RFA: char is unsigned on some architectures

RFA: char is unsigned on some architectures

[Jim Blandy]

ISO C specifies that `char' acts like either `signed char' or
`unsigned char'.  At the moment, GDB assumes that `char' is `signed
char'.  This patch allows the architecture to specify the correct
answer, and sets the value correctly for the IBM S/390 architecture.

2001-12-19  Jim Blandy  <jimb@redhat.com>

	* gdbarch.sh (TARGET_CHAR_SIGNED): New macro.
	* gdbarch.c, gdbarch.h: Regenerated.
	* gdbtypes.c (build_gdbtypes): If TARGET_CHAR_SIGNED is zero,
	set the TYPE_FLAG_UNSIGNED bit on the type.
	* s390-tdep.c (s390_gdbarch_init): On the S/390, characters
	are unsigned by default.

Index: gdb/gdbarch.sh
===================================================================
RCS file: /cvs/cvsfiles/devo/gdb/gdbarch.sh,v
retrieving revision 2.91
diff -c -r2.91 gdbarch.sh
*** gdb/gdbarch.sh	2001/12/11 00:04:41	2.91
--- gdb/gdbarch.sh	2001/12/19 22:35:08
***************
*** 384,389 ****
--- 384,392 ----
  # Number of bits in a BFD_VMA for the target object file format.
  v::TARGET_BFD_VMA_BIT:int:bfd_vma_bit::::8 * sizeof (void*):TARGET_ARCHITECTURE->bits_per_address::0
  #
+ # One if \`char' acts like \`signed char', zero if \`unsigned char'.
+ v::TARGET_CHAR_SIGNED:int:char_signed::::1:-1:1::%d::
+ #
  v::IEEE_FLOAT:int:ieee_float::::0:0::0:::
  #
  f::TARGET_READ_PC:CORE_ADDR:read_pc:ptid_t ptid:ptid::0:generic_target_read_pc::0
Index: gdb/gdbtypes.c
===================================================================
RCS file: /cvs/cvsfiles/devo/gdb/gdbtypes.c,v
retrieving revision 2.126
diff -c -r2.126 gdbtypes.c
*** gdb/gdbtypes.c	2001/12/11 00:04:41	2.126
--- gdb/gdbtypes.c	2001/12/19 22:35:09
***************
*** 2977,2983 ****
  	       "void", (struct objfile *) NULL);
    builtin_type_char =
      init_type (TYPE_CODE_INT, TARGET_CHAR_BIT / TARGET_CHAR_BIT,
! 	       TYPE_FLAG_NOSIGN,
  	       "char", (struct objfile *) NULL);
    builtin_type_true_char =
      init_type (TYPE_CODE_CHAR, TARGET_CHAR_BIT / TARGET_CHAR_BIT,
--- 2977,2984 ----
  	       "void", (struct objfile *) NULL);
    builtin_type_char =
      init_type (TYPE_CODE_INT, TARGET_CHAR_BIT / TARGET_CHAR_BIT,
! 	       (TYPE_FLAG_NOSIGN
!                 | (TARGET_CHAR_SIGNED ? 0 : TYPE_FLAG_UNSIGNED)),
  	       "char", (struct objfile *) NULL);
    builtin_type_true_char =
      init_type (TYPE_CODE_CHAR, TARGET_CHAR_BIT / TARGET_CHAR_BIT,
Index: gdb/s390-tdep.c
===================================================================
RCS file: /cvs/cvsfiles/devo/gdb/s390-tdep.c,v
retrieving revision 2.31
diff -c -r2.31 s390-tdep.c
*** gdb/s390-tdep.c	2001/12/19 22:33:10	2.31
--- gdb/s390-tdep.c	2001/12/19 22:35:10
***************
*** 1760,1765 ****
--- 1760,1766 ----
    gdbarch = gdbarch_alloc (&info, NULL);
  
    set_gdbarch_believe_pcc_promotion (gdbarch, 0);
+   set_gdbarch_char_signed (gdbarch, 0);
  
