[v1] LoongArch: Define the macro ASM_PREFERRED_EH_DATA_FORMAT by checking the assembler's support for eh_frame encoding.

Message ID 20220729024337.4182683-1-chenglulu@loongson.cn
State New, archived
Headers
Series [v1] LoongArch: Define the macro ASM_PREFERRED_EH_DATA_FORMAT by checking the assembler's support for eh_frame encoding. |

Commit Message

chenglulu July 29, 2022, 2:43 a.m. UTC
  The ASM_PREFERRED_EH_DATA_FORMAT macro before and after modification is as follows:

 #define ASM_PREFERRED_EH_DATA_FORMAT(CODE, GLOBAL) \
-  (((GLOBAL) ? DW_EH_PE_indirect : 0) | DW_EH_PE_absptr)
+  (((GLOBAL) ? DW_EH_PE_indirect : 0) | DW_EH_PE_pcrel | DW_EH_PE_sdata4)

Use the following tests to describe the effect of modifying this macro on the generated
assembly code:

	#include <iostream>
	#include <exception>

	using namespace std;
	int main()
	{

	  try {
		  throw 1;
	  }
	  catch (int i)
	  {
	    cout << i << endl;
	  }
	}

The main comparisons related to the eh_frame section are as follows:

                   OLD                                                                NEW
.LFB1997 = .                                                  |  .LFB1780 = .                                                                                                                              
          .cfi_startproc                                      |          .cfi_startproc
          .cfi_personality 0x80,DW.ref.__gxx_personality_v0   |          .cfi_personality 0x9b,DW.ref.__gxx_personality_v0                                                                                 
          .cfi_lsda 0,.LLSDA1997                              |          .cfi_lsda 0x1b,.LLSDA1780                                

If the assembly file generated by the new gcc is compiled with the binutils of version 2.38, the
following error will be reported:
test.s:74: Error: invalid or unsupported encoding in .cfi_personality
test.s:75: Error: invalid or unsupported encoding in .cfi_lsda

So I think I should judge whether binutils supports this encoding when gcc is configured, and then choose how to define
macro ASM_PREFERRED_EH_DATA_FORMAT.




------------
.eh_frame DW_EH_PE_pcrel encoding format is not supported by gas <= 2.39.
Check if the assembler support DW_EH_PE_PCREL encoding and define .eh_frame
encoding type.

gcc/ChangeLog:

	* config.in: Regenerate.
	* config/loongarch/loongarch.h (ASM_PREFERRED_EH_DATA_FORMAT):
	Select the value of the macro definition according to whether
	HAVE_AS_EH_FRAME_PCREL_ENCODING_SUPPORT is defined.
	* configure: Regenerate.
	* configure.ac: Reinstate HAVE_AS_EH_FRAME_PCREL_ENCODING_SUPPORT test.
---
 gcc/config.in                    |  8 +++++++-
 gcc/config/loongarch/loongarch.h |  5 +++++
 gcc/configure                    | 35 ++++++++++++++++++++++++++++++++
 gcc/configure.ac                 |  9 ++++++++
 4 files changed, 56 insertions(+), 1 deletion(-)
  

Comments

Xi Ruoyao July 29, 2022, 3:18 a.m. UTC | #1
On Fri, 2022-07-29 at 10:43 +0800, Lulu Cheng wrote:

> .eh_frame DW_EH_PE_pcrel encoding format is not supported by gas <= 2.39.
> Check if the assembler support DW_EH_PE_PCREL encoding and define .eh_frame
> encoding type.
> 
> gcc/ChangeLog:
> 
>         * config.in: Regenerate.
>         * config/loongarch/loongarch.h (ASM_PREFERRED_EH_DATA_FORMAT):
>         Select the value of the macro definition according to whether
>         HAVE_AS_EH_FRAME_PCREL_ENCODING_SUPPORT is defined.
>         * configure: Regenerate.
>         * configure.ac: Reinstate HAVE_AS_EH_FRAME_PCREL_ENCODING_SUPPORT test.

To me it looks a little strange to list regenerated config.in &
configure before configure.ac.  But maybe I'm wrong here if a
lexicographical order is preferred...

