rpms/gdb
3
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

- Fix callback-mode readline-6.0 regression for CTRL-C (for RHEL-6.0).

Browse Source
This commit is contained in:
Jan Kratochvil 2011-01-16 18:39:45 +01:00
parent 18d95bdf87
commit 5bf3960e62
2 changed files with 266 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

253
gdb-readline-6.0-signal.patch Normal file
View File

@ -0,0 +1,253 @@
http://sourceware.org/ml/gdb-patches/2009-11/msg00596.html
Subject: [gdb FYI-patch] callback-mode readline-6.0 regression
Hi Chet,
FSF GDB currently ships bundled with readline-5.2 which works fine.
But using --with-system-readline and readline-6.0-patchlevel4 has
a regression:
readline-5.2: Run `gdb -nx -q' and type CTRL-C:
(gdb) Quit
(gdb) _
readline-6.0: Run `gdb -nx -q' and type CTRL-C:
(gdb) _
= nothing happens (it gets buffered and executed later)
(It does also FAIL on gdb.gdb/selftest.exp.)
It is because GDB waits in its own poll() mainloop and readline uses via
rl_callback_handler_install and rl_callback_handler_remove. This way the
readline internal variable _rl_interrupt_immediately remains 0 and CTRL-C gets
only stored to _rl_caught_signal but not executed.
Seen in rl_signal_handler even if _rl_interrupt_immediately is set and
_rl_handle_signal is called then the signal is still stored to
_rl_caught_signal. In the _rl_interrupt_immediately case it should not be
stored when it was already processed.
rl_signal_handler does `_rl_interrupt_immediately = 0;' - while I am not aware
of its meaning it breaks the nest-counting of other routines which do
`_rl_interrupt_immediately++;' and `_rl_interrupt_immediately--;' possibly
creating problematic `_rl_interrupt_immediately == -1'.
`_rl_interrupt_immediately' is an internal variable, how it could be accessed
by a readline application? (OK, maybe it should not be used.)
Attaching a current GDB-side patch but it must access readline internal
variable _rl_caught_signal and it is generally just a workaround. Could you
please include support for signals in this asynchronous mode in readline-6.1?
I find it would be enough to make RL_CHECK_SIGNALS public?
GDB: No regressions on {x86_64,x86_64-m32,i686}-fedora12-linux-gnu.
But this is not a patch intended to be accepted.
Thanks,
Jan
gdb/
2009-11-29 Jan Kratochvil <jan.kratochvil@redhat.com>
* config.in, configure: Regenerate.
* configure.ac (for readline_echoing_p): Move inside $LIBS change.
(for _rl_caught_signal): New.
* event-loop.c: Include readline/readline.h.
(gdb_do_one_event) [HAVE_READLINE_CAUGHT_SIGNAL]: New.
gdb/testsuite/
2009-11-29 Jan Kratochvil <jan.kratochvil@redhat.com>
* gdb.gdb/selftest.exp (backtrace through signal handler): Move before
SIGINT pass, drop the timeout case.
(send SIGINT signal to child process): Use gdb_test.
(backtrace through readline handler): New.
Index: gdb-7.2/gdb/config.in
===================================================================
--- gdb-7.2.orig/gdb/config.in 2011-01-16 18:26:11.000000000 +0100
+++ gdb-7.2/gdb/config.in 2011-01-16 18:26:50.000000000 +0100
@@ -463,6 +463,9 @@
/* Define to 1 if wcwidth is declared even after undefining macros. */
#undef HAVE_RAW_DECL_WCWIDTH
+/* readline-6.0 workaround of blocked signals. */
+#undef HAVE_READLINE_CAUGHT_SIGNAL
+
/* Define to 1 if you have the `realpath' function. */
#undef HAVE_REALPATH
Index: gdb-7.2/gdb/configure.ac
===================================================================
--- gdb-7.2.orig/gdb/configure.ac 2011-01-16 18:26:11.000000000 +0100
+++ gdb-7.2/gdb/configure.ac 2011-01-16 18:26:26.000000000 +0100
@@ -779,17 +779,25 @@ if test "$with_system_readline" = yes; t
# readline-6.0 started to use the name `_rl_echoing_p'.
# `$(READLINE_DIR)/' of bundled readline would not resolve in configure.
- AC_MSG_CHECKING([for readline_echoing_p])
save_LIBS=$LIBS
LIBS="$LIBS $READLINE"
+ AC_MSG_CHECKING([for readline_echoing_p])
AC_LINK_IFELSE(AC_LANG_PROGRAM(,[[extern int readline_echoing_p;
return readline_echoing_p;]]),
[READLINE_ECHOING_P=yes],
[READLINE_ECHOING_P=no
AC_DEFINE([readline_echoing_p], [_rl_echoing_p],
[readline-6.0 started to use different name.])])
- LIBS="$save_LIBS"
AC_MSG_RESULT([$READLINE_ECHOING_P])
+ AC_MSG_CHECKING([for _rl_caught_signal])
+ AC_LINK_IFELSE(AC_LANG_PROGRAM(,[[extern int volatile _rl_caught_signal;
+ return _rl_caught_signal;]]),
+ [READLINE_CAUGHT_SIGNAL=yes
+ AC_DEFINE([HAVE_READLINE_CAUGHT_SIGNAL],,
+ [readline-6.0 workaround of blocked signals.])],
+ [READLINE_CAUGHT_SIGNAL=no])
+ AC_MSG_RESULT([$READLINE_CAUGHT_SIGNAL])
+ LIBS="$save_LIBS"
else
READLINE='$(READLINE_DIR)/libreadline.a'
READLINE_DEPS='$(READLINE)'
Index: gdb-7.2/gdb/event-loop.c
===================================================================
--- gdb-7.2.orig/gdb/event-loop.c 2010-05-25 17:48:43.000000000 +0200
+++ gdb-7.2/gdb/event-loop.c 2011-01-16 18:27:14.000000000 +0100