    set_gdbarch_frame_args_skip (gdbarch, 0);
    set_gdbarch_frame_args_address (gdbarch, s390_frame_args_address);
Index: gdb/gdbarch.c
===================================================================
RCS file: /cvs/cvsfiles/devo/gdb/gdbarch.c,v
retrieving revision 2.112
diff -c -r2.112 gdbarch.c
*** gdb/gdbarch.c	2001/12/11 00:04:41	2.112
--- gdb/gdbarch.c	2001/12/19 22:35:05
***************
*** 144,149 ****
--- 144,150 ----
    int ptr_bit;
    int addr_bit;
    int bfd_vma_bit;
+   int char_signed;
    int ieee_float;
    gdbarch_read_pc_ftype *read_pc;
    gdbarch_write_pc_ftype *write_pc;
***************
*** 285,290 ****
--- 286,292 ----
    8 * sizeof (void*),
    8 * sizeof (void*),
    8 * sizeof (void*),
+   1,
    0,
    0,
    0,
***************
*** 446,451 ****
--- 448,454 ----
    current_gdbarch->long_double_bit = 2*TARGET_DOUBLE_BIT;
    current_gdbarch->ptr_bit = TARGET_INT_BIT;
    current_gdbarch->bfd_vma_bit = TARGET_ARCHITECTURE->bits_per_address;
+   current_gdbarch->char_signed = -1;
    current_gdbarch->read_pc = generic_target_read_pc;
    current_gdbarch->write_pc = generic_target_write_pc;
    current_gdbarch->read_fp = generic_target_read_fp;
***************
*** 561,566 ****
--- 564,571 ----
    if (gdbarch->addr_bit == 0)
      gdbarch->addr_bit = TARGET_PTR_BIT;
    /* Skip verify of bfd_vma_bit, invalid_p == 0 */
+   if (gdbarch->char_signed == -1)
+     gdbarch->char_signed = 1;
    /* Skip verify of ieee_float, invalid_p == 0 */
    /* Skip verify of read_pc, invalid_p == 0 */
    /* Skip verify of write_pc, invalid_p == 0 */
***************
*** 1867,1872 ****
--- 1872,1885 ----
                        "gdbarch_dump: TARGET_BYTE_ORDER = %ld\n",
                        (long) TARGET_BYTE_ORDER);
  #endif
+ #ifdef TARGET_CHAR_SIGNED
+   fprintf_unfiltered (file,
+                       "gdbarch_dump: TARGET_CHAR_SIGNED # %s\n",
+                       XSTRING (TARGET_CHAR_SIGNED));
+   fprintf_unfiltered (file,
+                       "gdbarch_dump: TARGET_CHAR_SIGNED = %d\n",
+                       (long) TARGET_CHAR_SIGNED);
+ #endif
  #ifdef TARGET_DOUBLE_BIT
    fprintf_unfiltered (file,
                        "gdbarch_dump: TARGET_DOUBLE_BIT # %s\n",
***************
*** 2263,2268 ****
--- 2276,2299 ----
                           int bfd_vma_bit)
  {
    gdbarch->bfd_vma_bit = bfd_vma_bit;
+ }
+ 
+ int
+ gdbarch_char_signed (struct gdbarch *gdbarch)
+ {
+   if (gdbarch->char_signed == -1)
+     internal_error (__FILE__, __LINE__,
+                     "gdbarch: gdbarch_char_signed invalid");
+   if (gdbarch_debug >= 2)
+     fprintf_unfiltered (gdb_stdlog, "gdbarch_char_signed called\n");
+   return gdbarch->char_signed;
+ }
+ 
+ void
+ set_gdbarch_char_signed (struct gdbarch *gdbarch,
+                          int char_signed)
+ {
+   gdbarch->char_signed = char_signed;
  }
  
  int
Index: gdb/gdbarch.h
===================================================================
RCS file: /cvs/cvsfiles/devo/gdb/gdbarch.h,v
retrieving revision 2.104
diff -c -r2.104 gdbarch.h
*** gdb/gdbarch.h	2001/12/11 00:04:41	2.104
--- gdb/gdbarch.h	2001/12/19 22:35:06
***************
*** 287,292 ****
--- 287,310 ----
  #endif
  #endif
  
