Re: freebsd-5.4-stable panics
On Thu, 13 Oct 2005, Rob Watt wrote: > The test machine did panic. Unfortunately I was not running with > BREAK_TO_DEBUGGER. I will re-run the tests with all of the debugging > options we were using before, and then send you the trace info. Unfortunately I was not able to reproduce the panics with BREAK_TO_DEBUGGER compiled into the kernel. I have no log messages or core info from the first panic/reboot. We've been having other problems with our 5 machines, so it's possible (and likely) that the reboot I saw was unrelated to your patch. I ran our stress tests for about 14 more hours without incident. Before we were able to trigger the kern_proc bug on a 6.0 machine within 0-2 hours. We've only experienced the kern_proc bug in 6.0. The bugs we've experienced in 5.4 are multicast/network-threading related. So unfortunately the only thing we can really test for is whether the machine is still stable with your patch. I've had to put all of my test machines back into production running 6.0. If I can free a 5 SMP machine for more tests, I will do so, but at this point it would good if someone else can test the 5.4 patch. I can't post our simulations, but something similar to Antoine Pelisse's test program should be sufficient to see if the patch works. thanks for your help. - Rob Watt ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: freebsd-5.4-stable panics
On 12 Oct, Rob Watt wrote:
>> >> On Fri, 7 Oct 2005, Don Lewis wrote:
>> I MFC'ed the fix to RELENG_6 last week, but the patch didn't apply
>> cleanly to RELENG_5. I tweaked the patch for RELENG_5 and tested it on
>> a UP box. I'd like to get some testing on SMP hardware before I commit
>> it to RELENG_5, just to make sure that I don't destabilize -STABLE. I
>> do want to get the fix into RELENG_5, since this thread originated with
>> a complaint about 5.4-STABLE.
>
> I should be able to have a 5.4 machine available to test this tonight. Can
> you send me the tweaked patch?
I found a couple little nits that I fixed in this version:
Index: sys/kern/kern_proc.c
===
RCS file: /home/ncvs/src/sys/kern/kern_proc.c,v
retrieving revision 1.215.2.6
diff -u -r1.215.2.6 kern_proc.c
--- sys/kern/kern_proc.c22 Mar 2005 13:40:23 - 1.215.2.6
+++ sys/kern/kern_proc.c12 Oct 2005 19:13:14 -
@@ -72,6 +72,8 @@
static void doenterpgrp(struct proc *, struct pgrp *);
static void orphanpg(struct pgrp *pg);
+static void fill_kinfo_proc_only(struct proc *p, struct kinfo_proc *kp);
+static void fill_kinfo_thread(struct thread *td, struct kinfo_proc *kp);
static void pgadjustjobc(struct pgrp *pgrp, int entering);
static void pgdelete(struct pgrp *);
static int proc_ctor(void *mem, int size, void *arg, int flags);
@@ -601,33 +603,22 @@
}
}
#endif /* DDB */
-void
-fill_kinfo_thread(struct thread *td, struct kinfo_proc *kp);
/*
- * Fill in a kinfo_proc structure for the specified process.
+ * Clear kinfo_proc and fill in any information that is common
+ * to all threads in the process.
* Must be called with the target process locked.
*/
-void
-fill_kinfo_proc(struct proc *p, struct kinfo_proc *kp)
-{
- fill_kinfo_thread(FIRST_THREAD_IN_PROC(p), kp);
-}
-
-void
-fill_kinfo_thread(struct thread *td, struct kinfo_proc *kp)
+static void
+fill_kinfo_proc_only(struct proc *p, struct kinfo_proc *kp)
{
- struct proc *p;
struct thread *td0;
- struct ksegrp *kg;
struct tty *tp;
struct session *sp;
struct timeval tv;
struct ucred *cred;
struct sigacts *ps;
- p = td->td_proc;
-
bzero(kp, sizeof(*kp));
kp->ki_structsize = sizeof(*kp);
@@ -685,7 +676,8 @@
kp->ki_tsize = vm->vm_tsize;
kp->ki_dsize = vm->vm_dsize;
kp->ki_ssize = vm->vm_ssize;
- }
+ } else if (p->p_state == PRS_ZOMBIE)
+ kp->ki_stat = SZOMB;
if ((p->p_sflag & PS_INMEM) && p->p_stats) {
kp->ki_start = p->p_stats->p_start;
timevaladd(&kp->ki_start, &boottime);
@@ -704,71 +696,6 @@
kp->ki_nice = p->p_nice;
bintime2timeval(&p->p_runtime, &tv);
kp->ki_runtime = tv.tv_sec * (u_int64_t)100 + tv.tv_usec;
- if (p->p_state != PRS_ZOMBIE) {
-#if 0
- if (td == NULL) {
- /* XXXKSE: This should never happen. */
- printf("fill_kinfo_proc(): pid %d has no threads!\n",
- p->p_pid);
- mtx_unlock_spin(&sched_lock);
- return;
- }
-#endif
- if (td->td_wmesg != NULL) {
- strlcpy(kp->ki_wmesg, td->td_wmesg,
- sizeof(kp->ki_wmesg));
- }
- if (TD_ON_LOCK(td)) {
- kp->ki_kiflag |= KI_LOCKBLOCK;
- strlcpy(kp->ki_lockname, td->td_lockname,
- sizeof(kp->ki_lockname));
- }
-
- if (p->p_state == PRS_NORMAL) { /* XXXKSE very approximate */
- if (TD_ON_RUNQ(td) ||
- TD_CAN_RUN(td) ||
- TD_IS_RUNNING(td)) {
- kp->ki_stat = SRUN;
- } else if (P_SHOULDSTOP(p)) {
- kp->ki_stat = SSTOP;
- } else if (TD_IS_SLEEPING(td)) {
- kp->ki_stat = SSLEEP;
- } else if (TD_ON_LOCK(td)) {
- kp->ki_stat = SLOCK;
- } else {
- kp->ki_stat = SWAIT;
- }
- } else {
- kp->ki_stat = SIDL;
- }
-
- kg = td->td_ksegrp;
-
- /* things in the KSE GROUP */
- kp->ki_estcpu = kg->kg_estcpu;
- kp->ki_slptime = kg->kg_slptime;
- kp->ki_pri.pri_user = kg->kg_user_pri;
- kp->ki_pri.pri_class = kg->kg_pri_class;
-
- /* Things in the thread */
- kp->ki_wchan = td->td_wchan;
- kp->ki_pri.pri_level = td->td_priority;
- kp->ki_pri.pri_native = td->td_base_pri;
-
Re: freebsd-5.4-stable panics
> >> On Fri, 7 Oct 2005, Don Lewis wrote: > I MFC'ed the fix to RELENG_6 last week, but the patch didn't apply > cleanly to RELENG_5. I tweaked the patch for RELENG_5 and tested it on > a UP box. I'd like to get some testing on SMP hardware before I commit > it to RELENG_5, just to make sure that I don't destabilize -STABLE. I > do want to get the fix into RELENG_5, since this thread originated with > a complaint about 5.4-STABLE. I should be able to have a 5.4 machine available to test this tonight. Can you send me the tweaked patch? - Rob Watt ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: freebsd-5.4-stable panics
On 11 Oct, Rob Watt wrote: > On Mon, 10 Oct 2005, Rob Watt wrote: > >> Don, >> >> On Fri, 7 Oct 2005, Don Lewis wrote: >> >> > Both HEAD and RELENG_6 have been patched. I've tested the following >> > patch for RELENG_5 on a uniprocessor sparc64 box. I'd appreciate it if >> > anyone who was running into this problem on RELENG_5 with SMP hardare >> > could test it before I do the MFC. >> >> We have a machine running with those patches applied. We need to do some >> other tests on it today, but tonight we will run our threaded applications >> that trigger the kern_proc problem in top. We should have results tomorrow >> morning. > > Don, > > I had misunderstood what you had asked. I tested this on a 6.0 machine. I > could not crash an amd64 SMP box running 6.0-BETA5 with this patch. I do > not have a test box running RELENG_5 to try this patch on right now. If I > can setup a test box I will let you know our results, but that may take a > day or two. I MFC'ed the fix to RELENG_6 last week, but the patch didn't apply cleanly to RELENG_5. I tweaked the patch for RELENG_5 and tested it on a UP box. I'd like to get some testing on SMP hardware before I commit it to RELENG_5, just to make sure that I don't destabilize -STABLE. I do want to get the fix into RELENG_5, since this thread originated with a complaint about 5.4-STABLE. ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: freebsd-5.4-stable panics
On Mon, 10 Oct 2005, Rob Watt wrote: > Don, > > On Fri, 7 Oct 2005, Don Lewis wrote: > > > Both HEAD and RELENG_6 have been patched. I've tested the following > > patch for RELENG_5 on a uniprocessor sparc64 box. I'd appreciate it if > > anyone who was running into this problem on RELENG_5 with SMP hardare > > could test it before I do the MFC. > > We have a machine running with those patches applied. We need to do some > other tests on it today, but tonight we will run our threaded applications > that trigger the kern_proc problem in top. We should have results tomorrow > morning. Don, I had misunderstood what you had asked. I tested this on a 6.0 machine. I could not crash an amd64 SMP box running 6.0-BETA5 with this patch. I do not have a test box running RELENG_5 to try this patch on right now. If I can setup a test box I will let you know our results, but that may take a day or two. - Rob Watt ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: freebsd-5.4-stable panics
Don, On Fri, 7 Oct 2005, Don Lewis wrote: > Both HEAD and RELENG_6 have been patched. I've tested the following > patch for RELENG_5 on a uniprocessor sparc64 box. I'd appreciate it if > anyone who was running into this problem on RELENG_5 with SMP hardare > could test it before I do the MFC. We have a machine running with those patches applied. We need to do some other tests on it today, but tonight we will run our threaded applications that trigger the kern_proc problem in top. We should have results tomorrow morning. - Rob Watt ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: freebsd-5.4-stable panics
On 3 Oct, Rob Watt wrote:
> We noticed the patches from Don Lewis, but have not tested them yet. We
> weren't sure if we could just apply those patches against 6.0-BETA5, or
> whether we should wait for them to be MFC'd.
