* [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
@ 2010-05-13 13:27 Jan Kratochvil
2010-05-13 15:47 ` H.J. Lu
0 siblings, 1 reply; 19+ messages in thread
From: Jan Kratochvil @ 2010-05-13 13:27 UTC (permalink / raw)
To: gdb-patches
Hi,
gdb currently has needlessly huge .dynsym and thus even huge the (on-demand
loaded) LOAD segment. With the attached patch it gets smaller:
-rwxr-xr-x 1 4524488 gdb-7.1-19.fc13.x86_64/usr/bin/gdb
-rwxr-xr-x 1 4266728 gdb-7.1-19dynamiclist.fc13.x86_64/usr/bin/gdb
-rw-r--r-- 1 2364656 gdb-7.1-19.fc13.x86_64.rpm
-rw-r--r-- 1 2274300 gdb-7.1-19dynamiclist.fc13.x86_64.rpm
I am not sure with the gdbserver/Makefile.in new CDEPS there. I have not
found it defined anywhere while it is used as the gdbserver target dependency.
I hope on the build hosts where -rdynamic was supported even
-Wl,--dynamic-list must be supported.
ps_get_thread_area is not always defined while the proc-service.list file
lists it. GNU ld does not seem to be concerned by undefined symbols.
No regressions on {x86_64,x86_64-m32,i686}-fedora12-linux-gnu.
Thanks,
Jan
gdb/
2010-05-13 Jan Kratochvil <jan.kratochvil@redhat.com>
* Makefile.in (SFILES): Add proc-service.list.
* config/alpha/alpha-linux.mh (NAT_CDEPS): New.
(LOADLIBES): Replace -rdynamic by -Wl,--dynamic-list=(...).
* config/arm/linux.mh: Likewise.
* config/i386/linux.mh: Likewise.
* config/i386/linux64.mh: Likewise.
* config/ia64/linux.mh: Likewise.
* config/m32r/linux.mh: Likewise.
* config/m68k/linux.mh: Likewise.
* config/mips/linux.mh: Likewise.
* config/pa/linux.mh: Likewise.
* config/powerpc/linux.mh: Likewise.
* config/powerpc/ppc64-linux.mh: Likewise.
* config/s390/s390.mh: Likewise.
* config/sparc/linux.mh: Likewise.
* config/sparc/linux64.mh: Likewise.
* config/xtensa/linux.mh: Likewise.
* proc-service.list: New.
gdb/gdbserver/
2010-05-13 Jan Kratochvil <jan.kratochvil@redhat.com>
* Makefile.in (INTERNAL_CFLAGS): Change the subst to @DYNAMIC_LIST@.
(SFILES): Add $(srcdir)/proc-service.list.
(CDEPS): New.
* configure.ac: Change RDYNAMIC to DYNAMIC_LIST. Change the test to
-Wl,--dynamic-list.
* configure: Regenerate.
* proc-service.list: New.
--- a/gdb/Makefile.in
+++ b/gdb/Makefile.in
@@ -683,7 +683,7 @@ SFILES = ada-exp.y ada-lang.c ada-typeprint.c ada-valprint.c ada-tasks.c \
objc-exp.y objc-lang.c \
objfiles.c osabi.c observer.c osdata.c \
p-exp.y p-lang.c p-typeprint.c p-valprint.c parse.c printcmd.c \
- progspace.c \
+ proc-service.list progspace.c \
prologue-value.c psymtab.c \
regcache.c reggroups.c remote.c remote-fileio.c reverse.c \
scm-exp.c scm-lang.c scm-valprint.c \
--- a/gdb/config/alpha/alpha-linux.mh
+++ b/gdb/config/alpha/alpha-linux.mh
@@ -3,10 +3,11 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o corelow.o alpha-linux-nat.o \
fork-child.o proc-service.o linux-thread-db.o \
linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
# The dynamically loaded libthread_db needs access to symbols in the
# gdb executable.
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl -Wl,--dynamic-list=$(srcdir)/proc-service.list
# doublest.c currently assumes some properties of FP arithmetic
# on the host which require this.
--- a/gdb/config/arm/linux.mh
+++ b/gdb/config/arm/linux.mh
@@ -3,5 +3,6 @@
NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o arm-linux-nat.o \
proc-service.o linux-thread-db.o linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
-LOADLIBES= -ldl -rdynamic
+LOADLIBES= -ldl -Wl,--dynamic-list=$(srcdir)/proc-service.list
--- a/gdb/config/i386/linux.mh
+++ b/gdb/config/i386/linux.mh
@@ -5,7 +5,8 @@ NATDEPFILES= inf-ptrace.o fork-child.o \
i386-nat.o i386-linux-nat.o \
proc-service.o linux-thread-db.o \
linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
# The dynamically loaded libthread_db needs access to symbols in the
# gdb executable.
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl -Wl,--dynamic-list=$(srcdir)/proc-service.list
--- a/gdb/config/i386/linux64.mh
+++ b/gdb/config/i386/linux64.mh
@@ -3,7 +3,8 @@ NATDEPFILES= inf-ptrace.o fork-child.o \
i386-nat.o amd64-nat.o amd64-linux-nat.o linux-nat.o \
proc-service.o linux-thread-db.o linux-fork.o
NAT_FILE= config/nm-linux.h
+NAT_CDEPS = $(srcdir)/proc-service.list
# The dynamically loaded libthread_db needs access to symbols in the
# gdb executable.
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl -Wl,--dynamic-list=$(srcdir)/proc-service.list
--- a/gdb/config/ia64/linux.mh
+++ b/gdb/config/ia64/linux.mh
@@ -4,5 +4,6 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o corelow.o \
core-regset.o ia64-linux-nat.o \
proc-service.o linux-thread-db.o linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl -Wl,--dynamic-list=$(srcdir)/proc-service.list
--- a/gdb/config/m32r/linux.mh
+++ b/gdb/config/m32r/linux.mh
@@ -4,5 +4,6 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o corelow.o \
m32r-linux-nat.o proc-service.o linux-thread-db.o \
linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
-LOADLIBES= -ldl -rdynamic
+LOADLIBES= -ldl -Wl,--dynamic-list=$(srcdir)/proc-service.list
--- a/gdb/config/m68k/linux.mh
+++ b/gdb/config/m68k/linux.mh
@@ -4,7 +4,8 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o \
corelow.o m68klinux-nat.o \
proc-service.o linux-thread-db.o linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
# The dynamically loaded libthread_db needs access to symbols in the
# gdb executable.
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl -Wl,--dynamic-list=$(srcdir)/proc-service.list
--- a/gdb/config/mips/linux.mh
+++ b/gdb/config/mips/linux.mh
@@ -3,5 +3,6 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o mips-linux-nat.o \
linux-thread-db.o proc-service.o \
linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl -Wl,--dynamic-list=$(srcdir)/proc-service.list
--- a/gdb/config/pa/linux.mh
+++ b/gdb/config/pa/linux.mh
@@ -3,5 +3,6 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o corelow.o \
hppa-linux-nat.o proc-service.o linux-thread-db.o linux-nat.o \
linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl -Wl,--dynamic-list=$(srcdir)/proc-service.list
--- a/gdb/config/powerpc/linux.mh
+++ b/gdb/config/powerpc/linux.mh
@@ -6,5 +6,6 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o \
ppc-linux-nat.o proc-service.o linux-thread-db.o \
linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl -Wl,--dynamic-list=$(srcdir)/proc-service.list
--- a/gdb/config/powerpc/ppc64-linux.mh
+++ b/gdb/config/powerpc/ppc64-linux.mh
@@ -6,6 +6,7 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o \
ppc-linux-nat.o proc-service.o linux-thread-db.o \
linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
# The PowerPC has severe limitations on TOC size, and uses them even
# for non-PIC code. GDB overflows those tables when compiling with
@@ -15,4 +16,4 @@ MH_CFLAGS= -mminimal-toc
# The dynamically loaded libthread_db needs access to symbols in the
# gdb executable.
-LOADLIBES= -ldl -rdynamic
+LOADLIBES= -ldl -Wl,--dynamic-list=$(srcdir)/proc-service.list
--- a/gdb/config/s390/s390.mh
+++ b/gdb/config/s390/s390.mh
@@ -2,4 +2,5 @@
NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o corelow.o s390-nat.o \
linux-thread-db.o proc-service.o linux-nat.o linux-fork.o
-LOADLIBES = -ldl -rdynamic
+NAT_CDEPS = $(srcdir)/proc-service.list
+LOADLIBES = -ldl -Wl,--dynamic-list=$(srcdir)/proc-service.list
--- a/gdb/config/sparc/linux.mh
+++ b/gdb/config/sparc/linux.mh
@@ -4,7 +4,8 @@ NATDEPFILES= sparc-nat.o sparc-linux-nat.o \
corelow.o core-regset.o fork-child.o inf-ptrace.o \
proc-service.o linux-thread-db.o \
linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
# The dynamically loaded libthread_db needs access to symbols in the
# gdb executable.
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl -Wl,--dynamic-list=$(srcdir)/proc-service.list
