* [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior
@ 2020-01-29 14:11 Tom de Vries (Code Review)
2020-01-29 15:33 ` Mihails Strasuns (Code Review)
` (12 more replies)
0 siblings, 13 replies; 14+ messages in thread
From: Tom de Vries (Code Review) @ 2020-01-29 14:11 UTC (permalink / raw)
To: gdb-patches
Change URL: https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/759
......................................................................
[gdb/threads] Fix hang in stop_all_threads after killing inferior
Consider a two-threaded testcase a.out, sleeping in both its threads:
...
$ gdb -ex r --args a.out
Reading symbols from a.out...
Starting program: /data/gdb_versions/devel/a.out
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
[New Thread 0x7ffff77fe700 (LWP 31268)]
...
Typing ^C causing stop_all_threads to be executed, and if an external SIGKILL
(such as caused by killall -9 a.out) arrives at the start of stop_all_threads,
gdb hangs in stop_all_threads after giving this warning:
...
warning: unable to open /proc file '/proc/24938/status'
...
Using "set debug infrun 1" we can see in more detail where we hang:
...
infrun: stop_all_threads
infrun: stop_all_threads, pass=0, iterations=0
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, need stop
infrun: target_wait (-1.0.0, status) =
infrun: 10264.10268.0 [Thread 0x7ffff77fe700 (LWP 10268)],
infrun: status->kind = signalled, signal = GDB_SIGNAL_KILL
infrun: stop_all_threads status->kind = signalled, signal = GDB_SIGNAL_KILL \
Thread 0x7ffff77fe700 (LWP 10268)
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, already stopping
warning: unable to open /proc file '/proc/10264/status'
infrun: target_wait (-1.0.0, status) =
infrun: -1.0.0 [process -1],
infrun: status->kind = no-resumed
infrun: infrun_async(0)
infrun: stop_all_threads status->kind = no-resumed process -1
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, already stopping
infrun: stop_all_threads status->kind = no-resumed process -1
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, already stopping
infrun: stop_all_threads status->kind = no-resumed process -1
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, already stopping
<repeat>
......
So, we're hanging in the 'while (1)' loop in stop_all_threads as follows:
- thread t is tested, and both t->executing and t->stop_requested are found
to be 1
- consequently need_wait is set 1
- consequently wait_one is executed
- wait_one returns a TARGET_WAITKIND_NO_RESUMED event, which is handled by
continuing at the start of the loop
The loop actually starts with update_thread_list (), but that doesn't seem
to change the state of the threads.
Fix the hang by detecting the first sign of trouble: the
TARGET_WAITKIND_SIGNALLED event with signal GDB_SIGNAL_KILL, and breaking out
of the loop.
Build and reg-tested on x86_64-linux.
gdb/ChangeLog:
2020-01-29 Tom de Vries <tdevries@suse.de>
PR threads/25478
* infrun.c (stop_all_threads): Return when detecting event
TARGET_WAITKIND_SIGNALLED with signal GDB_SIGNAL_KILL.
Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
---
M gdb/infrun.c
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/gdb/infrun.c b/gdb/infrun.c
index 22de42c..e34ddc8 100644
--- a/gdb/infrun.c
+++ b/gdb/infrun.c
@@ -4772,7 +4772,10 @@
target_pid_to_str (event.ptid).c_str ());
}
- if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
+ if (event.ws.kind == TARGET_WAITKIND_SIGNALLED
+ && event.ws.value.sig == GDB_SIGNAL_KILL)
+ goto done;
+ else if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
|| event.ws.kind == TARGET_WAITKIND_THREAD_EXITED
|| event.ws.kind == TARGET_WAITKIND_EXITED
|| event.ws.kind == TARGET_WAITKIND_SIGNALLED)
@@ -4872,6 +4875,7 @@
}
}
+ done:
if (debug_infrun)
fprintf_unfiltered (gdb_stdlog, "infrun: stop_all_threads done\n");
}
--
Gerrit-Project: binutils-gdb
Gerrit-Branch: master
Gerrit-Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
Gerrit-Change-Number: 759
Gerrit-PatchSet: 1
Gerrit-Owner: Tom de Vries <tdevries@suse.de>
Gerrit-MessageType: newchange
^ permalink raw reply [flat|nested] 14+ messages in thread
* [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior
2020-01-29 14:11 [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior Tom de Vries (Code Review)
@ 2020-01-29 15:33 ` Mihails Strasuns (Code Review)
2020-01-29 16:13 ` Tom de Vries (Code Review)
` (11 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Mihails Strasuns (Code Review) @ 2020-01-29 15:33 UTC (permalink / raw)
To: Tom de Vries, gdb-patches; +Cc: Pedro Alves
Mihails Strasuns has posted comments on this change.
Change URL: https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/759
......................................................................
Patch Set 1:
This seems very similar to https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/133 - I don't think there should be any logic specific to SIGKILL here. It seems like a general race condition for early termination while `stop_all_threads` is being executed.
--
Gerrit-Project: binutils-gdb
Gerrit-Branch: master
Gerrit-Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
Gerrit-Change-Number: 759
Gerrit-PatchSet: 1
Gerrit-Owner: Tom de Vries <tdevries@suse.de>
Gerrit-Reviewer: Pedro Alves <palves@redhat.com>
Gerrit-Reviewer: Tom de Vries <tdevries@suse.de>
Gerrit-CC: Mihails Strasuns <mihails.strasuns@intel.com>
Gerrit-Comment-Date: Wed, 29 Jan 2020 14:39:15 +0000
Gerrit-HasComments: No
Gerrit-Has-Labels: No
Gerrit-MessageType: comment
^ permalink raw reply [flat|nested] 14+ messages in thread
* [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior
2020-01-29 14:11 [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior Tom de Vries (Code Review)
2020-01-29 15:33 ` Mihails Strasuns (Code Review)
@ 2020-01-29 16:13 ` Tom de Vries (Code Review)
2020-01-29 16:20 ` Tankut Baris Aktemur (Code Review)
` (10 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Tom de Vries (Code Review) @ 2020-01-29 16:13 UTC (permalink / raw)
To: gdb-patches; +Cc: Mihails Strasuns, Pedro Alves
Tom de Vries has posted comments on this change.
Change URL: https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/759
......................................................................
Patch Set 1:
> Patch Set 1:
>
> This seems very similar to https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/133
> - I don't think there should be any logic specific to SIGKILL here. It seems like a
> general race condition for early termination while `stop_all_threads` is being executed.
Thanks for noticing, that's useful.
FWIW, I looked over the review comments there, and noticed the suggestion to "leave the TARGET_WAITKIND_EXITED/TARGET_WAITKIND_SIGNALLED event pending", which I tried using this additional patch:
...
diff --git a/gdb/infrun.c b/gdb/infrun.c
index e34ddc83b45..c1035c25d7f 100644
--- a/gdb/infrun.c
+++ b/gdb/infrun.c
@@ -4774,7 +4774,11 @@ stop_all_threads (void)
if (event.ws.kind == TARGET_WAITKIND_SIGNALLED
&& event.ws.value.sig == GDB_SIGNAL_KILL)
- goto done;
+ {
+ thread_info *t = find_thread_ptid (event.target, event.ptid);
+ save_waitstatus (t, &event.ws);
+ goto done;
+ }
else if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
|| event.ws.kind == TARGET_WAITKIND_THREAD_EXITED
|| event.ws.kind == TARGET_WAITKIND_EXITED
...
But I didn't notice any difference in behaviour.
--
Gerrit-Project: binutils-gdb
Gerrit-Branch: master
Gerrit-Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
Gerrit-Change-Number: 759
Gerrit-PatchSet: 1
Gerrit-Owner: Tom de Vries <tdevries@suse.de>
Gerrit-Reviewer: Pedro Alves <palves@redhat.com>
Gerrit-Reviewer: Tom de Vries <tdevries@suse.de>
Gerrit-CC: Mihails Strasuns <mihails.strasuns@intel.com>
Gerrit-Comment-Date: Wed, 29 Jan 2020 16:02:12 +0000
Gerrit-HasComments: No
Gerrit-Has-Labels: No
Gerrit-MessageType: comment
^ permalink raw reply [flat|nested] 14+ messages in thread
* [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior
2020-01-29 14:11 [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior Tom de Vries (Code Review)
2020-01-29 15:33 ` Mihails Strasuns (Code Review)
2020-01-29 16:13 ` Tom de Vries (Code Review)
@ 2020-01-29 16:20 ` Tankut Baris Aktemur (Code Review)
2020-01-30 14:55 ` Tom de Vries (Code Review)
` (9 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Tankut Baris Aktemur (Code Review) @ 2020-01-29 16:20 UTC (permalink / raw)
To: Tom de Vries, gdb-patches; +Cc: Mihails Strasuns, Pedro Alves
Tankut Baris Aktemur has posted comments on this change.
Change URL: https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/759
......................................................................
Patch Set 1:
> FWIW, I looked over the review comments there, and noticed the suggestion to "leave the TARGET_WAITKIND_EXITED/TARGET_WAITKIND_SIGNALLED event pending"
For the continuation of the discussion, please see
https://sourceware.org/ml/gdb-patches/2020-01/msg00212.html
--
Gerrit-Project: binutils-gdb
Gerrit-Branch: master
Gerrit-Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
Gerrit-Change-Number: 759
Gerrit-PatchSet: 1
Gerrit-Owner: Tom de Vries <tdevries@suse.de>
Gerrit-Reviewer: Pedro Alves <palves@redhat.com>
Gerrit-Reviewer: Tom de Vries <tdevries@suse.de>
Gerrit-CC: Mihails Strasuns <mihails.strasuns@intel.com>
Gerrit-CC: Tankut Baris Aktemur <tankut.baris.aktemur@intel.com>
Gerrit-Comment-Date: Wed, 29 Jan 2020 16:13:34 +0000
Gerrit-HasComments: No
Gerrit-Has-Labels: No
Gerrit-MessageType: comment
^ permalink raw reply [flat|nested] 14+ messages in thread
* [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior
2020-01-29 14:11 [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior Tom de Vries (Code Review)
` (2 preceding siblings ...)
2020-01-29 16:20 ` Tankut Baris Aktemur (Code Review)
@ 2020-01-30 14:55 ` Tom de Vries (Code Review)
2020-01-30 16:30 ` [review v2] " Tom de Vries (Code Review)
` (8 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Tom de Vries (Code Review) @ 2020-01-30 14:55 UTC (permalink / raw)
To: gdb-patches; +Cc: Tankut Baris Aktemur, Mihails Strasuns, Pedro Alves
Tom de Vries has posted comments on this change.
Change URL: https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/759
......................................................................
Patch Set 1:
> FWIW, I looked over the review comments there, and noticed the suggestion to "leave the TARGET_WAITKIND_EXITED/TARGET_WAITKIND_SIGNALLED event pending", which I tried using this additional patch:
> ...
> diff --git a/gdb/infrun.c b/gdb/infrun.c
> index e34ddc83b45..c1035c25d7f 100644
> --- a/gdb/infrun.c
> +++ b/gdb/infrun.c
> @@ -4774,7 +4774,11 @@ stop_all_threads (void)
>
> if (event.ws.kind == TARGET_WAITKIND_SIGNALLED
> && event.ws.value.sig == GDB_SIGNAL_KILL)
> - goto done;
> + {
> + thread_info *t = find_thread_ptid (event.target, event.ptid);
> + save_waitstatus (t, &event.ws);
> + goto done;
> + }
> else if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
> || event.ws.kind == TARGET_WAITKIND_THREAD_EXITED
> || event.ws.kind == TARGET_WAITKIND_EXITED
> ...
>
> But I didn't notice any difference in behaviour.
I've realized from another use of save_waitstatus that I have to set t->resumed in order to have the saved wait status noticed.
With this additional patch, I did manage to observe a difference in behaviour:
...
@@ -4777,6 +4777,7 @@ stop_all_threads (void)
{
thread_info *t = find_thread_ptid (event.target, event.ptid);
save_waitstatus (t, &event.ws);
+ t->resumed = 1;
goto done;
}
else if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
...
The difference is in what happens when I continue after the ^C is handled:
- without the two patches gdb hangs, until I press enter, after which I get the
prompt
- with the two patches, gdb just presents the prompt
--
Gerrit-Project: binutils-gdb
Gerrit-Branch: master
Gerrit-Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
Gerrit-Change-Number: 759
Gerrit-PatchSet: 1
Gerrit-Owner: Tom de Vries <tdevries@suse.de>
Gerrit-Reviewer: Pedro Alves <palves@redhat.com>
Gerrit-Reviewer: Tom de Vries <tdevries@suse.de>
Gerrit-CC: Mihails Strasuns <mihails.strasuns@intel.com>
Gerrit-CC: Tankut Baris Aktemur <tankut.baris.aktemur@intel.com>
Gerrit-Comment-Date: Thu, 30 Jan 2020 14:52:10 +0000
Gerrit-HasComments: No
Gerrit-Has-Labels: No
Gerrit-MessageType: comment
^ permalink raw reply [flat|nested] 14+ messages in thread
* [review v2] [gdb/threads] Fix hang in stop_all_threads after killing inferior
2020-01-29 14:11 [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior Tom de Vries (Code Review)
` (3 preceding siblings ...)
2020-01-30 14:55 ` Tom de Vries (Code Review)
@ 2020-01-30 16:30 ` Tom de Vries (Code Review)
2020-01-30 17:36 ` Tankut Baris Aktemur (Code Review)
` (7 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Tom de Vries (Code Review) @ 2020-01-30 16:30 UTC (permalink / raw)
To: Pedro Alves, gdb-patches; +Cc: Mihails Strasuns, Tankut Baris Aktemur
Change URL: https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/759
......................................................................