Both HEAD and RELENG_6 have been patched. I've tested the following
patch for RELENG_5 on a uniprocessor sparc64 box. I'd appreciate it if
anyone who was running into this problem on RELENG_5 with SMP hardare
could test it before I do the MFC.
Index: sys/kern/kern_proc.c
===
RCS file: /home/ncvs/src/sys/kern/kern_proc.c,v
retrieving revision 1.215.2.6
diff -u -r1.215.2.6 kern_proc.c
--- sys/kern/kern_proc.c22 Mar 2005 13:40:23 - 1.215.2.6
+++ sys/kern/kern_proc.c7 Oct 2005 23:17:26 -
@@ -72,6 +72,8 @@
static void doenterpgrp(struct proc *, struct pgrp *);
static void orphanpg(struct pgrp *pg);
+static void fill_kinfo_proc_only(struct proc *p, struct kinfo_proc *kp);
+static void fill_kinfo_thread(struct thread *td, struct kinfo_proc *kp);
static void pgadjustjobc(struct pgrp *pgrp, int entering);
static void pgdelete(struct pgrp *);
static int proc_ctor(void *mem, int size, void *arg, int flags);
@@ -601,33 +603,22 @@
}
}
#endif /* DDB */
-void
-fill_kinfo_thread(struct thread *td, struct kinfo_proc *kp);
/*
- * Fill in a kinfo_proc structure for the specified process.
+ * Clear kinfo_proc and fill in any information that is common
+ * to all threads in the process.
* Must be called with the target process locked.
*/
-void
-fill_kinfo_proc(struct proc *p, struct kinfo_proc *kp)
-{
- fill_kinfo_thread(FIRST_THREAD_IN_PROC(p), kp);
-}
-
-void
-fill_kinfo_thread(struct thread *td, struct kinfo_proc *kp)
+static void
+fill_kinfo_proc_only(struct proc *p, struct kinfo_proc *kp)
{
- struct proc *p;
struct thread *td0;
- struct ksegrp *kg;
struct tty *tp;
struct session *sp;
struct timeval tv;
struct ucred *cred;
struct sigacts *ps;
- p = td->td_proc;
-
bzero(kp, sizeof(*kp));
kp->ki_structsize = sizeof(*kp);
@@ -685,7 +676,8 @@
kp->ki_tsize = vm->vm_tsize;
kp->ki_dsize = vm->vm_dsize;
kp->ki_ssize = vm->vm_ssize;
- }
+ } else if (p->p_state != PRS_ZOMBIE)
+ kp->ki_stat = SZOMB;
if ((p->p_sflag & PS_INMEM) && p->p_stats) {
kp->ki_start = p->p_stats->p_start;
timevaladd(&kp->ki_start, &boottime);
@@ -704,71 +696,6 @@
kp->ki_nice = p->p_nice;
bintime2timeval(&p->p_runtime, &tv);
kp->ki_runtime = tv.tv_sec * (u_int64_t)100 + tv.tv_usec;
- if (p->p_state != PRS_ZOMBIE) {
-#if 0
- if (td == NULL) {
- /* XXXKSE: This should never happen. */
- printf("fill_kinfo_proc(): pid %d has no threads!\n",
- p->p_pid);
- mtx_unlock_spin(&sched_lock);
- return;
- }
-#endif
- if (td->td_wmesg != NULL) {
- strlcpy(kp->ki_wmesg, td->td_wmesg,
- sizeof(kp->ki_wmesg));
- }
- if (TD_ON_LOCK(td)) {
- kp->ki_kiflag |= KI_LOCKBLOCK;
- strlcpy(kp->ki_lockname, td->td_lockname,
- sizeof(kp->ki_lockname));
- }
-
- if (p->p_state == PRS_NORMAL) { /* XXXKSE very approximate */
- if (TD_ON_RUNQ(td) ||
- TD_CAN_RUN(td) ||
- TD_IS_RUNNING(td)) {
- kp->ki_stat = SRUN;
- } else if (P_SHOULDSTOP(p)) {
- kp->ki_stat = SSTOP;
- } else if (TD_IS_SLEEPING(td)) {
- kp->ki_stat = SSLEEP;
- } else if (TD_ON_LOCK(td)) {
- kp->ki_stat = SLOCK;
- } else {
- kp->ki_stat = SWAIT;
- }
- } else {
- kp->ki_stat = SIDL;
- }
-
- kg = td->td_ksegrp;
-
- /* things in the KSE GROUP */
- kp->ki_estcpu = kg->kg_estcpu;
- kp->ki_slptime = kg->kg_slptime;
- kp->ki_pri.pri_user = kg->kg_user_pri;
- kp->ki_pri.pri_class = kg->kg_pri_class;
-
- /* Things in the thread */
- kp->ki_wchan = td->td_wchan;
- kp->ki_pri.pri_level = td->td_priority;
- kp->ki_pri.pri_native = td->td_base_pri;
- kp->ki_lastcpu = td->td_lastcpu;
- kp->ki_oncpu = td->td_oncpu;
- kp->ki_tdflags = td->td_flags;
- kp->ki_tid = td->td_tid;
Re: freebsd-5.4-stable panics
On 3 Oct, Rob Watt wrote:
>> It turns out that the sysctl buffer is already wired in one of the two
>> cases
>> that this function is called, so I moved the wiring up to the upper
> layer
>> in
>> the other case and cut out a bunch of the locking gymnastics as a
> result.
>> Can you try this patch?
>>
>> Index: kern_proc.c
>> ===
>> RCS file: /usr/cvs/src/sys/kern/kern_proc.c,v
>> retrieving revision 1.231
>> diff -u -r1.231 kern_proc.c
>> --- kern_proc.c 27 Sep 2005 18:03:15 - 1.231
>> +++ kern_proc.c 30 Sep 2005 17:04:57 -
>> @@ -875,22 +875,16 @@
>>
>> if (flags & KERN_PROC_NOTHREADS) {
>> fill_kinfo_proc(p, &kinfo_proc);
>> - PROC_UNLOCK(p);
>> error = SYSCTL_OUT(req, (caddr_t)&kinfo_proc,
>> sizeof(kinfo_proc));
>> - PROC_LOCK(p);
>> } else {
>> - _PHOLD(p);
>> FOREACH_THREAD_IN_PROC(p, td) {
>> fill_kinfo_thread(td, &kinfo_proc);
>> - PROC_UNLOCK(p);
>> error = SYSCTL_OUT(req, (caddr_t)&kinfo_proc,
>> sizeof(kinfo_proc));
>> - PROC_LOCK(p);
>> if (error)
>> break;
>> }
>> - _PRELE(p);
>> }
>> PROC_UNLOCK(p);
>> if (error)
>> @@ -932,6 +926,9 @@
>> if (oid_number == KERN_PROC_PID) {
>> if (namelen != 1)
>> return (EINVAL);
>> + error = sysctl_wire_old_buffer(req, 0);
>> + if (error)
>> + return (error);
>> p = pfind((pid_t)name[0]);
>> if (!p)
>> return (ESRCH);
>
> John,
>
> We tried this patch and were able to run our simulations (and top) for 3
> days straight without crashing. Since we were panicking every 3-6 hours
> before when running top, this seems to have fixed the problem.
>
> We noticed the patches from Don Lewis, but have not tested them yet. We
> weren't sure if we could just apply those patches against 6.0-BETA5, or
> whether we should wait for them to be MFC'd.
I haven't tried applying my patch to RELENG_5 yet, but hope to do so in
the next few days in preparation for doing a MFC. If any changes are
required, I can send you a copy of the patch.
___
freebsd-hackers@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-hackers
To unsubscribe, send any mail to "[EMAIL PROTECTED]"
re: freebsd-5.4-stable panics
> It turns out that the sysctl buffer is already wired in one of the two
> cases
> that this function is called, so I moved the wiring up to the upper
layer
> in
> the other case and cut out a bunch of the locking gymnastics as a
result.
> Can you try this patch?
>
> Index: kern_proc.c
> ===
> RCS file: /usr/cvs/src/sys/kern/kern_proc.c,v
> retrieving revision 1.231
> diff -u -r1.231 kern_proc.c
> --- kern_proc.c 27 Sep 2005 18:03:15 - 1.231
> +++ kern_proc.c 30 Sep 2005 17:04:57 -
> @@ -875,22 +875,16 @@
>
> if (flags & KERN_PROC_NOTHREADS) {
> fill_kinfo_proc(p, &kinfo_proc);
> - PROC_UNLOCK(p);
> error = SYSCTL_OUT(req, (caddr_t)&kinfo_proc,
> sizeof(kinfo_proc));
> - PROC_LOCK(p);
> } else {
> - _PHOLD(p);
> FOREACH_THREAD_IN_PROC(p, td) {
> fill_kinfo_thread(td, &kinfo_proc);
> - PROC_UNLOCK(p);
> error = SYSCTL_OUT(req, (caddr_t)&kinfo_proc,
> sizeof(kinfo_proc));
> - PROC_LOCK(p);
> if (error)
> break;
> }
> - _PRELE(p);
> }
> PROC_UNLOCK(p);
> if (error)
> @@ -932,6 +926,9 @@
> if (oid_number == KERN_PROC_PID) {
> if (namelen != 1)
> return (EINVAL);
> + error = sysctl_wire_old_buffer(req, 0);
> + if (error)
> + return (error);
> p = pfind((pid_t)name[0]);
> if (!p)
> return (ESRCH);
John,
We tried this patch and were able to run our simulations (and top) for 3
days straight without crashing. Since we were panicking every 3-6 hours
before when running top, this seems to have fixed the problem.