+ /* One if `char' acts like `signed char', zero if `unsigned char'. */
+ 
+ /* Default (value) for non- multi-arch platforms. */
+ #if (!GDB_MULTI_ARCH) && !defined (TARGET_CHAR_SIGNED)
+ #define TARGET_CHAR_SIGNED (1)
+ #endif
+ 
+ extern int gdbarch_char_signed (struct gdbarch *gdbarch);
+ extern void set_gdbarch_char_signed (struct gdbarch *gdbarch, int char_signed);
+ #if (GDB_MULTI_ARCH > GDB_MULTI_ARCH_PARTIAL) && defined (TARGET_CHAR_SIGNED)
+ #error "Non multi-arch definition of TARGET_CHAR_SIGNED"
+ #endif
+ #if GDB_MULTI_ARCH
+ #if (GDB_MULTI_ARCH > GDB_MULTI_ARCH_PARTIAL) || !defined (TARGET_CHAR_SIGNED)
+ #define TARGET_CHAR_SIGNED (gdbarch_char_signed (current_gdbarch))
+ #endif
+ #endif
+ 
  /* Default (value) for non- multi-arch platforms. */
  #if (!GDB_MULTI_ARCH) && !defined (IEEE_FLOAT)
  #define IEEE_FLOAT (0)

Re: RFA: char is unsigned on some architectures

[Andrew Cagney]

> ISO C specifies that `char' acts like either `signed char' or
> `unsigned char'.  At the moment, GDB assumes that `char' is `signed
> char'.  This patch allows the architecture to specify the correct
> answer, and sets the value correctly for the IBM S/390 architecture.
> 
> 2001-12-19  Jim Blandy  <jimb@redhat.com>
> 
> * gdbarch.sh (TARGET_CHAR_SIGNED): New macro.
> 	* gdbarch.c, gdbarch.h: Regenerated.
> 	* gdbtypes.c (build_gdbtypes): If TARGET_CHAR_SIGNED is zero,
> 	set the TYPE_FLAG_UNSIGNED bit on the type.
> 	* s390-tdep.c (s390_gdbarch_init): On the S/390, characters
> 	are unsigned by default.
> 


Hmm, the rs6000 and prodogy (Hmm, IBM ...) are also unsigned.  Want to 
add that ``obvious'' tweek?

Andrew

Re: RFA: char is unsigned on some architectures

[Jim Blandy]

Andrew Cagney <ac131313@cygnus.com> writes:
> Hmm, the rs6000 and prodogy (Hmm, IBM ...) are also unsigned.  Want to 
> add that ``obvious'' tweek?

Sure.

Thanks for the quick response.

Re: RFA: char is unsigned on some architectures

[Jim Blandy]

Andrew Cagney <ac131313@cygnus.com> writes:
> Hmm, the rs6000 and prodogy (Hmm, IBM ...) are also unsigned.  Want to 
> add that ``obvious'' tweek?

Okay, here's the patch I actually committed:

2001-12-19  Jim Blandy  <jimb@redhat.com>

	* gdbarch.sh (TARGET_CHAR_SIGNED): New macro.
	* gdbarch.c, gdbarch.h: Regenerated.
	* gdbtypes.c (build_gdbtypes): If TARGET_CHAR_SIGNED is zero,
	set the TYPE_FLAG_UNSIGNED bit on the type.
	* s390-tdep.c (s390_gdbarch_init): On the S/390, characters
	are unsigned by default.
	* rs6000-tdep.c (rs6000_gdbarch_init): Same for PowerPC and
	RS6000.