[gdb/threads] Fix hang in stop_all_threads after killing inferior
Consider a two-threaded testcase a.out, sleeping in both its threads:
...
$ gdb -ex r --args a.out
Reading symbols from a.out...
Starting program: /data/gdb_versions/devel/a.out
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
[New Thread 0x7ffff77fe700 (LWP 31268)]
...
Typing ^C causes stop_all_threads to be executed, and if an external SIGKILL
(such as caused by killall -9 a.out) arrives at the start of stop_all_threads,
gdb hangs in stop_all_threads after giving this warning:
...
warning: unable to open /proc file '/proc/24938/status'
...
Using "set debug infrun 1" we can see in more detail where we hang:
...
infrun: stop_all_threads
infrun: stop_all_threads, pass=0, iterations=0
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, need stop
infrun: target_wait (-1.0.0, status) =
infrun: 10264.10268.0 [Thread 0x7ffff77fe700 (LWP 10268)],
infrun: status->kind = signalled, signal = GDB_SIGNAL_KILL
infrun: stop_all_threads status->kind = signalled, signal = GDB_SIGNAL_KILL \
Thread 0x7ffff77fe700 (LWP 10268)
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, already stopping
warning: unable to open /proc file '/proc/10264/status'
infrun: target_wait (-1.0.0, status) =
infrun: -1.0.0 [process -1],
infrun: status->kind = no-resumed
infrun: infrun_async(0)
infrun: stop_all_threads status->kind = no-resumed process -1
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, already stopping
infrun: stop_all_threads status->kind = no-resumed process -1
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, already stopping
infrun: stop_all_threads status->kind = no-resumed process -1
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, already stopping
<repeat>
......
So, we're hanging in the 'while (1)' loop in stop_all_threads as follows:
- thread t is tested, and both t->executing and t->stop_requested are found
to be 1 (noted with 'executing, already stopping')
- consequently need_wait is set 1
- consequently wait_one is executed
- wait_one returns a TARGET_WAITKIND_NO_RESUMED event, which is handled by
continuing at the start of the loop
The loop actually starts with update_thread_list (), but that doesn't seem
to change the state of the threads.
Fix the hang by:
- detecting the first sign of trouble: the TARGET_WAITKIND_SIGNALLED event
with signal GDB_SIGNAL_KILL,
- making that event pending again, and
- breaking out of the loop.
This results in the ^C being handled without showing the user that the
test-case was killed:
...
^C
Thread 1 received signal SIGINT, Interrupt.
0x00007ffff78c50f0 in nanosleep () from /lib64/libc.so.6
(gdb)
...
But a subsequent continue does show that:
...
(gdb) c
Continuing.
Program terminated with signal SIGKILL, Killed.
The program no longer exists.
(gdb)
....
Build and reg-tested on x86_64-linux.
gdb/ChangeLog:
2020-01-29 Tom de Vries <tdevries@suse.de>
PR threads/25478
* infrun.c (stop_all_threads): Detecting event
TARGET_WAITKIND_SIGNALLED with signal GDB_SIGNAL_KILL, make event
pending again and return.
Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
---
M gdb/infrun.c
1 file changed, 10 insertions(+), 1 deletion(-)
diff --git a/gdb/infrun.c b/gdb/infrun.c
index 22de42c..2fd1066 100644
--- a/gdb/infrun.c
+++ b/gdb/infrun.c
@@ -4772,7 +4772,15 @@
target_pid_to_str (event.ptid).c_str ());
}
- if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
+ if (event.ws.kind == TARGET_WAITKIND_SIGNALLED
+ && event.ws.value.sig == GDB_SIGNAL_KILL)
+ {
+ thread_info *t = find_thread_ptid (event.target, event.ptid);
+ save_waitstatus (t, &event.ws);
+ t->resumed = 1;
+ goto done;
+ }
+ else if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
|| event.ws.kind == TARGET_WAITKIND_THREAD_EXITED
|| event.ws.kind == TARGET_WAITKIND_EXITED
|| event.ws.kind == TARGET_WAITKIND_SIGNALLED)
@@ -4872,6 +4880,7 @@
}
}
+ done:
if (debug_infrun)
fprintf_unfiltered (gdb_stdlog, "infrun: stop_all_threads done\n");
}
--
Gerrit-Project: binutils-gdb
Gerrit-Branch: master
Gerrit-Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
Gerrit-Change-Number: 759
Gerrit-PatchSet: 2
Gerrit-Owner: Tom de Vries <tdevries@suse.de>
Gerrit-Reviewer: Pedro Alves <palves@redhat.com>
Gerrit-Reviewer: Tom de Vries <tdevries@suse.de>
Gerrit-CC: Mihails Strasuns <mihails.strasuns@intel.com>
Gerrit-CC: Tankut Baris Aktemur <tankut.baris.aktemur@intel.com>
Gerrit-MessageType: newpatchset
^ permalink raw reply [flat|nested] 14+ messages in thread
* [review v2] [gdb/threads] Fix hang in stop_all_threads after killing inferior
2020-01-29 14:11 [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior Tom de Vries (Code Review)
` (4 preceding siblings ...)
2020-01-30 16:30 ` [review v2] " Tom de Vries (Code Review)
@ 2020-01-30 17:36 ` Tankut Baris Aktemur (Code Review)
2020-01-30 17:42 ` Tom de Vries (Code Review)
` (6 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Tankut Baris Aktemur (Code Review) @ 2020-01-30 17:36 UTC (permalink / raw)
To: Tom de Vries, gdb-patches; +Cc: Mihails Strasuns, Pedro Alves
Tankut Baris Aktemur has posted comments on this change.
Change URL: https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/759
......................................................................
Patch Set 2:
(1 comment)
| --- gdb/infrun.c
| +++ gdb/infrun.c
| @@ -4767,11 +4767,19 @@ stop_all_threads (void)
| if (debug_infrun)
| {
| fprintf_unfiltered (gdb_stdlog,
| "infrun: stop_all_threads %s %s\n",
| target_waitstatus_to_string (&event.ws).c_str (),
| target_pid_to_str (event.ptid).c_str ());
| }
|
| - if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
| + if (event.ws.kind == TARGET_WAITKIND_SIGNALLED
| + && event.ws.value.sig == GDB_SIGNAL_KILL)
PS2, Line 4776:
Why only this case? Wouldn't the same problem be seen if, say, a
different signal or an exited status was received?