We noticed the patches from Don Lewis, but have not tested them yet. We
weren't sure if we could just apply those patches against 6.0-BETA5, or
whether we should wait for them to be MFC'd.
-
Rob Watt
___
freebsd-hackers@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-hackers
To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: freebsd-5.4-stable panics
On 2 Oct, Don Lewis wrote: > It turns out that fill_kinfo_thread() grabs a bunch of locks to grab > things out of struct proc, which breaks badly if sched_lock is grabbed > before calling fill_kinfo_thread(). > > I refactored fill_kinfo_thread() into two functions, one of which > doesn't need any additional locks and only gathers per-thread data, and > a new function, fill_kinfo_proc_only(), which gathers the data that is > common to all theads and can be called before grabbing sched_lock. This > should be more efficient if there is more than one thread because the > per-process data is only gathered once, and only the per-thread data in > kinfo_proc is overwritten for each thread. [ snip ] After fixing a few whitespace nits and one minor buglet, I commited my patch to HEAD, in kern_proc.c 1.232. I hope to be able to MFC it soon. ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: freebsd-5.4-stable panics
On 1 Oct, Don Lewis wrote:
> On 30 Sep, John Baldwin wrote:
>> It turns out that the sysctl buffer is already wired in one of the two cases
>> that this function is called, so I moved the wiring up to the upper layer in
>> the other case and cut out a bunch of the locking gymnastics as a result.
>>
>> Index: kern_proc.c
>> ===
>> RCS file: /usr/cvs/src/sys/kern/kern_proc.c,v
>> retrieving revision 1.231
>> diff -u -r1.231 kern_proc.c
>> --- kern_proc.c 27 Sep 2005 18:03:15 - 1.231
>> +++ kern_proc.c 30 Sep 2005 17:04:57 -
>> @@ -875,22 +875,16 @@
>>
>> if (flags & KERN_PROC_NOTHREADS) {
>> fill_kinfo_proc(p, &kinfo_proc);
>> -PROC_UNLOCK(p);
>> error = SYSCTL_OUT(req, (caddr_t)&kinfo_proc,
>> sizeof(kinfo_proc));
>> -PROC_LOCK(p);
>> } else {
>> -_PHOLD(p);
>> FOREACH_THREAD_IN_PROC(p, td) {
>> fill_kinfo_thread(td, &kinfo_proc);
>> -PROC_UNLOCK(p);
>> error = SYSCTL_OUT(req, (caddr_t)&kinfo_proc,
>> sizeof(kinfo_proc));
>> -PROC_LOCK(p);
>> if (error)
>> break;
>> }
>> -_PRELE(p);
>> }
>> PROC_UNLOCK(p);
>> if (error)
>> @@ -932,6 +926,9 @@
>> if (oid_number == KERN_PROC_PID) {
>> if (namelen != 1)
>> return (EINVAL);
>> +error = sysctl_wire_old_buffer(req, 0);
>> +if (error)
>> +return (error);
>> p = pfind((pid_t)name[0]);
>> if (!p)
>> return (ESRCH);
>>
>
> sched_lock needs to be grabbed before the FOREACH_THREAD_IN_PROC loop.
>
> Can _PHOLD()/_PRELE() be dropped?
It turns out that fill_kinfo_thread() grabs a bunch of locks to grab
things out of struct proc, which breaks badly if sched_lock is grabbed
before calling fill_kinfo_thread().
I refactored fill_kinfo_thread() into two functions, one of which
doesn't need any additional locks and only gathers per-thread data, and
a new function, fill_kinfo_proc_only(), which gathers the data that is
common to all theads and can be called before grabbing sched_lock. This
should be more efficient if there is more than one thread because the
per-process data is only gathered once, and only the per-thread data in
kinfo_proc is overwritten for each thread.
Index: kern_proc.c
===
RCS file: /home/ncvs/src/sys/kern/kern_proc.c,v
retrieving revision 1.231
diff -u -r1.231 kern_proc.c
--- kern_proc.c 27 Sep 2005 18:03:15 - 1.231
+++ kern_proc.c 2 Oct 2005 08:48:56 -
@@ -73,6 +73,8 @@
static void doenterpgrp(struct proc *, struct pgrp *);
static void orphanpg(struct pgrp *pg);
+static void fill_kinfo_proc_only(struct proc *p, struct kinfo_proc *kp);
+static void fill_kinfo_thread(struct thread *td, struct kinfo_proc *kp);
static void pgadjustjobc(struct pgrp *pgrp, int entering);
static void pgdelete(struct pgrp *);
static int proc_ctor(void *mem, int size, void *arg, int flags);
@@ -596,33 +598,22 @@
}
}
#endif /* DDB */
-void
-fill_kinfo_thread(struct thread *td, struct kinfo_proc *kp);
/*
- * Fill in a kinfo_proc structure for the specified process.
+ * Clear kinfo_proc and fill in any information that is common
+ * to all threads in the process.
* Must be called with the target process locked.
*/
-void
-fill_kinfo_proc(struct proc *p, struct kinfo_proc *kp)
-{
- fill_kinfo_thread(FIRST_THREAD_IN_PROC(p), kp);
-}
-
-void
-fill_kinfo_thread(struct thread *td, struct kinfo_proc *kp)
+static void
+fill_kinfo_proc_only(struct proc *p, struct kinfo_proc *kp)
{
- struct proc *p;
struct thread *td0;
- struct ksegrp *kg;
struct tty *tp;
struct session *sp;
struct timeval tv;
struct ucred *cred;
struct sigacts *ps;
- p = td->td_proc;
-
bzero(kp, sizeof(*kp));
kp->ki_structsize = sizeof(*kp);
@@ -684,78 +675,14 @@
kp->ki_tsize = vm->vm_tsize;
kp->ki_dsize = vm->vm_dsize;
kp->ki_ssize = vm->vm_ssize;
- }
+ } else if (p->p_state == PRS_ZOMBIE)
+ kp->ki_stat = SZOMB;
kp->ki_sflag = p->p_sflag;
kp->ki_swtime = p->p_swtime;
kp->ki_pid = p->p_pid;
kp->ki_nice = p->p_nice;
bintime2timeval(&p->p_rux.rux_runtime, &tv);
kp->ki_runtime = tv.tv_sec * (u_int64_t)100 + tv.tv_usec;
- if (p->p_state != PRS_ZOMBIE) {
-#if 0
- if (td == NULL) {
- /* XXXKSE: This should never happen. */
- printf("fill_kinfo_proc(): pid %d has no threads!\n",
-
Re: freebsd-5.4-stable panics
On 30 Sep, John Baldwin wrote:
> On Friday 30 September 2005 11:25 am, Antoine Pelisse wrote:
>> On 9/30/05, John Baldwin <[EMAIL PROTECTED]> wrote:
>> > On Friday 30 September 2005 05:24 am, Antoine Pelisse wrote:
>> > > Hi Robert,
>> > > I don't think your patch is correct, the total linked list can be
>> > > broken
>> > >
>> > > while the lock is released, thus just passing the link may not be
>> > > enough I have submitted a PR[1] for this a month ago but nobody took
>> > > care of it yet Regards,
>> > > Antoine Pelisse
>> > >
>> > > [1] http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/84684
>> >
>> > I think this patch looks ok. Robert, can you get the original panic on
>> > this
>> > thread tested against this patch?
>>
>> I had a small program which could reproduce this panic in 10 seconds, it
>> was basically creating empty threads and calling kvm_getprocs() in the same
>> time. Anyway the patch was able to stop the program from panicing.
>> The panic is also reproducible in RELENG_6 and HEAD IIRC.
>
> It turns out that the sysctl buffer is already wired in one of the two cases
> that this function is called, so I moved the wiring up to the upper layer in
> the other case and cut out a bunch of the locking gymnastics as a result.
> Can you try this patch?
>
> Index: kern_proc.c
> ===
> RCS file: /usr/cvs/src/sys/kern/kern_proc.c,v
> retrieving revision 1.231
> diff -u -r1.231 kern_proc.c
> --- kern_proc.c 27 Sep 2005 18:03:15 - 1.231
> +++ kern_proc.c 30 Sep 2005 17:04:57 -
> @@ -875,22 +875,16 @@
>
> if (flags & KERN_PROC_NOTHREADS) {
> fill_kinfo_proc(p, &kinfo_proc);
> - PROC_UNLOCK(p);
> error = SYSCTL_OUT(req, (caddr_t)&kinfo_proc,
> sizeof(kinfo_proc));
> - PROC_LOCK(p);
> } else {
> - _PHOLD(p);
> FOREACH_THREAD_IN_PROC(p, td) {
> fill_kinfo_thread(td, &kinfo_proc);
> - PROC_UNLOCK(p);
> error = SYSCTL_OUT(req, (caddr_t)&kinfo_proc,
> sizeof(kinfo_proc));
> - PROC_LOCK(p);
> if (error)
> break;
> }
> - _PRELE(p);
> }
> PROC_UNLOCK(p);
> if (error)
> @@ -932,6 +926,9 @@
> if (oid_number == KERN_PROC_PID) {
> if (namelen != 1)
> return (EINVAL);
> + error = sysctl_wire_old_buffer(req, 0);
> + if (error)
> + return (error);
> p = pfind((pid_t)name[0]);
> if (!p)
> return (ESRCH);
>
sched_lock needs to be grabbed before the FOREACH_THREAD_IN_PROC loop.
