Skip to contentSkip to conversation

fix pthread cancellation in native skin

PATCH Applied Jan Kiszka 3 participants 7 messages Jan 28, 2006 – Feb 3, 2006

lore
Share & tools
lore

Jan Kiszka

author
#0 lore raw

PATCH 1/1 fix pthread cancellation in native skin

Hi,

Gilles' work on cancellation for the posix skin reminded me of this issue I once discovered in the native skin:

https://mail.gna.org/public/xenomai-core/2005-12/msg00014.html

I found out that this can easily be fixed by switching the pthread of a native task to PTHREAD_CANCEL_ASYNCHRONOUS. See attached patch.

At this chance I discovered that calling rt_task_delete for a task that was created and started with T_SUSP mode but was not yet resumed, locks up the system. More precisely: it raises a fatal warning when XENO_OPT_DEBUG is on. Might be the case that it just works on system without this switched on. Either this is a real bug, or the warning needs to be fixed. (Deleting a task after rt_task_suspend works.)

Jan

skins/native/task.c +6 -0
Index: src/skins/native/task.c
===================================================================
@@ -59,6 +59,9 @@
     param.sched_priority = sched_get_priority_max(SCHED_FIFO);
     pthread_setschedparam(pthread_self(),SCHED_FIFO,&param);
 
+    /* rt_task_delete requires asynchronous cancellation */
+    pthread_setcanceltype(PTHREAD_CANCEL_ASYNCHRONOUS, NULL);
+
     signal(SIGCHLD,&rt_task_sigharden);
 
     bulk.a1 = (u_long)iargs->task;
@@ -160,6 +163,9 @@
 {
     struct rt_arg_bulk bulk;
 
+    /* rt_task_delete requires asynchronous cancellation */
+    pthread_setcanceltype(PTHREAD_CANCEL_ASYNCHRONOUS, NULL);
+
     signal(SIGCHLD,&rt_task_sigharden);
 
     bulk.a1 = (u_long)task;
#1 lore raw
Quoted 9 lines · Jan Kiszka

Hi,

Gilles' work on cancellation for the posix skin reminded me of this issue I once discovered in the native skin:

https://mail.gna.org/public/xenomai-core/2005-12/msg00014.html

I found out that this can easily be fixed by switching the pthread of a native task to PTHREAD_CANCEL_ASYNCHRONOUS. See attached patch.

Applied, thanks.

Signature · 1 line
					    Gilles Chanteperdrix.
#2 lore raw
In reply to Jan Kiszka
Quoted 11 lines · Jan Kiszka

Hi,

Gilles' work on cancellation for the posix skin reminded me of this issue I once discovered in the native skin:

https://mail.gna.org/public/xenomai-core/2005-12/msg00014.html

I found out that this can easily be fixed by switching the pthread of a native task to PTHREAD_CANCEL_ASYNCHRONOUS. See attached patch.

Quoted 6 lines

At this chance I discovered that calling rt_task_delete for a task that was created and started with T_SUSP mode but was not yet resumed, locks up the system. More precisely: it raises a fatal warning when XENO_OPT_DEBUG is on. Might be the case that it just works on system without this switched on. Either this is a real bug, or the warning needs to be fixed. (Deleting a task after rt_task_suspend works.)

Actually, the fatal warning happens when starting with rt_task_start the task which was created with the T_SUSP bit. The task needs to wake-up in xnshadow_wait_barrier until it gets really suspended in primary mode by the final xnshadow_harden. This situation triggers the fatal, because the thread has the nucleus XNSUSP bit and is running in secondary mode.

This is not the only situation where a thread with a nucleus suspension bit need to run shortly in secondary mode: it also occurs when suspending with xnpod_suspend_thread() a thread running in secondary mode; the thread receives the SIGCHLD signal and need to execute shortly with the suspension bit set in order to cause a migration to primary mode.

So, the only case when we are sure that a user-space thread can not be scheduled by Linux seems to be when this thread does not have the XNRELAX bit.

Signature · 1 line
					    Gilles Chanteperdrix.
#3 lore raw
Quoted 10 lines · Gilles Chanteperdrix

This is not the only situation where a thread with a nucleus suspension bit need to run shortly in secondary mode: it also occurs when suspending with xnpod_suspend_thread() a thread running in secondary mode; the thread receives the SIGCHLD signal and need to execute shortly with the suspension bit set in order to cause a migration to primary mode.