| + {
| + thread_info *t = find_thread_ptid (event.target, event.ptid);
| + save_waitstatus (t, &event.ws);
| + t->resumed = 1;
| + goto done;
| + }
| + else if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
| || event.ws.kind == TARGET_WAITKIND_THREAD_EXITED
| || event.ws.kind == TARGET_WAITKIND_EXITED
--
Gerrit-Project: binutils-gdb
Gerrit-Branch: master
Gerrit-Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
Gerrit-Change-Number: 759
Gerrit-PatchSet: 2
Gerrit-Owner: Tom de Vries <tdevries@suse.de>
Gerrit-Reviewer: Pedro Alves <palves@redhat.com>
Gerrit-Reviewer: Tom de Vries <tdevries@suse.de>
Gerrit-CC: Mihails Strasuns <mihails.strasuns@intel.com>
Gerrit-CC: Tankut Baris Aktemur <tankut.baris.aktemur@intel.com>
Gerrit-Comment-Date: Thu, 30 Jan 2020 16:41:03 +0000
Gerrit-HasComments: Yes
Gerrit-Has-Labels: No
Gerrit-MessageType: comment
^ permalink raw reply [flat|nested] 14+ messages in thread
* [review v2] [gdb/threads] Fix hang in stop_all_threads after killing inferior
2020-01-29 14:11 [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior Tom de Vries (Code Review)
` (5 preceding siblings ...)
2020-01-30 17:36 ` Tankut Baris Aktemur (Code Review)
@ 2020-01-30 17:42 ` Tom de Vries (Code Review)
2020-01-30 17:43 ` Tom de Vries (Code Review)
` (5 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Tom de Vries (Code Review) @ 2020-01-30 17:42 UTC (permalink / raw)
To: gdb-patches; +Cc: Tankut Baris Aktemur, Mihails Strasuns, Pedro Alves
Tom de Vries has posted comments on this change.
Change URL: https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/759
......................................................................
Patch Set 2:
(1 comment)
| --- gdb/infrun.c
| +++ gdb/infrun.c
| @@ -4767,11 +4767,19 @@ stop_all_threads (void)
| if (debug_infrun)
| {
| fprintf_unfiltered (gdb_stdlog,
| "infrun: stop_all_threads %s %s\n",
| target_waitstatus_to_string (&event.ws).c_str (),
| target_pid_to_str (event.ptid).c_str ());
| }
|
| - if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
| + if (event.ws.kind == TARGET_WAITKIND_SIGNALLED
| + && event.ws.value.sig == GDB_SIGNAL_KILL)
PS2, Line 4776:
For now I'm just trying to solve the PR in the absolute minimal way.
| + {
| + thread_info *t = find_thread_ptid (event.target, event.ptid);
| + save_waitstatus (t, &event.ws);
| + t->resumed = 1;
| + goto done;
| + }
| + else if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
| || event.ws.kind == TARGET_WAITKIND_THREAD_EXITED
| || event.ws.kind == TARGET_WAITKIND_EXITED
--
Gerrit-Project: binutils-gdb
Gerrit-Branch: master
Gerrit-Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
Gerrit-Change-Number: 759
Gerrit-PatchSet: 2
Gerrit-Owner: Tom de Vries <tdevries@suse.de>
Gerrit-Reviewer: Pedro Alves <palves@redhat.com>
Gerrit-Reviewer: Tom de Vries <tdevries@suse.de>
Gerrit-CC: Mihails Strasuns <mihails.strasuns@intel.com>
Gerrit-CC: Tankut Baris Aktemur <tankut.baris.aktemur@intel.com>
Gerrit-Comment-Date: Thu, 30 Jan 2020 17:35:54 +0000
Gerrit-HasComments: Yes
Gerrit-Has-Labels: No
Comment-In-Reply-To: Tankut Baris Aktemur <tankut.baris.aktemur@intel.com>
Gerrit-MessageType: comment
^ permalink raw reply [flat|nested] 14+ messages in thread
* [review v2] [gdb/threads] Fix hang in stop_all_threads after killing inferior
2020-01-29 14:11 [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior Tom de Vries (Code Review)
` (6 preceding siblings ...)
2020-01-30 17:42 ` Tom de Vries (Code Review)
@ 2020-01-30 17:43 ` Tom de Vries (Code Review)
2020-01-30 17:52 ` [review v3] " Tom de Vries (Code Review)
` (4 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Tom de Vries (Code Review) @ 2020-01-30 17:43 UTC (permalink / raw)
To: gdb-patches; +Cc: Tankut Baris Aktemur, Mihails Strasuns, Pedro Alves
Tom de Vries has posted comments on this change.
Change URL: https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/759
......................................................................
Patch Set 2:
(1 comment)
| --- gdb/infrun.c
| +++ gdb/infrun.c
| @@ -4773,10 +4773,18 @@ stop_all_threads (void)
| }
|
| - if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
| + if (event.ws.kind == TARGET_WAITKIND_SIGNALLED
| + && event.ws.value.sig == GDB_SIGNAL_KILL)
| + {
| + thread_info *t = find_thread_ptid (event.target, event.ptid);
| + save_waitstatus (t, &event.ws);
| + t->resumed = 1;
| + goto done;
PS2, Line 4781:
I've realized that the 'goto done' is not a good idea, since
all_non_exited_threads can iterate over more than one inferior.
I'll upload a new version that removes the 'goto done.
| + }
| + else if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
| || event.ws.kind == TARGET_WAITKIND_THREAD_EXITED
| || event.ws.kind == TARGET_WAITKIND_EXITED
| || event.ws.kind == TARGET_WAITKIND_SIGNALLED)
| {
| /* All resumed threads exited
| or one thread/process exited/signalled. */
| }
--
Gerrit-Project: binutils-gdb
Gerrit-Branch: master
Gerrit-Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
Gerrit-Change-Number: 759
Gerrit-PatchSet: 2
Gerrit-Owner: Tom de Vries <tdevries@suse.de>
Gerrit-Reviewer: Pedro Alves <palves@redhat.com>
Gerrit-Reviewer: Tom de Vries <tdevries@suse.de>
Gerrit-CC: Mihails Strasuns <mihails.strasuns@intel.com>
Gerrit-CC: Tankut Baris Aktemur <tankut.baris.aktemur@intel.com>
Gerrit-Comment-Date: Thu, 30 Jan 2020 17:42:19 +0000
Gerrit-HasComments: Yes
Gerrit-Has-Labels: No
Gerrit-MessageType: comment
^ permalink raw reply [flat|nested] 14+ messages in thread
* [review v3] [gdb/threads] Fix hang in stop_all_threads after killing inferior
2020-01-29 14:11 [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior Tom de Vries (Code Review)
` (7 preceding siblings ...)
2020-01-30 17:43 ` Tom de Vries (Code Review)
@ 2020-01-30 17:52 ` Tom de Vries (Code Review)
2020-02-03 14:30 ` Tankut Baris Aktemur (Code Review)
` (3 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Tom de Vries (Code Review) @ 2020-01-30 17:52 UTC (permalink / raw)
To: Pedro Alves, gdb-patches; +Cc: Mihails Strasuns, Tankut Baris Aktemur
Change URL: https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/759
......................................................................
