" //
// Note that all the test purposes of a standalone test are executed // in the same process to simplify debugging (some debuggers may not // handle fork() calls properly by default).
//
// WAIT_TIME configuration parameter has no effect on the standalone tests. // They will not be interrupted, no matter how long they run. Otherwise they // could be stopped in the middle of the debugging process (because their time // had expired) which is probably not what you want. /////////////////////////////////////////////////////////////////////////////

#include <stdio.h>
#include <stdlib.h>
#include <errno.h>
#include <unistd.h>
#include <glib-object.h>

const char* test_name_ = "gobject_value_arrays";
const char* suite_subdir_ = "gobject-t2c";

// List of requirement catalogs to be loaded
const char* rcat_names_[] = {
"gobject_value_arrays",
NULL
};

const char* rel_href_path_ =

    "gobject-t2c/tests/gobject_value_arrays/gobject_value_arrays.html#%s%d\">";

char* t2c_href_tpl_ = NULL;
char* t2c_href_full_tpl_ = NULL;

// Name of the env. variable that is used to determine whether the REQ // implementation should output just IDs (value = 0, default) or hyperlinks // to the html representation, enclosed in some pseudotags for DTK Manager // (value != 0).
const char* t2c_gen_hlinks_name = "T2C_GEN_HLINKS";
int gen_hlinks = 0;

char* init_fail_reason_ = NULL;

/*

g_value_array_append            DONE
g_value_array_copy              DONE
g_value_array_free              DONE
g_value_array_get_nth           DONE
g_value_array_insert            DONE
g_value_array_new               DONE
g_value_array_prepend           DONE
g_value_array_remove            DONE
g_value_array_sort              DONE
g_value_array_sort_with_data    DONE

*/

//Global variables, used in get_value function
//They must be declared as global , because get_value can return pointers to these variables.

gint gintVar;
guint guintVar;
gint64 gint64Var;
guint64 huint64Var;
gchar gcharVar;
guchar gucharVar;
glong glongVar;
gulong gulongVar;
guchar gucharVar;
gfloat gfloatVar;
gdouble gdoubleVar;
gboolean gbooleanVar;

char* message; // set_value function generates this message , and it contains the unique text for each GType
char* sortMessage; //this message will be generated in cmp or cmp_data interfaces GType sortType;

//
//This function gets the GType and calls the correct set interface for this type //For example, it calls g_value_set_int for G_TYPE_INT.
//
void set_value(GType initType,GValue* value, void* vl )
{

    if(initType==G_TYPE_INT)
    {
         g_value_set_int(value,*((gint *)vl));
         message="The function should return %d,but returns %d";
         return;
    }
    else if(initType==G_TYPE_CHAR)
    {
         g_value_set_char(value,*((gchar *)vl));
         message="The function should return %c,but returns %c";
         return;
    }
    else if(initType==G_TYPE_UCHAR)
    {
         g_value_set_uchar(value,*((guchar *)vl));
         message="The  function should return %u,but returns %u";
         return;
    }
    else if(initType==G_TYPE_LONG)
    {
         g_value_set_long(value,*((glong *)vl));
         message="The  function should return %ld,but returns %ld";
         return;
    }
    else if(initType==G_TYPE_ULONG)
    {
         g_value_set_ulong(value,*((gulong *)vl));
         return;
    }
    else if(initType==G_TYPE_INT64)
    {
         g_value_set_int64(value,*((gint64 *)vl));
         return;
    }
    else if(initType==G_TYPE_UINT)
    {
         g_value_set_uint(value,*((guint *)vl));
         message="The function should return %u,but returns %u";
         return;
    }
    else if(initType==G_TYPE_BOOLEAN)
    {
         g_value_set_boolean(value,*((gboolean *)vl));
         return;
    }
    else if(initType==G_TYPE_DOUBLE)
    {
         g_value_set_double(value,*((gdouble *)vl));
         message="The function should return %f,but returns %f";
         return;
    }
    else if(initType==G_TYPE_FLOAT)
    {
         g_value_set_float(value,*((gfloat *)vl));
         message="The function should return %f,but returns %f";
         return;
    }
    return ;

