PR 17408 - assertion failure in switch_back_to_stepped_thread

This PR shows that GDB can easily trigger an assertion here, in
infrun.c:

 5392              /* Did we find the stepping thread?  */
 5393              if (tp->control.step_range_end)
 5394                {
 5395                  /* Yep.  There should only one though.  */
 5396                  gdb_assert (stepping_thread == NULL);
 5397
 5398                  /* The event thread is handled at the top, before we
 5399                     enter this loop.  */
 5400                  gdb_assert (tp != ecs->event_thread);
 5401
 5402                  /* If some thread other than the event thread is
 5403                     stepping, then scheduler locking can't be in effect,
 5404                     otherwise we wouldn't have resumed the current event
 5405                     thread in the first place.  */
 5406                  gdb_assert (!schedlock_applies (currently_stepping (tp)));
 5407
 5408                  stepping_thread = tp;
 5409                }

Like:

 gdb/infrun.c:5406: internal-error: switch_back_to_stepped_thread: Assertion `!schedlock_applies (1)' failed.

The way the assertion is written is assuming that with schedlock=step
we'll always leave threads other than the one with the stepping range
locked, while that's not true with the "next" command.  With schedlock
"step", other threads still run unlocked when "next" detects a
function call and steps over it.  Whether that makes sense or not,
still, it's documented that way in the manual.  If another thread hits
an event that doesn't cause a stop while the nexting thread steps over
a function call, we'll get here and fail the assertion.

The fix is just to adjust the assertion.  Even though we found the
stepping thread, we'll still step-over the breakpoint that just
triggered correctly.

Surprisingly, gdb.threads/schedlock.exp doesn't have any test that
steps over a function call.  This commits fixes that.  This ensures
that "next" doesn't switch focus to another thread, and checks whether
other threads run locked or not, depending on scheduler locking mode
and command.  There's a lot of duplication in that file that this ends
cleaning up.  There's more that could be cleaned up, but that would
end up an unrelated change, best done separately.

This new coverage in schedlock.exp happens to trigger the internal
error in question, like so:

 FAIL: gdb.threads/schedlock.exp: schedlock=step: cmd=next: call_function=1: next to increment (1) (GDB internal error)
 FAIL: gdb.threads/schedlock.exp: schedlock=step: cmd=next: call_function=1: next to increment (3) (GDB internal error)
 FAIL: gdb.threads/schedlock.exp: schedlock=step: cmd=next: call_function=1: next to increment (5) (GDB internal error)
 FAIL: gdb.threads/schedlock.exp: schedlock=step: cmd=next: call_function=1: next to increment (7) (GDB internal error)
 FAIL: gdb.threads/schedlock.exp: schedlock=step: cmd=next: call_function=1: next to increment (9) (GDB internal error)
 FAIL: gdb.threads/schedlock.exp: schedlock=step: cmd=next: call_function=1: next does not change thread (switched to thread 0)
 FAIL: gdb.threads/schedlock.exp: schedlock=step: cmd=next: call_function=1: current thread advanced - unlocked (wrong amount)

That's because we have more than one thread running the same loop, and
while one thread is stepping over a function call, the other thread
hits the step-resume breakpoint of the first, which needs to be
stepped over, and we end up in switch_back_to_stepped_thread exactly
in the problem case.

I think a simpler and more directed test is also useful, to not rely
on internal breakpoint magics.  So this commit also adds a test that
has a thread trip on a conditional breakpoint that doesn't cause a
user-visible stop while another thread is stepping over a call.  That
currently fails like this:

 FAIL: gdb.threads/next-bp-other-thread.exp: schedlock=step: next over function call (GDB internal error)

Tested on x86_64 Fedora 20.

gdb/
2014-10-29  Pedro Alves  <palves@redhat.com>

	PR gdb/17408
	* infrun.c (switch_back_to_stepped_thread): Use currently_stepping
	instead of assuming a thread with a stepping range is always
	stepping.

gdb/testsuite/
2014-10-29  Pedro Alves  <palves@redhat.com>

	PR gdb/17408
	* gdb.threads/schedlock.c (some_function): New function.
	(call_function): New global.
	(MAYBE_CALL_SOME_FUNCTION): New macro.
	(thread_function): Call it.
	* gdb.threads/schedlock.exp (get_args): Add description parameter,
	and use it instead of a global counter.  Adjust all callers.
	(get_current_thread): Use "find current thread" for test message
	here rather than having all callers pass down the same string.
	(goto_loop): New procedure, factored out from ...
	(my_continue): ... this.
	(step_ten_loops): Change parameter from test message to command to
	use.  Adjust.
	(list_count): Delete global.
	(check_result): New procedure, factored out from duplicate top
	level code.
	(continue tests): Wrap in with_test_prefix.
	(test_step): New procedure, factored out from duplicate top level
	code.
	(top level): Test "step" in combination with all scheduler-locking
	modes.  Test "next" in combination with all scheduler-locking
	modes, and in combination with stepping over a function call or
	not.
	* gdb.threads/next-bp-other-thread.c: New file.
	* gdb.threads/next-bp-other-thread.exp: New file.
This commit is contained in:
Pedro Alves
2014-10-29 18:15:39 +00:00
parent d3d4baedb6
commit 354204061c
5 changed files with 257 additions and 124 deletions