So, the only case when we are sure that a user-space thread can not be scheduled by Linux seems to be when this thread does not have the XNRELAX bit.

From all the bits in XNTHREAD_BLOCK_BITS, only when the XNPEND bit is set, a thread can not be running in secondary mode. Hence the proposed patch.

Signature · 1 line
					    Gilles Chanteperdrix.
nucleus/shadow.c +14 -3
Index: ksrc/nucleus/shadow.c
===================================================================
@@ -1543,14 +1543,25 @@
 
 #ifdef CONFIG_XENO_OPT_DEBUG
 	{
-	xnflags_t status = threadin->status & ~XNRELAX;
+	xnflags_t status = threadin->status;
 	int sigpending = signal_pending(next);
 
-        if (!(next->ptrace & PT_PTRACED) &&
+        if (!testbits(status, XNRELAX))
+            {
+            show_stack(xnthread_user_task(threadin),NULL);
+            xnpod_fatal("Hardened thread %s[%d] running in Linux domain?! (status=0x%lx, sig=%d, prev=%s[%d])",
+			threadin->name,
+			next->pid,
+			status,
+			sigpending,
+			prev->comm,
+			prev->pid);
+	    }
+        else if (!(next->ptrace & PT_PTRACED) &&
 	    /* Allow ptraced threads to run shortly in order to
 	       properly recover from a stopped state. */
 	    testbits(status,XNSTARTED) &&
-	    testbits(status,XNTHREAD_BLOCK_BITS))
+	    testbits(status,XNPEND))
 	    {
 	    show_stack(xnthread_user_task(threadin),NULL);
             xnpod_fatal("blocked thread %s[%d] rescheduled?! (status=0x%lx, sig=%d, prev=%s[%d])",
#4 lore raw
Quoted 8 lines · Gilles Chanteperdrix
Quoted 7 lines · Gilles Chanteperdrix

This is not the only situation where a thread with a nucleus suspension bit need to run shortly in secondary mode: it also occurs when suspending with xnpod_suspend_thread() a thread running in secondary mode; the thread receives the SIGCHLD signal and need to execute shortly with the suspension bit set in order to cause a migration to primary mode.

Quoted 8 lines
Quoted 3 lines

So, the only case when we are sure that a user-space thread can not be scheduled by Linux seems to be when this thread does not have the XNRELAX bit.

From all the bits in XNTHREAD_BLOCK_BITS, only when the XNPEND bit is set, a thread can not be running in secondary mode. Hence the proposed patch.

Almost ok, but XNDELAY might also be set alone, indicating a purely timed wait state (i.e. without sync object to pend on, so XNPEND is off).

Quoted 45 lines

------------------------------------------------------------------------

Index: ksrc/nucleus/shadow.c
===================================================================

nucleus/shadow.c +14 -4
@@ -1543,14 +1543,25 @@
 
 #ifdef CONFIG_XENO_OPT_DEBUG
 	{
-	xnflags_t status = threadin->status & ~XNRELAX;
+	xnflags_t status = threadin->status;
 	int sigpending = signal_pending(next);
 
-        if (!(next->ptrace & PT_PTRACED) &&
+        if (!testbits(status, XNRELAX))
+            {
+            show_stack(xnthread_user_task(threadin),NULL);
+            xnpod_fatal("Hardened thread %s[%d] running in Linux domain?! (status=0x%lx, sig=%d, prev=%s[%d])",
+			threadin->name,
+			next->pid,
+			status,
+			sigpending,
+			prev->comm,
+			prev->pid);
+	    }
+        else if (!(next->ptrace & PT_PTRACED) &&
 	    /* Allow ptraced threads to run shortly in order to
 	       properly recover from a stopped state. */
 	    testbits(status,XNSTARTED) &&
-	    testbits(status,XNTHREAD_BLOCK_BITS))
+	    testbits(status,XNPEND))
 	    {
 	    show_stack(xnthread_user_task(threadin),NULL);
             xnpod_fatal("blocked thread %s[%d] rescheduled?! (status=0x%lx, sig=%d, prev=%s[%d])",
 
 
------------------------------------------------------------------------

_______________________________________________
Xenomai-core mailing list
[email protected]
https://mail.gna.org/listinfo/xenomai-core