Index: gdb/gdbarch.c
===================================================================
RCS file: /cvs/cvsfiles/devo/gdb/gdbarch.c,v
retrieving revision 2.113
diff -c -r2.113 gdbarch.c
*** gdb/gdbarch.c	2001/12/19 21:20:29	2.113
--- gdb/gdbarch.c	2001/12/20 03:16:50
***************
*** 144,149 ****
--- 144,150 ----
    int ptr_bit;
    int addr_bit;
    int bfd_vma_bit;
+   int char_signed;
    int ieee_float;
    gdbarch_read_pc_ftype *read_pc;
    gdbarch_write_pc_ftype *write_pc;
***************
*** 285,290 ****
--- 286,292 ----
    8 * sizeof (void*),
    8 * sizeof (void*),
    8 * sizeof (void*),
+   1,
    0,
    0,
    0,
***************
*** 446,451 ****
--- 448,454 ----
    current_gdbarch->long_double_bit = 2*TARGET_DOUBLE_BIT;
    current_gdbarch->ptr_bit = TARGET_INT_BIT;
    current_gdbarch->bfd_vma_bit = TARGET_ARCHITECTURE->bits_per_address;
+   current_gdbarch->char_signed = -1;
    current_gdbarch->read_pc = generic_target_read_pc;
    current_gdbarch->write_pc = generic_target_write_pc;
    current_gdbarch->read_fp = generic_target_read_fp;
***************
*** 561,566 ****
--- 564,571 ----
    if (gdbarch->addr_bit == 0)
      gdbarch->addr_bit = TARGET_PTR_BIT;
    /* Skip verify of bfd_vma_bit, invalid_p == 0 */
+   if (gdbarch->char_signed == -1)
+     gdbarch->char_signed = 1;
    /* Skip verify of ieee_float, invalid_p == 0 */
    /* Skip verify of read_pc, invalid_p == 0 */
    /* Skip verify of write_pc, invalid_p == 0 */
***************
*** 1867,1872 ****
--- 1872,1885 ----
                        "gdbarch_dump: TARGET_BYTE_ORDER = %ld\n",
                        (long) TARGET_BYTE_ORDER);
  #endif
+ #ifdef TARGET_CHAR_SIGNED
+   fprintf_unfiltered (file,
+                       "gdbarch_dump: TARGET_CHAR_SIGNED # %s\n",
+                       XSTRING (TARGET_CHAR_SIGNED));
+   fprintf_unfiltered (file,
+                       "gdbarch_dump: TARGET_CHAR_SIGNED = %d\n",
+                       (long) TARGET_CHAR_SIGNED);
+ #endif
  #ifdef TARGET_DOUBLE_BIT
    fprintf_unfiltered (file,
                        "gdbarch_dump: TARGET_DOUBLE_BIT # %s\n",
***************
*** 2263,2268 ****
--- 2276,2299 ----
                           int bfd_vma_bit)
  {
    gdbarch->bfd_vma_bit = bfd_vma_bit;
+ }
+ 
+ int
+ gdbarch_char_signed (struct gdbarch *gdbarch)
+ {
+   if (gdbarch->char_signed == -1)
+     internal_error (__FILE__, __LINE__,
+                     "gdbarch: gdbarch_char_signed invalid");
+   if (gdbarch_debug >= 2)
+     fprintf_unfiltered (gdb_stdlog, "gdbarch_char_signed called\n");
+   return gdbarch->char_signed;
+ }
+ 
+ void
+ set_gdbarch_char_signed (struct gdbarch *gdbarch,
+                          int char_signed)
+ {
+   gdbarch->char_signed = char_signed;
  }
  
  int
Index: gdb/gdbarch.h
===================================================================
RCS file: /cvs/cvsfiles/devo/gdb/gdbarch.h,v
retrieving revision 2.104
diff -c -r2.104 gdbarch.h
*** gdb/gdbarch.h	2001/12/11 00:04:41	2.104
--- gdb/gdbarch.h	2001/12/20 03:16:51
***************
*** 287,292 ****
--- 287,310 ----
  #endif
  #endif
  
