Re: [PATCH v3 1/9] Add configure check to test if gcc supports attribute ifunc.

[Stefan Liebler <stli@linux.vnet.ibm.com>]

[-- Attachment #1: Type: text/plain, Size: 1370 bytes --]

On 08/29/2016 11:34 AM, Florian Weimer wrote:
> On 08/24/2016 04:04 PM, Stefan Liebler wrote:
>> +For multi-arch support it is recommended to use a GCC with
>> gnu-indirect-function
>> +support as it is used to generate ifunc'ed symbols with correct
>> +debug-information. This support can either be enabled by configuring
>> GCC with
>> +@samp{--enable-gnu-indirect-function} or by enabling it by default by
>> setting
>> +@samp{default_gnu_indirect_function} variable for a particular
>> architecture in
>> +gcc source file @file{gcc/config.gcc}.
>
> Maybe use this text instead?
> 6808c16e2d067c8137fa9eae6078dc55c5d355c6
> For multi-arch support it is recommended to use a GCC which has been
> built with support for GNU indirect functions.  This ensures that
> correct debugging information is generated for functions selected by
> IFUNC resolvers.  This support can either be enabled by configuring GCC
> with @samp{--enable-gnu-indirect-function}, or by enabling it by default
> by setting @samp{default_gnu_indirect_function} variable for a
> particular architecture in the GCC source file @file{gcc/config.gcc}.
>
Okay. I've updated install-files and I'll use this updated text also for 
the NEWS entry.
>
> The configure check looks okay to me.  Some existing checks use
> AC_TRY_COMPILE, others use the manual expansion you provided.
>
> Thanks,
> Florian
>


[-- Attachment #2: 20160830_ifunc_configure_check.patch --]
[-- Type: text/x-patch, Size: 8598 bytes --]

commit 3cfc378ca6f95feaa35b99b4b5897fa35dbff68f
Author: Stefan Liebler <stli@linux.vnet.ibm.com>
Date:   Wed Aug 24 10:18:14 2016 +0200

    Add configure check to test if gcc supports attribute ifunc.
    
    This patch adds a configure check to test if gcc supports attribute ifunc.
    The support can either be enabled in <gcc-src>/gcc/config.gcc for one
    architecture in general by setting default_gnu_indirect_function variable to yes
    or by configuring gcc with --enable-gnu-indirect-function.
    
    The next patch rewrites libc_ifunc macro to use gcc attribute ifunc instead
    of inline assembly to generate the IFUNC symbols due to false debuginfo.
    
    If gcc does not support attribute ifunc, the old approach for generating
    ifunc'ed symbols is used. Then the debug-information is false. Thus it is
    recommended to use a gcc with indirect function support (See notes in INSTALL).
    After this patch-series these inline assemblies for ifunc-handling are not
    scattered in multiple files but are used only indirect via ifunc-macros
    and can simply removed in libc-symbols.h in future.
    
    If glibc is configured with --enable-multi-arch and gcc does not support
    attribute ifunc, a configure warning is dumped!
    
    This NEWS entry will be added:
    * For multi-arch support it is recommended to use a GCC which has
      been built with support for GNU indirect functions.  This ensures
      that correct debugging information is generated for functions
      selected by IFUNC resolvers.  This support can either be enabled by
      configuring GCC with '--enable-gnu-indirect-function', or by
      enabling it by default by setting 'default_gnu_indirect_function'
      variable for a particular architecture in the GCC source file
      'gcc/config.gcc'.
    
    ChangeLog:
    
    	* config.h.in (HAVE_GCC_IFUNC): New undef.
    	* configure.ac: Add check if gcc supports attribute ifunc feature.
    	* configure: Regenerated.
    	* manual/install.texi: Add recommendation for gcc with
    	indirect-function support.
    	* INSTALL: Regenerated.

diff --git a/INSTALL b/INSTALL
index ec3445f..66be341 100644
--- a/INSTALL
+++ b/INSTALL
@@ -359,6 +359,15 @@ build the GNU C Library:
      better code.  As of release time, GCC 5.3 is the newest compiler
      verified to work to build the GNU C Library.
 
