From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (qmail 18468 invoked by alias); 27 Mar 2009 00:46:17 -0000 Received: (qmail 18460 invoked by uid 22791); 27 Mar 2009 00:46:15 -0000 X-SWARE-Spam-Status: No, hits=-2.2 required=5.0 tests=AWL,BAYES_00,SPF_PASS X-Spam-Check-By: sourceware.org Received: from smtp-out.google.com (HELO smtp-out.google.com) (216.239.45.13) by sourceware.org (qpsmtpd/0.43rc1) with ESMTP; Fri, 27 Mar 2009 00:46:06 +0000 Received: from wpaz13.hot.corp.google.com (wpaz13.hot.corp.google.com [172.24.198.77]) by smtp-out.google.com with ESMTP id n2R0k3FF011880; Thu, 26 Mar 2009 17:46:03 -0700 Received: from localhost (ruffy.mtv.corp.google.com [172.18.118.116]) by wpaz13.hot.corp.google.com with ESMTP id n2R0k1Zb031216; Thu, 26 Mar 2009 17:46:01 -0700 Received: by localhost (Postfix, from userid 67641) id 576E41C74E3; Thu, 26 Mar 2009 17:46:01 -0700 (PDT) To: pedro@codesourcery.com, gdb-patches@sourceware.org Subject: [RFA] fix crash when rerunning programs in non-stop mode Message-Id: <20090327004601.576E41C74E3@localhost> Date: Fri, 27 Mar 2009 01:23:00 -0000 From: dje@google.com (Doug Evans) X-System-Of-Record: true X-IsSubscribed: yes Mailing-List: contact gdb-patches-help@sourceware.org; run by ezmlm Precedence: bulk List-Id: List-Subscribe: List-Archive: List-Post: List-Help: , Sender: gdb-patches-owner@sourceware.org X-SW-Source: 2009-03/txt/msg00606.txt.bz2 Hi. I noticed a patch that recently went in to fix do_restore_current_thread_cleanup's handling of restoring the current thread if the program exited. Good timing because I've been playing with gdb crashes in this area. The patch does this: if (tp && is_exited (tp->ptid) && find_inferior_pid (ptid_get_pid (tp->ptid)) == NULL) restore_current_thread (null_ptid); else restore_current_thread (old->inferior_ptid); One problem is that calling is_exited is problematic if ! target_has_execution (this one I'm tripping over). Another problem is if tp == NULL, then old->inferior_ptid is restored (this is just an observation). This patch reverses the sense of what's being tested, i.e. check for whether old->inferior_ptid restorable instead of whether it's dead. Ok to check in? Included is a testcase to exercise the problem. 2009-03-26 Doug Evans * thread.c (do_restore_current_thread_cleanup): Redo test for whether to restore old->inferior_ptid. * gdb.mi/mi-nonstop-exit.exp: New file. * gdb.mi/non-stop-exit.c: New file. Index: thread.c =================================================================== RCS file: /cvs/src/src/gdb/thread.c,v retrieving revision 1.107 diff -u -p -u -p -r1.107 thread.c --- thread.c 26 Mar 2009 21:30:26 -0000 1.107 +++ thread.c 27 Mar 2009 00:32:30 -0000 @@ -924,11 +924,10 @@ do_restore_current_thread_cleanup (void then don't revert back to it, but instead simply drop back to no thread selected. */ if (tp - && is_exited (tp->ptid) - && find_inferior_pid (ptid_get_pid (tp->ptid)) == NULL) - restore_current_thread (null_ptid); - else + && find_inferior_pid (ptid_get_pid (tp->ptid)) != NULL) restore_current_thread (old->inferior_ptid); + else + restore_current_thread (null_ptid); /* The running state of the originally selected thread may have changed, so we have to recheck it here. */ Index: testsuite/gdb.mi/mi-nonstop-exit.exp =================================================================== RCS file: testsuite/gdb.mi/mi-nonstop-exit.exp diff -N testsuite/gdb.mi/mi-nonstop-exit.exp --- /dev/null 1 Jan 