[patch] make --disable-gdbmi work

[patch] make --disable-gdbmi work

[Aleksandar Ristovski]

[-- Attachment #1: Type: text/plain, Size: 1031 bytes --]

Hello,

I have attempted to create a 'minimal' gdb. I used the following configure
options (linux host, native):

--disable-gdbmi --disable-sim --disable-tui --disable-gdbtk

but it fails to build due to some unnecessary dependencies.

The attached patch fixes the problem.

There are other possible decoupling to be done. For example, there are a few 
changes needed to make REMOTE_OBS optional (or at least some of the files listed 
in it).

Thanks,
Aleksandar Ristovski
QNX Software Systems


ChangeLog:

2008-01-31  Aleksandar Ristovski  <aristovski@qnx.com>

	* Makefile.in (SUBDIR_MI_OBJS): Removed mi-common.o from the list.
	(SUBDIR_MI_SRCS): Removed mi-common.c from the list.
	(async_h, async.o): Added.
	(mi_common_h, mi-common-o): Removed.
	(COMMON_OBS): Added async.o.
	* async.c: New file. Content is copied from mi-common.c.
	* async.h: New file. Content is copied from mi-common.h.
	* breakpoint.c: Include async.h instead mi-common.h.
	* infrun.c: Likewise.
	* mi-common.c: Removed.
	* mi-common.h: Removed.
	





[-- Attachment #2: disable-gdbmi.diff --]
[-- Type: text/plain, Size: 10250 bytes --]

diff -x CVS -up -N -r clean/gdb/Makefile.in src/gdb/Makefile.in
--- clean/gdb/Makefile.in	2008-01-30 02:17:31.000000000 -0500
+++ src/gdb/Makefile.in	2008-01-31 10:31:17.000000000 -0500
@@ -186,14 +186,14 @@ SUBDIR_MI_OBS = \
 	mi-cmds.o mi-cmd-env.o mi-cmd-var.o mi-cmd-break.o mi-cmd-stack.o \
 	mi-cmd-file.o mi-cmd-disas.o mi-symbol-cmds.o mi-cmd-target.o \
 	mi-interp.o \
-	mi-main.o mi-parse.o mi-getopt.o mi-common.o
+	mi-main.o mi-parse.o mi-getopt.o
 SUBDIR_MI_SRCS = \
 	mi/mi-out.c mi/mi-console.c \
 	mi/mi-cmds.c mi/mi-cmd-env.c \
 	mi/mi-cmd-var.c mi/mi-cmd-break.c mi/mi-cmd-stack.c \
 	mi/mi-cmd-file.c mi/mi-cmd-disas.c mi/mi-symbol-cmds.c \
 	mi/mi-cmd-target.c mi/mi-interp.c \
-	mi/mi-main.c mi/mi-parse.c mi/mi-getopt.c mi/mi-common.c
+	mi/mi-main.c mi/mi-parse.c mi/mi-getopt.c
 SUBDIR_MI_DEPS =
 SUBDIR_MI_LDFLAGS=
 SUBDIR_MI_CFLAGS= \
@@ -588,6 +588,7 @@ TARGET_FLAGS_TO_PASS = \
 
 SFILES = ada-exp.y ada-lang.c ada-typeprint.c ada-valprint.c  \
 	addrmap.c \
+	async.c \
 	auxv.c ax-general.c ax-gdb.c \
 	bcache.c \
 	bfd-target.c \
@@ -729,6 +730,7 @@ annotate_h = annotate.h $(symtab_h) $(gd
 arch_utils_h = arch-utils.h
 arm_linux_tdep_h = arm-linux-tdep.h
 arm_tdep_h = arm-tdep.h
+async_h = async.h
 auxv_h = auxv.h
 ax_gdb_h = ax-gdb.h
 ax_h = ax.h $(doublest_h)
@@ -932,7 +934,6 @@ mi_getopt_h = $(srcdir)/mi/mi-getopt.h
 mi_main_h = $(srcdir)/mi/mi-main.h
 mi_out_h = $(srcdir)/mi/mi-out.h
 mi_parse_h = $(srcdir)/mi/mi-parse.h
-mi_common_h = $(srcdir)/mi/mi-common.h
 