@@ -37,6 +37,7 @@
#include "exceptions.h"
#include "gdb_assert.h"
#include "gdb_select.h"
+#include "readline/readline.h"
/* Tell create_file_handler what events we are interested in.
This is used by the select version of the event loop. */
@@ -416,6 +417,9 @@ gdb_do_one_event (void *data)
static int event_source_head = 0;
const int number_of_sources = 3;
int current = 0;
+#ifdef HAVE_READLINE_CAUGHT_SIGNAL
+ extern int volatile _rl_caught_signal;
+#endif
/* Any events already waiting in the queue? */
if (process_event ())
@@ -460,6 +464,16 @@ gdb_do_one_event (void *data)
if (gdb_wait_for_event (1) < 0)
return -1;
+#ifdef HAVE_READLINE_CAUGHT_SIGNAL
+ if (async_command_editing_p && RL_ISSTATE (RL_STATE_CALLBACK)
+ && _rl_caught_signal)
+ {
+ /* Call RL_CHECK_SIGNALS this way. */
+ rl_callback_handler_remove ();
+ rl_callback_handler_install (NULL, input_handler);
+ }
+#endif
+
/* Handle any new events occurred while waiting. */
if (process_event ())
return 1;
Index: gdb-7.2/gdb/testsuite/gdb.gdb/selftest.exp
===================================================================
--- gdb-7.2.orig/gdb/testsuite/gdb.gdb/selftest.exp 2011-01-16 18:26:08.000000000 +0100
+++ gdb-7.2/gdb/testsuite/gdb.gdb/selftest.exp 2011-01-16 18:30:41.000000000 +0100
@@ -433,6 +433,28 @@ proc test_with_self { executable } {
}
}
+ # get a stack trace with the poll function
+ #
+ # This fails on some linux systems for unknown reasons. On the
+ # systems where it fails, sometimes it works fine when run manually.
+ # The testsuite failures may not be limited to just aout systems.
+ setup_xfail "i*86-pc-linuxaout-gnu"
+ set description "backtrace through signal handler"
+ gdb_test_multiple "backtrace" "$description" {
+ -re "#0.*(read|poll).*in main \\(.*\\) at .*gdb\\.c.*$gdb_prompt $" {
+ pass "$description"
+ }
+ -re ".*$gdb_prompt $" {
+ # On the alpha, we hit the infamous problem about gdb
+ # being unable to get the frame pointer (mentioned in
+ # gdb/README). As it is intermittent, there is no way to
+ # XFAIL it which will give us an XPASS if the problem goes
+ # away.
+ setup_xfail "alpha*-*-osf*"
+ fail "$description"
+ }
+ }
+
set description "send SIGINT signal to child process"
gdb_test "signal SIGINT" \
"Continuing with signal SIGINT.*" \
@@ -443,10 +465,11 @@ proc test_with_self { executable } {
# This fails on some linux systems for unknown reasons. On the
# systems where it fails, sometimes it works fine when run manually.
# The testsuite failures may not be limited to just aout systems.
+ # Optional system readline may not have symbols to be shown.
setup_xfail "i*86-pc-linuxaout-gnu"
- set description "backtrace through signal handler"
+ set description "backtrace through readline handler"
gdb_test_multiple "backtrace" "$description" {
- -re "#0.*(read|poll).*in main \\(.*\\) at .*gdb\\.c.*$gdb_prompt $" {
+ -re "#0.*gdb_do_one_event.*in main \\(.*\\) at .*gdb\\.c.*$gdb_prompt $" {
pass "$description"
}
-re ".*$gdb_prompt $" {
Index: gdb-7.2/gdb/configure
===================================================================
--- gdb-7.2.orig/gdb/configure 2011-01-16 18:26:11.000000000 +0100
+++ gdb-7.2/gdb/configure 2011-01-16 18:26:51.000000000 +0100
@@ -10305,10 +10305,10 @@ if test "$with_system_readline" = yes; t
# readline-6.0 started to use the name `_rl_echoing_p'.
# `$(READLINE_DIR)/' of bundled readline would not resolve in configure.
- { $as_echo "$as_me:${as_lineno-$LINENO}: checking for readline_echoing_p" >&5
-$as_echo_n "checking for readline_echoing_p... " >&6; }
save_LIBS=$LIBS
LIBS="$LIBS $READLINE"
+ { $as_echo "$as_me:${as_lineno-$LINENO}: checking for readline_echoing_p" >&5
+$as_echo_n "checking for readline_echoing_p... " >&6; }
cat confdefs.h - <<_ACEOF >conftest.$ac_ext
/* end confdefs.h. */
@@ -10331,9 +10331,35 @@ $as_echo "#define readline_echoing_p _rl
fi
rm -f core conftest.err conftest.$ac_objext \
conftest$ac_exeext conftest.$ac_ext
- LIBS="$save_LIBS"
{ $as_echo "$as_me:${as_lineno-$LINENO}: result: $READLINE_ECHOING_P" >&5
$as_echo "$READLINE_ECHOING_P" >&6; }
+ { $as_echo "$as_me:${as_lineno-$LINENO}: checking for _rl_caught_signal" >&5
+$as_echo_n "checking for _rl_caught_signal... " >&6; }
+ cat confdefs.h - <<_ACEOF >conftest.$ac_ext
+/* end confdefs.h. */
+
+int
+main ()
+{
+extern int volatile _rl_caught_signal;
+ return _rl_caught_signal;
+ ;
+ return 0;
+}
+_ACEOF
+if ac_fn_c_try_link "$LINENO"; then :
+ READLINE_CAUGHT_SIGNAL=yes
+
+$as_echo "#define HAVE_READLINE_CAUGHT_SIGNAL /**/" >>confdefs.h
+
+else
+ READLINE_CAUGHT_SIGNAL=no
+fi
+rm -f core conftest.err conftest.$ac_objext \
+ conftest$ac_exeext conftest.$ac_ext
+ { $as_echo "$as_me:${as_lineno-$LINENO}: result: $READLINE_CAUGHT_SIGNAL" >&5
+$as_echo "$READLINE_CAUGHT_SIGNAL" >&6; }
+ LIBS="$save_LIBS"
else
READLINE='$(READLINE_DIR)/libreadline.a'
READLINE_DEPS='$(READLINE)'