+     For multi-arch support it is recommended to use a GCC which has
+     been built with support for GNU indirect functions.  This ensures
+     that correct debugging information is generated for functions
+     selected by IFUNC resolvers.  This support can either be enabled by
+     configuring GCC with '--enable-gnu-indirect-function', or by
+     enabling it by default by setting 'default_gnu_indirect_function'
+     variable for a particular architecture in the GCC source file
+     'gcc/config.gcc'.
+
      You can use whatever compiler you like to compile programs that use
      the GNU C Library.
 
diff --git a/config.h.in b/config.h.in
index 856ef6a..9f5c5ff 100644
--- a/config.h.in
+++ b/config.h.in
@@ -171,6 +171,9 @@
 /* Define to 1 if STT_GNU_IFUNC support actually works.  */
 #define HAVE_IFUNC 0
 
+/* Define if gcc supports attribute ifunc.  */
+#undef HAVE_GCC_IFUNC
+
 /* Define if the linker defines __ehdr_start.  */
 #undef HAVE_EHDR_START
 
diff --git a/configure b/configure
index 17625e1..8ffc2b5 100755
--- a/configure
+++ b/configure
@@ -3914,6 +3914,36 @@ fi
 { $as_echo "$as_me:${as_lineno-$LINENO}: result: $libc_cv_ld_gnu_indirect_function" >&5
 $as_echo "$libc_cv_ld_gnu_indirect_function" >&6; }
 
+# Check if gcc supports attribute ifunc as it is used in libc_ifunc macro.
+{ $as_echo "$as_me:${as_lineno-$LINENO}: checking for gcc attribute ifunc support" >&5
+$as_echo_n "checking for gcc attribute ifunc support... " >&6; }
+if ${libc_cv_gcc_indirect_function+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  cat > conftest.c <<EOF
+extern int func (int);
+int used_func (int a)
+{
+  return a;
+}
+static void *resolver ()
+{
+  return &used_func;
+}
+extern __typeof (func) func __attribute__ ((ifunc ("resolver")));
+EOF
+libc_cv_gcc_indirect_function=no
+if ${CC-cc} -c conftest.c -o conftest.o 1>&5 \
+   2>&5 ; then
+  if $READELF -s conftest.o | grep IFUNC >/dev/null 2>&5; then
+    libc_cv_gcc_indirect_function=yes
+  fi
+fi
+rm -f conftest*
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $libc_cv_gcc_indirect_function" >&5
+$as_echo "$libc_cv_gcc_indirect_function" >&6; }
+
 if test x"$libc_cv_ld_gnu_indirect_function" != xyes; then
   if test x"$multi_arch" = xyes; then
     as_fn_error $? "--enable-multi-arch support requires assembler and linker support" "$LINENO" 5
@@ -3921,6 +3951,13 @@ if test x"$libc_cv_ld_gnu_indirect_function" != xyes; then
     multi_arch=no
   fi
 fi
+if test x"$libc_cv_gcc_indirect_function" != xyes &&
+   test x"$multi_arch" = xyes; then
+  { $as_echo "$as_me:${as_lineno-$LINENO}: WARNING: --enable-multi-arch support recommends a gcc with gnu-indirect-function support.
+Please use a gcc which supports it by default or configure gcc with --enable-gnu-indirect-function" >&5
+$as_echo "$as_me: WARNING: --enable-multi-arch support recommends a gcc with gnu-indirect-function support.
+Please use a gcc which supports it by default or configure gcc with --enable-gnu-indirect-function" >&2;}
+fi
 multi_arch_d=
 if test x"$multi_arch" != xno; then
   multi_arch_d=/multiarch
@@ -6504,6 +6541,11 @@ if test x"$libc_cv_ld_gnu_indirect_function" = xyes; then
 
 fi
 
+if test x"$libc_cv_gcc_indirect_function" = xyes; then
+  $as_echo "#define HAVE_GCC_IFUNC 1" >>confdefs.h
+
+fi
+
 # This is far from the AC_ARG_ENABLE that sets it so that a sysdeps
 # configure fragment can override the value to prevent this AC_DEFINE.
 
