OpenE2K
/
binutils-gdb
12
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Sun Sep 10 10:24:48 1995 Michael Tiemann <tiemann@axon.cygnus.com> 

* tm-ppc-eabi.h (PC_IN_CALL_DUMMY): Redefine this to work with the
	simulator.  FIXME.

	* rs6000-tdep.c (push_dummy_frame): Calculate the correct link
	register offset from the current frame (don't assume it is always 8).
	(push_dummy_frame): Add comment about having only 4096 bytes of
	stack space in the simulator (by default).

	* remote-sim.c (gdbsim_create_inferior): Call
	`add_text_to_loadinfo' so that gdb can find TOC entries when
	calling functions in the inferior.

With this changes, it is now possible to correctly call inferior
functions in the PPC simulator.
This commit is contained in:
Michael Tiemann 1995-09-10 17:37:22 +00:00
parent 0384d2e3df
commit 3a4f97862f
3 changed files with 32 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
gdb/ChangeLog
View File

@ -1,3 +1,17 @@
Sun Sep 10 10:24:48 1995 Michael Tiemann <tiemann@axon.cygnus.com>
* tm-ppc-eabi.h (PC_IN_CALL_DUMMY): Redefine this to work with the
simulator. FIXME.
* rs6000-tdep.c (push_dummy_frame): Calculate the correct link
register offset from the current frame (don't assume it is always 8).
(push_dummy_frame): Add comment about having only 4096 bytes of
stack space in the simulator (by default).
* remote-sim.c (gdbsim_create_inferior): Call
`add_text_to_loadinfo' so that gdb can find TOC entries when
calling functions in the inferior.
Sun Sep 10 09:00:28 1995 Peter Schauer (pes@regent.e-technik.tu-muenchen.de)
* sparc-tdep.c (fill_fpregset): Fix incorrect FP_MAX_REGNUM

6
gdb/remote-sim.c
View File

@ -200,6 +200,12 @@ gdbsim_create_inferior (exec_file, args, env)
inferior_pid = 42;
insert_breakpoints (); /* Needed to get correct instruction in cache */
/* FIXME: This needs to be done in a target-specific fashion.
What you see here works for the PPC simulator tiemann 9/9/1995.
Also, the fact that we can find the addresses for breakpoints without
this mechanism suggests some redundant code that could be merged. */
add_text_to_loadinfo (entry_pt, entry_pt+exec_bfd->sections->_cooked_size);
proceed (entry_pt, TARGET_SIGNAL_DEFAULT, 0);
}

14
gdb/rs6000-tdep.c
View File

@ -445,6 +445,10 @@ push_dummy_frame ()
/* Same thing, target byte order. */
char pc_targ[4];
/* Needed to figure out where to save the dummy link area.
FIXME: There should be an easier way to do this, no? tiemann 9/9/95. */
struct rs6000_framedata fdata;
int ii;
target_fetch_registers (-1);
@ -463,6 +467,8 @@ push_dummy_frame ()
pc = read_register(PC_REGNUM);
store_address (pc_targ, 4, pc);
(void) skip_prologue (get_pc_function_start (pc) + FUNCTION_START_OFFSET, &fdata);
dummy_frame_addr [dummy_frame_count++] = sp;
/* Be careful! If the stack pointer is not decremented first, then kernel
@ -471,6 +477,10 @@ push_dummy_frame ()
before writing register values into the new frame, decrement and update
%sp first in order to secure your frame. */
/* FIXME: We don't check if the stack really has this much space.
This is a problem on the ppc simulator (which only grants one page
(4096 bytes) by default. */
write_register (SP_REGNUM, sp-DUMMY_FRAME_SIZE);
/* gdb relies on the state of current_frame. We'd better update it,
@ -479,7 +489,7 @@ push_dummy_frame ()
flush_cached_frames ();
/* save program counter in link register's space. */
write_memory (sp+8, pc_targ, 4);
write_memory (sp+fdata.lr_offset, pc_targ, 4);
/* save all floating point and general purpose registers here. */
@ -550,7 +560,7 @@ pop_dummy_frame ()
&registers[REGISTER_BYTE (FPLAST_REGNUM + ii)], 4);
read_memory (sp-(DUMMY_FRAME_SIZE-8),
&registers [REGISTER_BYTE(PC_REGNUM)], 4);
&registers [REGISTER_BYTE(PC_REGNUM)], 4);
/* when a dummy frame was being pushed, we had to decrement %sp first, in
order to secure astack space. Thus, saved %sp (or %r1) value, is not the