--- a/gdb/config/sparc/linux64.mh
+++ b/gdb/config/sparc/linux64.mh
@@ -5,7 +5,8 @@ NATDEPFILES= sparc-nat.o sparc64-nat.o sparc64-linux-nat.o \
fork-child.o inf-ptrace.o \
proc-service.o linux-thread-db.o \
linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
# The dynamically loaded libthread_db needs access to symbols in the
# gdb executable.
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl -Wl,--dynamic-list=$(srcdir)/proc-service.list
--- a/gdb/config/xtensa/linux.mh
+++ b/gdb/config/xtensa/linux.mh
@@ -4,5 +4,6 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o xtensa-linux-nat.o \
linux-thread-db.o proc-service.o linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl -Wl,--dynamic-list=$(srcdir)/proc-service.list
--- a/gdb/gdbserver/Makefile.in
+++ b/gdb/gdbserver/Makefile.in
@@ -95,12 +95,13 @@ INTERNAL_CFLAGS = $(WARN_CFLAGS) ${CFLAGS} ${GLOBAL_CFLAGS} \
# LDFLAGS is specifically reserved for setting from the command line
# when running make.
LDFLAGS = @LDFLAGS@
-INTERNAL_LDFLAGS = $(LDFLAGS) @RDYNAMIC@
+INTERNAL_LDFLAGS = $(LDFLAGS) @DYNAMIC_LIST@
# All source files that go into linking GDB remote server.
SFILES= $(srcdir)/gdbreplay.c $(srcdir)/inferiors.c \
- $(srcdir)/mem-break.c $(srcdir)/proc-service.c $(srcdir)/regcache.c \
+ $(srcdir)/mem-break.c $(srcdir)/proc-service.c \
+ $(srcdir)/proc-service.list $(srcdir)/regcache.c \
$(srcdir)/remote-utils.c $(srcdir)/server.c $(srcdir)/target.c \
$(srcdir)/thread-db.c $(srcdir)/utils.c \
$(srcdir)/linux-arm-low.c $(srcdir)/linux-cris-low.c \
@@ -133,6 +134,7 @@ OBS = inferiors.o regcache.o remote-utils.o server.o signals.o target.o \
GDBREPLAY_OBS = gdbreplay.o version.o
GDBSERVER_LIBS = @GDBSERVER_LIBS@
XM_CLIBS = @LIBS@
+CDEPS = $(srcdir)/proc-service.list
# XML files to compile in to gdbserver, if any.
XML_DIR = $(srcdir)/../features
--- a/gdb/gdbserver/configure.ac
+++ b/gdb/gdbserver/configure.ac
@@ -189,9 +189,12 @@ if test "$srv_linux_thread_db" = "yes"; then
if test "$ac_cv_lib_dl_dlopen" = "yes"; then
srv_libs="-ldl"
old_LDFLAGS="$LDFLAGS"
- LDFLAGS="$LDFLAGS -rdynamic"
- AC_TRY_LINK([], [], [RDYNAMIC=-rdynamic], [RDYNAMIC=])
- AC_SUBST(RDYNAMIC)
+ DYNAMIC_LIST="-Wl,--dynamic-list=${srcdir}/proc-service.list"
+ LDFLAGS="$LDFLAGS $DYNAMIC_LIST"
+ AC_TRY_LINK([], [],
+ [DYNAMIC_LIST='-Wl,--dynamic-list=$(srcdir)/proc-service.list'],
+ [DYNAMIC_LIST=])
+ AC_SUBST(DYNAMIC_LIST)
LDFLAGS="$old_LDFLAGS"
else
srv_libs="-lthread_db"
--- /dev/null
+++ b/gdb/gdbserver/proc-service.list
@@ -0,0 +1,30 @@
+/* -Wl,--dynamic-list symbols exported for libthread_db.
+
+ Copyright (C) 2010 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/>. */
+
+{
+ ps_get_thread_area;
+ ps_getpid;
+ ps_lgetfpregs;
+ ps_lgetregs;
+ ps_lsetfpregs;
+ ps_lsetregs;
+ ps_pdread;
+ ps_pdwrite;
+ ps_pglobal_lookup;
+};
--- /dev/null
+++ b/gdb/proc-service.list
@@ -0,0 +1,40 @@
+/* -Wl,--dynamic-list symbols exported for libthread_db.
+
+ Copyright (C) 2010 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/>. */
+
+{
+ ps_get_thread_area;
+ ps_getpid;
+ ps_lcontinue;
+ ps_lgetfpregs;
+ ps_lgetregs;
+ ps_lgetxregs;
+ ps_lgetxregsize;
+ ps_lsetfpregs;
+ ps_lsetregs;
+ ps_lsetxregs;
+ ps_lstop;
+ ps_pcontinue;
+ ps_pdread;
+ ps_pdwrite;
+ ps_pglobal_lookup;
+ ps_plog;
+ ps_pstop;
+ ps_ptread;
+ ps_ptwrite;
+};
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-05-13 13:27 [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list Jan Kratochvil
@ 2010-05-13 15:47 ` H.J. Lu
2010-05-16 21:08 ` Jan Kratochvil
0 siblings, 1 reply; 19+ messages in thread
From: H.J. Lu @ 2010-05-13 15:47 UTC (permalink / raw)
To: Jan Kratochvil; +Cc: gdb-patches
On Thu, May 13, 2010 at 6:18 AM, Jan Kratochvil
<jan.kratochvil@redhat.com> wrote:
> Hi,
>
> gdb currently has needlessly huge .dynsym and thus even huge the (on-demand
> loaded) LOAD segment. With the attached patch it gets smaller:
>
> -rwxr-xr-x 1 4524488 gdb-7.1-19.fc13.x86_64/usr/bin/gdb
> -rwxr-xr-x 1 4266728 gdb-7.1-19dynamiclist.fc13.x86_64/usr/bin/gdb
> -rw-r--r-- 1 2364656 gdb-7.1-19.fc13.x86_64.rpm
> -rw-r--r-- 1 2274300 gdb-7.1-19dynamiclist.fc13.x86_64.rpm
>
> I am not sure with the gdbserver/Makefile.in new CDEPS there. I have not
> found it defined anywhere while it is used as the gdbserver target dependency.
>
> I hope on the build hosts where -rdynamic was supported even
> -Wl,--dynamic-list must be supported.
>
> ps_get_thread_area is not always defined while the proc-service.list file
> lists it. GNU ld does not seem to be concerned by undefined symbols.
>
> No regressions on {x86_64,x86_64-m32,i686}-fedora12-linux-gnu.
>
>
FWIW, --dynamic-list was added in Sept., 2006 and a bug was fixed in Oct., 2006.
--
H.J.
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-05-13 15:47 ` H.J. Lu
@ 2010-05-16 21:08 ` Jan Kratochvil
2010-05-27 19:58 ` Tom Tromey
2010-06-13 19:58 ` [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list Andreas Schwab
0 siblings, 2 replies; 19+ messages in thread
From: Jan Kratochvil @ 2010-05-16 21:08 UTC (permalink / raw)
To: H.J. Lu; +Cc: gdb-patches
On Thu, 13 May 2010 15:26:47 +0200, H.J. Lu wrote:
> On Thu, May 13, 2010 at 6:18 AM, Jan Kratochvil <jan.kratochvil@redhat.com> wrote:
> > I hope on the build hosts where -rdynamic was supported even
> > -Wl,--dynamic-list must be supported.
>
> FWIW, --dynamic-list was added in Sept., 2006
Oops, thanks, therefore it now still fallbacks to -rdynamic. It is
reproducible on RHEL-4.8 (+derivations).
> and a bug was fixed in Oct., 2006.
You refer probably to:
[PATCH] Fix --dynamic-list
http://sourceware.org/ml/binutils/2006-10/msg00064.html
commit 5bab058dff8c0ff134868cc6fdcbe6afe58d7ebc
Author: Jakub Jelinek <jakub@redhat.com>
Date: Wed Oct 11 14:58:19 2006 +0000
I have tested this fix on ppc64 and it has no effect on the build with binutils
from that GIT hash. lang_append_dynamic_list gets called only once with all
the symbols at once:
Breakpoint 1, lang_append_dynamic_list (dynamic=0x1020a140) at ldlang.c:6961
6961 if (link_info.dynamic)
(gdb) p *dynamic
$1 = {next = 0x1020a0f0, pattern = 0x1020a120 ".ps_ptwrite", symbol = 0x1020a170 ".ps_ptwrite", symver = 0, script = 0, mask = 1}
(gdb) p *dynamic->next
$2 = {next = 0x1020a080, pattern = 0x1020a0d0 "ps_ptwrite", symbol = 0x1020a0d0 "ps_ptwrite", symver = 0, script = 0, mask = 1}
(gdb) p *dynamic->next->next
$3 = {next = 0x1020a030, pattern = 0x1020a060 ".ps_ptread", symbol = 0x1020a0b0 ".ps_ptread", symver = 0, script = 0, mask = 1}
Tested it on:
OpenSolaris 2009.06 snv_111b X86
gcc (GCC) 3.4.3 (csl-sol210-3_4-20050802)
$ gcc -rdynamic
gcc: unrecognized option `-rdynamic'
$ gcc -Wl,-export-dynamic ...
ld: fatal: entry point symbol `xport-dynamic' is undefined
$ gcc -Wl,--export-dynamic ...
usage: ld [-6:abc:d:e:f:h:il:mo:p:rstu:z:B:CD:F:GI:L:M:N:P:Q:R:S:VW:Y:?] file(s)
$ /usr/sfw/libexec/gcc/i386-pc-solaris2.11/3.4.3/collect2 --version
ld: Software Generation Utilities - Solaris Link Editors: 5.11-1.1652
$ ld --help|grep export-dynamic
GNU ld version 2.15
-E, --export-dynamic Export all dynamic symbols
But no special option is needed for threads debugging there; there also isn't
any one available for the Sun-provided gcc build. I can remove the Solaris
host simplification
* configure.ac (solaris*): Replace -Wl,-export-dynamic by $RDYNAMIC.
if there are any concerns. The patch replaces -Wl,-export-dynamic by -rdynamic.
it could also test -Wl,-export-dynamic as a second fallback after current
-rdynamic. -rdynamic was already used directly in linux .mh config/ files.