[gdb/threads] Fix hang in stop_all_threads after killing inferior
Consider a two-threaded testcase a.out, sleeping in both its threads:
...
$ gdb -ex r --args a.out
Reading symbols from a.out...
Starting program: /data/gdb_versions/devel/a.out
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
[New Thread 0x7ffff77fe700 (LWP 31268)]
...
Typing ^C causes stop_all_threads to be executed, and if an external SIGKILL
(such as caused by killall -9 a.out) arrives at the start of stop_all_threads,
gdb hangs in stop_all_threads after giving this warning:
...
warning: unable to open /proc file '/proc/24938/status'
...
Using "set debug infrun 1" we can see in more detail where we hang:
...
infrun: stop_all_threads
infrun: stop_all_threads, pass=0, iterations=0
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, need stop
infrun: target_wait (-1.0.0, status) =
infrun: 10264.10268.0 [Thread 0x7ffff77fe700 (LWP 10268)],
infrun: status->kind = signalled, signal = GDB_SIGNAL_KILL
infrun: stop_all_threads status->kind = signalled, signal = GDB_SIGNAL_KILL \
Thread 0x7ffff77fe700 (LWP 10268)
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, already stopping
warning: unable to open /proc file '/proc/10264/status'
infrun: target_wait (-1.0.0, status) =
infrun: -1.0.0 [process -1],
infrun: status->kind = no-resumed
infrun: infrun_async(0)
infrun: stop_all_threads status->kind = no-resumed process -1
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, already stopping
infrun: stop_all_threads status->kind = no-resumed process -1
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, already stopping
infrun: stop_all_threads status->kind = no-resumed process -1
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, already stopping
<repeat>
......
So, we're hanging in the 'while (1)' loop in stop_all_threads as follows:
- thread t is tested, and both t->executing and t->stop_requested are found
to be 1 (noted with 'executing, already stopping')
- consequently need_wait is set 1
- consequently wait_one is executed
- wait_one returns a TARGET_WAITKIND_NO_RESUMED event, which is handled by
continuing at the start of the loop
The loop actually starts with update_thread_list (), but that doesn't seem
to change the state of the threads.
Fix the hang by:
- detecting the first sign of trouble: the TARGET_WAITKIND_SIGNALLED event
with signal GDB_SIGNAL_KILL,
- making that event pending again,
- making sure the corresponding thread will not set need_wait again
(by setting t->executing == 0)
- making sure that the corresponding thread keeps t->resumed == 1 in the
the all_non_exited_threads loop
This results in the ^C being handled without showing the user that the
test-case was killed:
...
^C
Thread 1 received signal SIGINT, Interrupt.
0x00007ffff78c50f0 in nanosleep () from /lib64/libc.so.6
(gdb)
...
But a subsequent continue does show that:
...
(gdb) c
Continuing.
Program terminated with signal SIGKILL, Killed.
The program no longer exists.
(gdb)
....
Build and reg-tested on x86_64-linux.
gdb/ChangeLog:
2020-01-29 Tom de Vries <tdevries@suse.de>
PR threads/25478
* infrun.c (stop_all_threads): Detecting event
TARGET_WAITKIND_SIGNALLED with signal GDB_SIGNAL_KILL, make event
pending again, set t->executing to 0 and keep t->resumed set to 1.
Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
---
M gdb/infrun.c
1 file changed, 15 insertions(+), 2 deletions(-)
diff --git a/gdb/infrun.c b/gdb/infrun.c
index 22de42c..9587072 100644
--- a/gdb/infrun.c
+++ b/gdb/infrun.c
@@ -4749,7 +4749,12 @@
/* The thread may be not executing, but still be
resumed with a pending status to process. */
- t->resumed = 0;
+ if (t->suspend.waitstatus.kind == TARGET_WAITKIND_SIGNALLED
+ && t->suspend.waitstatus.value.sig == GDB_SIGNAL_KILL
+ && t->suspend.waitstatus_pending_p)
+ ;
+ else
+ t->resumed = 0;
}
}
@@ -4772,7 +4777,15 @@
target_pid_to_str (event.ptid).c_str ());
}
- if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
+ if (event.ws.kind == TARGET_WAITKIND_SIGNALLED
+ && event.ws.value.sig == GDB_SIGNAL_KILL)
+ {
+ thread_info *t = find_thread_ptid (event.target, event.ptid);
+ save_waitstatus (t, &event.ws);
+ t->resumed = 1;
+ t->executing = 0;
+ }
+ else if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
|| event.ws.kind == TARGET_WAITKIND_THREAD_EXITED
|| event.ws.kind == TARGET_WAITKIND_EXITED
|| event.ws.kind == TARGET_WAITKIND_SIGNALLED)
--
Gerrit-Project: binutils-gdb
Gerrit-Branch: master
Gerrit-Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
Gerrit-Change-Number: 759
Gerrit-PatchSet: 3
Gerrit-Owner: Tom de Vries <tdevries@suse.de>
Gerrit-Reviewer: Pedro Alves <palves@redhat.com>
Gerrit-Reviewer: Tom de Vries <tdevries@suse.de>
Gerrit-CC: Mihails Strasuns <mihails.strasuns@intel.com>
Gerrit-CC: Tankut Baris Aktemur <tankut.baris.aktemur@intel.com>
Gerrit-MessageType: newpatchset
^ permalink raw reply [flat|nested] 14+ messages in thread
* [review v3] [gdb/threads] Fix hang in stop_all_threads after killing inferior
2020-01-29 14:11 [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior Tom de Vries (Code Review)
` (8 preceding siblings ...)
2020-01-30 17:52 ` [review v3] " Tom de Vries (Code Review)
@ 2020-02-03 14:30 ` Tankut Baris Aktemur (Code Review)
2020-02-03 15:20 ` Tom de Vries (Code Review)
` (2 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Tankut Baris Aktemur (Code Review) @ 2020-02-03 14:30 UTC (permalink / raw)
To: Tom de Vries, gdb-patches; +Cc: Mihails Strasuns, Pedro Alves
Tankut Baris Aktemur has posted comments on this change.
Change URL: https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/759
......................................................................
Patch Set 3:
I've tried this patch in two scenarios. One is when multi-threading is
enabled (executable is multi.out), and the other is when the program is
single-threaded (executable is named single1.out and single2.out -- exactly
the same copies just to distinguish the processes). The test program is
sleep.c from the PR.
I start the multi-threaded case as follows:
$ gdb -ex "set prompt (master-gdb) " -ex "b stop_all_threads" -ex r -ex \
"shell killall -9 multi.out" -ex c --args gdb -ex "set debug infrun 1" \
-ex r --args multi.out
When gdb stops, I try "info inferiors" and "info threads" commands:
Thread 1 received signal SIGINT, Interrupt.