 #
 # gdb/tui/ headers
@@ -1072,7 +1073,8 @@ COMMON_OBS = $(DEPFILES) $(CONFIG_OBS) $
 	tramp-frame.o \
 	solib.o solib-null.o \
 	prologue-value.o memory-map.o xml-support.o \
-	target-descriptions.o target-memory.o xml-tdesc.o xml-builtin.o
+	target-descriptions.o target-memory.o xml-tdesc.o xml-builtin.o \
+	async.o
 
 TSOBS = inflow.o
 
@@ -1921,6 +1923,7 @@ arm-tdep.o: arm-tdep.c $(defs_h) $(frame
 arm-wince-tdep.o: arm-wince-tdep.c $(defs_h) $(osabi_h) \
 	$(gdbcore_h) $(target_h) $(solib_h) $(solib_target_h) \
 	$(gdb_string_h) $(arm_tdep_h)
+async.o: async.c $(defs_h) $(gdb_assert_h) $(async_h)
 auxv.o: auxv.c $(defs_h) $(target_h) $(gdbtypes_h) $(command_h) \
 	$(inferior_h) $(valprint_h) $(gdb_assert_h) $(auxv_h) \
 	$(elf_common_h)
@@ -3235,8 +3238,6 @@ mi-parse.o: $(srcdir)/mi/mi-parse.c $(de
 mi-symbol-cmds.o: $(srcdir)/mi/mi-symbol-cmds.c $(defs_h) $(mi_cmds_h) \
 	$(symtab_h) $(ui_out_h)
 	$(CC) -c $(INTERNAL_CFLAGS) $(srcdir)/mi/mi-symbol-cmds.c
-mi-common.o: $(srcdir)/mi/mi-common.c $(defs_h) $(mi_common_h)
-	$(CC) -c $(INTERNAL_CFLAGS) $(srcdir)/mi/mi-common.c
 
 #
 # gdb/signals/ dependencies
diff -x CVS -up -N -r clean/gdb/async.c src/gdb/async.c
--- clean/gdb/async.c	1969-12-31 19:00:00.000000000 -0500
+++ src/gdb/async.c	2008-01-31 10:33:51.000000000 -0500
@@ -0,0 +1,46 @@
+/* Implementation of common async functions.  
+   Copyright (C) 2005, 2007, 2008 Free Software Foundation, Inc.
+
+   This file is part of GDB.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 3 of the License, or
+   (at your option) any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
+
+#include "defs.h"
+#include "async.h"
+#include "gdb_assert.h"
+
+static const char * const async_reason_string_lookup[] =
+{
+  "breakpoint-hit",
+  "watchpoint-trigger",
+  "read-watchpoint-trigger",
+  "access-watchpoint-trigger",
+  "function-finished",
+  "location-reached",
+  "watchpoint-scope",
+  "end-stepping-range",
+  "exited-signalled",
+  "exited",
+  "exited-normally",
+  "signal-received",
+  NULL
+};
+
+const char *
+async_reason_lookup (enum async_reply_reason reason)
+{
+  gdb_assert (ARRAY_SIZE (async_reason_string_lookup) == EXEC_ASYNC_LAST + 1);
+  return async_reason_string_lookup[reason];
+}
+
diff -x CVS -up -N -r clean/gdb/async.h src/gdb/async.h
--- clean/gdb/async.h	1969-12-31 19:00:00.000000000 -0500
+++ src/gdb/async.h	2008-01-31 10:34:15.000000000 -0500
@@ -0,0 +1,44 @@
+/* Interface for common async routines.  
+   Copyright (C) 2005, 2007, 2008 Free Software Foundation, Inc.
+
+   This file is part of GDB.
+
+   This program is free software; you can redistribute it and/or modify
+   it under the terms of the GNU General Public License as published by
+   the Free Software Foundation; either version 3 of the License, or
+   (at your option) any later version.
+
+   This program is distributed in the hope that it will be useful,
+   but WITHOUT ANY WARRANTY; without even the implied warranty of
+   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+   GNU General Public License for more details.
+
+   You should have received a copy of the GNU General Public License
+   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
+
+#ifndef ASYNC_H
+#define ASYNC_H
+
+/* Represents the reason why GDB is sending an asynchronous command to the
+   front end.  NOTE: When modifing this, don't forget to update gdb.texinfo!  */
+enum async_reply_reason
+{
+  EXEC_ASYNC_BREAKPOINT_HIT = 0,
+  EXEC_ASYNC_WATCHPOINT_TRIGGER,
+  EXEC_ASYNC_READ_WATCHPOINT_TRIGGER,
+  EXEC_ASYNC_ACCESS_WATCHPOINT_TRIGGER,
+  EXEC_ASYNC_FUNCTION_FINISHED,
+  EXEC_ASYNC_LOCATION_REACHED,
+  EXEC_ASYNC_WATCHPOINT_SCOPE,
+  EXEC_ASYNC_END_STEPPING_RANGE,
+  EXEC_ASYNC_EXITED_SIGNALLED,
+  EXEC_ASYNC_EXITED,
+  EXEC_ASYNC_EXITED_NORMALLY,
+  EXEC_ASYNC_SIGNAL_RECEIVED,
+  /* This is here only to represent the number of enums.  */
+  EXEC_ASYNC_LAST
+};
+
+const char *async_reason_lookup (enum async_reply_reason reason);
+
+#endif
diff -x CVS -up -N -r clean/gdb/breakpoint.c src/gdb/breakpoint.c
--- clean/gdb/breakpoint.c	2008-01-29 12:52:47.000000000 -0500
+++ src/gdb/breakpoint.c	2008-01-31 09:24:26.000000000 -0500
@@ -56,7 +56,7 @@
 #include "top.h"
 