I do not have sparc solaris available to test the gdbserver change on solaris.
gdbserver is not supported on x86* solaris.
> > I am not sure with the gdbserver/Makefile.in new CDEPS there. Â I have not
> > found it defined anywhere while it is used as the gdbserver target dependency.
No regressions on {x86_64,x86_64-m32,i686}-fedora12-linux-gnu.
Thanks,
Jan
gdb/
2010-05-16 Jan Kratochvil <jan.kratochvil@redhat.com>
* Makefile.in (RDYNAMIC): New.
(SFILES): Add proc-service.list.
* config/alpha/alpha-linux.mh (NAT_CDEPS): New.
(LOADLIBES): Replace -rdynamic by $(RDYNAMIC).
* config/arm/linux.mh: Likewise.
* config/i386/linux.mh: Likewise.
* config/i386/linux64.mh: Likewise.
* config/ia64/linux.mh: Likewise.
* config/m32r/linux.mh: Likewise.
* config/m68k/linux.mh: Likewise.
* config/mips/linux.mh: Likewise.
* config/pa/linux.mh: Likewise.
* config/powerpc/linux.mh: Likewise.
* config/powerpc/ppc64-linux.mh: Likewise.
* config/s390/s390.mh: Likewise.
* config/sparc/linux.mh: Likewise.
* config/sparc/linux64.mh: Likewise.
* config/xtensa/linux.mh: Likewise.
* configure.ac: New RDYNAMIC on native host and GCC.
(solaris*): Replace -Wl,-export-dynamic by $RDYNAMIC.
* configure: Regenerate.
* proc-service.list: New.
gdb/gdbserver/
2010-05-16 Jan Kratochvil <jan.kratochvil@redhat.com>
* Makefile.in (SFILES): Add $(srcdir)/proc-service.list.
(CDEPS): New.
* configure.ac (RDYNAMIC): New AC_MSG_CHECKING wrapping. Test also
-Wl,--dynamic-list.
* configure: Regenerate.
* proc-service.list: New.
--- a/gdb/Makefile.in
+++ b/gdb/Makefile.in
@@ -157,6 +157,8 @@ GDB_WERROR_CFLAGS = $(WERROR_CFLAGS)
GDB_WARN_CFLAGS_NO_FORMAT = `echo " $(GDB_WARN_CFLAGS) " | sed "s/ -Wformat-nonliteral / /g"`
+RDYNAMIC = @RDYNAMIC@
+
# Where is the INTL library? Typically in ../intl.
INTL = @LIBINTL@
INTL_DEPS = @LIBINTL_DEP@
@@ -683,7 +685,7 @@ SFILES = ada-exp.y ada-lang.c ada-typeprint.c ada-valprint.c ada-tasks.c \
objc-exp.y objc-lang.c \
objfiles.c osabi.c observer.c osdata.c \
p-exp.y p-lang.c p-typeprint.c p-valprint.c parse.c printcmd.c \
- progspace.c \
+ proc-service.list progspace.c \
prologue-value.c psymtab.c \
regcache.c reggroups.c remote.c remote-fileio.c reverse.c \
scm-exp.c scm-lang.c scm-valprint.c \
--- a/gdb/config/alpha/alpha-linux.mh
+++ b/gdb/config/alpha/alpha-linux.mh
@@ -3,10 +3,11 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o corelow.o alpha-linux-nat.o \
fork-child.o proc-service.o linux-thread-db.o \
linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
# The dynamically loaded libthread_db needs access to symbols in the
# gdb executable.
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl $(RDYNAMIC)
# doublest.c currently assumes some properties of FP arithmetic
# on the host which require this.
--- a/gdb/config/arm/linux.mh
+++ b/gdb/config/arm/linux.mh
@@ -3,5 +3,6 @@
NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o arm-linux-nat.o \
proc-service.o linux-thread-db.o linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
-LOADLIBES= -ldl -rdynamic
+LOADLIBES= -ldl $(RDYNAMIC)
--- a/gdb/config/i386/linux.mh
+++ b/gdb/config/i386/linux.mh
@@ -5,7 +5,8 @@ NATDEPFILES= inf-ptrace.o fork-child.o \
i386-nat.o i386-linux-nat.o \
proc-service.o linux-thread-db.o \
linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
# The dynamically loaded libthread_db needs access to symbols in the
# gdb executable.
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl $(RDYNAMIC)
--- a/gdb/config/i386/linux64.mh
+++ b/gdb/config/i386/linux64.mh
@@ -3,7 +3,8 @@ NATDEPFILES= inf-ptrace.o fork-child.o \
i386-nat.o amd64-nat.o amd64-linux-nat.o linux-nat.o \
proc-service.o linux-thread-db.o linux-fork.o
NAT_FILE= config/nm-linux.h
+NAT_CDEPS = $(srcdir)/proc-service.list
# The dynamically loaded libthread_db needs access to symbols in the
# gdb executable.
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl $(RDYNAMIC)
--- a/gdb/config/ia64/linux.mh
+++ b/gdb/config/ia64/linux.mh
@@ -4,5 +4,6 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o corelow.o \
core-regset.o ia64-linux-nat.o \
proc-service.o linux-thread-db.o linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl $(RDYNAMIC)
--- a/gdb/config/m32r/linux.mh
+++ b/gdb/config/m32r/linux.mh
@@ -4,5 +4,6 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o corelow.o \
m32r-linux-nat.o proc-service.o linux-thread-db.o \
linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
-LOADLIBES= -ldl -rdynamic
+LOADLIBES= -ldl $(RDYNAMIC)
--- a/gdb/config/m68k/linux.mh
+++ b/gdb/config/m68k/linux.mh
@@ -4,7 +4,8 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o \
corelow.o m68klinux-nat.o \
proc-service.o linux-thread-db.o linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
# The dynamically loaded libthread_db needs access to symbols in the
# gdb executable.
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl $(RDYNAMIC)
--- a/gdb/config/mips/linux.mh
+++ b/gdb/config/mips/linux.mh
@@ -3,5 +3,6 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o mips-linux-nat.o \
linux-thread-db.o proc-service.o \
linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl $(RDYNAMIC)
--- a/gdb/config/pa/linux.mh
+++ b/gdb/config/pa/linux.mh
@@ -3,5 +3,6 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o corelow.o \
hppa-linux-nat.o proc-service.o linux-thread-db.o linux-nat.o \
linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl $(RDYNAMIC)
--- a/gdb/config/powerpc/linux.mh
+++ b/gdb/config/powerpc/linux.mh
@@ -6,5 +6,6 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o \
ppc-linux-nat.o proc-service.o linux-thread-db.o \
linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl $(RDYNAMIC)
--- a/gdb/config/powerpc/ppc64-linux.mh
+++ b/gdb/config/powerpc/ppc64-linux.mh
@@ -6,6 +6,7 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o \
ppc-linux-nat.o proc-service.o linux-thread-db.o \
linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
# The PowerPC has severe limitations on TOC size, and uses them even
# for non-PIC code. GDB overflows those tables when compiling with
@@ -15,4 +16,4 @@ MH_CFLAGS= -mminimal-toc
# The dynamically loaded libthread_db needs access to symbols in the
# gdb executable.
-LOADLIBES= -ldl -rdynamic
+LOADLIBES= -ldl $(RDYNAMIC)
--- a/gdb/config/s390/s390.mh
+++ b/gdb/config/s390/s390.mh
@@ -2,4 +2,5 @@
NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o corelow.o s390-nat.o \
linux-thread-db.o proc-service.o linux-nat.o linux-fork.o
-LOADLIBES = -ldl -rdynamic
+NAT_CDEPS = $(srcdir)/proc-service.list
+LOADLIBES = -ldl $(RDYNAMIC)
--- a/gdb/config/sparc/linux.mh
+++ b/gdb/config/sparc/linux.mh
@@ -4,7 +4,8 @@ NATDEPFILES= sparc-nat.o sparc-linux-nat.o \
corelow.o core-regset.o fork-child.o inf-ptrace.o \
proc-service.o linux-thread-db.o \
linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
# The dynamically loaded libthread_db needs access to symbols in the
# gdb executable.
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl $(RDYNAMIC)
--- a/gdb/config/sparc/linux64.mh
+++ b/gdb/config/sparc/linux64.mh
@@ -5,7 +5,8 @@ NATDEPFILES= sparc-nat.o sparc64-nat.o sparc64-linux-nat.o \
fork-child.o inf-ptrace.o \
proc-service.o linux-thread-db.o \
linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
# The dynamically loaded libthread_db needs access to symbols in the
# gdb executable.
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl $(RDYNAMIC)
--- a/gdb/config/xtensa/linux.mh
+++ b/gdb/config/xtensa/linux.mh
@@ -4,5 +4,6 @@ NAT_FILE= config/nm-linux.h
NATDEPFILES= inf-ptrace.o fork-child.o xtensa-linux-nat.o \
linux-thread-db.o proc-service.o linux-nat.o linux-fork.o
+NAT_CDEPS = $(srcdir)/proc-service.list
-LOADLIBES = -ldl -rdynamic
+LOADLIBES = -ldl $(RDYNAMIC)
--- a/gdb/configure.ac
+++ b/gdb/configure.ac
@@ -1312,6 +1312,30 @@ dnl Note that we only want this if we are both native (host == target),
dnl and not doing a canadian cross build (build == host).
if test ${build} = ${host} -a ${host} = ${target} ; then
+ if test "$GCC" = "yes" ; then
+ # The dynamically loaded libthread_db needs access to symbols in the gdb
+ # executable. Older GNU ld supports --export-dynamic but --dynamic-list
+ # may not be supported there.