0x00007ffff78a99d0 in nanosleep () from /lib/x86_64-linux-gnu/libc.so.6
infrun: infrun_async(0)
(gdb) i inferiors
Num Description Connection Executable
* 1 process 12979 1 (native) /path/to/multi.out
(gdb) i threads
Id Target Id Frame
* 1 Thread 0x7ffff7fe3740 (LWP 12979) 0x00007ffff78a99d0 in nanosleep () from /lib/x86_64-linux-gnu/libc.so.6
2 Thread 0x7ffff77c4700 (LWP 12996) Couldn't get registers: No such process.
(gdb) thread 2
[Switching to thread 2 (Thread 0x7ffff77c4700 (LWP 12996))]
Couldn't get registers: No such process.
(gdb) i threads
Couldn't get registers: No such process.
(gdb) thread 1
Couldn't get registers: No such process.
The second scenario is similar, except that instead of a multi-threaded
single process, we have two single-threaded processes. Started as
$ gdb -ex "set prompt (master-gdb) " -ex "b stop_all_threads" -ex r -ex c \
-ex c -ex "shell killall -9 single1.out" -ex c --args gdb -ex "set debug infrun 1" \
-ex start -ex "add-inferior -exec ./single2.out" -ex "inferior 2" -ex "start" \
-ex "set schedule-multiple on" -ex c --args ./single1.out
The behavior is this:
(gdb) info inferiors
Num Description Connection Executable
1 process 2266 1 (native) /path/to/single1.out
* 2 process 2282 1 (native) /path/to/single2.out
(gdb) info threads
Id Target Id Frame
1.1 process 2266 Couldn't get registers: No such process.
(gdb) inferior 1
[Switching to inferior 1 [process 2266] (/path/to/single1.out)]
[Switching to thread 1.1 (process 2266)]
Couldn't get registers: No such process.
(gdb) info inferiors
Num Description Connection Executable
Couldn't get registers: No such process.
(gdb) info threads
Couldn't get registers: No such process.
(gdb)
So, I think at this point it boils down to the discussion covered in
https://sourceware.org/ml/gdb-patches/2020-01/msg00212.html
and
https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/133
GDB is not able to handle disappeared processes gracefully. The
maintainers may consider the fix for the hanging behavior (i.e. the
infinite loop) OK and the weird post-behavior above as a separate problem
to be addressed later, but my understanding from Pedro's comment was that
the deeper problem of GDB not handling already-gone processes well shall be
addressed.
Thanks
-Baris
--
Gerrit-Project: binutils-gdb
Gerrit-Branch: master
Gerrit-Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
Gerrit-Change-Number: 759
Gerrit-PatchSet: 3
Gerrit-Owner: Tom de Vries <tdevries@suse.de>
Gerrit-Reviewer: Pedro Alves <palves@redhat.com>
Gerrit-Reviewer: Tom de Vries <tdevries@suse.de>
Gerrit-CC: Mihails Strasuns <mihails.strasuns@intel.com>
Gerrit-CC: Tankut Baris Aktemur <tankut.baris.aktemur@intel.com>
Gerrit-Comment-Date: Mon, 03 Feb 2020 14:30:20 +0000
Gerrit-HasComments: No
Gerrit-Has-Labels: No
Gerrit-MessageType: comment
^ permalink raw reply [flat|nested] 14+ messages in thread
* [review v3] [gdb/threads] Fix hang in stop_all_threads after killing inferior
2020-01-29 14:11 [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior Tom de Vries (Code Review)
` (9 preceding siblings ...)
2020-02-03 14:30 ` Tankut Baris Aktemur (Code Review)
@ 2020-02-03 15:20 ` Tom de Vries (Code Review)
2020-02-05 19:53 ` [review v4] " Tom de Vries (Code Review)
2020-02-05 19:53 ` Tom de Vries (Code Review)
12 siblings, 0 replies; 14+ messages in thread
From: Tom de Vries (Code Review) @ 2020-02-03 15:20 UTC (permalink / raw)
To: gdb-patches; +Cc: Tankut Baris Aktemur, Mihails Strasuns, Pedro Alves
Tom de Vries has posted comments on this change.
Change URL: https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/759
......................................................................
Patch Set 3:
> Patch Set 3:
> GDB is not able to handle disappeared processes gracefully.
Ack.
> The
> maintainers may consider the fix for the hanging behavior (i.e. the
> infinite loop) OK and the weird post-behavior above as a separate problem
That's my understanding from Pedro's comment: the weird post-behaviour is a pre-existing issue, independent of the hang that this patch fixes.
--
Gerrit-Project: binutils-gdb
Gerrit-Branch: master
Gerrit-Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
Gerrit-Change-Number: 759
Gerrit-PatchSet: 3
Gerrit-Owner: Tom de Vries <tdevries@suse.de>
Gerrit-Reviewer: Pedro Alves <palves@redhat.com>
Gerrit-Reviewer: Tom de Vries <tdevries@suse.de>
Gerrit-CC: Mihails Strasuns <mihails.strasuns@intel.com>
Gerrit-CC: Tankut Baris Aktemur <tankut.baris.aktemur@intel.com>
Gerrit-Comment-Date: Mon, 03 Feb 2020 15:20:12 +0000
Gerrit-HasComments: No
Gerrit-Has-Labels: No
Gerrit-MessageType: comment
^ permalink raw reply [flat|nested] 14+ messages in thread
* [review v4] [gdb/threads] Fix hang in stop_all_threads after killing inferior
2020-01-29 14:11 [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior Tom de Vries (Code Review)
` (10 preceding siblings ...)
2020-02-03 15:20 ` Tom de Vries (Code Review)
@ 2020-02-05 19:53 ` Tom de Vries (Code Review)
2020-02-05 19:53 ` Tom de Vries (Code Review)
12 siblings, 0 replies; 14+ messages in thread
From: Tom de Vries (Code Review) @ 2020-02-05 19:53 UTC (permalink / raw)
To: Pedro Alves, gdb-patches; +Cc: Mihails Strasuns, Tankut Baris Aktemur
Change URL: https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/759
......................................................................
[gdb/threads] Fix hang in stop_all_threads after killing inferior
Consider a two-threaded testcase a.out, sleeping in both its threads:
...
$ gdb -ex r --args a.out
Reading symbols from a.out...
Starting program: /data/gdb_versions/devel/a.out
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
[New Thread 0x7ffff77fe700 (LWP 31268)]
...
Typing ^C causes stop_all_threads to be executed, and if an external SIGKILL
(such as caused by killall -9 a.out) arrives at the start of stop_all_threads,
gdb hangs in stop_all_threads after giving this warning:
...
warning: unable to open /proc file '/proc/24938/status'
...