 #include "gdb-events.h"
-#include "mi/mi-common.h"
+#include "async.h"
 
 /* Prototypes for local functions. */
 
diff -x CVS -up -N -r clean/gdb/infrun.c src/gdb/infrun.c
--- clean/gdb/infrun.c	2008-01-29 17:47:19.000000000 -0500
+++ src/gdb/infrun.c	2008-01-31 09:24:42.000000000 -0500
@@ -47,7 +47,7 @@
 #include "main.h"
 
 #include "gdb_assert.h"
-#include "mi/mi-common.h"
+#include "async.h"
 
 /* Prototypes for local functions */
 
diff -x CVS -up -N -r clean/gdb/mi/mi-common.c src/gdb/mi/mi-common.c
--- clean/gdb/mi/mi-common.c	2008-01-01 17:53:14.000000000 -0500
+++ src/gdb/mi/mi-common.c	1969-12-31 19:00:00.000000000 -0500
@@ -1,51 +0,0 @@
-/* Interface for common GDB/MI data
-   Copyright (C) 2005, 2007, 2008 Free Software Foundation, Inc.
-
-   This file is part of GDB.
-
-   This program is free software; you can redistribute it and/or modify
-   it under the terms of the GNU General Public License as published by
-   the Free Software Foundation; either version 3 of the License, or
-   (at your option) any later version.
-
-   This program is distributed in the hope that it will be useful,
-   but WITHOUT ANY WARRANTY; without even the implied warranty of
-   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-   GNU General Public License for more details.
-
-   You should have received a copy of the GNU General Public License
-   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
-
-#include "defs.h"
-#include "mi-common.h"
-
-static const char * const async_reason_string_lookup[] =
-{
-  "breakpoint-hit",
-  "watchpoint-trigger",
-  "read-watchpoint-trigger",
-  "access-watchpoint-trigger",
-  "function-finished",
-  "location-reached",
-  "watchpoint-scope",
-  "end-stepping-range",
-  "exited-signalled",
-  "exited",
-  "exited-normally",
-  "signal-received",
-  NULL
-};
-
-const char *
-async_reason_lookup (enum async_reply_reason reason)
-{
-  return async_reason_string_lookup[reason];
-}
-
-void
-_initialize_gdb_mi_common (void)
-{
-  if (ARRAY_SIZE (async_reason_string_lookup) != EXEC_ASYNC_LAST + 1)
-    internal_error (__FILE__, __LINE__,
-		    _("async_reason_string_lookup is inconsistent"));
-}
diff -x CVS -up -N -r clean/gdb/mi/mi-common.h src/gdb/mi/mi-common.h
--- clean/gdb/mi/mi-common.h	2008-01-01 17:53:14.000000000 -0500
+++ src/gdb/mi/mi-common.h	1969-12-31 19:00:00.000000000 -0500
@@ -1,44 +0,0 @@
-/* Interface for common GDB/MI data
-   Copyright (C) 2005, 2007, 2008 Free Software Foundation, Inc.
-
-   This file is part of GDB.
-
-   This program is free software; you can redistribute it and/or modify
-   it under the terms of the GNU General Public License as published by
-   the Free Software Foundation; either version 3 of the License, or
-   (at your option) any later version.
-
-   This program is distributed in the hope that it will be useful,
-   but WITHOUT ANY WARRANTY; without even the implied warranty of
-   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
-   GNU General Public License for more details.
-
-   You should have received a copy of the GNU General Public License
-   along with this program.  If not, see <http://www.gnu.org/licenses/>.  */
-
-#ifndef MI_COMMON_H
-#define MI_COMMON_H
-
-/* Represents the reason why GDB is sending an asynchronous command to the
-   front end.  NOTE: When modifing this, don't forget to update gdb.texinfo!  */
-enum async_reply_reason
-{
-  EXEC_ASYNC_BREAKPOINT_HIT = 0,
-  EXEC_ASYNC_WATCHPOINT_TRIGGER,
-  EXEC_ASYNC_READ_WATCHPOINT_TRIGGER,
-  EXEC_ASYNC_ACCESS_WATCHPOINT_TRIGGER,
-  EXEC_ASYNC_FUNCTION_FINISHED,
-  EXEC_ASYNC_LOCATION_REACHED,
-  EXEC_ASYNC_WATCHPOINT_SCOPE,
-  EXEC_ASYNC_END_STEPPING_RANGE,
-  EXEC_ASYNC_EXITED_SIGNALLED,
-  EXEC_ASYNC_EXITED,
-  EXEC_ASYNC_EXITED_NORMALLY,
-  EXEC_ASYNC_SIGNAL_RECEIVED,
-  /* This is here only to represent the number of enums.  */
-  EXEC_ASYNC_LAST
-};
-
-const char *async_reason_lookup (enum async_reply_reason reason);
-
-#endif