+ AC_MSG_CHECKING(for the dynamic export flag)
+ old_LDFLAGS="$LDFLAGS"
+ # Older GNU ld supports --export-dynamic but --dynamic-list it does not.
+ RDYNAMIC="-Wl,--dynamic-list=${srcdir}/proc-service.list"
+ LDFLAGS="$LDFLAGS $RDYNAMIC"
+ AC_TRY_LINK([], [],
+ [found="-Wl,--dynamic-list"
+ RDYNAMIC='-Wl,--dynamic-list=$(srcdir)/proc-service.list'],
+ [RDYNAMIC="-rdynamic"
+ LDFLAGS="$old_LDFLAGS $RDYNAMIC"
+ AC_TRY_LINK([], [],
+ [found="-rdynamic"],
+ [found="no"
+ # Error on its usage by .mh file.
+ RDYNAMIC="no-rdynamic-available"])])
+ AC_SUBST(RDYNAMIC)
+ LDFLAGS="$old_LDFLAGS"
+ AC_MSG_RESULT($found)
+ fi
+
case ${host_os} in
hpux*)
AC_MSG_CHECKING(for HPUX/OSF thread support)
@@ -1342,19 +1366,7 @@ if test ${build} = ${host} -a ${host} = ${target} ; then
CONFIG_OBS="${CONFIG_OBS} sol-thread.o"
CONFIG_SRCS="${CONFIG_SRCS} sol-thread.c"
AC_CHECK_LIB(dl, dlopen)
- if test "$GCC" = "yes" ; then
- # The GNU linker requires the -export-dynamic option to make
- # all symbols visible in the dynamic symbol table.
- hold_ldflags=$LDFLAGS
- AC_MSG_CHECKING(for the ld -export-dynamic flag)
- LDFLAGS="${LDFLAGS} -Wl,-export-dynamic"
- AC_TRY_LINK(, [int i;], found=yes, found=no)
- LDFLAGS=$hold_ldflags
- AC_MSG_RESULT($found)
- if test $found = yes; then
- CONFIG_LDFLAGS="${CONFIG_LDFLAGS} -Wl,-export-dynamic"
- fi
- fi
+ CONFIG_LDFLAGS="${CONFIG_LDFLAGS} $RDYNAMIC"
# Sun randomly tweaked the prototypes in <proc_service.h>
# at one point.
AC_MSG_CHECKING(if <proc_service.h> is old)
--- a/gdb/gdbserver/Makefile.in
+++ b/gdb/gdbserver/Makefile.in
@@ -100,7 +100,8 @@ INTERNAL_LDFLAGS = $(LDFLAGS) @RDYNAMIC@
# All source files that go into linking GDB remote server.
SFILES= $(srcdir)/gdbreplay.c $(srcdir)/inferiors.c \
- $(srcdir)/mem-break.c $(srcdir)/proc-service.c $(srcdir)/regcache.c \
+ $(srcdir)/mem-break.c $(srcdir)/proc-service.c \
+ $(srcdir)/proc-service.list $(srcdir)/regcache.c \
$(srcdir)/remote-utils.c $(srcdir)/server.c $(srcdir)/target.c \
$(srcdir)/thread-db.c $(srcdir)/utils.c \
$(srcdir)/linux-arm-low.c $(srcdir)/linux-cris-low.c \
@@ -133,6 +134,7 @@ OBS = inferiors.o regcache.o remote-utils.o server.o signals.o target.o \
GDBREPLAY_OBS = gdbreplay.o version.o
GDBSERVER_LIBS = @GDBSERVER_LIBS@
XM_CLIBS = @LIBS@
+CDEPS = $(srcdir)/proc-service.list
# XML files to compile in to gdbserver, if any.
XML_DIR = $(srcdir)/../features
--- a/gdb/gdbserver/configure.ac
+++ b/gdb/gdbserver/configure.ac
@@ -188,11 +188,24 @@ USE_THREAD_DB=
if test "$srv_linux_thread_db" = "yes"; then
if test "$ac_cv_lib_dl_dlopen" = "yes"; then
srv_libs="-ldl"
+ AC_MSG_CHECKING(for the dynamic export flag)
old_LDFLAGS="$LDFLAGS"
- LDFLAGS="$LDFLAGS -rdynamic"
- AC_TRY_LINK([], [], [RDYNAMIC=-rdynamic], [RDYNAMIC=])
+ # Older GNU ld supports --export-dynamic but --dynamic-list may not be
+ # supported there.
+ RDYNAMIC="-Wl,--dynamic-list=${srcdir}/proc-service.list"
+ LDFLAGS="$LDFLAGS $RDYNAMIC"
+ AC_TRY_LINK([], [],
+ [found="-Wl,--dynamic-list"
+ RDYNAMIC='-Wl,--dynamic-list=$(srcdir)/proc-service.list'],
+ [RDYNAMIC="-rdynamic"
+ LDFLAGS="$old_LDFLAGS $RDYNAMIC"
+ AC_TRY_LINK([], [],
+ [found="-rdynamic"],
+ [found="no"
+ RDYNAMIC=""])])
AC_SUBST(RDYNAMIC)
LDFLAGS="$old_LDFLAGS"
+ AC_MSG_RESULT($found)
else
srv_libs="-lthread_db"
fi
--- /dev/null
+++ b/gdb/gdbserver/proc-service.list
@@ -0,0 +1,30 @@
+/* -Wl,--dynamic-list symbols exported for libthread_db.
+
+ Copyright (C) 2010 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/>. */
+
+{
+ ps_get_thread_area;
+ ps_getpid;
+ ps_lgetfpregs;
+ ps_lgetregs;
+ ps_lsetfpregs;
+ ps_lsetregs;
+ ps_pdread;
+ ps_pdwrite;
+ ps_pglobal_lookup;
+};
--- /dev/null
+++ b/gdb/proc-service.list
@@ -0,0 +1,40 @@
+/* -Wl,--dynamic-list symbols exported for libthread_db.
+
+ Copyright (C) 2010 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/>. */
+
+{
+ ps_get_thread_area;
+ ps_getpid;
+ ps_lcontinue;
+ ps_lgetfpregs;
+ ps_lgetregs;
+ ps_lgetxregs;
+ ps_lgetxregsize;
+ ps_lsetfpregs;
+ ps_lsetregs;
+ ps_lsetxregs;
+ ps_lstop;
+ ps_pcontinue;
+ ps_pdread;
+ ps_pdwrite;
+ ps_pglobal_lookup;
+ ps_plog;
+ ps_pstop;
+ ps_ptread;
+ ps_ptwrite;
+};
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-05-16 21:08 ` Jan Kratochvil
@ 2010-05-27 19:58 ` Tom Tromey
2010-05-28 23:11 ` Jan Kratochvil
2010-06-13 19:58 ` [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list Andreas Schwab
1 sibling, 1 reply; 19+ messages in thread
From: Tom Tromey @ 2010-05-27 19:58 UTC (permalink / raw)
To: Jan Kratochvil; +Cc: H.J. Lu, gdb-patches
>>>>> "Jan" == Jan Kratochvil <jan.kratochvil@redhat.com> writes:
Jan> gdb/
Jan> 2010-05-16 Jan Kratochvil <jan.kratochvil@redhat.com>
Jan> * Makefile.in (RDYNAMIC): New.
Jan> (SFILES): Add proc-service.list.
Jan> * config/alpha/alpha-linux.mh (NAT_CDEPS): New.
Jan> (LOADLIBES): Replace -rdynamic by $(RDYNAMIC).
Jan> * config/arm/linux.mh: Likewise.
Jan> * config/i386/linux.mh: Likewise.
Jan> * config/i386/linux64.mh: Likewise.
Jan> * config/ia64/linux.mh: Likewise.
Jan> * config/m32r/linux.mh: Likewise.
Jan> * config/m68k/linux.mh: Likewise.
Jan> * config/mips/linux.mh: Likewise.
Jan> * config/pa/linux.mh: Likewise.
Jan> * config/powerpc/linux.mh: Likewise.
Jan> * config/powerpc/ppc64-linux.mh: Likewise.
Jan> * config/s390/s390.mh: Likewise.
Jan> * config/sparc/linux.mh: Likewise.
Jan> * config/sparc/linux64.mh: Likewise.
Jan> * config/xtensa/linux.mh: Likewise.