Using "set debug infrun 1" we can see in more detail where we hang:
...
infrun: stop_all_threads
infrun: stop_all_threads, pass=0, iterations=0
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, need stop
infrun: target_wait (-1.0.0, status) =
infrun: 10264.10268.0 [Thread 0x7ffff77fe700 (LWP 10268)],
infrun: status->kind = signalled, signal = GDB_SIGNAL_KILL
infrun: stop_all_threads status->kind = signalled, signal = GDB_SIGNAL_KILL \
Thread 0x7ffff77fe700 (LWP 10268)
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, already stopping
warning: unable to open /proc file '/proc/10264/status'
infrun: target_wait (-1.0.0, status) =
infrun: -1.0.0 [process -1],
infrun: status->kind = no-resumed
infrun: infrun_async(0)
infrun: stop_all_threads status->kind = no-resumed process -1
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, already stopping
infrun: stop_all_threads status->kind = no-resumed process -1
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, already stopping
infrun: stop_all_threads status->kind = no-resumed process -1
infrun: Thread 0x7ffff7fa6740 (LWP 10264) not executing
infrun: Thread 0x7ffff77fe700 (LWP 10268) executing, already stopping
<repeat>
......
So, we're hanging in the 'while (1)' loop in stop_all_threads as follows:
- thread t is tested, and both t->executing and t->stop_requested are found
to be 1 (noted with 'executing, already stopping')
- consequently need_wait is set 1
- consequently wait_one is executed
- wait_one returns a TARGET_WAITKIND_NO_RESUMED event, which is handled by
continuing at the start of the loop
The loop actually starts with update_thread_list (), but that doesn't seem
to change the state of the threads.
Fix the hang by:
- detecting the first sign of trouble: the TARGET_WAITKIND_SIGNALLED event
with signal GDB_SIGNAL_KILL,
- making that event pending again,
- making sure the corresponding thread will not set need_wait again
(by setting t->executing == 0)
- making sure that the corresponding thread keeps t->resumed == 1 in the
the all_non_exited_threads loop
This results in the ^C being handled without showing the user that the
test-case was killed:
...
^C
Thread 1 received signal SIGINT, Interrupt.
0x00007ffff78c50f0 in nanosleep () from /lib64/libc.so.6
(gdb)
...
But a subsequent continue does show that:
...
(gdb) c
Continuing.
Program terminated with signal SIGKILL, Killed.
The program no longer exists.
(gdb)
....
Build and reg-tested on x86_64-linux.
gdb/ChangeLog:
2020-01-29 Tom de Vries <tdevries@suse.de>
PR threads/25478
* infrun.c (stop_all_threads): Detecting event
TARGET_WAITKIND_SIGNALLED with signal GDB_SIGNAL_KILL, make event
pending again, set t->executing to 0 and keep t->resumed set to 1.
gdb/testsuite/ChangeLog:
2020-02-05 Tom de Vries <tdevries@suse.de>
* gdb.threads/kill-in-stop-all-threads.c: New test.
* gdb.threads/kill-in-stop-all-threads.exp: New file.
Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
---
M gdb/infrun.c
A gdb/testsuite/gdb.threads/kill-in-stop-all-threads.c
A gdb/testsuite/gdb.threads/kill-in-stop-all-threads.exp
3 files changed, 182 insertions(+), 2 deletions(-)
diff --git a/gdb/infrun.c b/gdb/infrun.c
index 3e846f8..a722da5 100644
--- a/gdb/infrun.c
+++ b/gdb/infrun.c
@@ -4749,7 +4749,12 @@
/* The thread may be not executing, but still be
resumed with a pending status to process. */
- t->resumed = false;
+ if (t->suspend.waitstatus.kind == TARGET_WAITKIND_SIGNALLED
+ && t->suspend.waitstatus.value.sig == GDB_SIGNAL_KILL
+ && t->suspend.waitstatus_pending_p)
+ ;
+ else
+ t->resumed = false;
}
}
@@ -4772,7 +4777,15 @@
target_pid_to_str (event.ptid).c_str ());
}
- if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
+ if (event.ws.kind == TARGET_WAITKIND_SIGNALLED
+ && event.ws.value.sig == GDB_SIGNAL_KILL)
+ {
+ thread_info *t = find_thread_ptid (event.target, event.ptid);
+ save_waitstatus (t, &event.ws);
+ t->resumed = true;
+ t->executing = false;
+ }
+ else if (event.ws.kind == TARGET_WAITKIND_NO_RESUMED
|| event.ws.kind == TARGET_WAITKIND_THREAD_EXITED
|| event.ws.kind == TARGET_WAITKIND_EXITED
|| event.ws.kind == TARGET_WAITKIND_SIGNALLED)
diff --git a/gdb/testsuite/gdb.threads/kill-in-stop-all-threads.c b/gdb/testsuite/gdb.threads/kill-in-stop-all-threads.c
new file mode 100644
index 0000000..cddbd4d
--- /dev/null
+++ b/gdb/testsuite/gdb.threads/kill-in-stop-all-threads.c
@@ -0,0 +1,42 @@
+/* This testcase is part of GDB, the GNU debugger.
+
+ Copyright 2020 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 <signal.h>
+
+static void *
+fun (void *dummy)
+{
+ raise (SIGINT);
+ while (1)
+ sleep (1);
+
+ return NULL;
+}
+
+int
+main (void)
+{
+ pthread_t thread;
+ pthread_create (&thread, NULL, fun, NULL);
+
+ while (1)
+ sleep (1);
+
+ return 0;
+}
diff --git a/gdb/testsuite/gdb.threads/kill-in-stop-all-threads.exp b/gdb/testsuite/gdb.threads/kill-in-stop-all-threads.exp
new file mode 100644
index 0000000..2df509f
--- /dev/null
+++ b/gdb/testsuite/gdb.threads/kill-in-stop-all-threads.exp
@@ -0,0 +1,125 @@
+# Copyright 2020 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/>. */
+
+# This test-case starts gdb with a $binfile inferior, which sends a SIGINT to
+# itself. Then when gdb handles the SIGINT and starts executing
+# stop_all_threads, the inferior is killed by a SIGKILL.
+#
+# In order to time the SIGKILL, the gdb itself is run using another gdb, which
+# allows us to set a breakpoint on stop_all_threads. So, we have the following
+# hierarchy:
+# - master gdb
+# - inferior gdb
+# - $binfile
+#
+# This setup make this a non-standard test-case.
+#
+
+standard_testfile
+
+if {[build_executable "failed to build" $testfile $srcfile \
+ {debug pthreads}] == -1} {
+ return -1
+}
+
+# Setup master gdb, with inferior gdb as executable.
+clean_restart $GDB
+
+# Set master gdb to have a different prompt, to make it easier to distinguish
+# between prompts of master gdb and inferior gdb.
+set saved_gdb_prompt $gdb_prompt
+set master_gdb_prompt "\\(master-gdb\\)"
+set gdb_prompt $master_gdb_prompt
+gdb_test_no_output "set prompt $master_gdb_prompt "
+
+# Set the arguments for the inferior gdb.