Can _PHOLD()/_PRELE() be dropped?
___
freebsd-hackers@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-hackers
To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: freebsd-5.4-stable panics
On 30 Sep, Antoine Pelisse wrote: > Hi Robert, > I don't think your patch is correct, the total linked list can be broken > while the lock is released, thus just passing the link may not be enough > I have submitted a PR[1] for this a month ago but nobody took care of it yet There are two problems with your patch: sched_lock needs to be held while iterating over the threads sysctl_kern_proc() calls sysctl_out_proc() multiple times in a loop in the !KERN_PROC_PID case, so the buffer needs to be wired before calling sysctl_out_proc(). Is _PHOLD()/_PRELE() needed if we don't drop PROC_LOCK? Passing a size estimate to sysctl_wire_old_buffer() is desirable, but sysctl_out_proc() would need some restructuring to do this correctly. ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: freebsd-5.4-stable panics
On 9/30/05, John Baldwin <[EMAIL PROTECTED]> wrote:
>
> On Friday 30 September 2005 11:25 am, Antoine Pelisse wrote:
> > On 9/30/05, John Baldwin <[EMAIL PROTECTED]> wrote:
> > > On Friday 30 September 2005 05:24 am, Antoine Pelisse wrote:
> > > > Hi Robert,
> > > > I don't think your patch is correct, the total linked list can be
> > > > broken
> > > >
> > > > while the lock is released, thus just passing the link may not be
> > > > enough I have submitted a PR[1] for this a month ago but nobody took
> > > > care of it yet Regards,
> > > > Antoine Pelisse
> > > >
> > > > [1] http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/84684
> > >
> > > I think this patch looks ok. Robert, can you get the original panic on
> > > this
> > > thread tested against this patch?
> >
> > I had a small program which could reproduce this panic in 10 seconds, it
> > was basically creating empty threads and calling kvm_getprocs() in the
> same
> > time. Anyway the patch was able to stop the program from panicing.
> > The panic is also reproducible in RELENG_6 and HEAD IIRC.
>
> It turns out that the sysctl buffer is already wired in one of the two
> cases
> that this function is called, so I moved the wiring up to the upper layer
> in
> the other case and cut out a bunch of the locking gymnastics as a result.
> Can you try this patch?
>
> Index: kern_proc.c
> ===
> RCS file: /usr/cvs/src/sys/kern/kern_proc.c,v
> retrieving revision 1.231
> diff -u -r1.231 kern_proc.c
> --- kern_proc.c 27 Sep 2005 18:03:15 - 1.231
> +++ kern_proc.c 30 Sep 2005 17:04:57 -
> @@ -875,22 +875,16 @@
>
> if (flags & KERN_PROC_NOTHREADS) {
> fill_kinfo_proc(p, &kinfo_proc);
> - PROC_UNLOCK(p);
> error = SYSCTL_OUT(req, (caddr_t)&kinfo_proc,
> sizeof(kinfo_proc));
> - PROC_LOCK(p);
> } else {
> - _PHOLD(p);
> FOREACH_THREAD_IN_PROC(p, td) {
> fill_kinfo_thread(td, &kinfo_proc);
> - PROC_UNLOCK(p);
> error = SYSCTL_OUT(req, (caddr_t)&kinfo_proc,
> sizeof(kinfo_proc));
> - PROC_LOCK(p);
> if (error)
> break;
> }
> - _PRELE(p);
> }
> PROC_UNLOCK(p);
> if (error)
> @@ -932,6 +926,9 @@
> if (oid_number == KERN_PROC_PID) {
> if (namelen != 1)
> return (EINVAL);
> + error = sysctl_wire_old_buffer(req, 0);
> + if (error)
> + return (error);
> p = pfind((pid_t)name[0]);
> if (!p)
> return (ESRCH);
>
> --
> John Baldwin <[EMAIL PROTECTED]> <>< http://www.FreeBSD.org/~jhb/
> "Power Users Use the Power to Serve" = http://www.FreeBSD.org
>
Hi John,
I'm sorry I can't test it right now, I'm in a foreign country for three
months and can only connect to the internet through the university
connection.
I'll be back home mid-december. Maybe Rob should take care of testing it.
Regards,
Antoine Pelisse
___
freebsd-hackers@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-hackers
To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: freebsd-5.4-stable panics
On Friday 30 September 2005 11:25 am, Antoine Pelisse wrote:
> On 9/30/05, John Baldwin <[EMAIL PROTECTED]> wrote:
> > On Friday 30 September 2005 05:24 am, Antoine Pelisse wrote:
> > > Hi Robert,
> > > I don't think your patch is correct, the total linked list can be
> > > broken
> > >
> > > while the lock is released, thus just passing the link may not be
> > > enough I have submitted a PR[1] for this a month ago but nobody took
> > > care of it yet Regards,
> > > Antoine Pelisse
> > >
> > > [1] http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/84684
> >
> > I think this patch looks ok. Robert, can you get the original panic on
> > this
> > thread tested against this patch?
>
> I had a small program which could reproduce this panic in 10 seconds, it
> was basically creating empty threads and calling kvm_getprocs() in the same
> time. Anyway the patch was able to stop the program from panicing.
> The panic is also reproducible in RELENG_6 and HEAD IIRC.
It turns out that the sysctl buffer is already wired in one of the two cases
that this function is called, so I moved the wiring up to the upper layer in
the other case and cut out a bunch of the locking gymnastics as a result.
Can you try this patch?
Index: kern_proc.c
===
RCS file: /usr/cvs/src/sys/kern/kern_proc.c,v
retrieving revision 1.231
diff -u -r1.231 kern_proc.c
--- kern_proc.c 27 Sep 2005 18:03:15 - 1.231
+++ kern_proc.c 30 Sep 2005 17:04:57 -
@@ -875,22 +875,16 @@
if (flags & KERN_PROC_NOTHREADS) {
fill_kinfo_proc(p, &kinfo_proc);
- PROC_UNLOCK(p);
error = SYSCTL_OUT(req, (caddr_t)&kinfo_proc,
sizeof(kinfo_proc));
- PROC_LOCK(p);
} else {
- _PHOLD(p);
FOREACH_THREAD_IN_PROC(p, td) {
fill_kinfo_thread(td, &kinfo_proc);
- PROC_UNLOCK(p);
error = SYSCTL_OUT(req, (caddr_t)&kinfo_proc,
sizeof(kinfo_proc));
- PROC_LOCK(p);
if (error)
break;
}
- _PRELE(p);
}
PROC_UNLOCK(p);
if (error)
@@ -932,6 +926,9 @@
if (oid_number == KERN_PROC_PID) {
if (namelen != 1)
return (EINVAL);
+ error = sysctl_wire_old_buffer(req, 0);
+ if (error)
+ return (error);
p = pfind((pid_t)name[0]);
if (!p)
return (ESRCH);
--
John Baldwin <[EMAIL PROTECTED]> <>< http://www.FreeBSD.org/~jhb/
"Power Users Use the Power to Serve" = http://www.FreeBSD.org
___
freebsd-hackers@freebsd.org mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-hackers
To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: freebsd-5.4-stable panics
On Friday 30 September 2005 05:24 am, Antoine Pelisse wrote:
> Hi Robert,
> I don't think your patch is correct, the total linked list can be broken
> while the lock is released, thus just passing the link may not be enough
> I have submitted a PR[1] for this a month ago but nobody took care of it
> yet Regards,
> Antoine Pelisse
>
> [1] http://www.freebsd.org/cgi/query-pr.cgi?pr=kern/84684
I think this patch looks ok. Robert, can you get the original panic on this
thread tested against this patch?
> On 9/29/05, Robert Watson <[EMAIL PROTECTED]> wrote:
> > On Thu, 29 Sep 2005, Rob Watt wrote:
> > > On Thu, 29 Sep 2005, Robert Watson wrote:
> > >> Could you dump the contents of *td and *td->td_proc for me? I'm quite
> > >> interested to know what the value in td->td_proc->p_state is, among
> >
> > other
> >
> > >> things. If I could also have you generate a dump of the KSE group
> > >> structures in td->td_proc->p_ksegrps and the threads in
> > >> td->td_proc->p_threads.
> > >
> > > I've attached a file with many of the values you have asked for. We
> > > looked at some of the threads referenced by td->td_proc->p_threads, but
> > > we weren't sure we were walking the list correctly. Do you have any
> > > tips
> > >
> > > for walking those thread lists?
> > >
> > >> Could you tell me if the program named by p->p_comm is linked against
> > >> a threading library? If it's a custom app, you may already know, and
> > >> if not, you can run ldd on the application to see what it is linked
> > >> against.
> > >
> > > The programs named by p->p_comm is linked against the pthreads library.
> >
> > This seems to be enough information to at least track this down a bit:
> > td_ksegrp is NULL, rather than a corrupt value, which suggests that the
> > thread is incompletely initialized. Other hints that this are the case
> > are that td_critnest is 1 (as is set when it is allocated), and the state
> > is TDS_INACTIVE. Some other fields are set though, such as td_oncpu,
> > which is normally initialized to NOCPU.
