diff options
| author | Jim Kingdon <jkingdon@engr.sgi.com> | 1993-07-14 15:12:05 +0000 |
|---|---|---|
| committer | Jim Kingdon <jkingdon@engr.sgi.com> | 1993-07-14 15:12:05 +0000 |
| commit | 5af4f5f6f13163c6d9a516fc7cf7383d824d14b4 (patch) | |
| tree | b126a6a9fc89606345b6c1fb61b41425b2e7cb2b /gdb/stabsread.c | |
| parent | b2eed00fa9d4dbee676d1f96dda1fce088eee5bf (diff) | |
| download | gdb-5af4f5f6f13163c6d9a516fc7cf7383d824d14b4.zip gdb-5af4f5f6f13163c6d9a516fc7cf7383d824d14b4.tar.gz gdb-5af4f5f6f13163c6d9a516fc7cf7383d824d14b4.tar.bz2 | |
Tue Jul 13 14:03:48 1993 Jim Kingdon (kingdon@lioth.cygnus.com)
* stabsread.c (define_symbol): Make the caddr_t hack apply to `function
returning foo' as well as `pointer to foo'.
* remote.c [REMOTE_BREAKPOINT]: Use for breakpoint insn if defined.
* config/m68k/tm-m68k.h: Define it.
* mem-break.c, breakpoint.c: Improve comments.
Tue Jul 13 13:35:31 1993 Frederic Pierresteguy (F.Pierresteguy@frcl.bull.fr)
* config/m68k/tm-dpx2.h: Replace "tm-68k.h" with "m68k/tm-m68k.h".
* config/m68k/xm-dpx2.h: Define HAVE_TERMIOS not HAVE_TERMIO.
Diffstat (limited to 'gdb/stabsread.c')
| -rw-r--r-- | gdb/stabsread.c | 30 |
1 files changed, 16 insertions, 14 deletions
diff --git a/gdb/stabsread.c b/gdb/stabsread.c index 7e8f10a..a31427a 100644 --- a/gdb/stabsread.c +++ b/gdb/stabsread.c @@ -1037,25 +1037,27 @@ define_symbol (valu, string, desc, type, objfile) if (TYPE_NAME (SYMBOL_TYPE (sym)) == NULL) { - if (TYPE_CODE (SYMBOL_TYPE (sym)) == TYPE_CODE_PTR) + if (TYPE_CODE (SYMBOL_TYPE (sym)) == TYPE_CODE_PTR + || TYPE_CODE (SYMBOL_TYPE (sym)) == TYPE_CODE_FUNC) { - /* If we are giving a name to a type such as "pointer - to foo", we better not set the TYPE_NAME. If the - program contains "typedef char *caddr_t;", we don't - want all variables of type char * to print as - caddr_t. This is not just a consequence of GDB's - type management; PCC and GCC (at least through - version 2.4) both output variables of either type - char * or caddr_t with the type number defined in - the 't' symbol for caddr_t. If a future compiler - cleans this up it GDB is not ready for it yet, but - if it becomes ready we somehow need to disable this - check (without breaking the PCC/GCC2.4 case). + /* If we are giving a name to a type such as "pointer to + foo" or "function returning foo", we better not set + the TYPE_NAME. If the program contains "typedef char + *caddr_t;", we don't want all variables of type char + * to print as caddr_t. This is not just a + consequence of GDB's type management; PCC and GCC (at + least through version 2.4) both output variables of + either type char * or caddr_t with the type number + defined in the 't' symbol for caddr_t. If a future + compiler cleans this up it GDB is not ready for it + yet, but if it becomes ready we somehow need to + disable this check (without breaking the PCC/GCC2.4 + case). Sigh. Fortunately, this check seems not to be necessary - for anything except pointers. */ + for anything except pointers or functions. */ } else TYPE_NAME (SYMBOL_TYPE (sym)) = SYMBOL_NAME (sym); |