Re: [patch] make --disable-gdbmi work

[Daniel Jacobowitz]

On Thu, Jan 31, 2008 at 03:50:27PM -0500, Aleksandar Ristovski wrote:
> The attached patch fixes the problem.

My opinion is that we don't want this functionality any more.  MI used
to be optional, but it is a central part of GDB nowadays, and
conditional compilation is hard to maintain.

-- 
Daniel Jacobowitz
CodeSourcery

Re: [patch] make --disable-gdbmi work

[Aleksandar Ristovski]

Daniel Jacobowitz wrote:
> On Thu, Jan 31, 2008 at 03:50:27PM -0500, Aleksandar Ristovski wrote:
>> The attached patch fixes the problem.
> 
> My opinion is that we don't want this functionality any more.  MI used
> to be optional, but it is a central part of GDB nowadays, and
> conditional compilation is hard to maintain.
> 

It certainly adds some effort to maintenance, but it's not as bad as it may 
sound, especially since the change is not extensive at all (see the patch).

For embedded systems, it is important to have as small a binary as possible, and 
if we can reduce it with reasonable effort, I would say, why not?

Re: [patch] make --disable-gdbmi work

[Thiago Jung Bauermann]

On Thu, 2008-01-31 at 15:50 -0500, Aleksandar Ristovski wrote:
> I have attempted to create a 'minimal' gdb. I used the following configure
> options (linux host, native):
> 
> --disable-gdbmi --disable-sim --disable-tui --disable-gdbtk

You're possibly aware of it already, but since GDB 6.7 there's also
--with-expat=no to avoid linking with libexpat.
-- 
[]'s
Thiago Jung Bauermann
Software Engineer
IBM Linux Technology Center

Re: [patch] make --disable-gdbmi work

[Doug Evans]

On Jan 31, 2008 1:01 PM, Aleksandar Ristovski <aristovski@qnx.com> wrote:
> Daniel Jacobowitz wrote:
> > On Thu, Jan 31, 2008 at 03:50:27PM -0500, Aleksandar Ristovski wrote:
> >> The attached patch fixes the problem.
> >
> > My opinion is that we don't want this functionality any more.  MI used
> > to be optional, but it is a central part of GDB nowadays, and
> > conditional compilation is hard to maintain.
> >
>
> It certainly adds some effort to maintenance, but it's not as bad as it may
> sound, especially since the change is not extensive at all (see the patch).
>
> For embedded systems, it is important to have as small a binary as possible, and
> if we can reduce it with reasonable effort, I would say, why not?

[fwiw]

The patch appeals to the minimalist side in me.  And given its trivial
nature, it could go in without formally committing to the intended
goal.  If later things get unwieldy we can revisit whether to support
minimal builds then.

btw, do we have automated builds+tests for gdb?