> >
> > > (kgdb) p *td
> > > $1 = {td_proc = 0xff004aa9f000, td_ksegrp = 0x0, td_plist =
> > > {tqe_next = 0xff 00b4798000,
> > > tqe_prev = 0xff00a97ae010}, td_kglist = {tqe_next =
> > > 0xff00b4798000,
> > > tqe_prev = 0xff00a97ae020}, td_slpq = {tqe_next = 0x0, tqe_prev
> > > = 0x ff001fac7c10}, td_lockq = {
> > > tqe_next = 0xff00a97ae000, tqe_prev = 0xb6797a70},
> > > td_runq = {tq e_next = 0x0,
> > > tqe_prev = 0x80608180}, td_selq = {tqh_first = 0x0, tqh_last
> > > = 0xfff fff00633112c0},
> > > td_sleepqueue = 0xff00382b0400, td_turnstile = 0xff00c1712900,
> > > td_umtx q = 0xff00d1207080,
> > > td_tid = 100253, td_flags = 16777216, td_inhibitors = 0, td_pflags =
> > > 128, td_d upfd = 0, td_wchan = 0x0,
> > > td_wmesg = 0x0, td_lastcpu = 2 '\002', td_oncpu = 2 '\002',
> > > td_owepreempt = 0 '\0', td_locks = 0,
> > > td_blocked = 0x0, td_ithd = 0x0, td_lockname = 0x0, td_contested =
> > > {lh_first =
> > > 0x0}, td_sleeplocks = 0x0,
> > > td_intr_nesting_level = 0, td_pinned = 0, td_mailbox = 0x0, td_ucred =
> > > 0xf f00ad18f200,
> > > td_standin = 0x0, td_upcall = 0x0, td_sticks = 0, td_uuticks = 0,
> > > td_usticks =
> > > 0, td_intrval = 0,
> > > td_oldsigmask = {__bits = {0, 0, 0, 0}}, td_sigmask = {__bits =
> > > {4294967295, 4 294967295, 4294967295,
> > > 4294967295}}, td_siglist = {__bits = {0, 0, 0, 0}}, td_generation
> > > = 14, td _sigstk = {ss_sp = 0x0,
> > > ss_size = 0, ss_flags = 0}, td_kflags = 0, td_xsig = 0,
> > > td_profil_addr = 0, td_profil_ticks = 0,
> > > td_base_pri = 182 '\u', td_priority = 182 '\u', td_pcb =
> > > 0xb68 dcd10, td_state = TDS_INACTIVE,
> > > td_retval = {1, 29309280}, td_slpcallout = {c_links = {sle = {sle_next
> > > = 0x0},
> > > tqe = {tqe_next = 0x0,
> > > tqe_prev = 0xff001fac7d80}}, c_time = 55907602, c_arg =
> > > 0xff0063 311260,
> > > c_func = 0x802e32a0 , c_mtx = 0x0, c_flags =
> > > 16}, td _frame = 0xb68dcc40,
> > > td_kstack_obj = 0xff0087f93d20, td_kstack = 18446744072477315072,
> > > td_kstac k_pages = 4,
> > > td_altkstack_obj = 0x0, td_altkstack = 0, td_altkstack_pages = 0,
> > > td_critnest = 1, td_md = {
> > > md_spinlock_count = 1, md_saved_flags = 582}, td_sched =
> > > 0xff0063311488}
> >
> > I'm not familiar with the internals of the thread and KSE life cycle
> > here,
> >
> > so I think we'll need to look to those more familiar with this to
> > understand what of two things may be going on:
> >
> > (1) Is the fact that td_ksegrp != NULL an invariant for a connected
> > thread, and that kern_proc is relying on that but the thread code is
> > failing to implement it safely?
> >
> > (2) Is td_ksegrp sometimes left legitimately as NULL as part of the
> > thread life cycle, and that kern_proc incorrectly assumes that it is
> > never NULL when hooked up to a thread.
> >
> > This suggests a possible work-around of simply testing td_ksegr
Re: freebsd-5.4-stable panics
On Thu, 29 Sep 2005, Robert Watson wrote: > Could you dump the contents of *td and *td->td_proc for me? I'm quite > interested to know what the value in td->td_proc->p_state is, among other > things. If I could also have you generate a dump of the KSE group > structures in td->td_proc->p_ksegrps and the threads in > td->td_proc->p_threads. I've attached a file with many of the values you have asked for. We looked at some of the threads referenced by td->td_proc->p_threads, but we weren't sure we were walking the list correctly. Do you have any tips for walking those thread lists? > > Could you tell me if the program named by p->p_comm is linked against a > threading library? If it's a custom app, you may already know, and if > not, you can run ldd on the application to see what it is linked against. > The programs named by p->p_comm is linked against the pthreads library. > Depending on how much time you have available, it might make sense for me > to grab from you a copy of your source tree, compiled kernel with debug > symbols, and core dump. We can upload the source, kernel etc somewhere, but uncompressed that is about 5G of data. What is the best way to get that to you? Thanks. - Rob Watt 6.0-BETA5.kgdb.out Description: Binary data ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: freebsd-5.4-stable panics
Robert,
We have gotten some more information from our type1 crash:
>sh lockedvnods
Locked vnodes
>sh alllocks
Process 2204 (dataplay) thread 0xff00b1726a000 (100214)
exclusive sleep mutex inp (udpinp) f = 0 (0xff00cc90fcc8) locked @
/usr/src/sys/netinet/udp_usrreq.c:762
Process 62 (pagedaemon) thread 0xff00e358c280 (100049)
exclusive sleep mutex UMA lock r = 0 (0x8068bf80) locked @
/usr/src/sys/vm/uma_core.c:1491
exclusive sleep mutex Giant r = 0 (0x8062ed80) locked @
/usr/src/sys/vm/vm_pageout.c:717
Process 48 (swi1:net) thread 0xff00e3597780 (100027)
exclusive sleep mutex IPFW static rules r = 0 (0x8067ae50) locked
@ /usr/src/sys/netinet/ip_fw2.c:149
>sh pcpu
cpuid=0
currthread = 0xff00e358c280: pid 63 "pagedaemon"
currpcb = 0xb34e3d10
fpcurrthread= none
idle thread = 0xff00e35b6000: pid 14 (idle cpu0)
spin locks held =
>sh pcpu 1
cpuid=1
currthread = 0xff00e358b3c80: pid 13 "idle cpu1"
currpcb = 0xffb34e7d10
fpcurrthread= none
idle thread = 0xff00e358b3c80: pid 13 (idle cpu1)
spin locks held =
>sh pcpu 2
cpuid=2
currthread = 0xff00e35e4000: pid 2715 "bonnie"
currpcb = 0xffb636dd10
fpcurrthread= none
idle thread = 0xff00e35b3a00: pid 12 (idle cpu2)
spin locks held =
>sh pcpu 3
cpuid=3
currthread = 0xff00e35aea00: pid 40 "irq27: em1 em2"
currpcb = 0xffb34b6d10
fpcurrthread= none
idle thread = 0xff00e35b3780: pid 11 (idle cpu0)
spin locks held =
I have attached the core output as type1-core.2.txt, but unfortunately it
does not help us determine the area of code that triggered the exception.
If I can get more DDB output from the type2 crash I will post it.
There is some encouraging news: since we stopped running top, our
6.0-BETA5 test machine has not crashed (its been running tests now for
over 26 hours).
We also started running tests on a dual single-core machine running
5-STABLE. That machine has been running for 50 hours without crashing.
This means that we are now only hitting these bugs with dual dual-core
machines running 5-STABLE.
-
Rob WattDDB:
>sh lockedvnods
Locked vnodes
>sh alllocks
Process 2204 (dataplay) thread 0xff00b1726a000 (100214)
exclusive sleep mutex inp (udpinp) f = 0 (0xff00cc90fcc8) locked @
/usr/src/sys/netinet/udp_usrreq.c:762
Process 62 (pagedaemon) thread 0xff00e358c280 (100049)
exclusive sleep mutex UMA lock r = 0 (0x8068bf80) locked @
/usr/src/sys/vm/uma_core.c:1491
exclusive sleep mutex Giant r = 0 (0x8062ed80) locked @
/usr/src/sys/vm/vm_pageout.c:717
Process 48 (swi1:net) thread 0xff00e3597780 (100027)
exclusive sleep mutex IPFW static rules r = 0 (0x8067ae50) locked @
/usr/src/sys/netinet/ip_fw2.c:149
>sh pcpu
cpuid=0
currthread = 0xff00e358c280: pid 63 "pagedaemon"
currpcb = 0xb34e3d10
fpcurrthread= none
idle thread = 0xff00e35b6000: pid 14 (idle cpu0)
spin locks held =
>sh pcpu 1
cpuid=1
currthread = 0xff00e358b3c80: pid 13 "idle cpu1"
currpcb = 0xffb34e7d10
fpcurrthread= none
idle thread = 0xff00e358b3c80: pid 13 "idle cpu1"
spin locks held =
>sh pcpu 2
cpuid=2
currthread = 0xff00e35e4000: pid 2715 "bonnie"
currpcb = 0xffb636dd10
fpcurrthread= none
idle thread = 0xff00e35b3a00: pid 12 "idle cpu2"
spin locks held =
>sh pcpu 3
cpuid=3
currthread = 0xff00e35aea00: pid 40 "irq27: em1 em2"
currpcb = 0xffb34b6d10
fpcurrthread= none
idle thread = 0xff00e35b3780: pid 11 "idle cpu0"
spin locks held =
KGDB:
Unread portion of the kernel message buffer:
panic: No TID bitmap?
cpuid = 0
KDB: enter: panic
#0 doadump () at pcpu.h:167
167 pcpu.h: No such file or directory.
in pcpu.h
(kgdb) bt
#0 doadump () at pcpu.h:167
#1 0x801924f6 in db_fncall (dummy1=0, dummy2=0, dummy3=0, dummy4=0x0)
at /usr/src/sys/ddb/db_command.c:531
#2 0x80192985 in db_command_loop () at
/usr/src/sys/ddb/db_command.c:349
#3 0x80194833 in db_trap (type=-1286719648, code=0) at
/usr/src/sys/ddb/db_main.c:221
#4 0x802cb8f0 in kdb_trap (type=3, code=0, tf=0x0) at
/usr/src/sys/kern/subr_kdb.c:470
#5 0x804169dc in trap (frame=
{tf_rdi = 0, tf_rsi = -2136928256, tf_rdx = 0, tf_rcx = 523776, tf_r8 =
-1286719440, tf_r9 = 10, tf_rax = 18, tf_rbx = -2142686258, tf_rbp =
-1286719200, tf_r10 = 20765, tf_r11 = 0, tf_r12 = 0, tf_r13 = 256, tf_r14 =
-1095697382784, tf_r15 = 768605, tf_trapno = 3, tf_addr = 0, tf_flags = 256,
tf_err = 0, tf_rip = -2144554161, tf_cs = 8, tf_rflags = 642, tf_rsp =
-1286719200, tf_ss = 16}) at /usr/src/sys/amd64/amd64/trap.c:431
#6 0x804046fb in calltrap () at
/usr/src/sys/amd64/amd64/exception.S:171
#7 0x in ?? ()
#8 0x80a11000 in ?? ()
#9 0x in ?? ()
#10 0x0007fe00 in
Re: freebsd-5.4-stable panics
On Thu, 29 Sep 2005, Rob Watt wrote:
On Thu, 29 Sep 2005, Robert Watson wrote:
Could you dump the contents of *td and *td->td_proc for me? I'm quite
interested to know what the value in td->td_proc->p_state is, among other
things. If I could also have you generate a dump of the KSE group
structures in td->td_proc->p_ksegrps and the threads in
td->td_proc->p_threads.