View File

@@ -5462,7 +5462,7 @@ switch_back_to_stepped_thread (struct execution_control_state *ecs)
stepping, then scheduler locking can't be in effect,
otherwise we wouldn't have resumed the current event
thread in the first place. */
gdb_assert (!schedlock_applies (1));
gdb_assert (!schedlock_applies (currently_stepping (tp)));
stepping_thread = tp;
}

View File

@@ -0,0 +1,45 @@
/* This testcase is part of GDB, the GNU debugger.
Copyright 2014 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 <http://www.gnu.org/licenses/>. */
#include <pthread.h>
#include <unistd.h>
#include <stdlib.h>
/* Always zero, used in breakpoint condition. */
volatile int global_zero;
void *
child_function (void *arg)
{
while (1)
{
usleep (1); /* set breakpoint child here */
}
pthread_exit (NULL);
}
int
main (void)
{
pthread_t child_thread;
int res;
res = pthread_create (&child_thread, NULL, child_function, NULL);
sleep (2); /* set wait-thread breakpoint here */
exit (EXIT_SUCCESS);
}

View File

@@ -0,0 +1,54 @@
# Copyright (C) 2014 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 <http://www.gnu.org/licenses/>.
# Test that GDB behaves correctly when a "next" steps over a call, and
# another thread hits a breakpoint that doesn't cause a user visible
# stop (and so needs to be stepped over). GDB used to trip on an
# invalid assertion - PR17408.
standard_testfile
if {[prepare_for_testing "failed to prepare" $testfile $srcfile {debug pthreads}] == -1} {
return -1
}
# Test all "set scheduler-locking" variants.
foreach schedlock {"off" "step" "on" } {
with_test_prefix "schedlock=$schedlock" {
clean_restart $binfile
if ![runto_main] {
continue
}
gdb_breakpoint [gdb_get_line_number "set wait-thread breakpoint here"]
gdb_continue_to_breakpoint "run to wait-thread breakpoint"
gdb_test "info threads" "2 .*\\\* 1.*" "info threads shows all threads"
delete_breakpoints
gdb_breakpoint [gdb_get_line_number "set breakpoint child here"]
# Give it a condition that always fails.
gdb_test "condition \$bpnum global_zero == 1" ".*"
gdb_test_no_output "set scheduler-locking $schedlock"
# While stepping over the sleep call, the other thread hits a
# breakpoint that doesn't cause a user visible stop (and so
# needs to be stepped over). The next should complete as if
# that breakpoint never triggered.
gdb_test "next" "EXIT_SUCCESS.*" "next over function call"
}
}

View File

@@ -48,6 +48,28 @@ int main() {
exit(EXIT_SUCCESS);
}
void some_function (void) {
/* Sleep a bit to give the other threads a chance to run, if not
locked. This also ensure that even if the compiler optimizes out
or inlines some_function, there's still be some function that
needs to be stepped over. */
usleep (1);
}
/* When testing "next", this is set to have the loop call
some_function, which GDB should step over. When testing "step",
that would step into the function, which is not what we want. */
volatile int call_function = 0;
/* Call some_function if CALL_FUNCTION is set. This is wrapped in a
macro so that it's a single source line in the main loop. */
#define MAYBE_CALL_SOME_FUNCTION() \
do \
{ \
if (call_function) \
some_function (); \
} while (0)
void *thread_function(void *arg) {
int my_number = (long) arg;
int *myp = (int *) &args[my_number];
@@ -56,7 +78,7 @@ void *thread_function(void *arg) {
while (*myp > 0)
{
/* schedlock.exp: main loop. */
(*myp) ++;
MAYBE_CALL_SOME_FUNCTION(); (*myp) ++;
}
pthread_exit(NULL);