Re: [patch] make --disable-gdbmi work

[Vladimir Prus]

Aleksandar Ristovski wrote:

> Daniel Jacobowitz wrote:
>> On Thu, Jan 31, 2008 at 03:50:27PM -0500, Aleksandar Ristovski wrote:
>>> The attached patch fixes the problem.
>> 
>> My opinion is that we don't want this functionality any more.  MI used
>> to be optional, but it is a central part of GDB nowadays, and
>> conditional compilation is hard to maintain.
>> 
> 
> It certainly adds some effort to maintenance, but it's not as bad as it
> may sound, especially since the change is not extensive at all (see the
> patch).
> 
> For embedded systems, it is important to have as small a binary as
> possible, and if we can reduce it with reasonable effort, I would say, why
> not?

Only if you're running gdb itself on an embedded target, which might not
be the best idea -- you'd need the binaries with debug symbols on the
embedded system, and you'd need memory where gdb can store same debug symbols
read from files, and so on.

- Volodya

Re: [patch] make --disable-gdbmi work

[Aleksandar Ristovski]

Doug Evans wrote:
> On Jan 31, 2008 1:01 PM, Aleksandar Ristovski <aristovski@qnx.com> wrote:
>> Daniel Jacobowitz wrote:
>>> On Thu, Jan 31, 2008 at 03:50:27PM -0500, Aleksandar Ristovski wrote:
>>>> The attached patch fixes the problem.
>>> My opinion is that we don't want this functionality any more.  MI used
>>> to be optional, but it is a central part of GDB nowadays, and
>>> conditional compilation is hard to maintain.
>>>
>> It certainly adds some effort to maintenance, but it's not as bad as it may
>> sound, especially since the change is not extensive at all (see the patch).
>>
>> For embedded systems, it is important to have as small a binary as possible, and
>> if we can reduce it with reasonable effort, I would say, why not?
> 
> [fwiw]
> 
> The patch appeals to the minimalist side in me.  And given its trivial
> nature, it could go in without formally committing to the intended
> goal.  If later things get unwieldy we can revisit whether to support
> minimal builds then.
> 
> btw, do we have automated builds+tests for gdb?
> 

Yes, Vladimir correctly pointed out - my argument about reducing the code size 
is not really a good one...

I still think the patch is useful, if for no other reason, to keep dependencies 
single directional (mi -> gdb).

Re: [patch] make --disable-gdbmi work

[Daniel Jacobowitz]

On Fri, Feb 01, 2008 at 02:06:11PM -0500, Aleksandar Ristovski wrote:
> Yes, Vladimir correctly pointed out - my argument about reducing the code 
> size is not really a good one...
>
> I still think the patch is useful, if for no other reason, to keep 
> dependencies single directional (mi -> gdb).

That's the part I don't think is useful.  The dependencies shouldn't
be single-directional; I don't see why we should keep MI out of the
rest of GDB.

-- 
Daniel Jacobowitz
CodeSourcery

Re: [patch] make --disable-gdbmi work

[Daniel Jacobowitz]

On Fri, Feb 01, 2008 at 09:39:47AM -0800, Doug Evans wrote:
> btw, do we have automated builds+tests for gdb?

Not publicly; various people run them and occasionally post the
results.  Ours are having some technical difficulties lately.

-- 
Daniel Jacobowitz
CodeSourcery

Re: [patch] make --disable-gdbmi work

[Aleksandar Ristovski]

Daniel Jacobowitz wrote:
> On Fri, Feb 01, 2008 at 02:06:11PM -0500, Aleksandar Ristovski wrote:
>> Yes, Vladimir correctly pointed out - my argument about reducing the code 
>> size is not really a good one...
>>
>> I still think the patch is useful, if for no other reason, to keep 
>> dependencies single directional (mi -> gdb).
> 
> That's the part I don't think is useful.  The dependencies shouldn't
> be single-directional; I don't see why we should keep MI out of the
> rest of GDB.
> 
Without wanting to get into a lengthy discussion, I have to disagree with this: 
keeping things modular is important, and one of the cleanest ways is keeping 
dependencies single directional whenever possible. Modularity may not be the 
goal of keeping things modular, but extensibility and maintainability are.

Ok, I got into the 'lengthy discussion' mode, but I quickly deleted what I 
started typing (about functional interface in gdb, MI using it blah blah...)...