+gdb_test_no_output "set args $INTERNAL_GDBFLAGS $GDBFLAGS $binfile"
+
+# Run to main in master gdb.
+if ![runto_main] then {
+ fail "can't run to main"
+ return 0
+}
+
+# Set a breakpoint for the inferior gdb in master gdb.
+if {[gdb_breakpoint "stop_all_threads" no-message] != 1 } {
+ # If we cannot set this breakpoint, there no point in trying further.
+ # Bail out.
+ return -1
+}
+
+# Continue from main in master gdb to initial inferior gdb prompt.
+set gdb_prompt $saved_gdb_prompt
+gdb_test "continue" "Continuing\..*Reading symbols from.*" \
+ "continue from gdb main"
+
+# Start $binfile
+set gdb_prompt $master_gdb_prompt
+gdb_test "start" "Starting program.*Breakpoint 2, stop_all_threads \\(\\).*" \
+ "start inferior"
+
+# We run into stop_all_threads breakpoint in master gdb here, but it's too
+# early, continue to inferior gdb prompt.
+set gdb_prompt $saved_gdb_prompt
+gdb_continue_to_breakpoint "continue past stop_all_threads bp" ".*$srcfile:.*"
+
+# Get the pid of the $binfile process
+set pid -1
+gdb_test_multiple "info inferior 1" "get inferior pid" {
+ -re -wrap "process (\[0-9\]*).*" {
+ set pid $expect_out(1,string)
+ pass $gdb_test_name
+ }
+}
+if { $pid == -1 } {
+ return -1
+}
+
+# Continue $binfile. The $binfile will then trigger SIGINT, which will
+# trigger stop_all_threads in inferior gdb, which will cause the breakpoint
+# to hit in master gdb.
+set gdb_prompt $master_gdb_prompt
+gdb_continue_to_breakpoint "continue to stop_all_threads bp" ".*infrun.c:.*"
+
+# Kill inferior in master gdb, while we're at the start of stop_all_threads in
+# inferior gdb.
+gdb_test_no_output "shell kill -9 $pid"
+
+# Continue from master gdb prompt to inferior gdb prompt.
+set gdb_prompt $saved_gdb_prompt
+gdb_test_multiple "continue" "continue to raise sigint" {
+ -re "warning: unable to open /proc file" {
+ # This warning is the first sign of trouble before we start hanging.
+ # Fail and bail out now, instead of waiting for a timeout.
+ fail $gdb_test_name
+ return -1
+ }
+ -re -wrap "Continuing\.\[\r\n\]+Thread \[0-9\]+ received signal SIGINT, Interrupt\..*" {
+ pass $gdb_test_name
+ }
+}
+
+# Check that the SIGKILL of $binfile is reported by inferior gdb.
+set killed_msg [multi_line "Program terminated with signal SIGKILL, Killed\." \
+ "The program no longer exists\."]
+gdb_test "continue" "Continuing\.\[\r\n\]+$killed_msg" "continue to inferior exit"
+
+# Check that inferior gdb exits normally
+set gdb_prompt $master_gdb_prompt
+gdb_test "quit" "$inferior_exited_re normally\\\]"
+
+# Restore gdb prompt in master gdb.
+set gdb_prompt $saved_gdb_prompt
+gdb_test_no_output "set prompt $saved_gdb_prompt "
--
Gerrit-Project: binutils-gdb
Gerrit-Branch: master
Gerrit-Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
Gerrit-Change-Number: 759
Gerrit-PatchSet: 4
Gerrit-Owner: Tom de Vries <tdevries@suse.de>
Gerrit-Reviewer: Pedro Alves <palves@redhat.com>
Gerrit-Reviewer: Tom de Vries <tdevries@suse.de>
Gerrit-CC: Mihails Strasuns <mihails.strasuns@intel.com>
Gerrit-CC: Tankut Baris Aktemur <tankut.baris.aktemur@intel.com>
Gerrit-MessageType: newpatchset
^ permalink raw reply [flat|nested] 14+ messages in thread
* [review v4] [gdb/threads] Fix hang in stop_all_threads after killing inferior
2020-01-29 14:11 [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior Tom de Vries (Code Review)
` (11 preceding siblings ...)
2020-02-05 19:53 ` [review v4] " Tom de Vries (Code Review)
@ 2020-02-05 19:53 ` Tom de Vries (Code Review)
12 siblings, 0 replies; 14+ messages in thread
From: Tom de Vries (Code Review) @ 2020-02-05 19:53 UTC (permalink / raw)
To: gdb-patches; +Cc: Tankut Baris Aktemur, Mihails Strasuns, Pedro Alves
Tom de Vries has posted comments on this change.
Change URL: https://gnutoolchain-gerrit.osci.io/r/c/binutils-gdb/+/759
......................................................................
Patch Set 4:
Added test-case.
--
Gerrit-Project: binutils-gdb
Gerrit-Branch: master
Gerrit-Change-Id: Ibe1f29251fe2ff1c1991f041babbe18373c113b1
Gerrit-Change-Number: 759
Gerrit-PatchSet: 4
Gerrit-Owner: Tom de Vries <tdevries@suse.de>
Gerrit-Reviewer: Pedro Alves <palves@redhat.com>
Gerrit-Reviewer: Tom de Vries <tdevries@suse.de>
Gerrit-CC: Mihails Strasuns <mihails.strasuns@intel.com>
Gerrit-CC: Tankut Baris Aktemur <tankut.baris.aktemur@intel.com>
Gerrit-Comment-Date: Wed, 05 Feb 2020 19:53:46 +0000
Gerrit-HasComments: No
Gerrit-Has-Labels: No
Gerrit-MessageType: comment
^ permalink raw reply [flat|nested] 14+ messages in thread
end of thread, other threads:[~2020-02-05 19:53 UTC | newest]
Thread overview: 14+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2020-01-29 14:11 [review] [gdb/threads] Fix hang in stop_all_threads after killing inferior Tom de Vries (Code Review)
2020-01-29 15:33 ` Mihails Strasuns (Code Review)
2020-01-29 16:13 ` Tom de Vries (Code Review)
2020-01-29 16:20 ` Tankut Baris Aktemur (Code Review)
2020-01-30 14:55 ` Tom de Vries (Code Review)
2020-01-30 16:30 ` [review v2] " Tom de Vries (Code Review)
2020-01-30 17:36 ` Tankut Baris Aktemur (Code Review)
2020-01-30 17:42 ` Tom de Vries (Code Review)
2020-01-30 17:43 ` Tom de Vries (Code Review)
2020-01-30 17:52 ` [review v3] " Tom de Vries (Code Review)
2020-02-03 14:30 ` Tankut Baris Aktemur (Code Review)
2020-02-03 15:20 ` Tom de Vries (Code Review)
2020-02-05 19:53 ` [review v4] " Tom de Vries (Code Review)
2020-02-05 19:53 ` Tom de Vries (Code Review)
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).