From: Jan Kratochvil <jan.kratochvil@redhat.com>
To: Luis Machado <luisgpm@linux.vnet.ibm.com>
Cc: gdb-patches ml <gdb-patches@sourceware.org>
Subject: Re: [patch] Single stepping over atomic instruction sets for GDB 6.6
Date: Sat, 28 Apr 2007 18:16:00 -0000 [thread overview]
Message-ID: <20070428175641.GA2136@host0.dyn.jankratochvil.net> (raw)
In-Reply-To: <1177352371.15414.29.camel@localhost>
[-- Attachment #1: Type: text/plain, Size: 1240 bytes --]
On Mon, 23 Apr 2007 20:19:31 +0200, Luis Machado wrote:
...
> This is the patch for handling the sigle stepping of atomic instruction
> sets previously posted on this list refreshed for the stable version 6.6
> of GDB, in case users of this version would like to use it.
It fails for me on the attached testcase on gdb-6.6.ppc64 in 99% cases with
infrun.c:1567: internal-error: handle_inferior_event: Assertion `SOFTWARE_SINGLE_STEP_P () && singlestep_breakpoints_inserted_p' failed.
(tested on 4x CPU POWER5+, kernel-2.6.18-8.1.1.el5.ppc64)
The attached patch ON TOP OF YOURS fixes this problem (mostly, see below).
This patch has not been ported to HEAD as the last CVS version of your patch
I found is
http://sourceware.org/ml/gdb-patches/2007-03/msg00145.html
which is no longer applicable to HEAD, please update it.
--
This fix is not 100%, though, it now passes only in 93% cases, in the remaining
7% of cases it will prematurely FAIL due to (missed breakpoint)
Program exited normally.
I believe it is not a failure of this fix but some unrelated PPC port problem.
Unfortunately this "Program exited normally." FAIL becomes unreproducible while
trying to debug it (`set debug infrun 1'), there is some race.
Regards,
Jan
[-- Attachment #2: gdb-6.6-stepping_past_singlestep.patch --]
[-- Type: text/plain, Size: 7508 bytes --]
2007-04-28 Jan Kratochvil <jan.kratochvil@redhat.com>
* infrun.c (handle_inferior_event): Handle
(STEPPING_PAST_SINGLESTEP_BREAKPOINT
&& !SINGLESTEP_BREAKPOINTS_INSERTED_P).
2007-04-28 Jan Kratochvil <jan.kratochvil@redhat.com>
* gdb.threads/atomic-step-mass-join.c,
gdb.threads/atomic-step-mass-join.exp: New files.
diff -u -rup gdb-6.6-orig/gdb/infrun.c gdb-6.6/gdb/infrun.c
--- gdb-6.6-orig/gdb/infrun.c 2006-10-18 18:56:13.000000000 +0200
+++ gdb-6.6/gdb/infrun.c 2007-04-28 19:23:08.000000000 +0200
@@ -1561,10 +1561,17 @@ handle_inferior_event (struct execution_
if (stepping_past_singlestep_breakpoint)
{
- gdb_assert (SOFTWARE_SINGLE_STEP_P ()
- && singlestep_breakpoints_inserted_p);
- gdb_assert (ptid_equal (singlestep_ptid, ecs->ptid));
- gdb_assert (!ptid_equal (singlestep_ptid, saved_singlestep_ptid));
+ gdb_assert (SOFTWARE_SINGLE_STEP_P ());
+ /* SINGLESTEP_BREAKPOINTS_INSERTED_P is 0 if the another thread which hit
+ the singlestep breakpoint intended for SAVED_SINGLESTEP_PTID is not
+ right now at the atomic sequence. SINGLESTEP_PTID is not defined in
+ such case. SINGLESTEP_BREAKPOINTS_INSERTED_P is 1 when both the
+ SAVED_SINGLESTEP_PTID thread and the SINGLESTEP_PTID thread skip over
+ an atomic sequence, possibly a different one. */
+ gdb_assert (!singlestep_breakpoints_inserted_p
+ || ptid_equal (singlestep_ptid, ecs->ptid));
+ gdb_assert (!singlestep_breakpoints_inserted_p
+ || !ptid_equal (singlestep_ptid, saved_singlestep_ptid));
stepping_past_singlestep_breakpoint = 0;
@@ -1575,9 +1582,13 @@ handle_inferior_event (struct execution_
{
if (debug_infrun)
fprintf_unfiltered (gdb_stdlog, "infrun: stepping_past_singlestep_breakpoint\n");
- /* Pull the single step breakpoints out of the target. */
- (void) SOFTWARE_SINGLE_STEP (0, 0);
- singlestep_breakpoints_inserted_p = 0;
+
+ if (singlestep_breakpoints_inserted_p)
+ {
+ /* Pull the single step breakpoints out of the target. */
+ (void) SOFTWARE_SINGLE_STEP (0, 0);
+ singlestep_breakpoints_inserted_p = 0;
+ }
ecs->random_signal = 0;
diff -u -rup gdb-6.6-orig/gdb/testsuite/gdb.threads/atomic-step-mass-join.c gdb-6.6/gdb/testsuite/gdb.threads/atomic-step-mass-join.c
--- gdb-6.6-orig/gdb/testsuite/gdb.threads/atomic-step-mass-join.c 2007-04-28 19:43:46.000000000 +0200
+++ gdb-6.6/gdb/testsuite/gdb.threads/atomic-step-mass-join.c 2007-04-28 19:24:15.000000000 +0200
@@ -0,0 +1,57 @@
+/* Test case for stepping over RISC atomic code constructs.