View File

@@ -30,8 +30,10 @@ if {[gdb_compile_pthreads "${srcdir}/${subdir}/${srcfile}" "${binfile}" executab
# Now we can proceed with the real testing.
proc get_args { } {
global list_count
# Get the current contents of the `args` array in the test program.
# Description is appended to the test message.
proc get_args { description } {
global gdb_prompt
global NUM
@@ -40,10 +42,10 @@ proc get_args { } {
append pattern ", (\[0-9\]+)"
}
gdb_test_multiple "print args" "listed args ($list_count)" {
set test "listed args ($description)"
gdb_test_multiple "print args" $test {
-re "\\\$\[0-9\]+ = {$pattern}.*$gdb_prompt $" {
set list_count [expr $list_count + 1]
pass "listed args ($list_count)"
pass $test
set result ""
for {set i 1} {[expr $i <= $NUM]} {incr i} {
@@ -75,46 +77,62 @@ proc stop_process { description } {
proc get_current_thread { description } {
global gdb_prompt
gdb_test_multiple "bt" "$description" {
set test "find current thread ($description)"
gdb_test_multiple "bt" $test {
-re "thread_function \\(arg=0x(\[0-9\])\\).*$gdb_prompt $" {
pass $description
pass $test
return $expect_out(1,string)
}
}
return ""
}
# Make sure we're stopped in the loop, in one of the non-main threads.
proc goto_loop { msg } {
gdb_breakpoint [concat [gdb_get_line_number "schedlock.exp: main loop"] " if arg != 0"]
set test "return to loop"
if {$msg != ""} {
set test "$test ($msg)"
}
gdb_continue_to_breakpoint $test
delete_breakpoints
}
proc my_continue { msg } {
gdb_test_multiple "continue" "continuing ($msg)" {
set test "continue ($msg)"
gdb_test_multiple "continue" $test {
-re "Continuing" {
pass "continue ($msg)"
pass $test
}
}
stop_process "stop all threads ($msg)"
# Make sure we're in one of the non-main looping threads.
gdb_breakpoint [concat [gdb_get_line_number "schedlock.exp: main loop"] " if arg != 0"]
gdb_continue_to_breakpoint "return to loop ($msg)"
delete_breakpoints
goto_loop $msg
}
proc step_ten_loops { msg } {
# Use CMD to step the loop 10 times. CMD may be "step" or "next".
proc step_ten_loops { cmd } {
global gdb_prompt
for {set i 0} {[expr $i < 10]} {set i [expr $i + 1]} {
set other_step 0
gdb_test_multiple "step" "step to increment ($msg $i)" {
set test "$cmd to increment ($i)"
gdb_test_multiple $cmd $test {
-re ".*myp\\) \\+\\+;\[\r\n\]+$gdb_prompt $" {
pass "step to increment ($msg $i)"
pass $test
}
-re "$gdb_prompt $" {
if {$other_step == 0} {
set other_step 1
send_gdb "step\n"
send_gdb "$cmd\n"
exp_continue
} else {
fail "step to increment ($msg $i)"
fail $test
# FIXME cascade?
}
}
@@ -158,15 +176,12 @@ gdb_test_multiple "set scheduler-locking off" "scheduler locking set to none" {
gdb_breakpoint [gdb_get_line_number "schedlock.exp: last thread start"]
gdb_continue_to_breakpoint "all threads started"
global list_count
set list_count 0
set start_args [get_args]
set start_args [get_args "before initial"]
# First make sure that all threads are alive.
my_continue "initial"
set cont_args [get_args]
set cont_args [get_args "after initial"]
set bad 0
for {set i 0} {[expr $i < $NUM]} {set i [expr $i + 1]} {
@@ -180,125 +195,122 @@ if { $bad == 0 } {
fail "all threads alive ($bad/$NUM did not run)"
}
# We can't change threads, unfortunately, in current GDB. Use
# whichever we stopped in.
set curthread [get_current_thread "find current thread (1)"]
# Compare the previous thread and args with the current thread and
# args. Check that we didn't switch threads, and that the threads
# incremented their args counter the amounts expected. CMD is the
# command being tested. BEFORE_THREAD is the thread that was selected
# before the command was run. BEFORE_ARGS is the value of the
# thread's args before the command was run. LOCKED indicates whether
# we expect threads other than the selected thread remained locked.
proc check_result { cmd before_thread before_args locked } {
global NUM
# Make sure we're still in the same thread.
set newthread [get_current_thread "after"]
# Test stepping without scheduler locking.
gdb_test_no_output "set scheduler-locking off"
step_ten_loops "unlocked"
# Make sure we're still in the same thread.
set newthread [get_current_thread "find current thread (2)"]
if {$curthread == $newthread} {
pass "step without lock does not change thread"
} else {
fail "step without lock does not change thread (switched to thread $newthread)"
}
set start_args $cont_args
set cont_args [get_args]
set num_other_threads 0
for {set i 0} {[expr $i < $NUM]} {set i [expr $i + 1]} {
if {[lindex $start_args $i] == [lindex $cont_args $i]} {
if {$i == $curthread} {
fail "current thread stepped (didn't run)"