}

void* get_value(GType initType,GValue* value )
{

    if(initType==G_TYPE_INT)
    {
         gintVar=g_value_get_int(value);
         return  (void*)(&gintVar);
    }
    else if(initType==G_TYPE_CHAR)
    {
         gcharVar=g_value_get_char(value);
         return  (void*)(&gcharVar);
    }
    else if(initType==G_TYPE_UCHAR)
    {
         gucharVar=g_value_get_uchar(value);
         return  (void*)(&gucharVar);
    }
    else if(initType==G_TYPE_LONG)
    {
         glongVar=g_value_get_long(value);
         return  (void*)(&glongVar);
    }
    else if(initType==G_TYPE_ULONG)
    {
         gulongVar=g_value_get_ulong(value);
         return  (void*)(&gulongVar);
    }
    else if(initType==G_TYPE_INT64)
    {
         gint64Var=g_value_get_int64(value);
         return  (void*)(&gint64Var);
    }
    else if(initType==G_TYPE_UINT)
    {
         guintVar=g_value_get_uint(value);
         return  (void*)(&gintVar);
    }
    else if(initType==G_TYPE_BOOLEAN)
    {
         gbooleanVar=g_value_get_boolean(value);
         return  (void*)(&gbooleanVar);
    }
    else if(initType==G_TYPE_DOUBLE)
    {
         gdoubleVar=g_value_get_double(value);
         return  (void*)(&gdoubleVar);
    }
    else if(initType==G_TYPE_FLOAT)
    {
         gfloatVar=g_value_get_float(value);
         return  (void*)(&gfloatVar);
    }
    return (void *)NULL;

}

///////////////////////////////////////////////////////////////////////// //compare function for g_value_array_sort

gint cmp (gconstpointer a, gconstpointer b)

{
    if(sortType==G_TYPE_INT)
    {
        GValue* aValue=(GValue*)a;
        gint a1=g_value_get_int(aValue);

        GValue* bValue=(GValue*)b;
        gint b1=g_value_get_int(bValue);

        sortMessage="After sorting array, the %d-th element of array should be %d, but is %d";

        if (a1 - b1 < 0)
            return -1;
        if (a1 - b1 > 0)
            return 1;
        return 0;
    }

    if(sortType==G_TYPE_CHAR)
    {
        GValue* aValue=(GValue*)a;
        gchar a1=g_value_get_char(aValue);

        GValue* bValue=(GValue*)b;
        gchar b1=g_value_get_char(bValue);

        sortMessage="After sorting array, the %d-th element of array should be %c, but is %c";

        if (a1 - b1 < 0)
            return -1;
        if (a1 - b1 > 0)
            return 1;
        return 0;
    }

    if(sortType==G_TYPE_UCHAR)
    {
        GValue* aValue=(GValue*)a;
        guchar a1=g_value_get_uchar(aValue);

        GValue* bValue=(GValue*)b;
        guchar b1=g_value_get_uchar(bValue);

        sortMessage="After sorting array, the %d-th element of array should be %u, but is %u";

        if (a1 - b1 < 0)
            return -1;
        if (a1 - b1 > 0)
            return 1;
        return 0;
    }

    if(sortType==G_TYPE_DOUBLE)
    {
        GValue* aValue=(GValue*)a;
        gdouble a1=g_value_get_double(aValue);

        GValue* bValue=(GValue*)b;
        gdouble b1=g_value_get_double(bValue);

        sortMessage="After sorting array, the %d-th element of array should be %f, but is %f";

        if (a1 - b1 < 0)
            return -1;
        if (a1 - b1 > 0)
            return 1;
        return 0;
    }
    
    return -1;

}

gint cmp_data (gconstpointer a, gconstpointer b,gpointer userData) {

    if(sortType==G_TYPE_INT)
    {
        GValue* aValue=(GValue*)a;
        gint a1=g_value_get_int(aValue);

        GValue* bValue=(GValue*)b;
        gint b1=g_value_get_int(bValue);

        GValue* userValue=(GValue*)userData;
        gint user=g_value_get_int(userValue);

        sortMessage="After sorting array, the %d-th element of array should be %d, but is %d";

        if (a1 - b1 < user)
            return -1;
        if (a1 - b1 > user)
            return 1;
        return 0;


