This is the mail archive of the
gdb-patches@sources.redhat.com
mailing list for the GDB project.
[RFA]: ARM: Fix parameter handling of FP types in arm_push_arguments()
- From: Corinna Vinschen <vinschen at redhat dot com>
- To: gdb-patches <gdb-patches at sources dot redhat dot com>
- Date: Fri, 8 Feb 2002 16:49:56 +0100
- Subject: [RFA]: ARM: Fix parameter handling of FP types in arm_push_arguments()
- Reply-to: gdb-patches <gdb-patches at sources dot redhat dot com>
Hi,
while running the testsuite on ARM, I found that the handling
of type `float' as argument type in the function arm_push_arguments()
is pretty buggy.
For some reason the size of float arguments was always given as
sizeof(float) even when K&R style functions are called which
expect the float arguments with sizeof(double).
Up to this point one would expect that calling K&R functions with
float arguments is broken.
As a matter of fact, arm_push_arguments() coerced float args
always to double types so that K&R functions were happy while
calling ANSI style functions getting float args were broken.
The below fix is doing two things now.
- It adds the macro COERCE_FLOAT_TO_DOUBLE() to config/arm/tm-arm.h,
calling standard_coerce_float_to_double(). This results in
floats being coerced to double already *before* arm_push_arguments()
is called.
- It eliminates the now useless (and wrong) special float handling
from arm_push_arguments().
Corinna
2002-02-08 Corinna Vinschen <vinschen@redhat.com>
* arm-tdep.c (arm_push_arguments): Eliminate special float
type handling.
* config/arm/tm-arm.h (COERCE_FLOAT_TO_DOUBLE): Define to
call standard_coerce_float_to_double().
Index: arm-tdep.c
===================================================================
RCS file: /cvs/src/src/gdb/arm-tdep.c,v
retrieving revision 1.36
diff -u -p -r1.36 arm-tdep.c
--- arm-tdep.c 2002/02/06 15:21:16 1.36
+++ arm-tdep.c 2002/02/08 15:32:18
@@ -1440,12 +1440,7 @@ arm_push_arguments (int nargs, struct va
arg_type = check_typedef (VALUE_TYPE (args[argnum]));
len = TYPE_LENGTH (arg_type);
- /* ANSI C code passes float arguments as integers, K&R code
- passes float arguments as doubles. Correct for this here. */
- if (TYPE_CODE_FLT == TYPE_CODE (arg_type) && REGISTER_SIZE == len)
- nstack_size += FP_REGISTER_VIRTUAL_SIZE;
- else
- nstack_size += len;
+ nstack_size += len;
}
/* Allocate room on the stack, and initialize our stack frame
@@ -1482,21 +1477,6 @@ arm_push_arguments (int nargs, struct va
typecode = TYPE_CODE (arg_type);
val = (char *) VALUE_CONTENTS (args[argnum]);
- /* ANSI C code passes float arguments as integers, K&R code
- passes float arguments as doubles. The .stabs record for
- for ANSI prototype floating point arguments records the
- type as FP_INTEGER, while a K&R style (no prototype)
- .stabs records the type as FP_FLOAT. In this latter case
- the compiler converts the float arguments to double before
- calling the function. */
- if (TYPE_CODE_FLT == typecode && REGISTER_SIZE == len)
- {
- DOUBLEST dblval;
- dblval = extract_floating (val, len);
- len = TARGET_DOUBLE_BIT / TARGET_CHAR_BIT;
- val = alloca (len);
- store_floating (val, len, dblval);
- }
#if 1
/* I don't know why this code was disable. The only logical use
for a function pointer is to call that function, so setting
Index: config/arm/tm-arm.h
===================================================================
RCS file: /cvs/src/src/gdb/config/arm/tm-arm.h,v
retrieving revision 1.23
diff -u -p -r1.23 tm-arm.h
--- tm-arm.h 2002/02/06 15:21:17 1.23
+++ tm-arm.h 2002/02/08 15:32:20
@@ -430,4 +430,6 @@ extern int arm_pc_is_thumb (bfd_vma mema
a Thumb function. */
extern int arm_pc_is_thumb_dummy (bfd_vma memaddr);
+#define COERCE_FLOAT_TO_DOUBLE(formal, actual) (standard_coerce_float_to_double (formal, actual))
+
#endif /* TM_ARM_H */
--
Corinna Vinschen
Cygwin Developer
Red Hat, Inc.
mailto:vinschen@redhat.com