Jan> * configure.ac: New RDYNAMIC on native host and GCC.
Jan> (solaris*): Replace -Wl,-export-dynamic by $RDYNAMIC.
Jan> * configure: Regenerate.
Jan> * proc-service.list: New.
Jan> gdb/gdbserver/
Jan> 2010-05-16 Jan Kratochvil <jan.kratochvil@redhat.com>
Jan> * Makefile.in (SFILES): Add $(srcdir)/proc-service.list.
Jan> (CDEPS): New.
Jan> * configure.ac (RDYNAMIC): New AC_MSG_CHECKING wrapping. Test also
Jan> -Wl,--dynamic-list.
Jan> * configure: Regenerate.
Jan> * proc-service.list: New.
This is ok. Thanks.
Tom
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-05-27 19:58 ` Tom Tromey
@ 2010-05-28 23:11 ` Jan Kratochvil
2010-06-05 2:48 ` Doug Evans
0 siblings, 1 reply; 19+ messages in thread
From: Jan Kratochvil @ 2010-05-28 23:11 UTC (permalink / raw)
To: Tom Tromey; +Cc: H.J. Lu, gdb-patches
On Thu, 27 May 2010 21:54:34 +0200, Tom Tromey wrote:
> >>>>> "Jan" == Jan Kratochvil <jan.kratochvil@redhat.com> writes:
>
> Jan> gdb/
> Jan> 2010-05-16 Jan Kratochvil <jan.kratochvil@redhat.com>
> Jan> gdb/gdbserver/
> Jan> 2010-05-16 Jan Kratochvil <jan.kratochvil@redhat.com>
> This is ok. Thanks.
Checked-in:
http://sourceware.org/ml/gdb-cvs/2010-05/msg00247.html
Thanks,
Jan
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-05-28 23:11 ` Jan Kratochvil
@ 2010-06-05 2:48 ` Doug Evans
2010-06-05 12:45 ` Jan Kratochvil
2012-01-12 19:28 ` [patch] -Wl,--dynamic-list: Test also PYTHON_LIBS, not just PYTHON_CFLAGS [Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list] Jan Kratochvil
0 siblings, 2 replies; 19+ messages in thread
From: Doug Evans @ 2010-06-05 2:48 UTC (permalink / raw)
To: Jan Kratochvil; +Cc: Tom Tromey, gdb-patches
On Fri, May 28, 2010 at 11:52 AM, Jan Kratochvil
<jan.kratochvil@redhat.com> wrote:
> On Thu, 27 May 2010 21:54:34 +0200, Tom Tromey wrote:
>> >>>>> "Jan" == Jan Kratochvil <jan.kratochvil@redhat.com> writes:
>>
>> Jan> gdb/
>> Jan> 2010-05-16 Jan Kratochvil <jan.kratochvil@redhat.com>
>> Jan> gdb/gdbserver/
>> Jan> 2010-05-16 Jan Kratochvil <jan.kratochvil@redhat.com>
>> This is ok. Thanks.
>
> Checked-in:
> http://sourceware.org/ml/gdb-cvs/2010-05/msg00247.html
I've been playing with this a bit and got the following:
"ImportError: /usr/grte/v1/k8-linux/lib/python2.4/lib-dynload/itertools.so:
undefined symbol: PyExc_StopIteration"
when I did "import itertools".
I was statically linking libpythonX.Y.a.
I'm not sure what the right fix is. IWBN if the python folks have
anticipated this and provide something like proc-service.list for
python.
I don't know all the python symbols that need to be exported ... maybe
there's a simple regexp (or a small number of them) that could find
them all.
[OTOH it might be a bit too fragile.]
Alternatively, one could disable this feature when python is being
linked statically.
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-06-05 2:48 ` Doug Evans
@ 2010-06-05 12:45 ` Jan Kratochvil
2010-06-08 15:23 ` Doug Evans
2012-01-12 19:28 ` [patch] -Wl,--dynamic-list: Test also PYTHON_LIBS, not just PYTHON_CFLAGS [Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list] Jan Kratochvil
1 sibling, 1 reply; 19+ messages in thread
From: Jan Kratochvil @ 2010-06-05 12:45 UTC (permalink / raw)
To: Doug Evans; +Cc: Tom Tromey, gdb-patches
On Sat, 05 Jun 2010 04:48:33 +0200, Doug Evans wrote:
> "ImportError: /usr/grte/v1/k8-linux/lib/python2.4/lib-dynload/itertools.so:
> undefined symbol: PyExc_StopIteration"
>
> when I did "import itertools".
>
> I was statically linking libpythonX.Y.a.
>
> I'm not sure what the right fix is.
This is an open Python bug: http://bugs.python.org/issue4434
(I was not aware of it.)
I doubt how useful is to depend on .so modules while having .a libpython
linked in but I understand my patch broke compatibility with such case.
With libpythonX.Y.so.1.0 the problem does not happen as all the symbols are
present in its .dynsym. There is an inefficiency Python-2.6.4/Include/pyport.h
uses dllimport and dllimport declarations for the API but it has no support
for gcc __attribute__((visibility)) there and thus libpythonX.Y.so.1.0's
.dynsym is excessively large there.
I was unable to test it as Fedora (13) does not ship libpythonX.Y.a (and it
failed to build one when I tried to). Could you test it, please?
If some existing python tests were already broken I can drop this new one.
No regressions on {x86_64,x86_64-m32,i686}-fedora13-linux-gnu.
Thanks,
Jan
gdb/
2010-06-05 Jan Kratochvil <jan.kratochvil@redhat.com>
* configure.ac <"${have_libpython}" != no>: New workaround of
python#4434.
* configure: Regenerate.
gdb/testsuite/
2010-06-05 Jan Kratochvil <jan.kratochvil@redhat.com>
* gdb.python/python.exp (pythonX.Y/lib-dynload/*.so): New.
--- a/gdb/configure.ac
+++ b/gdb/configure.ac
@@ -1470,16 +1470,39 @@ if test ${build} = ${host} -a ${host} = ${target} ; then
# Older GNU ld supports --export-dynamic but --dynamic-list it does not.
RDYNAMIC="-Wl,--dynamic-list=${srcdir}/proc-service.list"
LDFLAGS="$LDFLAGS $RDYNAMIC"
- AC_TRY_LINK([], [],
- [found="-Wl,--dynamic-list"
- RDYNAMIC='-Wl,--dynamic-list=$(srcdir)/proc-service.list'],
- [RDYNAMIC="-rdynamic"
- LDFLAGS="$old_LDFLAGS $RDYNAMIC"
- AC_TRY_LINK([], [],
- [found="-rdynamic"],
- [found="no"
- # Error on its usage by .mh file.
- RDYNAMIC="no-rdynamic-available"])])
+ dynamic_list=false
+ if test "${have_libpython}" = no; then
+ AC_TRY_LINK([], [], [dynamic_list=true])
+ else
+ # Workaround http://bugs.python.org/issue4434 where static
+ # libpythonX.Y.a would get its symbols required for
+ # pythonX.Y/lib-dynload/*.so modules hidden by -Wl,--dynamic-list.
+ # Problem does not happen for the recommended libpythonX.Y.so linkage.
+ old_CFLAGS="$CFLAGS"
+ CFLAGS="$CFLAGS $PYTHON_CFLAGS"
+ AC_RUN_IFELSE(
+ AC_LANG_PROGRAM(
+ [#include "]${have_libpython}[/Python.h"],
+ [int err;
+ Py_Initialize ();
+ err = PyRun_SimpleString ("import itertools\n");
+ Py_Finalize ();
+ return err == 0 ? 0 : 1;]),
+ [dynamic_list=true], [], [true])
+ CFLAGS="$old_CFLAGS"
+ fi
+ if $dynamic_list;then
+ found="-Wl,--dynamic-list"
+ RDYNAMIC='-Wl,--dynamic-list=$(srcdir)/proc-service.list'
+ else
+ RDYNAMIC="-rdynamic"
+ LDFLAGS="$old_LDFLAGS $RDYNAMIC"
+ AC_TRY_LINK([], [],
+ [found="-rdynamic"],
+ [found="no"
+ # Error on its usage by .mh file.
+ RDYNAMIC="no-rdynamic-available"])
+ fi
AC_SUBST(RDYNAMIC)
LDFLAGS="$old_LDFLAGS"
AC_MSG_RESULT($found)
--- a/gdb/configure
+++ b/gdb/configure
@@ -14199,7 +14199,9 @@ $as_echo_n "checking for the dynamic export flag... " >&6; }
# Older GNU ld supports --export-dynamic but --dynamic-list it does not.
RDYNAMIC="-Wl,--dynamic-list=${srcdir}/proc-service.list"
LDFLAGS="$LDFLAGS $RDYNAMIC"
- cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+ dynamic_list=false
+ if test "${have_libpython}" = no; then
+ cat confdefs.h - <<_ACEOF >conftest.$ac_ext
/* end confdefs.h. */
int
@@ -14211,12 +14213,51 @@ main ()
}
_ACEOF
if ac_fn_c_try_link "$LINENO"; then :
- found="-Wl,--dynamic-list"
- RDYNAMIC='-Wl,--dynamic-list=$(srcdir)/proc-service.list'
+ dynamic_list=true
+fi
+rm -f core conftest.err conftest.$ac_objext \
+ conftest$ac_exeext conftest.$ac_ext
+ else
+ # Workaround http://bugs.python.org/issue4434 where static
+ # libpythonX.Y.a would get its symbols required for
+ # pythonX.Y/lib-dynload/*.so modules hidden by -Wl,--dynamic-list.