+ /* One if `char' acts like `signed char', zero if `unsigned char'. */
+ 
+ /* Default (value) for non- multi-arch platforms. */
+ #if (!GDB_MULTI_ARCH) && !defined (TARGET_CHAR_SIGNED)
+ #define TARGET_CHAR_SIGNED (1)
+ #endif
+ 
+ extern int gdbarch_char_signed (struct gdbarch *gdbarch);
+ extern void set_gdbarch_char_signed (struct gdbarch *gdbarch, int char_signed);
+ #if (GDB_MULTI_ARCH > GDB_MULTI_ARCH_PARTIAL) && defined (TARGET_CHAR_SIGNED)
+ #error "Non multi-arch definition of TARGET_CHAR_SIGNED"
+ #endif
+ #if GDB_MULTI_ARCH
+ #if (GDB_MULTI_ARCH > GDB_MULTI_ARCH_PARTIAL) || !defined (TARGET_CHAR_SIGNED)
+ #define TARGET_CHAR_SIGNED (gdbarch_char_signed (current_gdbarch))
+ #endif
+ #endif
+ 
  /* Default (value) for non- multi-arch platforms. */
  #if (!GDB_MULTI_ARCH) && !defined (IEEE_FLOAT)
  #define IEEE_FLOAT (0)
Index: gdb/gdbarch.sh
===================================================================
RCS file: /cvs/cvsfiles/devo/gdb/gdbarch.sh,v
retrieving revision 2.92
diff -c -r2.92 gdbarch.sh
*** gdb/gdbarch.sh	2001/12/19 21:20:29	2.92
--- gdb/gdbarch.sh	2001/12/20 03:16:51
***************
*** 384,389 ****
--- 384,392 ----
  # Number of bits in a BFD_VMA for the target object file format.
  v::TARGET_BFD_VMA_BIT:int:bfd_vma_bit::::8 * sizeof (void*):TARGET_ARCHITECTURE->bits_per_address::0
  #
+ # One if \`char' acts like \`signed char', zero if \`unsigned char'.
+ v::TARGET_CHAR_SIGNED:int:char_signed::::1:-1:1::%d::
+ #
  v::IEEE_FLOAT:int:ieee_float::::0:0::0:::
  #
  f::TARGET_READ_PC:CORE_ADDR:read_pc:ptid_t ptid:ptid::0:generic_target_read_pc::0
Index: gdb/gdbtypes.c
===================================================================
RCS file: /cvs/cvsfiles/devo/gdb/gdbtypes.c,v
retrieving revision 2.127
diff -c -r2.127 gdbtypes.c
*** gdb/gdbtypes.c	2001/12/19 21:20:29	2.127
--- gdb/gdbtypes.c	2001/12/20 03:16:53
***************
*** 3029,3035 ****
  	       "void", (struct objfile *) NULL);
    builtin_type_char =
      init_type (TYPE_CODE_INT, TARGET_CHAR_BIT / TARGET_CHAR_BIT,
! 	       TYPE_FLAG_NOSIGN,
  	       "char", (struct objfile *) NULL);
    builtin_type_true_char =
      init_type (TYPE_CODE_CHAR, TARGET_CHAR_BIT / TARGET_CHAR_BIT,
--- 3029,3036 ----
  	       "void", (struct objfile *) NULL);
    builtin_type_char =
      init_type (TYPE_CODE_INT, TARGET_CHAR_BIT / TARGET_CHAR_BIT,
! 	       (TYPE_FLAG_NOSIGN
!                 | (TARGET_CHAR_SIGNED ? 0 : TYPE_FLAG_UNSIGNED)),
  	       "char", (struct objfile *) NULL);
    builtin_type_true_char =
      init_type (TYPE_CODE_CHAR, TARGET_CHAR_BIT / TARGET_CHAR_BIT,
Index: gdb/rs6000-tdep.c
===================================================================
RCS file: /cvs/cvsfiles/devo/gdb/rs6000-tdep.c,v
retrieving revision 2.129
diff -c -r2.129 rs6000-tdep.c
*** gdb/rs6000-tdep.c	2001/12/19 21:20:29	2.129
--- gdb/rs6000-tdep.c	2001/12/20 03:16:54
***************
*** 2298,2303 ****
--- 2298,2304 ----
    set_gdbarch_float_bit (gdbarch, 4 * TARGET_CHAR_BIT);
    set_gdbarch_double_bit (gdbarch, 8 * TARGET_CHAR_BIT);
    set_gdbarch_long_double_bit (gdbarch, 8 * TARGET_CHAR_BIT);
+   set_gdbarch_char_signed (gdbarch, 0);
  
