diff mbox

make it possible for the target to rename ".tm_clone_table"

Message ID 89484C79-C459-4F0B-A2EF-53CBE78AA6B3@sandoe-acoustics.co.uk
State New
Headers show

Commit Message

Iain Sandoe Nov. 19, 2011, 12:04 p.m. UTC 
On 18 Nov 2011, at 22:06, Joseph S. Myers wrote:

> TM_CLONE_TABLE_SECTION_NAME would better be a target hook, not a  
> macro.

... done as below ...

It should be possible for Rainer to define  
TARGET_ASM_TM_CLONE_TABLE_SECTION_NAME to NULL, and have varasm use  
the normal data section for tm_clone_table sections.    One might wish  
to re-adjust testsuite/gcc.dg/tm/20100615.c if that is done.


OK for trunk?
Iain

gcc:

	* target.def (tm_clone_table_section_name): New hook.
	* doc/tm.texi.in (TARGET_ASM_TM_CLONE_TABLE_SECTION_NAME): Define.
	* doc/tm.texi: Regenerate.
	* varasm.c (dump_tm_clone_pairs): Use target  
tm_clone_table_section_name if
	available, if that is NULL fall back to the data section.
	* config/darwin.h (TARGET_ASM_TM_CLONE_TABLE_SECTION_NAME): New.


gcc/testsuite:

	* gcc.dg/tm/20100615.c: Adjust for Darwin tm_clone_table section name.

Comments

Mike Stump Nov. 19, 2011, 5:39 p.m. UTC | #1 
On Nov 19, 2011, at 4:04 AM, Iain Sandoe wrote:
> On 18 Nov 2011, at 22:06, Joseph S. Myers wrote:
> 
>> TM_CLONE_TABLE_SECTION_NAME would better be a target hook, not a macro.
> 
> ... done as below ...
> 
> It should be possible for Rainer to define TARGET_ASM_TM_CLONE_TABLE_SECTION_NAME to NULL, and have varasm use the normal data section for tm_clone_table sections.    One might wish to re-adjust testsuite/gcc.dg/tm/20100615.c if that is done.
> 
> 
> OK for trunk?

Darwin bits Ok.  One step closer...
Richard Henderson Nov. 19, 2011, 6:43 p.m. UTC | #2 
On 11/19/2011 02:04 AM, Iain Sandoe wrote:
> It should be possible for Rainer to define
> TARGET_ASM_TM_CLONE_TABLE_SECTION_NAME to NULL, and have varasm use
> the normal data section for tm_clone_table sections.    One might
> wish to re-adjust testsuite/gcc.dg/tm/20100615.c if that is done.

Well, not really.  OSF is going to need an extension to collect2 in
order to create a single table, or changes to other bits of the 
compiler to invoke _ITM_registerTMCloneTable for each object file's
table.

... actually, there seems to be some sort of merge error.  I don't
see the crtstuff.c changes that called _ITM_registerTMCloneTable in
the first place.

> +/* Set if the target wants to override the default ".tm_clone_table"
> +   section name.  */
> +DEFHOOKPOD
> +(tm_clone_table_section_name,
> + "Define this macro to specify the section name that should be used\
> +  for transactional memory clone tables.  If this is defined to NULL\
> +  the normal data section will be used for the tables instead.\
> +  The default section name is @code{\".tm_clone_table\"}.",
> + const char *, ".tm_clone_table")

I'd prefer we return a section object instead.  I.e. pull the
call to get_named_section into the hook as well.


r~
diff mbox

Patch

Index: gcc/doc/tm.texi.in
===================================================================
--- gcc/doc/tm.texi.in	(revision 181497)
+++ gcc/doc/tm.texi.in	(working copy)
@@ -7013,6 +7013,8 @@  otherwise.
 
 @hook TARGET_ASM_MERGEABLE_RODATA_PREFIX
 
+@hook TARGET_ASM_TM_CLONE_TABLE_SECTION_NAME
+
 @hook TARGET_ASM_SELECT_RTX_SECTION
 Return the section into which a constant @var{x}, of mode @var{mode},
 should be placed.  You can assume that @var{x} is some kind of
Index: gcc/target.def
===================================================================
--- gcc/target.def	(revision 181497)
+++ gcc/target.def	(working copy)
@@ -305,6 +305,16 @@  section names for mergeable constant data.  Define
 the string if a different section name should be used.",
  const char *, ".rodata")
 
+/* Set if the target wants to override the default ".tm_clone_table"
+   section name.  */
+DEFHOOKPOD
+(tm_clone_table_section_name,
+ "Define this macro to specify the section name that should be used\
+  for transactional memory clone tables.  If this is defined to NULL\
+  the normal data section will be used for the tables instead.\
+  The default section name is @code{\".tm_clone_table\"}.",
+ const char *, ".tm_clone_table")
+
 /* Output a constructor for a symbol with a given priority.  */
 DEFHOOK
 (constructor,
Index: gcc/testsuite/gcc.dg/tm/20100615.c
===================================================================
--- gcc/testsuite/gcc.dg/tm/20100615.c	(revision 181497)
+++ gcc/testsuite/gcc.dg/tm/20100615.c	(working copy)
@@ -3,7 +3,8 @@ 
 
 /* Since the non TM version of new_node() gets optimized away, it
    shouldn't appear in the clone table either.  */
-/* { dg-final { scan-assembler-not "tm_clone_table" } } */
+/* { dg-final { scan-assembler-not "tm_clone_table" { target { ! *-*-darwin*  } } } } */
+/* { dg-final { scan-assembler-not "__DATA,__tm_clone_table" { target *-*-darwin*  } } } */
 
 #define NULL 0
 extern void *malloc (__SIZE_TYPE__);
Index: gcc/varasm.c
===================================================================
--- gcc/varasm.c	(revision 181497)
+++ gcc/varasm.c	(working copy)
@@ -5961,7 +5961,13 @@  dump_tm_clone_pairs (VEC(tm_alias_pair,heap) *tm_a
 
       if (!switched)
 	{
-	  switch_to_section (get_named_section (NULL, ".tm_clone_table", 3));
+	  if (targetm.asm_out.tm_clone_table_section_name)
+	    switch_to_section (get_named_section 
+				(NULL, 
+				 targetm.asm_out.tm_clone_table_section_name,
+				 3));
+	  else
+	    switch_to_section (data_section);
 	  assemble_align (POINTER_SIZE);
 	  switched = true;
 	}
Index: gcc/config/darwin.h
===================================================================
--- gcc/config/darwin.h	(revision 181497)
+++ gcc/config/darwin.h	(working copy)
@@ -692,6 +692,10 @@  extern GTY(()) section * darwin_sections[NUM_DARWI
 #define TARGET_ASM_UNIQUE_SECTION darwin_unique_section
 #undef  TARGET_ASM_FUNCTION_RODATA_SECTION
 #define TARGET_ASM_FUNCTION_RODATA_SECTION default_no_function_rodata_section
+
+#undef  TARGET_ASM_TM_CLONE_TABLE_SECTION_NAME
+#define TARGET_ASM_TM_CLONE_TABLE_SECTION_NAME "__DATA,__tm_clone_table"
+
 #undef  TARGET_ASM_RELOC_RW_MASK
 #define TARGET_ASM_RELOC_RW_MASK machopic_reloc_rw_mask