+ # Problem does not happen for the recommended libpythonX.Y.so linkage.
+ old_CFLAGS="$CFLAGS"
+ CFLAGS="$CFLAGS $PYTHON_CFLAGS"
+ if test "$cross_compiling" = yes; then :
+ true
else
- RDYNAMIC="-rdynamic"
- LDFLAGS="$old_LDFLAGS $RDYNAMIC"
- cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+ cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h. */
+#include "${have_libpython}/Python.h"
+int
+main ()
+{
+int err;
+ Py_Initialize ();
+ err = PyRun_SimpleString ("import itertools\n");
+ Py_Finalize ();
+ return err == 0 ? 0 : 1;
+ ;
+ return 0;
+}
+_ACEOF
+if ac_fn_c_try_run "$LINENO"; then :
+ dynamic_list=true
+fi
+rm -f core *.core core.conftest.* gmon.out bb.out conftest$ac_exeext \
+ conftest.$ac_objext conftest.beam conftest.$ac_ext
+fi
+
+ CFLAGS="$old_CFLAGS"
+ fi
+ if $dynamic_list;then
+ found="-Wl,--dynamic-list"
+ RDYNAMIC='-Wl,--dynamic-list=$(srcdir)/proc-service.list'
+ else
+ RDYNAMIC="-rdynamic"
+ LDFLAGS="$old_LDFLAGS $RDYNAMIC"
+ cat confdefs.h - <<_ACEOF >conftest.$ac_ext
/* end confdefs.h. */
int
@@ -14231,14 +14272,12 @@ if ac_fn_c_try_link "$LINENO"; then :
found="-rdynamic"
else
found="no"
- # Error on its usage by .mh file.
- RDYNAMIC="no-rdynamic-available"
-fi
-rm -f core conftest.err conftest.$ac_objext \
- conftest$ac_exeext conftest.$ac_ext
+ # Error on its usage by .mh file.
+ RDYNAMIC="no-rdynamic-available"
fi
rm -f core conftest.err conftest.$ac_objext \
conftest$ac_exeext conftest.$ac_ext
+ fi
LDFLAGS="$old_LDFLAGS"
{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $found" >&5
--- a/gdb/testsuite/gdb.python/python.exp
+++ b/gdb/testsuite/gdb.python/python.exp
@@ -80,3 +80,6 @@ gdb_test "source $srcdir/$subdir/source2.py" "yes"
gdb_test "python print gdb.current_objfile()" "None"
gdb_test "python print gdb.objfiles()" "\\\[\\\]"
+
+# Test http://bugs.python.org/issue4434 workaround in configure.ac
+gdb_test "python import itertools; print 'IMPOR'+'TED'" "IMPORTED" "pythonX.Y/lib-dynload/*.so"
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-06-05 12:45 ` Jan Kratochvil
@ 2010-06-08 15:23 ` Doug Evans
2010-06-08 16:08 ` Jan Kratochvil
0 siblings, 1 reply; 19+ messages in thread
From: Doug Evans @ 2010-06-08 15:23 UTC (permalink / raw)
To: Jan Kratochvil; +Cc: Tom Tromey, gdb-patches
On Sat, Jun 5, 2010 at 5:45 AM, Jan Kratochvil
<jan.kratochvil@redhat.com> wrote:
> I was unable to test it as Fedora (13) does not ship libpythonX.Y.a (and it
> failed to build one when I tried to). Could you test it, please?
>
> If some existing python tests were already broken I can drop this new one.
>
> No regressions on {x86_64,x86_64-m32,i686}-fedora13-linux-gnu.
>
>
> Thanks,
> Jan
>
>
> gdb/
> 2010-06-05 Jan Kratochvil <jan.kratochvil@redhat.com>
>
> * configure.ac <"${have_libpython}" != no>: New workaround of
> python#4434.
> * configure: Regenerate.
>
> gdb/testsuite/
> 2010-06-05 Jan Kratochvil <jan.kratochvil@redhat.com>
>
> * gdb.python/python.exp (pythonX.Y/lib-dynload/*.so): New.
I tested this and it fixes things. Thanks.
There is the issue of using AC_RUN_IFELSE and cross-compilation.
Python doesn't have good support for cross-compilation anyways, so
it's fine with me to punt on this for now.
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-06-08 15:23 ` Doug Evans
@ 2010-06-08 16:08 ` Jan Kratochvil
2010-06-08 16:12 ` Doug Evans
0 siblings, 1 reply; 19+ messages in thread
From: Jan Kratochvil @ 2010-06-08 16:08 UTC (permalink / raw)
To: Doug Evans; +Cc: Tom Tromey, gdb-patches
On Tue, 08 Jun 2010 17:22:54 +0200, Doug Evans wrote:
> On Sat, Jun 5, 2010 at 5:45 AM, Jan Kratochvil <jan.kratochvil@redhat.com> wrote:
> > gdb/
> > Â Â Â Â * configure.ac <"${have_libpython}" != no>: New workaround of
> > Â Â Â Â python#4434.
> > Â Â Â Â * configure: Regenerate.
> > gdb/testsuite/
> > Â Â Â Â * gdb.python/python.exp (pythonX.Y/lib-dynload/*.so): New.
>
> I tested this and it fixes things. Thanks.
Considering it as your check-in approval, going to check it in in a moment.
> There is the issue of using AC_RUN_IFELSE and cross-compilation.
> Python doesn't have good support for cross-compilation anyways, so
> it's fine with me to punt on this for now.
In this case the current code fallbacks to the larger but more compatible
-rdynamic (and thus this my recent optimization does not take place).
+ [dynamic_list=true], [], [true])
-> ^^^^^^
+ if test "$cross_compiling" = yes; then :
+ true
->
kept there the pre-set value:
dynamic_list=false
Thanks,
Jan
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-06-08 16:08 ` Jan Kratochvil
@ 2010-06-08 16:12 ` Doug Evans
2010-06-08 17:37 ` Jan Kratochvil
0 siblings, 1 reply; 19+ messages in thread
From: Doug Evans @ 2010-06-08 16:12 UTC (permalink / raw)
To: Jan Kratochvil; +Cc: Tom Tromey, gdb-patches
On Tue, Jun 8, 2010 at 9:07 AM, Jan Kratochvil
<jan.kratochvil@redhat.com> wrote:
> On Tue, 08 Jun 2010 17:22:54 +0200, Doug Evans wrote:
>> On Sat, Jun 5, 2010 at 5:45 AM, Jan Kratochvil <jan.kratochvil@redhat.com> wrote:
>> > gdb/
>> > * configure.ac <"${have_libpython}" != no>: New workaround of
>> > python#4434.
>> > * configure: Regenerate.
>> > gdb/testsuite/
>> > * gdb.python/python.exp (pythonX.Y/lib-dynload/*.so): New.
>>
>> I tested this and it fixes things. Thanks.
>
> Considering it as your check-in approval, going to check it in in a moment.
Yep.
>> There is the issue of using AC_RUN_IFELSE and cross-compilation.
>> Python doesn't have good support for cross-compilation anyways, so
>> it's fine with me to punt on this for now.
>
> In this case the current code fallbacks to the larger but more compatible
> -rdynamic (and thus this my recent optimization does not take place).
>
> + [dynamic_list=true], [], [true])
> -> ^^^^^^
> + if test "$cross_compiling" = yes; then :
> + true
> ->
> kept there the pre-set value:
> dynamic_list=false
Yeah, saw that. "works for me".
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-06-08 16:12 ` Doug Evans
@ 2010-06-08 17:37 ` Jan Kratochvil
0 siblings, 0 replies; 19+ messages in thread
From: Jan Kratochvil @ 2010-06-08 17:37 UTC (permalink / raw)
To: Doug Evans; +Cc: Tom Tromey, gdb-patches
On Tue, 08 Jun 2010 18:12:39 +0200, Doug Evans wrote:
> On Tue, Jun 8, 2010 at 9:07 AM, Jan Kratochvil
> > Considering it as your check-in approval, going to check it in in a moment.
>
> Yep.
Checked-in:
http://sourceware.org/ml/gdb-cvs/2010-06/msg00069.html
Thanks,
Jan
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-05-16 21:08 ` Jan Kratochvil
2010-05-27 19:58 ` Tom Tromey
@ 2010-06-13 19:58 ` Andreas Schwab
2010-06-13 20:17 ` Andreas Schwab
1 sibling, 1 reply; 19+ messages in thread
From: Andreas Schwab @ 2010-06-13 19:58 UTC (permalink / raw)
To: Jan Kratochvil; +Cc: H.J. Lu, gdb-patches
Jan Kratochvil <jan.kratochvil@redhat.com> writes:
> --- a/gdb/configure.ac
> +++ b/gdb/configure.ac
> @@ -1312,6 +1312,30 @@ dnl Note that we only want this if we are both native (host == target),
> dnl and not doing a canadian cross build (build == host).
>
> if test ${build} = ${host} -a ${host} = ${target} ; then
> + if test "$GCC" = "yes" ; then
> + # The dynamically loaded libthread_db needs access to symbols in the gdb
> + # executable. Older GNU ld supports --export-dynamic but --dynamic-list
> + # may not be supported there.
