[PATCH] Add libcc1 v1 compatibility to C compile feature

[PATCH] Add libcc1 v1 compatibility to C compile feature

[Keith Seitz]

From: Alexandre Oliva <aoliva@redhat.com>

This patch adds v1 compatibiltiy to the C compile feature.  The only change
in v1 concerns the handling of integer types, which permits GDB to specify the
built-in name for the type.

As far as I know, the C frontend is still on v0, so this patch is purely
precautionary. [By default C++ compile uses the equivalent of the C
frontend's int_type and float_type (aka the "v1" versions).]

gdb/ChangeLog:

	* compile/compile-c-types.c (convert_int, convert_float):
	Update for C FE v1.
---
 gdb/compile/compile-c-types.c | 28 +++++++++++++++++++++++-----
 1 file changed, 23 insertions(+), 5 deletions(-)

diff --git a/gdb/compile/compile-c-types.c b/gdb/compile/compile-c-types.c
index 19669a2edc..212cfe66be 100644
--- a/gdb/compile/compile-c-types.c
+++ b/gdb/compile/compile-c-types.c
@@ -273,9 +273,22 @@ convert_func (struct compile_c_instance *context, struct type *type)
 static gcc_type
 convert_int (struct compile_c_instance *context, struct type *type)
 {
-  return C_CTX (context)->c_ops->int_type_v0 (C_CTX (context),
-					      TYPE_UNSIGNED (type),
-					      TYPE_LENGTH (type));
+  if (C_CTX (context)->c_ops->c_version >= GCC_C_FE_VERSION_1)
+    {
+      if (TYPE_NOSIGN (type))
+	{
+	  gdb_assert (TYPE_LENGTH (type) == 1);
+	  return C_CTX (context)->c_ops->char_type (C_CTX (context));
+	}
+      return C_CTX (context)->c_ops->int_type (C_CTX (context),
+					       TYPE_UNSIGNED (type),
+					       TYPE_LENGTH (type),
+					       TYPE_NAME (type));
+    }
+  else
+    return C_CTX (context)->c_ops->int_type_v0 (C_CTX (context),
+						TYPE_UNSIGNED (type),
+						TYPE_LENGTH (type));
 }
 
 /* Convert a floating-point type to its gcc representation.  */
@@ -283,8 +296,13 @@ convert_int (struct compile_c_instance *context, struct type *type)
 static gcc_type
 convert_float (struct compile_c_instance *context, struct type *type)
 {
-  return C_CTX (context)->c_ops->float_type_v0 (C_CTX (context),
-						TYPE_LENGTH (type));
+  if (C_CTX (context)->c_ops->c_version >= GCC_C_FE_VERSION_1)
+    return C_CTX (context)->c_ops->float_type (C_CTX (context),
+					       TYPE_LENGTH (type),
+					       TYPE_NAME (type));
+  else
+    return C_CTX (context)->c_ops->float_type_v0 (C_CTX (context),
+						  TYPE_LENGTH (type));
 }
 
 /* Convert the 'void' type to its gcc representation.  */
-- 
2.13.6

Re: [PATCH] Add libcc1 v1 compatibility to C compile feature

[Pedro Alves]

On 02/22/2018 09:04 PM, Keith Seitz wrote:
> From: Alexandre Oliva <aoliva@redhat.com>
> 
> This patch adds v1 compatibiltiy to the C compile feature.  The only change
> in v1 concerns the handling of integer types, which permits GDB to specify the
> built-in name for the type.
> 
> As far as I know, the C frontend is still on v0, so this patch is purely
> precautionary. [By default C++ compile uses the equivalent of the C
> frontend's int_type and float_type (aka the "v1" versions).]
> 
> gdb/ChangeLog:
> 
> 	* compile/compile-c-types.c (convert_int, convert_float):
> 	Update for C FE v1.

OK.

Thanks,
Pedro Alves

Re: [PATCH] Add libcc1 v1 compatibility to C compile feature

[Keith Seitz]

On 04/09/2018 05:39 AM, Pedro Alves wrote:
> On 02/22/2018 09:04 PM, Keith Seitz wrote:
>> From: Alexandre Oliva <aoliva@redhat.com>
>>
>> This patch adds v1 compatibiltiy to the C compile feature.  The only change
>> in v1 concerns the handling of integer types, which permits GDB to specify the
>> built-in name for the type.
>>
>> As far as I know, the C frontend is still on v0, so this patch is purely
>> precautionary. [By default C++ compile uses the equivalent of the C
>> frontend's int_type and float_type (aka the "v1" versions).]
>>
>> gdb/ChangeLog:
>>
>> 	* compile/compile-c-types.c (convert_int, convert_float):
>> 	Update for C FE v1.
> 
> OK.

I've pushed that. Thank you!

Keith