I've attached a file with many of the values you have asked for. We
looked at some of the threads referenced by td->td_proc->p_threads, but
we weren't sure we were walking the list correctly. Do you have any tips
for walking those thread lists?
Could you tell me if the program named by p->p_comm is linked against a
threading library? If it's a custom app, you may already know, and if
not, you can run ldd on the application to see what it is linked
against.
The programs named by p->p_comm is linked against the pthreads library.
This seems to be enough information to at least track this down a bit:
td_ksegrp is NULL, rather than a corrupt value, which suggests that the
thread is incompletely initialized. Other hints that this are the case
are that td_critnest is 1 (as is set when it is allocated), and the state
is TDS_INACTIVE. Some other fields are set though, such as td_oncpu,
which is normally initialized to NOCPU.
(kgdb) p *td
$1 = {td_proc = 0xff004aa9f000, td_ksegrp = 0x0, td_plist =
{tqe_next = 0xff 00b4798000,
tqe_prev = 0xff00a97ae010}, td_kglist = {tqe_next =
0xff00b4798000,
tqe_prev = 0xff00a97ae020}, td_slpq = {tqe_next = 0x0, tqe_prev
= 0x ff001fac7c10}, td_lockq = {
tqe_next = 0xff00a97ae000, tqe_prev = 0xb6797a70},
td_runq = {tq e_next = 0x0,
tqe_prev = 0x80608180}, td_selq = {tqh_first = 0x0, tqh_last
= 0xfff fff00633112c0},
td_sleepqueue = 0xff00382b0400, td_turnstile = 0xff00c1712900,
td_umtx q = 0xff00d1207080,
td_tid = 100253, td_flags = 16777216, td_inhibitors = 0, td_pflags =
128, td_d upfd = 0, td_wchan = 0x0,
td_wmesg = 0x0, td_lastcpu = 2 '\002', td_oncpu = 2 '\002',
td_owepreempt = 0 '\0', td_locks = 0,
td_blocked = 0x0, td_ithd = 0x0, td_lockname = 0x0, td_contested =
{lh_first =
0x0}, td_sleeplocks = 0x0,
td_intr_nesting_level = 0, td_pinned = 0, td_mailbox = 0x0, td_ucred =
0xf f00ad18f200,
td_standin = 0x0, td_upcall = 0x0, td_sticks = 0, td_uuticks = 0,
td_usticks =
0, td_intrval = 0,
td_oldsigmask = {__bits = {0, 0, 0, 0}}, td_sigmask = {__bits =
{4294967295, 4 294967295, 4294967295,
4294967295}}, td_siglist = {__bits = {0, 0, 0, 0}}, td_generation
= 14, td _sigstk = {ss_sp = 0x0,
ss_size = 0, ss_flags = 0}, td_kflags = 0, td_xsig = 0,
td_profil_addr = 0, td_profil_ticks = 0,
td_base_pri = 182 '\u', td_priority = 182 '\u', td_pcb =
0xb68 dcd10, td_state = TDS_INACTIVE,
td_retval = {1, 29309280}, td_slpcallout = {c_links = {sle = {sle_next
= 0x0},
tqe = {tqe_next = 0x0,
tqe_prev = 0xff001fac7d80}}, c_time = 55907602, c_arg =
0xff0063 311260,
c_func = 0x802e32a0 , c_mtx = 0x0, c_flags =
16}, td _frame = 0xb68dcc40,
td_kstack_obj = 0xff0087f93d20, td_kstack = 18446744072477315072,
td_kstac k_pages = 4,
td_altkstack_obj = 0x0, td_altkstack = 0, td_altkstack_pages = 0,
td_critnest = 1, td_md = {
md_spinlock_count = 1, md_saved_flags = 582}, td_sched =
0xff0063311488}
I'm not familiar with the internals of the thread and KSE life cycle here,
so I think we'll need to look to those more familiar with this to
understand what of two things may be going on:
(1) Is the fact that td_ksegrp != NULL an invariant for a connected
thread, and that kern_proc is relying on that but the thread code is
failing to implement it safely?
(2) Is td_ksegrp sometimes left legitimately as NULL as part of the thread
life cycle, and that kern_proc incorrectly assumes that it is never
NULL when hooked up to a thread.
This suggests a possible work-around of simply testing td_ksegrp for NULL
in kern_proc in order to avoid this, while attempting to resolve whether
an invariant is violated (or incorrectly assumed), which might require
some serious thinking and a solution that is non-trivial. Something like
the following might work in the mean time:
Index: kern_proc.c
===
RCS file: /home/ncvs/src/sys/kern/kern_proc.c,v
retrieving revision 1.231
diff -u -r1.231 kern_proc.c
--- kern_proc.c 27 Sep 2005 18:03:15 - 1.231
+++ kern_proc.c 29 Sep 2005 20:50:33 -
@@ -882,6 +882,8 @@
} else {
_PHOLD(p);
FOREACH_THREAD_IN_PROC(p, td) {
+ if (td->td_ksegrp == NULL)
+ continue;
fill_kinfo_thread(td, &kinfo_proc);
PROC_UNLOCK(p);
error = SYSCTL_OUT(req, (caddr_t)&kinfo_pro
Re: freebsd-5.4-stable panics
On Wed, 28 Sep 2005, Rob Watt wrote: We re-compiled the kernel with 'options KDB_STOP_NMI', and were able to get a much more full analysis of what was happening on the 6-BETA5 crash. Great. We crashed in top again, and it does look like we may have hit a kern_proc bug. This sounds good, or at least, promising. in the attached file type3-core.txt you can see that it hits an exception in: 0x802b897a is in fill_kinfo_thread (/usr/src/sys/kern/kern_proc.c:736). 731 } 732 733 kg = td->td_ksegrp; 734 735 /* things in the KSE GROUP */ 736 kp->ki_estcpu = kg->kg_estcpu; 737 kp->ki_slptime = kg->kg_slptime; 738 kp->ki_pri.pri_user = kg->kg_user_pri; 739 kp->ki_pri.pri_class = kg->kg_pri_class; 740 (kgdb) frame 8 #8 0x802b897a in fill_kinfo_thread (td=0xff0063311260, kp=0xb62d8510) at /usr/src/sys/kern/kern_proc.c:733 733 kg = td->td_ksegrp; (kgdb) p kg->kg_estcpu Cannot access memory at address 0x173 (kgdb) p td->td_ksegrp $1 = (struct ksegrp *) 0x0 (kgdb) p kp->ki_estcpu $2 = 0 (kgdb) p kg $4 = (struct ksegrp *) 0x12b it seems that kg is an invalid pointer. Could you dump the contents of *td and *td->td_proc for me? I'm quite interested to know what the value in td->td_proc->p_state is, among other things. If I could also have you generate a dump of the KSE group structures in td->td_proc->p_ksegrps and the threads in td->td_proc->p_threads. Could you tell me if the program named by p->p_comm is linked against a threading library? If it's a custom app, you may already know, and if not, you can run ldd on the application to see what it is linked against. Depending on how much time you have available, it might make sense for me to grab from you a copy of your source tree, compiled kernel with debug symbols, and core dump. We have started our tests again without running top. Hope you have a great vacation. It was brief but very enjoyable, and quite disconnected :-). Thanks, Robert ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: freebsd-5.4-stable panics
Robert, On Tue, 27 Sep 2005, Robert Watson wrote: > Great. As mentioned I'll be offline for about the next 48 hours, but back > after then. If we can get a nice clean crash out of this, would really be > best. If it's top panicking, it could well be due to a bug in the process > monitoring code, in kern_proc. We've run into bugs a few times there in > the past, generally associated with threading or races in process > creation/teardown, in which partially initialized (or torn down) processes > are accessed by another thread and are in an unexpected state. We re-compiled the kernel with 'options KDB_STOP_NMI', and were able to get a much more full analysis of what was happening on the 6-BETA5 crash. We crashed in top again, and it does look like we may have hit a kern_proc bug. in the attached file type3-core.txt you can see that it hits an exception in: 0x802b897a is in fill_kinfo_thread (/usr/src/sys/kern/kern_proc.c:736). 731 } 732 733 kg = td->td_ksegrp; 734 735 /* things in the KSE GROUP */ 736 kp->ki_estcpu = kg->kg_estcpu; 737 kp->ki_slptime = kg->kg_slptime; 738 kp->ki_pri.pri_user = kg->kg_user_pri; 739 kp->ki_pri.pri_class = kg->kg_pri_class; 740 (kgdb) frame 8 #8 0x802b897a in fill_kinfo_thread (td=0xff0063311260, kp=0xb62d8510) at /usr/src/sys/kern/kern_proc.c:733 733 kg = td->td_ksegrp; (kgdb) p kg->kg_estcpu Cannot access memory at address 0x173 (kgdb) p td->td_ksegrp $1 = (struct ksegrp *) 0x0 (kgdb) p kp->ki_estcpu $2 = 0 (kgdb) p kg $4 = (struct ksegrp *) 0x12b it seems that kg is an invalid pointer. We have started our tests again without running top. Hope you have a great vacation. - Rob Watt type3-core.txt Description: Binary data Copyright (c) 1992-2005 The FreeBSD Project. Copyright (c) 1979, 1980, 1983, 1986, 1988, 1989, 1991, 1992, 1993, 1994 The Regents of the University of California. All rights reserved. FreeBSD 6.0-BETA5 #1: Tue Sep 27 17:38:32 EDT 2005 [EMAIL PROTECTED]:/usr/obj/usr/src/sys/LOCAL-DEBUG-NMI WARNING: WITNESS option enabled, expect reduced performance. Timecounter "i8254" frequency 1193182 Hz quality 0 CPU: Dual Core AMD Opteron(tm) Processor 275 (2190.05-MHz K8-class CPU) Origin = "AuthenticAMD" Id = 0x20f12 Stepping = 2 Features=0x178bfbff Features2=0x1 AMD Features=0xe2500800,LM,3DNow+,3DNow> Hyperthreading: 2 logical CPUs real memory = 3942580224 (3759 MB) avail memory = 3807399936 (3631 MB) ACPI APIC Table: FreeBSD/SMP: Multiprocessor System Detected: 4 CPUs cpu0 (BSP): APIC ID: 0 cpu1 (AP): APIC ID: 1 cpu2 (AP): APIC ID: 2 cpu3 (AP): APIC ID: 3 MADT: Forcing active-low polarity and level trigger for SCI ioapic0 irqs 0-23 on motherboard ioapic1 irqs 24-27 on motherboard ioapic2 irqs 28-31 on motherboard acpi0: on motherboard acpi0: Power Button (fixed) pci_link0: irq 10 on acpi0 pci_link1: irq 5 on acpi0 pci_link2: irq 11 on acpi0 pci_link3: irq 9 on acpi0 Timecounter "ACPI-fast" frequency 3579545 Hz quality 1000 acpi_timer0: <24-bit timer at 3.579545MHz> port 0x1008-0x100b