    set_gdbarch_use_generic_dummy_frames (gdbarch, 1);
    set_gdbarch_call_dummy_length (gdbarch, 0);
Index: gdb/s390-tdep.c
===================================================================
RCS file: /cvs/cvsfiles/devo/gdb/s390-tdep.c,v
retrieving revision 2.32
diff -c -r2.32 s390-tdep.c
*** gdb/s390-tdep.c	2001/12/19 23:08:16	2.32
--- gdb/s390-tdep.c	2001/12/20 03:16:55
***************
*** 1763,1768 ****
--- 1763,1769 ----
    gdbarch = gdbarch_alloc (&info, NULL);
  
    set_gdbarch_believe_pcc_promotion (gdbarch, 0);
+   set_gdbarch_char_signed (gdbarch, 0);
  
    set_gdbarch_frame_args_skip (gdbarch, 0);
    set_gdbarch_frame_args_address (gdbarch, s390_frame_args_address);

Re: RFA: char is unsigned on some architectures

[Lars Brinkhoff]

Andrew Cagney <ac131313@cygnus.com> writes:
> > ISO C specifies that `char' acts like either `signed char' or
> > `unsigned char'.  At the moment, GDB assumes that `char' is `signed
> > char'.  This patch allows the architecture to specify the correct
> > answer, and sets the value correctly for the IBM S/390 architecture.
> > 2001-12-19  Jim Blandy  <jimb@redhat.com>
> > * gdbarch.sh (TARGET_CHAR_SIGNED): New macro.
> > 	* gdbarch.c, gdbarch.h: Regenerated.
> > 	* gdbtypes.c (build_gdbtypes): If TARGET_CHAR_SIGNED is zero,
> > 	set the TYPE_FLAG_UNSIGNED bit on the type.
> > 	* s390-tdep.c (s390_gdbarch_init): On the S/390, characters
> > 	are unsigned by default.
> 
> Hmm, the rs6000 and prodogy (Hmm, IBM ...) are also unsigned.

How about ARM?

-- 
Lars Brinkhoff          http://lars.nocrew.org/     Linux, GCC, PDP-10
Brinkhoff Consulting    http://www.brinkhoff.se/    programming

Re: RFA: char is unsigned on some architectures

[Jim Blandy]

Lars Brinkhoff <lars.spam@nocrew.org> writes:
> Andrew Cagney <ac131313@cygnus.com> writes:
> > > ISO C specifies that `char' acts like either `signed char' or
> > > `unsigned char'.  At the moment, GDB assumes that `char' is `signed
> > > char'.  This patch allows the architecture to specify the correct
> > > answer, and sets the value correctly for the IBM S/390 architecture.
> > > 2001-12-19  Jim Blandy  <jimb@redhat.com>
> > > * gdbarch.sh (TARGET_CHAR_SIGNED): New macro.
> > > 	* gdbarch.c, gdbarch.h: Regenerated.
> > > 	* gdbtypes.c (build_gdbtypes): If TARGET_CHAR_SIGNED is zero,
> > > 	set the TYPE_FLAG_UNSIGNED bit on the type.
> > > 	* s390-tdep.c (s390_gdbarch_init): On the S/390, characters
> > > 	are unsigned by default.
> > 
> > Hmm, the rs6000 and prodogy (Hmm, IBM ...) are also unsigned.
> 
> How about ARM?

