with the values. Instead a DEPRECATED_FIX_CALL_DUMMY replacement
(PUSH_DUMMY_BREAKPOINT?) should just do everything. */
#ifdef GDB_TARGET_IS_HPPA
- real_pc = DEPRECATED_FIX_CALL_DUMMY (dummy1, start_sp, funaddr, nargs, args,
- value_type, using_gcc);
+ (*real_pc) = DEPRECATED_FIX_CALL_DUMMY (dummy1, start_sp, funaddr, nargs,
+ args, value_type, using_gcc);
#else
if (DEPRECATED_FIX_CALL_DUMMY_P ())
{
vector. Hence this direct call.
A follow-on change is to modify this interface so that it takes
- thread OR frame OR tpid as a parameter, and returns a dummy
+ thread OR frame OR ptid as a parameter, and returns a dummy
frame handle. The handle can then be used further down as a
- parameter SAVE_DUMMY_FRAME_TOS. Hmm, thinking about it, since
- everything is ment to be using generic dummy frames, why not
- even use some of the dummy frame code to here - do a regcache
- dup and then pass the duped regcache, along with all the other
- stuff, at one single point.
+ parameter to generic_save_dummy_frame_tos(). Hmm, thinking
+ about it, since everything is ment to be using generic dummy
+ frames, why not even use some of the dummy frame code to here -
+ do a regcache dup and then pass the duped regcache, along with
+ all the other stuff, at one single point.
In fact, you can even save the structure's return address in the
dummy frame and fix one of those nasty lost struct return edge
On a RISC architecture, a void parameterless generic dummy
frame (i.e., no parameters, no result) typically does not
need to push anything the stack and hence can leave SP and
- FP. Similarly, a framelss (possibly leaf) function does
+ FP. Similarly, a frameless (possibly leaf) function does
not push anything on the stack and, hence, that too can
leave FP and SP unchanged. As a consequence, a sequence of
void parameterless generic dummy frame calls to frameless
|| (INNER_THAN (2, 1) && sp >= old_sp));
}
else
- /* FIXME: cagney/2002-09-18: Hey, you loose! Who knows how
- badly aligned the SP is! Further, per comment above, if the
- generic dummy frame ends up empty (because nothing is pushed)
- GDB won't be able to correctly perform back traces. If a
- target is having trouble with backtraces, first thing to do
- is add FRAME_ALIGN() to its architecture vector. After that,
- try adding SAVE_DUMMY_FRAME_TOS() and modifying
- DEPRECATED_FRAME_CHAIN so that when the next outer frame is a
- generic dummy, it returns the current frame's base. */
+ /* FIXME: cagney/2002-09-18: Hey, you loose!
+
+ Who knows how badly aligned the SP is! Further, per comment
+ above, if the generic dummy frame ends up empty (because
+ nothing is pushed) GDB won't be able to correctly perform
+ back traces. If a target is having trouble with backtraces,
+ first thing to do is add FRAME_ALIGN() to the architecture
+ vector. If that fails, try unwind_dummy_id(). */
sp = old_sp;
}
if (DEPRECATED_DUMMY_WRITE_SP_P ())
DEPRECATED_DUMMY_WRITE_SP (sp);
- if (SAVE_DUMMY_FRAME_TOS_P ())
- SAVE_DUMMY_FRAME_TOS (sp);
+ if (gdbarch_unwind_dummy_id_p (current_gdbarch))
+ {
+ /* Sanity. The exact same SP value is returned by
+ PUSH_DUMMY_CALL, saved as the dummy-frame TOS, and used by
+ unwind_dummy_id to form the frame ID's stack address. */
+ gdb_assert (DEPRECATED_USE_GENERIC_DUMMY_FRAMES);
+ generic_save_dummy_frame_tos (sp);
+ }
+ else if (DEPRECATED_SAVE_DUMMY_FRAME_TOS_P ())
+ DEPRECATED_SAVE_DUMMY_FRAME_TOS (sp);
/* Now proceed, having reached the desired place. */
clear_proceed_status ();
set_momentary_breakpoint. We need to give the breakpoint a
frame ID so that the breakpoint code can correctly re-identify
the dummy breakpoint. */
- /* The assumption here is that push_dummy_call() returned the
- stack part of the frame ID. Unfortunatly, many older
- architectures were, via a convoluted mess, relying on the
- poorly defined and greatly overloaded DEPRECATED_TARGET_READ_FP
- or DEPRECATED_FP_REGNUM to supply the value. */
- if (DEPRECATED_TARGET_READ_FP_P ())
- frame = frame_id_build (DEPRECATED_TARGET_READ_FP (), sal.pc);
- else if (DEPRECATED_FP_REGNUM >= 0)
- frame = frame_id_build (read_register (DEPRECATED_FP_REGNUM), sal.pc);
+ if (gdbarch_unwind_dummy_id_p (current_gdbarch))
+ {
+ /* Sanity. The exact same SP value is returned by
+ PUSH_DUMMY_CALL, saved as the dummy-frame TOS, and used by
+ unwind_dummy_id to form the frame ID's stack address. */
+ gdb_assert (DEPRECATED_USE_GENERIC_DUMMY_FRAMES);
+ frame = frame_id_build (sp, sal.pc);
+ }
else
- frame = frame_id_build (sp, sal.pc);
+ {
+ /* The assumption here is that push_dummy_call() returned the
+ stack part of the frame ID. Unfortunatly, many older
+ architectures were, via a convoluted mess, relying on the
+ poorly defined and greatly overloaded
+ DEPRECATED_TARGET_READ_FP or DEPRECATED_FP_REGNUM to supply
+ the value. */
+ if (DEPRECATED_TARGET_READ_FP_P ())
+ frame = frame_id_build (DEPRECATED_TARGET_READ_FP (), sal.pc);
+ else if (DEPRECATED_FP_REGNUM >= 0)
+ frame = frame_id_build (read_register (DEPRECATED_FP_REGNUM), sal.pc);
+ else
+ frame = frame_id_build (sp, sal.pc);
+ }
bpt = set_momentary_breakpoint (sal, frame, bp_call_dummy);
bpt->disposition = disp_del;
}