on acpi0 cpu0: on acpi0 acpi_throttle0: on cpu0 cpu1: on acpi0 cpu2: on acpi0 cpu3: on acpi0 pcib0: port 0xcf8-0xcff on acpi0 pci0: on pcib0 pcib1: at device 6.0 on pci0 pci3: on pcib1 ohci0: mem 0xfeafc000-0xfeafcfff irq 19 at device 0.0 on pci3 ohci0: [GIANT-LOCKED] usb0: OHCI version 1.0, legacy support usb0: on ohci0 usb0: USB revision 1.0 uhub0: AMD OHCI root hub, class 9/0, rev 1.00/1.00, addr 1 uhub0: 3 ports with 3 removable, self powered ohci1: mem 0xfeafd000-0xfeafdfff irq 19 at device 0.1 on pci3 ohci1: [GIANT-LOCKED] usb1: OHCI version 1.0, legacy support usb1: on ohci1 usb1: USB revision 1.0 uhub1: AMD OHCI root hub, class 9/0, rev 1.00/1.00, addr 1 uhub1: 3 ports with 3 removable, self powered pci3: at device 6.0 (no driver attached) fxp0: port 0xbc00-0xbc3f mem 0xfeafb000-0xfeafbfff,0xfeaa-0xfeab irq 18 at device 8.0 on pci3 miibus0: on fxp0 inphy0: on miibus0 inphy0: 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, auto fxp0: Ethernet address: 00:e0:81:31:89:1c isab0: at device 7.0 on pci0 isa0: on isab0 atapci0: port 0x1f0-0x1f7,0x3f6,0x170-0x177,0x376,0xffa0-0xffaf at device 7.1 on pci0 ata0: on atapci0 ata1: on atapci0 pci0: at device 7.2 (no driver attached) pci0: at device 7.3 (no driver attached) pcib2: at device 10.0 on pci0 pci2: on pcib2 em0: port 0x8880-0x88bf mem 0xfc8c-0xfc8d,0xfc80-0xfc83 irq 26 at device 2.0 on pci2 em0: Ethernet address: 00:04:23:ba:d0:42 em0: Speed:N/A Duplex:N/A em1: port 0x8c00-0x8c3f mem 0xfc8e-0xfc8f,0xfc88-0xfc8b irq 27 at device 2.1 on pci2 em1: Ethernet address: 00:04:23:ba:d0:43 em1: Speed:N/A Duplex:N/A em2: port 0x8480-0x84bf mem 0xfc78-0xfc79,0xfc74-0xfc77 irq 27 at device 3.0 on pci2 em2: E
Re: freebsd-5.4-stable panics
On Sun, 25 Sep 2005, Robert Watson wrote: > > On Fri, 23 Sep 2005, Jason Carroll wrote: > 5B > > There seem to be 2 types of crashes we see with pretty different stack > > traces. What I'll call a type 1 crash, I believe, is often caused by > > one of the triggers I mention above. A type 2 crash appears to happen > > spontaneously after the machine has been running for a while. > > > > I poked around using kgdb in a core file from a type 2 crash, and it > > appeared the system hung closing sockets (specifically cleaning up > > multicast state i think) while cleaning up one of our multicast > > applications (note the trace through sys_exit). There's no reason this > > application should have been exiting unless it encountered some kind of > > error. > > Sounds nasty. It's possible the two panics are related, especially if > they involve a race in the multicast code, which could result in treading > on other kernel memory, potentially leading to the thread related panic. > My leaning would be that they are unrelated, but since we may be able to > eliminate the multicast one (see below), that would be a good starting > point. > > There are some other known stability nits in 6.x which are being worked > on, but in general the network stack stability is higher in 6.x than 5.x > when it comes to multicast due to the work I reference above. If you run > into any stability problems relating to the file system, set > debug.mpsafevfs=0 in loader.conf -- there are a few bug fixes relating to > running out of disk space or hitting quota limits that are fixed in HEAD, > but not yet backported to 6.x. Robert, Thanks for your quick response and suggestions. We have now experienced an additional type of crash. Type 3 is from 6.0-BETA5, it did not enter the debugger at all and we could not generate a core. Unfortunately the 6-BETA crash was completely different from everything we've seen so far. The panic was related to a page fault and 'top' was the active process. We are trying again to run our tests on 6.0, but if we keep encountering other bugs, then those other bugs may prevent us from determining if multicast is the problem. We also ran our applications in 5-STABLE without reading from or writing to disk (ie we ran the multicast data streams on a remote machine, and we told our listener/rebroadcaster apps not to write to disk). In this configuration we were able to run for 4 days without crashing. A few hours before the crash we had introduced disk activity (bonnie in a constant loop with 1G test file size). This crash was a type 1, and we were not able to save a core. The longest we had gone before without a crash was 6 hours, so it is possible that either load, or disk activity help trigger the bugs we have seen. files attached: kernel-conf.txt (6.0 kernel) type3-core.txt (copy of panic output to console) We will update you with more info from our 6.0 tests when we have it. We are in a bind right now. All modern hardware (ie emt64/amd64) only seems to work with versions of freebsd that aren't stable when running our applications. Many vendors do not even sell server hardware that is purely i386. We never encountered these types of problems on freebsd 4.x, and many of our 120+ i386 class machines that are running 4.x are showing their age and need to be replaced. Assuming that the problems we are experiencing are purely related to ths OS, we now don't have an OS to run on the newer hardware we've been buying. We really need to find a way to patch these problems or find a version of freebsd that supports our platform and is stable. Obviously we appreciate the hard work that all of you on the freebsd team do, and we are happy to do whatever we can to help squash these bugs. - Rob Watt# # GENERIC -- Generic kernel configuration file for FreeBSD/amd64 # # For more information on this file, please read the handbook section on # Kernel Configuration Files: # # http://www.FreeBSD.org/doc/en_US.ISO8859-1/books/handbook/kernelconfig-config.html # # The handbook is also available locally in /usr/share/doc/handbook # if you've installed the doc distribution, otherwise always see the # FreeBSD World Wide Web server (http://www.FreeBSD.org/) for the # latest information. # # An exhaustive list of options and more detailed explanations of the # device lines is also present in the ../../conf/NOTES and NOTES files. # If you are in doubt as to the purpose or necessity of a line, check first # in NOTES. # # $FreeBSD: src/sys/amd64/conf/GENERIC,v 1.421.2.11.2.1 2005/04/09 17:28:37 kensmith Exp $ machine amd64 cpu HAMMER ident CUSTOM # To statically compile in device wiring instead of /boot/device.hints #hints "GENERIC.hints" # Default places to look for devices. makeoptions DEBUG=-g options KDB options DDB options BREAK_TO_DEBUGGER options INVARIANTS options INVARIANT_SUPPORT options WITNESS options WITNESS_S
Re: freebsd-5.4-stable panics
On 9/27/05, Robert Watson <[EMAIL PROTECTED]> wrote: > > On Tue, 27 Sep 2005, Rob Watt wrote: > > > Is this an SMP box? If so, could you try compiling options KDB_STOP_NMI > into your kernel -- you'll also need to set debug.kdb.stop_cpus_with_nmi=1 > in either loader.conf or at runtime with sysctls. This is a dual-core dual 275 processor box. I have compiled the nmi options into the kernel and we are now using that to test. > > The trap information you've provided indicates that it is likely a data > NULL pointer dereference in the kernel (faulting address is a small > increment above NULL). The instruction pointer looks valid -- if you have > a debugging copy of the kernel, could you load it into gdb and show me > what line number / piece of code it's in? you can use "l > *803b88ca" to generate that, even without a live debugger session > this is the piece of code that was referenced by the ip: (gdb) l *0x803b88ca 0x803b88ca is in nfsrv_lookup (/usr/src/sys/nfsserver/nfs_serv.c:670). 665 NFSD_UNLOCK(); 666 mtx_lock(&Giant); /* VFS */ 667 if (dirp) 668 vrele(dirp); 669 NDFREE(&nd, NDF_ONLY_PNBUF); 670 if (ndp->ni_startdir) 671 vrele(ndp->ni_startdir); 672 if (ndp->ni_vp) 673 vput(ndp->ni_vp); 674 mtx_unlock(&Giant); /* VFS */ we are not running nfsd (although we do use nfs and nfsiod), and none of our processes should have been accessing nfs. Our processes are run from an nfs mount but do not access any nfs mounted files. > > Do you have a testbed or set of test hosts set up so you can > non-disruptively test change sets, btw? > yes we have 3 dual dual-core machines and 1 dual single-core machine that we can use to test with. Thanks! - Rob Watt ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: freebsd-5.4-stable panics
