panic sleeping without a mutex in usb_task_thread
On my laptop (but not the build machine; the latter may not have any USB devices) I get the following panic after builing, installing, trying to boot -CURRENT from today. (CVSup log below panic trace.) an0: Ethernet address 00:40:96:32:19:a9 bpf: an0 attached panic: sleeping without a mutex Debugger(panic) Stopped at Debugger+0x40: xorl%eax,%eax db trace Debugger(c03d647c) at Debugger+0x40 panic(c03d6cac,c021ffec,c8e4e000,c8e4e100,c8e4e000) at panic+0x70 msleep(c0442d98,0,5c,c03c7784,0) at msleep+0x23 usb_task_thread(0,c8e6ad48,c8e4e100,c021ffec,0) at usb_task_thread+0x23 fork_exit(c021ffec,0,c8e6ad48) at fork_exit+0x8c fork_trampoline() at fork_trampoline+0x37 dbshow locks dbshow witness [scrolls off screen] db [Above was hand-transcribed; I don't have a serial console on my laptop, and I'm not running VMware.] Recent CVSup history: freebeast(4.5-S)[1] tail /var/log/cvsup-history.log CVSup begin from cvsup14.freebsd.org at Thu Apr 4 03:47:06 PST 2002 CVSup ended from cvsup14.freebsd.org at Thu Apr 4 03:53:56 PST 2002 CVSup begin from cvsup14.freebsd.org at Fri Apr 5 03:47:02 PST 2002 CVSup ended from cvsup14.freebsd.org at Fri Apr 5 03:54:13 PST 2002 CVSup begin from cvsup14.freebsd.org at Sat Apr 6 03:47:02 PST 2002 CVSup ended from cvsup14.freebsd.org at Sat Apr 6 03:54:32 PST 2002 CVSup begin from cvsup14.freebsd.org at Sun Apr 7 03:47:03 PDT 2002 CVSup ended from cvsup14.freebsd.org at Sun Apr 7 03:53:32 PDT 2002 CVSup begin from cvsup14.freebsd.org at Mon Apr 8 03:47:02 PDT 2002 CVSup ended from cvsup14.freebsd.org at Mon Apr 8 03:53:43 PDT 2002 freebeast(4.5-S)[2] I can leave it in this condition for a while, if it would be helpful to poke around at all. Cheers, david -- David H. Wolfskill [EMAIL PROTECTED] Based on my experience as a computing professional, I consider the use of Microsoft products as components of computing systems to be just as advisable as using green wood to frame a house... and expect similar results. To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
RE: panic sleeping without a mutex in usb_task_thread
On 08-Apr-2002 David Wolfskill wrote: On my laptop (but not the build machine; the latter may not have any USB devices) I get the following panic after builing, installing, trying to boot -CURRENT from today. (CVSup log below panic trace.) an0: Ethernet address 00:40:96:32:19:a9 bpf: an0 attached panic: sleeping without a mutex Debugger(panic) Stopped atDebugger+0x40: xorl%eax,%eax db trace Debugger(c03d647c) at Debugger+0x40 panic(c03d6cac,c021ffec,c8e4e000,c8e4e100,c8e4e000) at panic+0x70 msleep(c0442d98,0,5c,c03c7784,0) at msleep+0x23 usb_task_thread(0,c8e6ad48,c8e4e100,c021ffec,0) at usb_task_thread+0x23 fork_exit(c021ffec,0,c8e6ad48) at fork_exit+0x8c fork_trampoline() at fork_trampoline+0x37 dbshow locks dbshow witness [scrolls off screen] db [Above was hand-transcribed; I don't have a serial console on my laptop, and I'm not running VMware.] Recent CVSup history: freebeast(4.5-S)[1] tail /var/log/cvsup-history.log CVSup begin from cvsup14.freebsd.org at Thu Apr 4 03:47:06 PST 2002 CVSup ended from cvsup14.freebsd.org at Thu Apr 4 03:53:56 PST 2002 CVSup begin from cvsup14.freebsd.org at Fri Apr 5 03:47:02 PST 2002 CVSup ended from cvsup14.freebsd.org at Fri Apr 5 03:54:13 PST 2002 CVSup begin from cvsup14.freebsd.org at Sat Apr 6 03:47:02 PST 2002 CVSup ended from cvsup14.freebsd.org at Sat Apr 6 03:54:32 PST 2002 CVSup begin from cvsup14.freebsd.org at Sun Apr 7 03:47:03 PDT 2002 CVSup ended from cvsup14.freebsd.org at Sun Apr 7 03:53:32 PDT 2002 CVSup begin from cvsup14.freebsd.org at Mon Apr 8 03:47:02 PDT 2002 CVSup ended from cvsup14.freebsd.org at Mon Apr 8 03:53:43 PDT 2002 freebeast(4.5-S)[2] I can leave it in this condition for a while, if it would be helpful to poke around at all. show witness isn't useful to most people so I would avoid it unless someone explicitly asks for it. The problem here is likely due to the usb_task_thread() not locking Giant when it starts up. -- John Baldwin [EMAIL PROTECTED]http://www.FreeBSD.org/~jhb/ Power Users Use the Power to Serve! - http://www.FreeBSD.org/ To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
Re: panic sleeping without a mutex in usb_task_thread
On Mon, Apr 08, 2002 at 12:33:40PM -0400, John Baldwin wrote:
show witness isn't useful to most people so I would avoid it unless someone
explicitly asks for it. The problem here is likely due to the
usb_task_thread() not locking Giant when it starts up.