14
gdb.spec
View File

@ -27,7 +27,7 @@ Version: 7.2
# The release always contains a leading reserved number, start it at 1.
# `upstream' is not a part of `name' to stay fully rpm dependencies compatible for the testing.
Release: 34%{?_with_upstream:.upstream}%{dist}
Release: 35%{?_with_upstream:.upstream}%{dist}
License: GPLv3+ and GPLv3+ with exceptions and GPLv2+ and GPLv2+ with exceptions and GPL+ and LGPLv2+ and GFDL and BSD and Public Domain
Group: Development/Debuggers
@ -471,6 +471,9 @@ Patch381: gdb-simultaneous-step-resume-breakpoint-test.patch
#=push+work: It should be in glibc: libc-alpha: <20091004161706.GA27450@.*>
Patch382: gdb-core-open-vdso-warning.patch
# Fix callback-mode readline-6.0 regression for CTRL-C (for RHEL-6.0).
Patch390: gdb-readline-6.0-signal.patch
# Fix syscall restarts for amd64->i386 biarch.
#=push
Patch391: gdb-x86_64-i386-syscall-restart.patch
@ -1002,8 +1005,14 @@ rm -f gdb/jv-exp.c gdb/m2-exp.c gdb/objc-exp.c gdb/p-exp.c
%patch549 -p1
%patch550 -p1
%patch390 -p1
%patch393 -p1
%patch335 -p1
readline="$(readlink -f %{_libdir}/libreadline.so)"
if [ "$readline" = "${readline%/libreadline.so.6.0}" ]
then
%patch390 -p1 -R
fi
%if 0%{!?el5:1}
%patch393 -p1 -R
%patch335 -p1 -R
@ -1392,6 +1401,9 @@ fi
%endif
%changelog
* Sun Jan 16 2011 Jan Kratochvil <jan.kratochvil@redhat.com> - 7.2-35.fc14
- Fix callback-mode readline-6.0 regression for CTRL-C (for RHEL-6.0).
* Sat Jan 15 2011 Jan Kratochvil <jan.kratochvil@redhat.com> - 7.2-34.fc14
- [vla] Support Fortran vector slices and subsets (BZ 609782).