/* snip */

> +    gcc_GAS_CHECK_FEATURE([eh_frame pcrel encoding support],
> +      gcc_cv_as_loongarch_eh_frame_pcrel_encoding_support,,
> +      [.LFB1780 = .
> +       .cfi_startproc
> +       .cfi_personality 0x9b,DW.ref.__gxx_personality_v0
> +       .cfi_lsda 0x1b,.LLSDA1780
> +       .cfi_endproc],,

I think the conftest content can be simplified to:

    .cfi_startproc
    .cfi_personality 0x9b,a
    .cfi_lsda 0x1b,b
    .cfi_endproc


> +      [AC_DEFINE(HAVE_AS_EH_FRAME_PCREL_ENCODING_SUPPORT, 1,
> +         [Define if your assembler supports eh_frame pcrel encoding.])])
>      ;;
>      s390*-*-*)
>      gcc_GAS_CHECK_FEATURE([.gnu_attribute support],
  
chenglulu July 29, 2022, 3:51 a.m. UTC | #2
在 2022/7/29 上午11:18, Xi Ruoyao 写道:
> On Fri, 2022-07-29 at 10:43 +0800, Lulu Cheng wrote:
>
>> .eh_frame DW_EH_PE_pcrel encoding format is not supported by gas <= 2.39.
>> Check if the assembler support DW_EH_PE_PCREL encoding and define .eh_frame
>> encoding type.
>>
>> gcc/ChangeLog:
>>
>>          * config.in: Regenerate.
>>          * config/loongarch/loongarch.h (ASM_PREFERRED_EH_DATA_FORMAT):
>>          Select the value of the macro definition according to whether
>>          HAVE_AS_EH_FRAME_PCREL_ENCODING_SUPPORT is defined.
>>          * configure: Regenerate.
>>          * configure.ac: Reinstate HAVE_AS_EH_FRAME_PCREL_ENCODING_SUPPORT test.
> To me it looks a little strange to list regenerated config.in &
> configure before configure.ac.  But maybe I'm wrong here if a
> lexicographical order is preferred...

This information is generated by me through the git gcc-commit-mklog 
command, then I didn't move the sequence.

>
> /* snip */
>
>> +    gcc_GAS_CHECK_FEATURE([eh_frame pcrel encoding support],
>> +      gcc_cv_as_loongarch_eh_frame_pcrel_encoding_support,,
>> +      [.LFB1780 = .
>> +       .cfi_startproc
>> +       .cfi_personality 0x9b,DW.ref.__gxx_personality_v0
>> +       .cfi_lsda 0x1b,.LLSDA1780
>> +       .cfi_endproc],,
> I think the conftest content can be simplified to:
>
>      .cfi_startproc
>      .cfi_personality 0x9b,a
>      .cfi_lsda 0x1b,b
>      .cfi_endproc
>
This one looks more concise, I'll change it
  
Xi Ruoyao July 29, 2022, 3:55 a.m. UTC | #3
On Fri, 2022-07-29 at 11:51 +0800, Lulu Cheng wrote:
> > > gcc/ChangeLog:
> > > 
> > >         * config.in: Regenerate.
> > >         * config/loongarch/loongarch.h (ASM_PREFERRED_EH_DATA_FORMAT):
> > >         Select the value of the macro definition according to whether
> > >         HAVE_AS_EH_FRAME_PCREL_ENCODING_SUPPORT is defined.
> > >         * configure: Regenerate.
> > >         * configure.ac: Reinstate HAVE_AS_EH_FRAME_PCREL_ENCODING_SUPPORT test.

> > To me it looks a little strange to list regenerated config.in &
> > configure before configure.ac.  But maybe I'm wrong here if a
> > lexicographical order is preferred...

> This information is generated by me through the git gcc-commit-mklog command,
> then I didn't move the sequence.

I have no objection then.
  

Patch

diff --git a/gcc/config.in b/gcc/config.in
index 16bb963b45b..413b2bd36cb 100644
--- a/gcc/config.in
+++ b/gcc/config.in
@@ -404,13 +404,19 @@ 
 #endif
 
 
+/* Define if your assembler supports eh_frame pcrel encoding. */
+#ifndef USED_FOR_TARGET
+#undef HAVE_AS_EH_FRAME_PCREL_ENCODING_SUPPORT
+#endif
+
+
 /* Define if your assembler supports the R_PPC64_ENTRY relocation. */
 #ifndef USED_FOR_TARGET
 #undef HAVE_AS_ENTRY_MARKERS
 #endif
 
 
-/* Define if your assembler supports explicit relocations. */
+/* Define if your assembler supports explicit relocation. */
 #ifndef USED_FOR_TARGET
 #undef HAVE_AS_EXPLICIT_RELOCS
 #endif
diff --git a/gcc/config/loongarch/loongarch.h b/gcc/config/loongarch/loongarch.h
index 89a5bd728fe..8b1288961e4 100644
--- a/gcc/config/loongarch/loongarch.h
+++ b/gcc/config/loongarch/loongarch.h
@@ -1127,8 +1127,13 @@  struct GTY (()) machine_function
 };
 #endif
 