    }


    if(sortType==G_TYPE_CHAR)
    {
        GValue* aValue=(GValue*)a;
        gchar a1=g_value_get_char(aValue);

        GValue* bValue=(GValue*)b;
        gchar b1=g_value_get_char(bValue);

        GValue* userValue=(GValue*)userData;
        gchar user=g_value_get_char(userValue);

        sortMessage="After sorting array, the %d-th element of array should be %c, but is %c";

        if (a1 - b1 < user)
            return -1;
        if (a1 - b1 > user)
            return 1;
        return 0;
    }

    if(sortType==G_TYPE_UCHAR)
    {
        GValue* aValue=(GValue*)a;
        guchar a1=g_value_get_uchar(aValue);

        GValue* bValue=(GValue*)b;
        guchar b1=g_value_get_uchar(bValue);

        GValue* userValue=(GValue*)userData;
        guchar user=g_value_get_uchar(userValue);

        sortMessage="After sorting array, the %d-th element of array should be %u, but is %u";

        if (a1 - b1 < user)
            return -1;
        if (a1 - b1 > user)
            return 1;
        return 0;
    }

    if(sortType==G_TYPE_DOUBLE)
    {
        GValue* aValue=(GValue*)a;
        gdouble a1=g_value_get_double(aValue);

        GValue* bValue=(GValue*)b;
        gdouble b1=g_value_get_double(bValue);

        GValue* userValue=(GValue*)userData;
        gdouble user=g_value_get_double(userValue);

        sortMessage="After sorting array, the %d-th element of array should be %f, but is %f";

        if (a1 - b1 < user)
            return -1;
        if (a1 - b1 > user)
            return  1;
        return 0;
    }
    
    return -1;

}

void main ()
{
#define COUNT 6

#define TYPE            G_TYPE_INT
#define BUILT_IN_TYPE   gint
#define DATA_ARRAY      {2,1,5,4,3,6}
#define RESULT_ARRAY    {1,2,3,4,5,6}
#define USER_DATA       1

    int i;
    gboolean check=TRUE;
    gboolean check_count=TRUE;
    sortType=TYPE;

    GValueArray*  array=NULL;
    g_type_init();
    array=g_value_array_new (COUNT);

    BUILT_IN_TYPE  data_array  []=DATA_ARRAY;
    BUILT_IN_TYPE  result_array[]=RESULT_ARRAY;

    GValue a_value[COUNT]={{0,}};

    BUILT_IN_TYPE user=USER_DATA;
    GValue userValue={0};
    g_value_init (&userValue, TYPE);
    set_value(TYPE,&userValue,(void*)(&user));




    for(i=0;i<COUNT;i++)
    {
        g_value_init (&a_value[i], TYPE);
        set_value(TYPE,&a_value[i],(void*)(&data_array[i]));

    }

    for(i=0;i<COUNT;i++)
    {
        array=g_value_array_append(array,&a_value[i]);
    }
    
    if(array->n_values!=COUNT)
    {
        printf("Before sorting the array should contain %d values, but contains %d",COUNT,array->n_values);
    }

    g_value_array_sort_with_data(array,cmp_data,(void*)&userValue);
    
    if(array->n_values!=COUNT)
    {
        printf("After sorting the array should contain %d values, but contains %d",COUNT,array->n_values);
    }
    
    
    /*
     * number of values contained in the array
     */


    for(i=0;i<COUNT;i++)
    {
	
        GValue* retValue=g_value_array_get_nth(array,i);
        void *ret1=get_value(TYPE,retValue);
        BUILT_IN_TYPE ret= *((BUILT_IN_TYPE *)ret1);

        if(ret!=result_array[i])
        {
		printf ("ret is %d ", ret);
		printf ("result_array[%d] is %d ", i, result_array[i]);
		printf("FAILED !!! \n");
        }

    }
    for (i=0; i < sizeof(result_array)/sizeof(result_array[0]); i++)
    {
        printf("%d ", result_array[i]);
    }
}

I reviewed the origin patch from Martin Jansa and found that my patch is different from him. My patch is to remove the check of qsort_r and the origin had still this check. I also update my patch because it have two file:// in SRC_URI.
Thank Martin Jansa.