On Tue, 27 Sep 2005, Rob Watt wrote: this is the piece of code that was referenced by the ip: (gdb) l *0x803b88ca 0x803b88ca is in nfsrv_lookup (/usr/src/sys/nfsserver/nfs_serv.c:670). 665 NFSD_UNLOCK(); 666 mtx_lock(&Giant); /* VFS */ 667 if (dirp) 668 vrele(dirp); 669 NDFREE(&nd, NDF_ONLY_PNBUF); 670 if (ndp->ni_startdir) 671 vrele(ndp->ni_startdir); 672 if (ndp->ni_vp) 673 vput(ndp->ni_vp); 674 mtx_unlock(&Giant); /* VFS */ we are not running nfsd (although we do use nfs and nfsiod), and none of our processes should have been accessing nfs. Our processes are run from an nfs mount but do not access any nfs mounted files. That code is in the NFS server lookup code, so should be called as a result of a lookup by a remote client. If the NFS server is not in use on the machine, this is most likely this is a quirk of gdb and instruction pointers, a run-time kernel/compile-time kernel mismatch, or something really nasty. ndp should really never be NULL there, as it's used frequently prior to that point. Let's hope for one of the former few options. Do you have a testbed or set of test hosts set up so you can non-disruptively test change sets, btw? yes we have 3 dual dual-core machines and 1 dual single-core machine that we can use to test with. Great. As mentioned I'll be offline for about the next 48 hours, but back after then. If we can get a nice clean crash out of this, would really be best. If it's top panicking, it could well be due to a bug in the process monitoring code, in kern_proc. We've run into bugs a few times there in the past, generally associated with threading or races in process creation/teardown, in which partially initialized (or torn down) processes are accessed by another thread and are in an unexpected state. Thanks, Robert N M Watson ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: freebsd-5.4-stable panics
On Tue, 27 Sep 2005, Rob Watt wrote: Thanks for your quick response and suggestions. We have now experienced an additional type of crash. Type 3 is from 6.0-BETA5, it did not enter the debugger at all and we could not generate a core. Is this an SMP box? If so, could you try compiling options KDB_STOP_NMI into your kernel -- you'll also need to set debug.kdb.stop_cpus_with_nmi=1 in either loader.conf or at runtime with sysctls. This will probably become the default at some point -- in the mean time, the default when entering the debugger on one CPU is to generate an IPI to the other CPUs telling them "go into the debugger". This works fine unless the CPU has interrupts disabled, such as if it's holding a spin lock in the scheduler, in which case the system will deadlock because that CPU won't acknowledge the IPI. With the above option, a non-maskable interrupt is used to signal the other CPUs into the debugger, which gets into the debugger much more reliably. The trap information you've provided indicates that it is likely a data NULL pointer dereference in the kernel (faulting address is a small increment above NULL). The instruction pointer looks valid -- if you have a debugging copy of the kernel, could you load it into gdb and show me what line number / piece of code it's in? you can use "l *803b88ca" to generate that, even without a live debugger session or core. If you can get into DDB with the above, generally good starting point debugging information (ideally gathered with a serial console) is: trace # current thread trace show pcpu # current CPU data show pcpu 0 # CPU 0 data show pcpu 1 # CPU 1 data ... # Any other CPUs ps# process listing show lockedvnods # VFS locking information If you have WITNESS compiled in, also: show alllocks Unfortunately the 6-BETA crash was completely different from everything we've seen so far. The panic was related to a page fault and 'top' was the active process. We are trying again to run our tests on 6.0, but if we keep encountering other bugs, then those other bugs may prevent us from determining if multicast is the problem. Let's see if we can get whatever this first bug you're hitting is fixed and see if we can get to the next original problems. We also ran our applications in 5-STABLE without reading from or writing to disk (ie we ran the multicast data streams on a remote machine, and we told our listener/rebroadcaster apps not to write to disk). In this configuration we were able to run for 4 days without crashing. A few hours before the crash we had introduced disk activity (bonnie in a constant loop with 1G test file size). This crash was a type 1, and we were not able to save a core. The longest we had gone before without a crash was 6 hours, so it is possible that either load, or disk activity help trigger the bugs we have seen. I'm heading off on a vacation for two days, and will be offline for that period, but if we can't easily get through solving 6.x problems on the host, I can backport a subset of the multicast fixes to 5.x and we can see if that fixes things up. It may make sense to do this anyway, but I may not have an opportunity to go through the development and testing on that until after 6.0 is out the door. files attached: kernel-conf.txt (6.0 kernel) type3-core.txt (copy of panic output to console) We will update you with more info from our 6.0 tests when we have it. We are in a bind right now. All modern hardware (ie emt64/amd64) only seems to work with versions of freebsd that aren't stable when running our applications. Many vendors do not even sell server hardware that is purely i386. We never encountered these types of problems on freebsd 4.x, and many of our 120+ i386 class machines that are running 4.x are showing their age and need to be replaced. Assuming that the problems we are experiencing are purely related to ths OS, we now don't have an OS to run on the newer hardware we've been buying. We really need to find a way to patch these problems or find a version of freebsd that supports our platform and is stable. Obviously we appreciate the hard work that all of you on the freebsd team do, and we are happy to do whatever we can to help squash these bugs. Hopefully we can get this fixed up as soon as possible. Do you have a testbed or set of test hosts set up so you can non-disruptively test change sets, btw? Thanks, Robert N M Watson ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: freebsd-5.4-stable panics
On Fri, 23 Sep 2005, Jason Carroll wrote: 5B There seem to be 2 types of crashes we see with pretty different stack traces. What I'll call a type 1 crash, I believe, is often caused by one of the triggers I mention above. A type 2 crash appears to happen spontaneously after the machine has been running for a while. I poked around using kgdb in a core file from a type 2 crash, and it appeared the system hung closing sockets (specifically cleaning up multicast state i think) while cleaning up one of our multicast applications (note the trace through sys_exit). There's no reason this application should have been exiting unless it encountered some kind of error. I'm attaching: dmesg.txt kernel-conf.txt (kernel config file) type1-core.txt (a kgdb bt from a type1/triggered crash) type2-core.txt (a kgdb bt from a type2/spontaneous crash) I'm happy to dig for more information, recompile with different options, apply patches, or do anything else that might help get this problem diagnosed and fixed! Hi there Jason! Sounds nasty. It's possible the two panics are related, especially if they involve a race in the multicast code, which could result in treading on other kernel memory, potentially leading to the thread related panic. My leaning would be that they are unrelated, but since we may be able to eliminate the multicast one (see below), that would be a good starting point. In the 6.x branch, quite a bit of work has been done to improve locking in the multicast code, and several important races have been fixed relating to IP multicast. These races tended to turn up on the following sorts of situations: (1) Multi-threaded appplications changing the multicast properties, such as membership, or a particular socket in parallel. (2) Changes to multicast membership during high multicast I/O load on the socket. For example, adding or deleting multicast groups on socket on CPU 0 while a packet is delivered to the same socket on CPU 1. (3) Removal of real or synthetic interfaces involved in active multicast, such as removal of pccards, vlans, etc during multicast I/O, or with sockets bound to the interfaces. These changes are not currently scheduled for a backport to 5.x, because they change the kernel network device driver API and ABI, requiring changes to and recompiling of third party device drivers. A subset could be backported, subject to some limitations, but it would be good to confirm whether these changes actually affect the problems you're seeing before working through that. All the changes should appear in the most recent snapshot, BETA5. Make sure to turn off extra kernel debugging features, such as WITNESS, INVARIANTS, and user space malloc debugging, if you start running into performance problems -- they have a big performance impact, although can be quite helpful in testing. Normally we turn these off during the release candidate portion of the release cycle. There are some other known stability nits in 6.x which are being worked on, but in general the network stack stability is higher in 6.x than 5.x when it comes to multicast due to the work I reference above. If you run into any stability problems relating to the file system, set debug.mpsafevfs=0 in loader.conf -- there are a few bug fixes relating to running out of disk space or hitting quota limits that are fixed in HEAD, but not yet backported to 6.x. Thanks, Robert N M Watson ___ freebsd-hackers@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-hackers To unsubscribe, send any mail to "[EMAIL PROTECTED]"