> + AC_MSG_CHECKING(for the dynamic export flag)
Why only for non-canadian-cross builds? Previously -rdynamic was used
for any native configuration, now you can no longer cross-build a
working gdb.
Andreas.
--
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756 01D3 44D5 214B 8276 4ED5
"And now for something completely different."
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-06-13 19:58 ` [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list Andreas Schwab
@ 2010-06-13 20:17 ` Andreas Schwab
2010-06-15 17:31 ` Tom Tromey
0 siblings, 1 reply; 19+ messages in thread
From: Andreas Schwab @ 2010-06-13 20:17 UTC (permalink / raw)
To: Jan Kratochvil; +Cc: H.J. Lu, gdb-patches
Tested on ppc-linux.
Andreas.
2010-06-13 Andreas Schwab <schwab@linux-m68k.org>
* configure.ac: Check for RDYNAMIC also for cross builds.
* configure: Regenerate.
--- configure.ac.~1.122.~ 2010-06-12 23:13:19.000000000 +0200
+++ configure.ac 2010-06-13 22:09:54.000000000 +0200
@@ -1453,14 +1453,7 @@ aix*)
;;
esac
-
-dnl For certain native configurations, we need to check whether thread
-dnl support can be built in or not.
-dnl
-dnl Note that we only want this if we are both native (host == target),
-dnl and not doing a canadian cross build (build == host).
-
-if test ${build} = ${host} -a ${host} = ${target} ; then
+if test "${gdb_native}" = yes; then
if test "$GCC" = "yes" ; then
# The dynamically loaded libthread_db needs access to symbols in the gdb
# executable. Older GNU ld supports --export-dynamic but --dynamic-list
@@ -1507,7 +1500,15 @@ if test ${build} = ${host} -a ${host} =
LDFLAGS="$old_LDFLAGS"
AC_MSG_RESULT($found)
fi
+fi
+dnl For certain native configurations, we need to check whether thread
+dnl support can be built in or not.
+dnl
+dnl Note that we only want this if we are both native (host == target),
+dnl and not doing a canadian cross build (build == host).
+
+if test ${build} = ${host} -a ${host} = ${target} ; then
case ${host_os} in
hpux*)
AC_MSG_CHECKING(for HPUX/OSF thread support)
--
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756 01D3 44D5 214B 8276 4ED5
"And now for something completely different."
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-06-13 20:17 ` Andreas Schwab
@ 2010-06-15 17:31 ` Tom Tromey
2010-06-15 19:09 ` Jan Kratochvil
2010-06-15 20:36 ` Jan Kratochvil
0 siblings, 2 replies; 19+ messages in thread
From: Tom Tromey @ 2010-06-15 17:31 UTC (permalink / raw)
To: Andreas Schwab; +Cc: Jan Kratochvil, H.J. Lu, gdb-patches
>>>>> "Andreas" == Andreas Schwab <schwab@linux-m68k.org> writes:
Andreas> 2010-06-13 Andreas Schwab <schwab@linux-m68k.org>
Andreas> * configure.ac: Check for RDYNAMIC also for cross builds.
Andreas> * configure: Regenerate.
This is ok. Thanks.
Tom
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-06-15 17:31 ` Tom Tromey
@ 2010-06-15 19:09 ` Jan Kratochvil
2010-06-15 20:36 ` Jan Kratochvil
1 sibling, 0 replies; 19+ messages in thread
From: Jan Kratochvil @ 2010-06-15 19:09 UTC (permalink / raw)
To: Tom Tromey; +Cc: Andreas Schwab, H.J. Lu, gdb-patches
On Tue, 15 Jun 2010 19:31:01 +0200, Tom Tromey wrote:
> >>>>> "Andreas" == Andreas Schwab <schwab@linux-m68k.org> writes:
>
> Andreas> 2010-06-13 Andreas Schwab <schwab@linux-m68k.org>
> Andreas> * configure.ac: Check for RDYNAMIC also for cross builds.
> Andreas> * configure: Regenerate.
>
> This is ok. Thanks.
I am preparing a modification of this patch but this patch is already
a regression fix, just incomplete. I will post the second part as a second
patch on top of it.
Thanks,
Jan
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-06-15 17:31 ` Tom Tromey
2010-06-15 19:09 ` Jan Kratochvil
@ 2010-06-15 20:36 ` Jan Kratochvil
2010-06-21 18:26 ` Tom Tromey
1 sibling, 1 reply; 19+ messages in thread
From: Jan Kratochvil @ 2010-06-15 20:36 UTC (permalink / raw)
To: Tom Tromey; +Cc: Andreas Schwab, H.J. Lu, gdb-patches
On Tue, 15 Jun 2010 19:31:01 +0200, Tom Tromey wrote:
> >>>>> "Andreas" == Andreas Schwab <schwab@linux-m68k.org> writes:
>
> Andreas> 2010-06-13 Andreas Schwab <schwab@linux-m68k.org>
> Andreas> * configure.ac: Check for RDYNAMIC also for cross builds.
> Andreas> * configure: Regenerate.
>
> This is ok. Thanks.
A patch on top of this one is attached.
Original code had `-rdynamic' hardcoded in the .mh files. .mh files get
included iff we have ${gdb_native}. It means two things:
(1) The `-rdynamic' upgrade to `-Wl,--dynamic-list' makes sense only iff
we have ${gdb_native}.
(2) Iff we have ${gdb_native} we should always provide at least `-rdynamic'.
Current code with my change for non-$GCC configurations provided empty
RDYNAMIC and thus breaking non-$GCC threading support.
(OTOH there is a notice -rdynamic' is required only for $GCC thus probably
nothing got broken.)
Moreover as the functionality is really tested (by either AC_RUN_IFELSE or
AC_TRY_LINK) it should not be needed to check $GCC. On non-$GCC the
functionality test should always fail anyway. Removed the $GCC check.
This patch is more a cleanup only / safer change from the original code before
`-Wl,--dynamic-list' introduction. I am not aware on which configuration this
patch below can change anything. No problem dropping it.
Tested build on x86_64-fedora13-linux-gnu and x86_64-rhel4-linux-gnu.
Thanks,
Jan
2010-06-15 Jan Kratochvil <jan.kratochvil@redhat.com>
* configure.ac: Always set RDYNAMIC at least as `-rdynamic'.
* configure: Regenerate.
--- a/gdb/configure.ac
+++ b/gdb/configure.ac
@@ -1453,54 +1453,47 @@ aix*)
;;
esac
+AC_MSG_CHECKING(for the dynamic export flag)
+dynamic_list=false
if test "${gdb_native}" = yes; then
- if test "$GCC" = "yes" ; then
- # The dynamically loaded libthread_db needs access to symbols in the gdb
- # executable. Older GNU ld supports --export-dynamic but --dynamic-list
- # may not be supported there.
- AC_MSG_CHECKING(for the dynamic export flag)
- old_LDFLAGS="$LDFLAGS"
- # Older GNU ld supports --export-dynamic but --dynamic-list it does not.
- RDYNAMIC="-Wl,--dynamic-list=${srcdir}/proc-service.list"
- LDFLAGS="$LDFLAGS $RDYNAMIC"
- dynamic_list=false
- if test "${have_libpython}" = no; then
- AC_TRY_LINK([], [], [dynamic_list=true])
- else
- # Workaround http://bugs.python.org/issue4434 where static
- # libpythonX.Y.a would get its symbols required for
- # pythonX.Y/lib-dynload/*.so modules hidden by -Wl,--dynamic-list.
- # Problem does not happen for the recommended libpythonX.Y.so linkage.
- old_CFLAGS="$CFLAGS"
- CFLAGS="$CFLAGS $PYTHON_CFLAGS"
- AC_RUN_IFELSE(
- AC_LANG_PROGRAM(
- [#include "]${have_libpython}[/Python.h"],
- [int err;
- Py_Initialize ();
- err = PyRun_SimpleString ("import itertools\n");
- Py_Finalize ();
- return err == 0 ? 0 : 1;]),
- [dynamic_list=true], [], [true])
- CFLAGS="$old_CFLAGS"
- fi
- if $dynamic_list;then
- found="-Wl,--dynamic-list"
- RDYNAMIC='-Wl,--dynamic-list=$(srcdir)/proc-service.list'
- else
- RDYNAMIC="-rdynamic"
- LDFLAGS="$old_LDFLAGS $RDYNAMIC"
- AC_TRY_LINK([], [],
- [found="-rdynamic"],
- [found="no"
- # Error on its usage by .mh file.
- RDYNAMIC="no-rdynamic-available"])
- fi
- AC_SUBST(RDYNAMIC)
- LDFLAGS="$old_LDFLAGS"
- AC_MSG_RESULT($found)
+ # The dynamically loaded libthread_db needs access to symbols in the gdb
+ # executable. Older GNU ld supports --export-dynamic but --dynamic-list
+ # may not be supported there.
+ old_LDFLAGS="$LDFLAGS"
+ # Older GNU ld supports --export-dynamic but --dynamic-list it does not.
+ RDYNAMIC="-Wl,--dynamic-list=${srcdir}/proc-service.list"
+ LDFLAGS="$LDFLAGS $RDYNAMIC"
+ if test "${have_libpython}" = no; then
+ AC_TRY_LINK([], [], [dynamic_list=true])
+ else
+ # Workaround http://bugs.python.org/issue4434 where static
+ # libpythonX.Y.a would get its symbols required for
+ # pythonX.Y/lib-dynload/*.so modules hidden by -Wl,--dynamic-list.