+
+ Copyright 2007
+ 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 2 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, write to the Free Software
+ Foundation, Inc., 59 Temple Place - Suite 330,
+ Boston, MA 02111-1307, USA. */
+
+#include <pthread.h>
+#include <assert.h>
+#include <unistd.h>
+
+static volatile int var;
+
+static void *start (void *arg)
+{
+ /* Do not start quitting before all the threads were spawn. */
+ sleep (5);
+
+ return arg;
+}
+
+#define LENGTH(x) (sizeof (x) / sizeof (x)[0])
+
+int main (void)
+{
+ pthread_t thread[20];
+ int i, j;
+
+ var = 0;
+ for (j = 0; j < LENGTH (thread); j++)
+ {
+ i = pthread_create (&thread[j], NULL, start, NULL); /* _create_ */
+ assert (i == 0);
+ }
+ for (j = 0; j < LENGTH (thread); j++)
+ {
+ i = pthread_join (thread[j], NULL); /* _delete_ */
+ assert (i == 0);
+ }
+
+ return 0; /* _quit_ */
+}
diff -u -rup gdb-6.6-orig/gdb/testsuite/gdb.threads/atomic-step-mass-join.exp gdb-6.6/gdb/testsuite/gdb.threads/atomic-step-mass-join.exp
--- gdb-6.6-orig/gdb/testsuite/gdb.threads/atomic-step-mass-join.exp 2007-04-28 19:43:46.000000000 +0200
+++ gdb-6.6/gdb/testsuite/gdb.threads/atomic-step-mass-join.exp 2007-04-28 19:42:25.000000000 +0200
@@ -0,0 +1,93 @@
+# atomic-step-mass-join.exp -- Test case for stepping over RISC atomic code constructs.
+# Copyright (C) 2007 Free Software Foundation, Inc.
+
+# 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 2 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, write to the Free Software
+# Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA. */
+
+# Please email any bugs, comments, and/or additions to this file to:
+# bug-gdb@prep.ai.mit.edu
+
+set testfile atomic-step-mass-join
+set srcfile ${testfile}.c
+set binfile ${objdir}/${subdir}/${testfile}
+
+if [istarget "*-*-linux"] then {
+ set target_cflags "-D_MIT_POSIX_THREADS"
+} else {
+ set target_cflags ""
+}
+
+if {[gdb_compile_pthreads "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug}] != "" } {
+ return -1
+}
+
+gdb_exit
+gdb_start
+gdb_reinitialize_dir $srcdir/$subdir
+
+gdb_load ${binfile}
+if ![runto_main] then {
+ fail "Can't run to main"
+ return 0
+}
+
+# With a watchpoint we would crash on pthread_create(3):
+# Program received signal SIGSTOP, Stopped (signal).
+# [Switching to Thread 4160681152 (LWP 979)]
+# 0x0f55c368 in clone () from /lib/libc.so.6
+# Still it is not applicable so far due to:
+# [New Thread 4139709632 (LWP 9022)]^M
+# 0x0f354a28 in __nptl_create_event () from /lib/libpthread.so.0^M
+# ptrace: No such process.^M
+#gdb_test "handle SIGSTOP nostop noprint pass" \
+# ".*SIGSTOP +No\tNo\tYes\t+Stopped \\(signal\\)" \
+# "Disown SIGSTOP"
+
+set line [gdb_get_line_number "_delete_"]
+gdb_test "tbreak $line" \
+ "Breakpoint (\[0-9\]+) at .*$srcfile, line $line\..*" \
+ "set breakpoint at delete"
+gdb_test "c" \
+ ".*/\\* _delete_ \\*/.*" \
+ "run till _delete_"
+
+# Without a software watchpoint no single-stepping would be used.
+set test "Start (software) watchpoint"
+gdb_test_multiple "watch var" $test {
+ -re "Watchpoint \[0-9\]+: var.*$gdb_prompt $" {
+ pass $test
+ }
+ -re "Hardware watchpoint \[0-9\]+: var.*$gdb_prompt $" {
+ # We do not test the goal but still the whole testcase should pass.
+ unsupported $test
+ }
+}
+
+###gdb_test "set debug infrun 1" ""
+
+set line [gdb_get_line_number "_quit_"]
+gdb_test "tbreak $line" \
+ "Breakpoint \[0-9\]+ at .*$srcfile, line $line\..*" \
+ "set breakpoint at _quit_"
+set timeout_old $timeout
+# 20 threads may take about 45 seconds. */
+set timeout 180
+gdb_test "c" \
+ ".*/\\* _quit_ \\*/.*" \
+ "run till _quit_"
+set timeout $timeout_old
+
+gdb_test "c" \
+ ".*Program exited normally\\." \
+ "run till program exit"
next prev parent reply other threads:[~2007-04-28 17:58 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-04-23 21:06 Luis Machado
2007-04-28 18:16 ` Jan Kratochvil [this message]
2007-04-28 23:37 ` Luis Machado
2007-04-29 15:50 ` Jan Kratochvil
2007-04-30 20:26 ` Luis Machado
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20070428175641.GA2136@host0.dyn.jankratochvil.net \
--to=jan.kratochvil@redhat.com \
--cc=gdb-patches@sourceware.org \
--cc=luisgpm@linux.vnet.ibm.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox