# Copyright 1999, 2001, 2002, 2003, 2004, 2005, 2006 # 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@gnu.org #### Dining Philosophers, on LinuxThreads - Jim Blandy #### #### At the moment, GDB's support for LinuxThreads is pretty #### idiosyncratic --- GDB's output doesn't look much like the output #### it produces for other thread implementations, messages appear at #### different times, etc. So these tests are specific to LinuxThreads. #### #### However, if all goes well, Linux will soon have a libthread_db #### interface, and GDB will manage it the same way it does other #### libthread_db-based systems. Then, we can adjust this file to #### work with any such system. ### Other things we ought to test: ### stepping a thread while others are running ### killing and restarting ### quitting gracefully if $tracelevel then { strace $tracelevel } set prms_id 0 set bug_id 0 # This only works with Linux configurations. if ![istarget *-*-linux-gnu*] then { return } set testfile "linux-dp" set srcfile ${testfile}.c set binfile ${objdir}/${subdir}/${testfile} if {[gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}" executable {debug libs=-lpthread}] != ""} { return -1 } gdb_start gdb_reinitialize_dir $srcdir/$subdir gdb_load ${binfile} send_gdb "set print sevenbit-strings\n" ; gdb_expect -re "$gdb_prompt $" runto_main # There should be no threads initially. gdb_test "info threads" "" "info threads 1" # Try stepping over the thread creation function. gdb_breakpoint [gdb_get_line_number "linuxthreads.exp: create philosopher"] set expect_manager -1 for {set i 0} {$i < 5} {incr i} { gdb_continue_to_breakpoint "about to create philosopher: $i" send_gdb "info threads\n" set threads_before {} gdb_expect { -re "info threads\r\n" { exp_continue } -re "^. +(\[0-9\]+ Thread \[-0-9\]+) \[^\n\]*\n" { verbose -log "found thread $expect_out(1,string)" 2 lappend threads_before $expect_out(1,string) exp_continue } -re "^\[^\n\]*\n" { verbose -log "skipping line" 2 } -re "^$gdb_prompt $" { } timeout { fail "(timeout) info threads before: $i" } } send_gdb "next\n" set threads_created 0 gdb_expect { -re "^next\r\n" { exp_continue } -re "^ *\[_!\] \[0-9\]* \[_!\]\r\n" { # Ignore program output. exp_continue -continue_timer } -re "^\\\[New \[^\]\n\]+\\\]\[^\n\]+\n" { incr threads_created exp_continue } -re "^189\[^\n\]+\n" { exp_continue } -re "^$gdb_prompt $" { } -re "Program received signal.*(Unknown signal|SIGUSR|Real-time event).*$gdb_prompt $" { # It would be nice if we could catch the message that GDB prints # when it first notices that the thread library doesn't support # debugging, or if we could explicitly ask GDB somehow. unsupported "This GDB does not support threads on this system." return -1 } -re "$gdb_prompt $" { } timeout { fail "(timeout) create philosopher: $i" } } if { $threads_created == 0 } { # Not all targets announce new threads as they are created. # For example, the GDB # remote protocol target only finds out about threads when # they actually report some event like a breakpoint hit, # or when the user types 'info threads'. unsupported "create philosopher: $i" } elseif { $threads_created == 1 } { if { $expect_manager < 0 } { set expect_manager 0 } pass "create philosopher: $i" } elseif { !$i && $threads_created == 2 } { # Two threads are created the first time in LinuxThreads, # where the second is the manager thread. In NPTL, there is none. set expect_manager 1 pass "create philosopher: $i" } else { fail "create philosopher: $i" } send_gdb "info threads\n" set threads_after {} gdb_expect { -re "info threads\r\n" { exp_continue } -re "^. +(\[0-9\]+ Thread \[-0-9\]+) \[^\n\]*\n" { set name $expect_out(1,string) for {set j 0} {$j != [llength $threads_before] } {incr j} { if {$name == [lindex $threads_before $j]} { set threads_before [lreplace $threads_before $j $j] set name "" break } } if { $name != "" } { lappend threads_after $name } exp_continue } -re "^\[^\n\]*\n" { verbose -log "skipping line" 2 } -re "^$gdb_prompt $" { if { [llength $threads_before] != 0 } { fail "info threads after: $i" } elseif { !$i && [llength $threads_after] == 2 } { set expect_manager 1 pass "info threads after: $i" } elseif { [llength $threads_after] == 1 } { if { $expect_manager < 0 } { set expect_manager 0 } pass "info threads after: $i" } else { fail "info threads after: $i" } } timeout { fail "(timeout) info threads after: $i" } } } set nthreads 6 # Run until there are some threads. gdb_breakpoint [gdb_get_line_number "linuxthreads.exp: info threads 2"] gdb_continue_to_breakpoint "main thread's sleep" set info_threads_ptn "" for {set i $nthreads} {$i > 0} {incr i -1} { append info_threads_ptn "$i Thread .*" } append info_threads_ptn "\[\r\n\]+$gdb_prompt $" set info_threads_manager_ptn "[expr $nthreads + 1] Thread .*$info_threads_ptn" gdb_test_multiple "info threads" "info threads 2" { -re "$info_threads_manager_ptn" { # We did see a manager thread. Check that against what we expected. switch -exact -- $expect_manager { -1 { # We weren't sure whether to expect a manager thread. pass "info threads 2" } 1 { # We were expecting a manager thread. pass "info threads 2" } 0 { # We were not expecting to see the manager thread. fail "info threads 2" } } set expect_manager 1 incr nthreads } -re "$info_threads_ptn" { # We did not see a manager thread. Check that against what we # expected. switch -exact -- $expect_manager { -1 { # We weren't sure whether to expect a manager thread. # Don't expect it from here on out. pass "info threads 2" } 1 { # We were expecting a manager thread, but we didn't see one. fail "info threads 2" } 0 { # We were not expecting to see the manager thread. pass "info threads 2" } } set expect_manager 0 } } # Try setting a thread-specific breakpoint. gdb_breakpoint "print_philosopher thread 5" gdb_continue_to_breakpoint "thread 5's print" # When there is no debugging info available for the thread library, # the backtrace entry for philosopher's caller looks like: # #1 0x4001c548 in pthread_create () from /lib/libpthread.so.0 # If you do have debug info, the output obviously depends more on the # exact library in use; under NPTL, you get: # #2 0x0012b7fc in start_thread (arg=0x21) at pthread_create.c:264 gdb_test "where" "print_philosopher.*philosopher.* \(from .*libpthread\|at pthread_create\).*" \ "first thread-specific breakpoint hit" # Make sure it's catching the right thread. Try hitting the # breakpoint ten times, and make sure we don't get anyone else. set only_five 1 for {set i 0} {$only_five > 0 && $i < 10} {incr i} { gdb_continue_to_breakpoint "thread 5's print, pass: $i" send_gdb "info threads\n" gdb_expect { -re "\\* 5 Thread .* print_philosopher .*\r\n$gdb_prompt $" { # Okay this time. } -re ".*$gdb_prompt $" { set only_five 0 } timeout { set only_five -1 } } } set name "thread-specific breakpoint is thread-specific" if {$only_five == 1} { pass $name } if {$only_five == 0} { fail $name } if {$only_five == -1} { fail "$name (timeout)" } ### Select a particular thread. proc select_thread {thread} { global gdb_prompt send_gdb "thread $thread\n" gdb_expect { -re "\\\[Switching to thread .*\\\].*\r\n$gdb_prompt $" { pass "selected thread: $thread" } -re "$gdb_prompt $" { fail "selected thread: $thread" } timeout { fail "selected thread: $thread (timeout)" } } } ### Select THREAD, check for a plausible backtrace, and make sure ### we're actually selecting a different philosopher each time. ### Return true if the thread had a stack which was not only ### acceptable, but interesting. SEEN should be an array in which ### SEEN(N) exists iff we have found philosopher number N before. set main_seen 0 set manager_seen 0 proc check_philosopher_stack {thread seen_name} { global gdb_prompt upvar $seen_name seen global main_seen global expect_manager manager_seen set name "philosopher is distinct: $thread" set interesting 0 select_thread $thread send_gdb "where\n" gdb_expect { -re ".* in philosopher \\(data=(0x\[0-9a-f\]+).*\r\n$gdb_prompt $" { set data $expect_out(1,string) if {[info exists seen($data)]} { fail $name } else { pass $name set seen($data) yep } set interesting 1 } -re ".* in __pthread_manager \\(.*$gdb_prompt $" { if {$manager_seen == 1} { fail "manager thread is distinct: $thread" } else { set manager_seen 1 pass "manager thread is distinct: $thread" } set interesting 1 } -re "pthread_start_thread.*\r\n$gdb_prompt $" { ## Maybe the thread hasn't started yet. pass $name } -re ".* in main \\(.*$gdb_prompt $" { if {$main_seen == 1} { fail "main is distinct: $thread" } else { set main_seen 1 pass "main is distinct: $thread" } set interesting 1 } -re " in \\?\\?.*\r\n$gdb_prompt $" { ## Sometimes we can't get a backtrace. I'm going to call ## this a pass, since we do verify that at least one ## thread was interesting, so we can get more consistent ## test suite totals. But in my heart, I think it should ## be an xfail. pass $name } -re "$gdb_prompt $" { fail $name } timeout { fail "$name (timeout)" } } return $interesting } set any_interesting 0 array set seen {} unset seen for {set i 1} {$i <= $nthreads} {incr i} { if [check_philosopher_stack $i seen] { set any_interesting 1 } } if {$any_interesting} { pass "found an interesting thread" } else { fail "found an interesting thread" } if {$manager_seen == $expect_manager} { pass "manager thread found (not found) when expected" } else { fail "manager thread found (not found) when expected" }