+ # Problem does not happen for the recommended libpythonX.Y.so linkage.
+ old_CFLAGS="$CFLAGS"
+ CFLAGS="$CFLAGS $PYTHON_CFLAGS"
+ AC_RUN_IFELSE(
+ AC_LANG_PROGRAM(
+ [#include "]${have_libpython}[/Python.h"],
+ [int err;
+ Py_Initialize ();
+ err = PyRun_SimpleString ("import itertools\n");
+ Py_Finalize ();
+ return err == 0 ? 0 : 1;]),
+ [dynamic_list=true], [], [true])
+ CFLAGS="$old_CFLAGS"
fi
+ LDFLAGS="$old_LDFLAGS"
+fi
+if $dynamic_list; then
+ found="-Wl,--dynamic-list"
+ RDYNAMIC='-Wl,--dynamic-list=$(srcdir)/proc-service.list'
+else
+ found="-rdynamic"
+ RDYNAMIC="-rdynamic"
fi
+AC_SUBST(RDYNAMIC)
+AC_MSG_RESULT($found)
dnl For certain native configurations, we need to check whether thread
dnl support can be built in or not.
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-06-15 20:36 ` Jan Kratochvil
@ 2010-06-21 18:26 ` Tom Tromey
2010-06-23 10:13 ` Jan Kratochvil
0 siblings, 1 reply; 19+ messages in thread
From: Tom Tromey @ 2010-06-21 18:26 UTC (permalink / raw)
To: Jan Kratochvil; +Cc: Andreas Schwab, H.J. Lu, gdb-patches
>>>>> "Jan" == Jan Kratochvil <jan.kratochvil@redhat.com> writes:
Jan> 2010-06-15 Jan Kratochvil <jan.kratochvil@redhat.com>
Jan> * configure.ac: Always set RDYNAMIC at least as `-rdynamic'.
Jan> * configure: Regenerate.
This is ok. Thanks.
Tom
^ permalink raw reply [flat|nested] 19+ messages in thread
* Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list
2010-06-21 18:26 ` Tom Tromey
@ 2010-06-23 10:13 ` Jan Kratochvil
0 siblings, 0 replies; 19+ messages in thread
From: Jan Kratochvil @ 2010-06-23 10:13 UTC (permalink / raw)
To: Tom Tromey; +Cc: Andreas Schwab, H.J. Lu, gdb-patches
On Mon, 21 Jun 2010 20:26:30 +0200, Tom Tromey wrote:
> >>>>> "Jan" == Jan Kratochvil <jan.kratochvil@redhat.com> writes:
>
> Jan> 2010-06-15 Jan Kratochvil <jan.kratochvil@redhat.com>
> Jan> * configure.ac: Always set RDYNAMIC at least as `-rdynamic'.
> Jan> * configure: Regenerate.
>
> This is ok. Thanks.
Checked-in:
http://sourceware.org/ml/gdb-cvs/2010-06/msg00155.html
Thanks,
Jan
^ permalink raw reply [flat|nested] 19+ messages in thread
* [patch] -Wl,--dynamic-list: Test also PYTHON_LIBS, not just PYTHON_CFLAGS [Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list]
2010-06-05 2:48 ` Doug Evans
2010-06-05 12:45 ` Jan Kratochvil
@ 2012-01-12 19:28 ` Jan Kratochvil
1 sibling, 0 replies; 19+ messages in thread
From: Jan Kratochvil @ 2012-01-12 19:28 UTC (permalink / raw)
To: Doug Evans; +Cc: Tom Tromey, gdb-patches
Hello Doug,
On Sat, 05 Jun 2010 04:48:33 +0200, Doug Evans wrote:
> I've been playing with this a bit and got the following:
>
> "ImportError: /usr/grte/v1/k8-linux/lib/python2.4/lib-dynload/itertools.so:
> undefined symbol: PyExc_StopIteration"
>
> when I did "import itertools".
>
> I was statically linking libpythonX.Y.a.
I found current FSF GDB HEAD always uses -rdynamic with libpython as it
falsely thinks -Wl,--dynamic-list is incompatible:
configure:14652: gcc -o conftest -m64 -ggdb2 -pipe -Wall -Wp,-D_FORTIFY_SOURCE=2 -fexceptions -fstack-protector --param=ssp-buffer-size=4 -fno-strict-aliasing -DNDEBUG -fwrapv -lmcheck -Wl,--dynamic-list=./proc-service.list conftest.c -ldl -lncurses -lz -lm >&5
/tmp/ccgMCqKR.o: In function `main':
/home/jkratoch/redhat/gdb-clean/gdb/conftest.c:212: undefined reference to `Py_Initialize'
/home/jkratoch/redhat/gdb-clean/gdb/conftest.c:213: undefined reference to `PyRun_SimpleStringFlags'
/home/jkratoch/redhat/gdb-clean/gdb/conftest.c:214: undefined reference to `Py_Finalize'
It does not make a difference now as with Python it would use -Xlinker
-export-dynamic on top of it anyway due to Python issue10112. But after it
gets fixed in Python GDB could become smaller (+its ld and ld.so faster) with
this patch.
It is mostly obvious it should use all the PYTHON_* compilation flags and not
just PYTHON_CFLAGS. But I would welcome a test on your static libpython.a
config if it does not break anything.
Thanks,
Jan
2012-01-12 Jan Kratochvil <jan.kratochvil@redhat.com>
* configure.ac [${gdb_native} && ${have_libpython}]: Use also
PYTHON_CPPFLAGS and PYTHON_LIBS.
* configure: Regenerate.
--- gdb/configure 2 Jan 2012 02:28:56 -0000 1.336
+++ gdb/configure 12 Jan 2012 18:43:25 -0000
@@ -14631,6 +14631,10 @@ rm -f core conftest.err conftest.$ac_obj
# Problem does not happen for the recommended libpythonX.Y.so linkage.
old_CFLAGS="$CFLAGS"
CFLAGS="$CFLAGS $PYTHON_CFLAGS"
+ old_CPPFLAGS="$CPPFLAGS"
+ CPPFLAGS="$CPPFLAGS $PYTHON_CPPFLAGS"
+ old_LIBS="$LIBS"
+ LIBS="$LIBS $PYTHON_LIBS"
if test "$cross_compiling" = yes; then :
true
else
@@ -14657,6 +14661,8 @@ rm -f core *.core core.conftest.* gmon.o
fi
CFLAGS="$old_CFLAGS"
+ CPPFLAGS="$old_CPPFLAGS"
+ LIBS="$old_LIBS"
fi
LDFLAGS="$old_LDFLAGS"
fi
--- gdb/configure.ac 4 Jan 2012 08:17:00 -0000 1.152
+++ gdb/configure.ac 12 Jan 2012 18:43:27 -0000
@@ -1576,6 +1576,10 @@ if test "${gdb_native}" = yes; then
# Problem does not happen for the recommended libpythonX.Y.so linkage.
old_CFLAGS="$CFLAGS"
CFLAGS="$CFLAGS $PYTHON_CFLAGS"
+ old_CPPFLAGS="$CPPFLAGS"
+ CPPFLAGS="$CPPFLAGS $PYTHON_CPPFLAGS"
+ old_LIBS="$LIBS"
+ LIBS="$LIBS $PYTHON_LIBS"
AC_RUN_IFELSE(
AC_LANG_PROGRAM(
[#include "]${have_libpython}[/Python.h"],
@@ -1586,6 +1590,8 @@ if test "${gdb_native}" = yes; then
return err == 0 ? 0 : 1;]),
[dynamic_list=true], [], [true])
CFLAGS="$old_CFLAGS"
+ CPPFLAGS="$old_CPPFLAGS"
+ LIBS="$old_LIBS"
fi
LDFLAGS="$old_LDFLAGS"
fi
^ permalink raw reply [flat|nested] 19+ messages in thread
end of thread, other threads:[~2012-01-12 18:48 UTC | newest]
Thread overview: 19+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2010-05-13 13:27 [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list Jan Kratochvil
2010-05-13 15:47 ` H.J. Lu
2010-05-16 21:08 ` Jan Kratochvil
2010-05-27 19:58 ` Tom Tromey
2010-05-28 23:11 ` Jan Kratochvil
2010-06-05 2:48 ` Doug Evans
2010-06-05 12:45 ` Jan Kratochvil
2010-06-08 15:23 ` Doug Evans
2010-06-08 16:08 ` Jan Kratochvil
2010-06-08 16:12 ` Doug Evans
2010-06-08 17:37 ` Jan Kratochvil
2012-01-12 19:28 ` [patch] -Wl,--dynamic-list: Test also PYTHON_LIBS, not just PYTHON_CFLAGS [Re: [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list] Jan Kratochvil
2010-06-13 19:58 ` [patch] Smaller gdb: -rdynamic -> -Wl,--dynamic-list Andreas Schwab
2010-06-13 20:17 ` Andreas Schwab
2010-06-15 17:31 ` Tom Tromey
2010-06-15 19:09 ` Jan Kratochvil
2010-06-15 20:36 ` Jan Kratochvil
2010-06-21 18:26 ` Tom Tromey
2010-06-23 10:13 ` Jan Kratochvil
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).