Signature · 1 line
Philippe.
#5 lore raw
Quoted 16 lines · Philippe Gerum
Quoted 14 lines · Gilles Chanteperdrix
Quoted 1 line · Gilles Chanteperdrix

This is not the only situation where a thread with a nucleus

suspension

Quoted 3 lines

bit need to run shortly in secondary mode: it also occurs when suspending with xnpod_suspend_thread() a thread running in secondary mode; the thread receives the SIGCHLD signal and need to execute

shortly

Quoted 3 lines

with the suspension bit set in order to cause a migration to primary mode.

Quoted 1 line

So, the only case when we are sure that a user-space thread can

not be

Quoted 2 lines

scheduled by Linux seems to be when this thread does not have the XNRELAX bit.

Quoted 7 lines
Quoted 4 lines

From all the bits in XNTHREAD_BLOCK_BITS, only when the XNPEND bit is set, a thread can not be running in secondary mode. Hence the proposed patch.

Almost ok, but XNDELAY might also be set alone, indicating a purely timed wait state (i.e. without sync object to pend on, so XNPEND is off).

Forget about this: XNDELAY might also be a transient bit like XNSUSP, so you are right, we cannot test it there.

Quoted 50 lines
Quoted 47 lines

------------------------------------------------------------------------

Index: ksrc/nucleus/shadow.c
===================================================================

nucleus/shadow.c (working copy) renamed from nucleus/shadow.c (revision 507) +14 -3
@@ -1543,14 +1543,25 @@
 
 #ifdef CONFIG_XENO_OPT_DEBUG
     {
-    xnflags_t status = threadin->status & ~XNRELAX;
+    xnflags_t status = threadin->status;
     int sigpending = signal_pending(next);
 
-        if (!(next->ptrace & PT_PTRACED) &&
+        if (!testbits(status, XNRELAX))
+            {
+            show_stack(xnthread_user_task(threadin),NULL);
+            xnpod_fatal("Hardened thread %s[%d] running in Linux 
 domain?! (status=0x%lx, sig=%d, prev=%s[%d])",
+            threadin->name,
+            next->pid,
+            status,
+            sigpending,
+            prev->comm,
+            prev->pid);
+        }
+        else if (!(next->ptrace & PT_PTRACED) &&
         /* Allow ptraced threads to run shortly in order to
            properly recover from a stopped state. */
         testbits(status,XNSTARTED) &&
-        testbits(status,XNTHREAD_BLOCK_BITS))
+        testbits(status,XNPEND))
         {
         show_stack(xnthread_user_task(threadin),NULL);

xnpod_fatal("blocked thread %s[%d] rescheduled?!
(status=0x%lx, sig=%d, prev=%s[%d])",

------------------------------------------------------------------------

_______________________________________________
Xenomai-core mailing list
[email protected]
https://mail.gna.org/listinfo/xenomai-core

Signature · 1 line
Philippe.
#6 lore raw
Quoted 22 lines · Philippe Gerum
Quoted 21 lines · Philippe Gerum
Quoted 18 lines · Gilles Chanteperdrix
Quoted 1 line · Gilles Chanteperdrix

This is not the only situation where a thread with a nucleus

suspension

Quoted 3 lines

bit need to run shortly in secondary mode: it also occurs when suspending with xnpod_suspend_thread() a thread running in secondary mode; the thread receives the SIGCHLD signal and need to execute

shortly

Quoted 3 lines

with the suspension bit set in order to cause a migration to primary mode.

Quoted 1 line

So, the only case when we are sure that a user-space thread can

not be

Quoted 2 lines

scheduled by Linux seems to be when this thread does not have the XNRELAX bit.

From all the bits in XNTHREAD_BLOCK_BITS, only when the XNPEND bit is set, a thread can not be running in secondary mode. Hence the proposed patch.

Quoted 6 lines
Quoted 3 lines

Almost ok, but XNDELAY might also be set alone, indicating a purely timed wait state (i.e. without sync object to pend on, so XNPEND is off).

Forget about this: XNDELAY might also be a transient bit like XNSUSP, so you are right, we cannot test it there.

Ok. The patch is applied to svn trunk.

Signature · 1 line
					    Gilles Chanteperdrix.