diff --git a/configure.ac b/configure.ac
index 33bcd62..643002f 100644
--- a/configure.ac
+++ b/configure.ac
@@ -634,6 +634,30 @@ if ${CC-cc} $CFLAGS $CPPFLAGS $LDFLAGS \
 fi
 rm -f conftest*])
 
+# Check if gcc supports attribute ifunc as it is used in libc_ifunc macro.
+AC_CACHE_CHECK([for gcc attribute ifunc support],
+	       libc_cv_gcc_indirect_function, [dnl
+cat > conftest.c <<EOF
+extern int func (int);
+int used_func (int a)
+{
+  return a;
+}
+static void *resolver ()
+{
+  return &used_func;
+}
+extern __typeof (func) func __attribute__ ((ifunc ("resolver")));
+EOF
+libc_cv_gcc_indirect_function=no
+if ${CC-cc} -c conftest.c -o conftest.o 1>&AS_MESSAGE_LOG_FD \
+   2>&AS_MESSAGE_LOG_FD ; then
+  if $READELF -s conftest.o | grep IFUNC >/dev/null 2>&AS_MESSAGE_LOG_FD; then
+    libc_cv_gcc_indirect_function=yes
+  fi
+fi
+rm -f conftest*])
+
 if test x"$libc_cv_ld_gnu_indirect_function" != xyes; then
   if test x"$multi_arch" = xyes; then
     AC_MSG_ERROR([--enable-multi-arch support requires assembler and linker support])
@@ -641,6 +665,11 @@ if test x"$libc_cv_ld_gnu_indirect_function" != xyes; then
     multi_arch=no
   fi
 fi
+if test x"$libc_cv_gcc_indirect_function" != xyes &&
+   test x"$multi_arch" = xyes; then
+  AC_MSG_WARN([--enable-multi-arch support recommends a gcc with gnu-indirect-function support.
+Please use a gcc which supports it by default or configure gcc with --enable-gnu-indirect-function])
+fi
 multi_arch_d=
 if test x"$multi_arch" != xno; then
   multi_arch_d=/multiarch
@@ -1770,6 +1799,10 @@ if test x"$libc_cv_ld_gnu_indirect_function" = xyes; then
   AC_DEFINE(HAVE_IFUNC)
 fi
 
+if test x"$libc_cv_gcc_indirect_function" = xyes; then
+  AC_DEFINE(HAVE_GCC_IFUNC)
+fi
+
 # This is far from the AC_ARG_ENABLE that sets it so that a sysdeps
 # configure fragment can override the value to prevent this AC_DEFINE.
 AC_SUBST(use_nscd)
diff --git a/manual/install.texi b/manual/install.texi
index 79ee45f..5f5705e 100644
--- a/manual/install.texi
+++ b/manual/install.texi
@@ -402,6 +402,14 @@ the newest version of the compiler that is known to work for building
 release time, GCC 5.3 is the newest compiler verified to work to build
 @theglibc{}.
 
+For multi-arch support it is recommended to use a GCC which has been built with
+support for GNU indirect functions.  This ensures that correct debugging
+information is generated for functions selected by IFUNC resolvers.  This
+support can either be enabled by configuring GCC with
+@samp{--enable-gnu-indirect-function}, or by enabling it by default by setting
+@samp{default_gnu_indirect_function} variable for a particular architecture in
+the GCC source file @file{gcc/config.gcc}.
+
 You can use whatever compiler you like to compile programs that use
 @theglibc{}.