set test "$cmd does not change thread"
if {$before_thread == $newthread} {
pass "$test"
} else {
fail "$test (switched to thread $newthread)"
}
} else {
if {$i == $curthread} {
if {[lindex $start_args $i] == [expr [lindex $cont_args $i] - 10]} {
pass "current thread stepped"
set after_args [get_args "after"]
set test "current thread advanced"
if { $locked } {
set test "$test - locked"
} else {
set test "$test - unlocked"
}
set num_other_threads 0
for {set i 0} {$i < $NUM} {incr i} {
if {[lindex $before_args $i] == [lindex $after_args $i]} {
if {$i == $before_thread} {
fail "$test (didn't run)"
}
} else {
fail "current thread stepped (wrong amount)"
if {$i == $before_thread} {
if {$cmd == "continue"
|| [lindex $before_args $i] == [expr [lindex $after_args $i] - 10]} {
pass "$test"
} else {
fail "$test (wrong amount)"
}
} else {
incr num_other_threads
}
}
}
if { $locked } {
gdb_assert {$num_other_threads == 0} "other threads didn't run - locked"
} else {
set num_other_threads [expr $num_other_threads + 1]
gdb_assert {$num_other_threads > 0} "other threads ran - unlocked"
}
}
}
if {$num_other_threads > 0} {
pass "other threads ran - unlocked"
} else {
fail "other threads ran - unlocked"
}
# Test continue with scheduler locking
gdb_test "set scheduler-locking on" ""
with_test_prefix "schedlock=on: cmd=continue" {
# Use whichever we stopped in.
set curthread [get_current_thread "before"]
my_continue "with lock"
# Test continue with scheduler locking.
gdb_test "set scheduler-locking on" ""
# Make sure we're still in the same thread.
set newthread [get_current_thread "find current thread (3)"]
if {$curthread == $newthread} {
pass "continue with lock does not change thread"
} else {
fail "continue with lock does not change thread (switched to thread $newthread)"
my_continue "with lock"
check_result "continue" $curthread $cont_args 1
}
set start_args $cont_args
set cont_args [get_args]
# Test stepping/nexting with different modes of scheduler locking.
proc test_step { schedlock cmd call_function } {
global NUM
set num_other_threads 0
for {set i 0} {[expr $i < $NUM]} {set i [expr $i + 1]} {
if {[lindex $start_args $i] == [lindex $cont_args $i]} {
if {$i == $curthread} {
fail "current thread ran (didn't run)"
gdb_test_no_output "set scheduler-locking off"
goto_loop ""
set curthread [get_current_thread "before"]
# No need to set to off again. This avoids a duplicate message.
if {$schedlock != "off"} {
gdb_test_no_output "set scheduler-locking $schedlock"
}
} else {
if {$i == $curthread} {
pass "current thread ran"
gdb_test "print call_function = $call_function" \
" = $call_function"
set before_args [get_args "before"]
step_ten_loops $cmd
# "next" lets other threads run while stepping over functions.
if { $schedlock == "on" || ($schedlock == "step" && !$call_function) } {
set locked 1
} else {
incr num_other_threads
set locked 0
}
}
}
if {$num_other_threads > 0} {
fail "other threads didn't run - locked"
} else {
pass "other threads didn't run - locked"
check_result $cmd $curthread $before_args $locked
}
# Test stepping with scheduler locking
step_ten_loops "locked"
# Make sure we're still in the same thread.
set newthread [get_current_thread "find current thread (2)"]
if {$curthread == $newthread} {
pass "step with lock does not change thread"
} else {
fail "step with lock does not change thread (switched to thread $newthread)"
}
set start_args $cont_args
set cont_args [get_args]
set num_other_threads 0
for {set i 0} {[expr $i < $NUM]} {set i [expr $i + 1]} {
if {[lindex $start_args $i] == [lindex $cont_args $i]} {
if {$i == $curthread} {
fail "current thread stepped locked (didn't run)"
}
} else {
if {$i == $curthread} {
if {[lindex $start_args $i] == [expr [lindex $cont_args $i] - 10]} {
pass "current thread stepped locked"
} else {
fail "current thread stepped locked (wrong amount)"
# Test stepping/nexting with different modes of scheduler locking.
foreach schedlock {"off" "step" "on"} {
with_test_prefix "schedlock=$schedlock" {
with_test_prefix "cmd=step" {
test_step $schedlock "step" 0
}
with_test_prefix "cmd=next" {
# With "next", and schedlock "step", threads run unlocked
# when stepping over a function call. This exercises both
# with and without a function call. Without a function
# call "next" should behave just like "step".
foreach call_function {0 1} {
with_test_prefix "call_function=$call_function" {
test_step $schedlock "next" $call_function
}
}
}
} else {
incr num_other_threads
}
}
}
if {$num_other_threads > 0} {
fail "other threads didn't run - step locked"
} else {
pass "other threads didn't run - step locked"
}
return 0