We probably want this:
Index: usb.c
===
RCS file: /home/ncvs/src/sys/dev/usb/usb.c,v
retrieving revision 1.75
diff -u -5 -r1.75 usb.c
--- usb.c 7 Apr 2002 14:21:32 - 1.75
+++ usb.c 8 Apr 2002 17:58:38 -
-423,10 +423,14
usb_task_thread(void *arg)
{
struct usb_task *task;
int s;
+#ifdef __FreeBSD__
+ mtx_lock(Giant);
+#endif
+
DPRINTF((usb_task_thread: start\n));
s = splusb();
for (;;) {
task = TAILQ_FIRST(usb_all_tasks);
Joe
msg37102/pgp0.pgp
Description: PGP signature
Re: panic sleeping without a mutex in usb_task_thread
On 08-Apr-2002 Josef Karthauser wrote:
On Mon, Apr 08, 2002 at 12:33:40PM -0400, John Baldwin wrote:
show witness isn't useful to most people so I would avoid it unless someone
explicitly asks for it. The problem here is likely due to the
usb_task_thread() not locking Giant when it starts up.
We probably want this:
Probably unless some USB specific locks are added instead, but this is the
easier fix for the time being.
Index: usb.c
===
RCS file: /home/ncvs/src/sys/dev/usb/usb.c,v
retrieving revision 1.75
diff -u -5 -r1.75 usb.c
--- usb.c 7 Apr 2002 14:21:32 - 1.75
+++ usb.c 8 Apr 2002 17:58:38 -
@@ -423,10 +423,14 @@
usb_task_thread(void *arg)
{
struct usb_task *task;
int s;
+#ifdef __FreeBSD__
+ mtx_lock(Giant);
+#endif
+
DPRINTF((usb_task_thread: start\n));
s = splusb();
for (;;) {
task = TAILQ_FIRST(usb_all_tasks);
Joe
--
John Baldwin [EMAIL PROTECTED]http://www.FreeBSD.org/~jhb/
Power Users Use the Power to Serve! - http://www.FreeBSD.org/
To Unsubscribe: send mail to [EMAIL PROTECTED]
with unsubscribe freebsd-current in the body of the message
Re: panic sleeping without a mutex in usb_task_thread
On Mon, Apr 08, 2002 at 01:37:55PM -0400, John Baldwin wrote: On 08-Apr-2002 Josef Karthauser wrote: On Mon, Apr 08, 2002 at 12:33:40PM -0400, John Baldwin wrote: show witness isn't useful to most people so I would avoid it unless someone explicitly asks for it. The problem here is likely due to the usb_task_thread() not locking Giant when it starts up. We probably want this: Probably unless some USB specific locks are added instead, but this is the easier fix for the time being. Cool. I've committed this; where should I look to get a low down on what to lock, when and how? Is there anything other than source to refer to? Joe msg37104/pgp0.pgp Description: PGP signature
Re: panic sleeping without a mutex in usb_task_thread
On 08-Apr-2002 Josef Karthauser wrote: On Mon, Apr 08, 2002 at 01:37:55PM -0400, John Baldwin wrote: On 08-Apr-2002 Josef Karthauser wrote: On Mon, Apr 08, 2002 at 12:33:40PM -0400, John Baldwin wrote: show witness isn't useful to most people so I would avoid it unless someone explicitly asks for it. The problem here is likely due to the usb_task_thread() not locking Giant when it starts up. We probably want this: Probably unless some USB specific locks are added instead, but this is the easier fix for the time being. Cool. I've committed this; where should I look to get a low down on what to lock, when and how? Is there anything other than source to refer to? Nothing besides the source really. If it's walking a list you might use a lock to protect the list for example, but then you might need to conditionally get Giant while performing an action on the list. Joe -- John Baldwin [EMAIL PROTECTED]http://www.FreeBSD.org/~jhb/ Power Users Use the Power to Serve! - http://www.FreeBSD.org/ To Unsubscribe: send mail to [EMAIL PROTECTED] with unsubscribe freebsd-current in the body of the message