1970 00:00:00 -0000 +++ testsuite/gdb.mi/mi-nonstop-exit.exp 27 Mar 2009 00:40:04 -0000 @@ -0,0 +1,106 @@ +# Copyright 2009 +# 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 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 . + +# This only works with native configurations +if {![isnative]} { + return +} + +load_lib mi-support.exp +set MIFLAGS "-i=mi" + +gdb_exit +if {[mi_gdb_start]} { + continue +} + +# +# Start here +# +set testfile "non-stop-exit" +set srcfile "$testfile.c" +set binfile "$objdir/$subdir/mi-$testfile" + +set options [list debug incdir=$objdir] +if {[gdb_compile_pthreads "$srcdir/$subdir/$srcfile" $binfile executable $options] != "" } { + return -1 +} + +mi_gdb_reinitialize_dir $srcdir/$subdir +mi_gdb_load $binfile + +set supported 0 +send_gdb "-gdb-show non-stop\n" +gdb_expect { + -re ".*\\^done,value=\"off\",supported=\"(\[^\"\]+)\"\r\n$mi_gdb_prompt$" { + if { $expect_out(1,string) == "1" } { + set supported 1 + } + } + -re ".$mi_gdb_prompt$" { + } +} + +mi_gdb_test "-gdb-set non-stop 1" ".*" +mi_gdb_test "-gdb-set target-async 1" ".*" +detect_async + +mi_gdb_test "200-break-insert -t main" ".*" + +set created "=thread-created,id=\"$decimal\"\r\n" +set running "\\*running,thread-id=\"$decimal\"\r\n" + +set notifs "($created)*($running)*($library_loaded_re)*" + +# Note: presently, we skip this test on non-native targets, +# so 'run' is OK. As soon as we start to run this on remote +# target, the logic from mi_run_cmd will have to be refactored. +send_gdb "-exec-run\n" +gdb_expect { + -re "\\^running\r\n$notifs$mi_gdb_prompt" { + } + -re "\\^error,msg=\"The target does not support running in non-stop mode.\"" { + verbose -log "Non-stop mode not supported, skipping all tests" + return + } + -re "\r\n$mi_gdb_prompt" { + perror "Cannot start target (unknown output after running)" + return -1 + } + timeout { + perror "Cannot start target (timeout)" + return -1 + } +} +mi_expect_stop "breakpoint-hit" main ".*" ".*" "\[0-9\]+" { "" "disp=\"del\"" } "run to main" + +mi_send_resuming_command "exec-continue" "continue" + +mi_expect_stop "exited-normally" "" "" "" "" "" "finished exec continue" + +# Run the program again. + +mi_gdb_test "200-break-insert -t main" ".*" + +mi_send_resuming_command "exec-run" "run (2)" + +mi_expect_stop "breakpoint-hit" main ".*" ".*" "\[0-9\]+" { "" "disp=\"del\"" } "run to main (2)" + +mi_send_resuming_command "exec-continue" "continue (2)" + +mi_expect_stop "exited-normally" "" "" "" "" "" "finished exec continue (2)" + +mi_gdb_exit Index: testsuite/gdb.mi/non-stop-exit.c =================================================================== RCS file: testsuite/gdb.mi/non-stop-exit.c diff -N testsuite/gdb.mi/non-stop-exit.c --- /dev/null 1 Jan 1970 00:00:00 -0000 +++ testsuite/gdb.mi/non-stop-exit.c 27 Mar 2009 00:40:04 -0000 @@ -0,0 +1,54 @@ +/* Test program exit in non-stop mode. + Copyright 2009 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 . */ + +#include +#include + +#define NTHREADS 4 +void* thread_function (void*); + +void * +thread_function (void *arg) +{ + int x = * (int *) arg; + + printf ("Thread <%d> executing\n", x); + + return NULL; +} + +int +main () +{ + pthread_t thread_id[NTHREADS]; + int args[NTHREADS]; + int i; + + for (i = 0; i < NTHREADS; ++i) + { + args[i] = i; + pthread_create (&thread_id[i], NULL, thread_function, &args[i]); + } + + for (i = 0; i < NTHREADS; ++i) + { + pthread_join (thread_id[i], NULL); + } + + return 0; +}