[patch] Arguments of GTY

[patch] Arguments of GTY

[Masaki Muranaka]

Hello All,

It was failed to compile gcc-3.4.5 with binutils (CVS head).
It seems the format of GTY was modified.

Follows are minimum patches enough to build gcc-core-3.4.5.
I checked to build with binutils, gdb+dejagnu.

- - - - - -
2005-05-06	Masaki Muranaka (monaka@monami-software.com)

	* include/hashtab.h Modify the arguments in GTY.
	* include/splay-tree.h Likewise.

- - - - - -

Index: include/hashtab.h
===================================================================
RCS file: /cvs/src/src/include/hashtab.h,v
retrieving revision 1.21
diff -u -p -r1.21 hashtab.h
--- include/hashtab.h   30 Jun 2005 23:02:54 -0000      1.21
+++ include/hashtab.h   6 May 2006 16:35:47 -0000
@@ -108,7 +108,7 @@ struct htab GTY(())
    htab_del del_f;
    /* Table itself.  */
-  void ** GTY ((use_param, length ("%h.size"))) entries;
+  void ** GTY ((use_param (""), length ("%h.size"))) entries;
    /* Current size (in entries) of the hash table.  */
    size_t size;
@@ -132,7 +132,7 @@ struct htab GTY(())
    htab_free free_f;
    /* Alternate allocate/free functions, which take an extra  
argument.  */
-  void * GTY((skip)) alloc_arg;
+  void * GTY((skip (""))) alloc_arg;
    htab_alloc_with_arg alloc_with_arg_f;
    htab_free_with_arg free_with_arg_f;
Index: include/splay-tree.h
===================================================================
RCS file: /cvs/src/src/include/splay-tree.h,v
retrieving revision 1.13
diff -u -p -r1.13 splay-tree.h
--- include/splay-tree.h        10 May 2005 10:21:08 -0000      1.13
+++ include/splay-tree.h        6 May 2006 16:35:47 -0000
@@ -81,21 +81,21 @@ typedef void (*splay_tree_deallocate_fn)
struct splay_tree_node_s GTY(())
{
    /* The key.  */
-  splay_tree_key GTY ((use_param1)) key;
+  splay_tree_key GTY ((use_param1 (""))) key;
    /* The value.  */
-  splay_tree_value GTY ((use_param2)) value;
+  splay_tree_value GTY ((use_param2 (""))) value;
    /* The left and right children, respectively.  */
-  splay_tree_node GTY ((use_params)) left;
-  splay_tree_node GTY ((use_params)) right;
+  splay_tree_node GTY ((use_params (""))) left;
+  splay_tree_node GTY ((use_params (""))) right;
};
/* The splay tree itself.  */
struct splay_tree_s GTY(())
{
    /* The root of the tree.  */
-  splay_tree_node GTY ((use_params)) root;
+  splay_tree_node GTY ((use_params (""))) root;
    /* The comparision function.  */
    splay_tree_compare_fn comp;
@@ -109,7 +109,7 @@ struct splay_tree_s GTY(())
    /* Allocate/free functions, and a data pointer to pass to them.  */
    splay_tree_allocate_fn allocate;
    splay_tree_deallocate_fn deallocate;
-  void * GTY((skip)) allocate_data;
+  void * GTY((skip (""))) allocate_data;
};
typedef struct splay_tree_s *splay_tree;

--
Masaki Muranaka
Monami software

Re: [patch] Arguments of GTY

[Daniel Jacobowitz]

On Sun, May 07, 2006 at 02:14:54AM +0900, Masaki Muranaka wrote:
> Hello All,
> 
> It was failed to compile gcc-3.4.5 with binutils (CVS head).
> It seems the format of GTY was modified.
> 
> Follows are minimum patches enough to build gcc-core-3.4.5.
> I checked to build with binutils, gdb+dejagnu.

It sounds like you are trying to mix and match the include/ headers
from binutils with an older GCC source tree.  Is that right?

Don't do it.  GCC has its own copy of these headers and should be built
using its own copy.

-- 
Daniel Jacobowitz
CodeSourcery

Re: [patch] Arguments of GTY

[Masaki Muranaka]

On 2006/05/07, at 5:41, Daniel Jacobowitz wrote:
> It sounds like you are trying to mix and match the include/ headers
> from binutils with an older GCC source tree.  Is that right?

Yes.

> Don't do it.  GCC has its own copy of these headers and should be  
> built
> using its own copy.

Then, why does it still have GTY?

--
Masaki Muranaka
Monami software

Re: [patch] Arguments of GTY

[Daniel Jacobowitz]

On Sun, May 07, 2006 at 10:14:55AM +0900, Masaki Muranaka wrote:
> 
> On 2006/05/07, at 5:41, Daniel Jacobowitz wrote:
> >It sounds like you are trying to mix and match the include/ headers
> >from binutils with an older GCC source tree.  Is that right?
> 
> Yes.
> 
> >Don't do it.  GCC has its own copy of these headers and should be  
> >built
> >using its own copy.
> 
> Then, why does it still have GTY?

The files are kept synchronized in the development HEAD.  If you're
using HEAD, you can generally combine the two directories, but you're
using two different versions from several years apart.

-- 
Daniel Jacobowitz
CodeSourcery

Re: [patch] Arguments of GTY

[Masaki Muranaka]

On 2006/05/07, at 23:00, Daniel Jacobowitz wrote:
>> Then, why does it still have GTY?
>
> The files are kept synchronized in the development HEAD.  If you're
> using HEAD, you can generally combine the two directories, but you're
> using two different versions from several years apart.

Thanks for your information.
I see I had some misunderstandings.
I believed that Sourceware's HEAD supports GCC3.4
branch also, because it was maintained till quite recent.

--
Masaki Muranaka
Monami software