This is the mail archive of the
gdb-patches@sources.redhat.com
mailing list for the GDB project.
[commit] Add a SENTINEL_FRAME type
- From: Andrew Cagney <cagney at gnu dot org>
- To: gdb-patches at sources dot redhat dot com
- Date: Sat, 01 May 2004 15:47:12 -0400
- Subject: [commit] Add a SENTINEL_FRAME type
To quote s390-tdep.c:
- if (get_frame_type (next_frame) == NORMAL_FRAME
- /* For some reason, sentinel frames are NORMAL_FRAMEs
- -- but they have negative frame level. */
- && frame_relative_level (next_frame) >= 0)
and sentinel-frame.c:
- /* Should the sentinel frame be given a special type? */
- NORMAL_FRAME,
There is no reason for having sentinel frame type set to NORMAL_FRAME,
and their attributes (e.g., address in block) are nothing like normal
frames.
This adds an explicit SENTINEL_FRAME type and then the fact that
SENTINEL_FRAME != NORMAL_FRAME to simplify some of the more obvious
sentinel frame tests.
committed,
Andrew
2004-05-01 Andrew Cagney <cagney@redhat.com>
* frame.c (create_sentinel_frame): Set type to SENTINEL_FRAME.
* dummy-frame.c (dummy_frame_this_id): Use get_frame_type, instead
of frame_relative_level.
* sentinel-frame.c (sentinel_frame_unwinder): Set unwinder's type
to SENTINEL_FRAME.
* frame.h (enum frame_type): Add SENTINEL_FRAME.
* s390-tdep.c (s390_prologue_frame_unwind_cache): Delete calls to
frame_relative_level.
Index: dummy-frame.c
===================================================================
RCS file: /cvs/src/src/gdb/dummy-frame.c,v
retrieving revision 1.30
diff -p -u -r1.30 dummy-frame.c
--- dummy-frame.c 3 Apr 2004 21:22:10 -0000 1.30
+++ dummy-frame.c 1 May 2004 19:33:43 -0000
@@ -353,7 +353,7 @@ dummy_frame_this_id (struct frame_info *
determine the dummy frame's ID. */
(*this_id) = gdbarch_unwind_dummy_id (current_gdbarch, next_frame);
}
- else if (frame_relative_level (next_frame) < 0)
+ else if (get_frame_type (next_frame) == SENTINEL_FRAME)
{
/* We're unwinding a sentinel frame, the PC of which is pointing
at a stack dummy. Fake up the dummy frame's ID using the
Index: frame.c
===================================================================
RCS file: /cvs/src/src/gdb/frame.c,v
retrieving revision 1.176
diff -p -u -r1.176 frame.c
--- frame.c 1 May 2004 19:02:11 -0000 1.176
+++ frame.c 1 May 2004 19:33:45 -0000
@@ -767,7 +767,7 @@ static struct frame_info *
create_sentinel_frame (struct regcache *regcache)
{
struct frame_info *frame = FRAME_OBSTACK_ZALLOC (struct frame_info);
- frame->type = NORMAL_FRAME;
+ frame->type = SENTINEL_FRAME;
frame->level = -1;
/* Explicitly initialize the sentinel frame's cache. Provide it
with the underlying regcache. In the future additional
Index: frame.h
===================================================================
RCS file: /cvs/src/src/gdb/frame.h,v
retrieving revision 1.130
diff -p -u -r1.130 frame.h
--- frame.h 1 May 2004 19:02:11 -0000 1.130
+++ frame.h 1 May 2004 19:33:45 -0000
@@ -355,7 +355,10 @@ enum frame_type
DUMMY_FRAME,
/* In a signal handler, various OSs handle this in various ways.
The main thing is that the frame may be far from normal. */
- SIGTRAMP_FRAME
+ SIGTRAMP_FRAME,
+ /* Sentinel or registers frame. This frame obtains register values
+ direct from the inferior's registers. */
+ SENTINEL_FRAME
};
extern enum frame_type get_frame_type (struct frame_info *);
Index: s390-tdep.c
===================================================================
RCS file: /cvs/src/src/gdb/s390-tdep.c,v
retrieving revision 1.130
diff -p -u -r1.130 s390-tdep.c
--- s390-tdep.c 22 Mar 2004 22:33:33 -0000 1.130
+++ s390-tdep.c 1 May 2004 19:33:46 -0000
@@ -1803,10 +1803,10 @@ s390_prologue_frame_unwind_cache (struct
/* If the next frame is a NORMAL_FRAME, this frame *cannot* have frame
size zero. This is only possible if the next frame is a sentinel
frame, a dummy frame, or a signal trampoline frame. */
- if (get_frame_type (next_frame) == NORMAL_FRAME
- /* For some reason, sentinel frames are NORMAL_FRAMEs
- -- but they have negative frame level. */
- && frame_relative_level (next_frame) >= 0)
+ /* FIXME: cagney/2004-05-01: This sanity check shouldn't be
+ needed, instead the code should simpliy rely on its
+ analysis. */
+ if (get_frame_type (next_frame) == NORMAL_FRAME)
return 0;
/* If we really have a frameless function, %r14 must be valid
@@ -1850,9 +1850,9 @@ s390_prologue_frame_unwind_cache (struct
treat it as frameless if we're currently within the function epilog
code at a point where the frame pointer has already been restored.
This can only happen in an innermost frame. */
- if (size > 0
- && (get_frame_type (next_frame) != NORMAL_FRAME
- || frame_relative_level (next_frame) < 0))
+ /* FIXME: cagney/2004-05-01: This sanity check shouldn't be needed,
+ instead the code should simpliy rely on its analysis. */
+ if (size > 0 && get_frame_type (next_frame) != NORMAL_FRAME)
{
/* See the comment in s390_in_function_epilogue_p on why this is
not completely reliable ... */
Index: sentinel-frame.c
===================================================================
RCS file: /cvs/src/src/gdb/sentinel-frame.c,v
retrieving revision 1.8
diff -p -u -r1.8 sentinel-frame.c
--- sentinel-frame.c 8 Jun 2003 18:27:14 -0000 1.8
+++ sentinel-frame.c 1 May 2004 19:33:46 -0000
@@ -83,8 +83,7 @@ sentinel_frame_this_id (struct frame_inf
const struct frame_unwind sentinel_frame_unwinder =
{
- /* Should the sentinel frame be given a special type? */
- NORMAL_FRAME,
+ SENTINEL_FRAME,
sentinel_frame_this_id,
sentinel_frame_prev_register
};