I don't know anything about the ARM.  If you do, well, it's a one-line
patch.  :)

Re: RFA: char is unsigned on some architectures

[Richard Earnshaw]

> > ISO C specifies that `char' acts like either `signed char' or
> > `unsigned char'.  At the moment, GDB assumes that `char' is `signed
> > char'.  This patch allows the architecture to specify the correct
> > answer, and sets the value correctly for the IBM S/390 architecture.
> > 
> > 2001-12-19  Jim Blandy  <jimb@redhat.com>
> > 
> > * gdbarch.sh (TARGET_CHAR_SIGNED): New macro.
> > 	* gdbarch.c, gdbarch.h: Regenerated.
> > 	* gdbtypes.c (build_gdbtypes): If TARGET_CHAR_SIGNED is zero,
> > 	set the TYPE_FLAG_UNSIGNED bit on the type.
> > 	* s390-tdep.c (s390_gdbarch_init): On the S/390, characters
> > 	are unsigned by default.
> > 
> 
> 
> Hmm, the rs6000 and prodogy (Hmm, IBM ...) are also unsigned.  Want to 
> add that ``obvious'' tweek?
> 
> Andrew
> 

And the ARM.

R.

Re: RFA: char is unsigned on some architectures

[Eli Zaretskii]

Jim Blandy wrote:
> 
> ISO C specifies that `char' acts like either `signed char' or
> `unsigned char'.  At the moment, GDB assumes that `char' is `signed
> char'.  This patch allows the architecture to specify the correct
> answer, and sets the value correctly for the IBM S/390 architecture.
> 
> 2001-12-19  Jim Blandy  <jimb@redhat.com>
> 
>         * gdbarch.sh (TARGET_CHAR_SIGNED): New macro.

This new macro should be documented in gdbint.texinfo, I think.

Re: RFA: char is unsigned on some architectures

[Jim Blandy]

Eli Zaretskii <eliz@is.elta.co.il> writes:
> > 2001-12-19  Jim Blandy  <jimb@redhat.com>
> > 
> >         * gdbarch.sh (TARGET_CHAR_SIGNED): New macro.
> 
> This new macro should be documented in gdbint.texinfo, I think.

Certainly.  How is this?

Index: gdbint.texinfo
===================================================================
RCS file: /cvs/src/src/gdb/doc/gdbint.texinfo,v
retrieving revision 1.48
diff -c -c -b -F'^(' -r1.48 gdbint.texinfo
*** gdbint.texinfo	2001/12/15 16:53:24	1.48
--- gdbint.texinfo	2001/12/20 09:13:22
***************
*** 3501,3506 ****
--- 3501,3517 ----
  @findex TARGET_CHAR_BIT
  Number of bits in a char; defaults to 8.
  
+ @item TARGET_CHAR_SIGNED
+ @findex TARGET_CHAR_SIGNED
+ Non-zero if @code{char} is normally signed on this architecture; zero if
+ it should be unsigned.
+ 
+ The ISO C standard requires the compiler to treat @code{char} as
+ equivalent to either @code{signed char} or @code{unsigned char}; any
+ character in the standard execution set is supposed to be positive.
+ Most compilers treat @code{char} as signed, but @code{char} is unsigned
+ on the IBM S/390, RS6000, and PowerPC targets.
+ 
  @item TARGET_COMPLEX_BIT
  @findex TARGET_COMPLEX_BIT
  Number of bits in a complex number; defaults to @code{2 * TARGET_FLOAT_BIT}.

Re: RFA: char is unsigned on some architectures

[Eli Zaretskii]

Jim Blandy wrote:
> 
> Eli Zaretskii <eliz@is.elta.co.il> writes:
> > > 2001-12-19  Jim Blandy  <jimb@redhat.com>
> > >
> > >         * gdbarch.sh (TARGET_CHAR_SIGNED): New macro.
> >
> > This new macro should be documented in gdbint.texinfo, I think.
> 
> Certainly.  How is this?
> 
> Index: gdbint.texinfo

Perfect!  Thanks.