+#ifdef HAVE_AS_EH_FRAME_PCREL_ENCODING_SUPPORT
+#define ASM_PREFERRED_EH_DATA_FORMAT(CODE, GLOBAL) \
+  (((GLOBAL) ? DW_EH_PE_indirect : 0) | DW_EH_PE_pcrel | DW_EH_PE_sdata4)
+#else
 #define ASM_PREFERRED_EH_DATA_FORMAT(CODE, GLOBAL) \
   (((GLOBAL) ? DW_EH_PE_indirect : 0) | DW_EH_PE_absptr)
+#endif
 
 /* Do emit .note.GNU-stack by default.  */
 #ifndef NEED_INDICATE_EXEC_STACK
diff --git a/gcc/configure b/gcc/configure
index 7eb9479ae8e..1e1a264b92a 100755
--- a/gcc/configure
+++ b/gcc/configure
@@ -28836,6 +28836,41 @@  if test $gcc_cv_as_loongarch_explicit_relocs = yes; then
 
 $as_echo "#define HAVE_AS_EXPLICIT_RELOCS 1" >>confdefs.h
 
+fi
+
+    { $as_echo "$as_me:${as_lineno-$LINENO}: checking assembler for eh_frame pcrel encoding support" >&5
+$as_echo_n "checking assembler for eh_frame pcrel encoding support... " >&6; }
+if ${gcc_cv_as_loongarch_eh_frame_pcrel_encoding_support+:} false; then :
+  $as_echo_n "(cached) " >&6
+else
+  gcc_cv_as_loongarch_eh_frame_pcrel_encoding_support=no
+  if test x$gcc_cv_as != x; then
+    $as_echo '.LFB1780 = .
+       .cfi_startproc
+       .cfi_personality 0x9b,DW.ref.__gxx_personality_v0
+       .cfi_lsda 0x1b,.LLSDA1780
+       .cfi_endproc' > conftest.s
+    if { ac_try='$gcc_cv_as $gcc_cv_as_flags  -o conftest.o conftest.s >&5'
+  { { eval echo "\"\$as_me\":${as_lineno-$LINENO}: \"$ac_try\""; } >&5
+  (eval $ac_try) 2>&5
+  ac_status=$?
+  $as_echo "$as_me:${as_lineno-$LINENO}: \$? = $ac_status" >&5
+  test $ac_status = 0; }; }
+    then
+	gcc_cv_as_loongarch_eh_frame_pcrel_encoding_support=yes
+    else
+      echo "configure: failed program was" >&5
+      cat conftest.s >&5
+    fi
+    rm -f conftest.o conftest.s
+  fi
+fi
+{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $gcc_cv_as_loongarch_eh_frame_pcrel_encoding_support" >&5
+$as_echo "$gcc_cv_as_loongarch_eh_frame_pcrel_encoding_support" >&6; }
+if test $gcc_cv_as_loongarch_eh_frame_pcrel_encoding_support = yes; then
+
+$as_echo "#define HAVE_AS_EH_FRAME_PCREL_ENCODING_SUPPORT 1" >>confdefs.h
+
 fi
 
     ;;
diff --git a/gcc/configure.ac b/gcc/configure.ac
index e5f708c6b38..eb9e79ef8e3 100644
--- a/gcc/configure.ac
+++ b/gcc/configure.ac
@@ -5302,6 +5302,15 @@  x:
       [a:pcalau12i $t0,%pc_hi20(a)],,
       [AC_DEFINE(HAVE_AS_EXPLICIT_RELOCS, 1,
 	  [Define if your assembler supports explicit relocation.])])
+    gcc_GAS_CHECK_FEATURE([eh_frame pcrel encoding support],
+      gcc_cv_as_loongarch_eh_frame_pcrel_encoding_support,,
+      [.LFB1780 = .
+       .cfi_startproc
+       .cfi_personality 0x9b,DW.ref.__gxx_personality_v0
+       .cfi_lsda 0x1b,.LLSDA1780
+       .cfi_endproc],,
+      [AC_DEFINE(HAVE_AS_EH_FRAME_PCREL_ENCODING_SUPPORT, 1,
+	  [Define if your assembler supports eh_frame pcrel encoding.])])
     ;;
     s390*-*-*)
     gcc_GAS_CHECK_FEATURE([.gnu_attribute support],