Re: panic: integer divide fault with vinum
On Sat, 10 Apr 1999, Greg Lehey wrote: ... In fact, it's not necessary. It happens any time you detach the last subdisk from a striped or RAID-5 plex. Ups... Correct. What happened to the stack traces? Don't bother, though, I've reproduced it here. Expect a fix in a couple of hours. Sorry, the stack traces had to be abandoned due to sleep :-) And many thanks for fixing the problem that quickly. Bye! Michael Reifenberger Plaut Software GmbH, R/3 Basis To Unsubscribe: send mail to majord...@freebsd.org with unsubscribe freebsd-current in the body of the message
Re: panic: integer divide fault with vinum
On Friday, 9 April 1999 at 15:01:51 +0200, Michael Reifenberger wrote:
Hi,
next try.
Please forgive if gets too stupid :-)
# cat x
drive d1 device /dev/da1d
drive d2 device /dev/da2d
volume raid
plex org stripe 256k
sd length 50m drive d1
sd length 50m drive d2
(please note again the misspelling of striped)
# vinum create x
1: drive d1 device /dev/da1d
** 1 Can't initialize drive d1: Device not configured
You need to configure it. Use disklabel -e.
2: drive d2 device /dev/da2d
** 2 Can't initialize drive d2: Device not configured
4: plex org stripe 256k
** 4 Invalid plex organization: Invalid argument
As advertised. But this isn't your primary fault, it's the one above.
5: sd length 50m drive d1
** 5 Unnamed sd is not associated with a plex: Invalid argument
Since you don't have a plex, you don't have a subdisk.
6: sd length 50m drive d2
** 6 Unnamed sd is not associated with a plex: Invalid argument
Configuration summary
Drives: 0 (4 configured)
Volumes:1 (4 configured)
Plexes: 0 (8 configured)
Subdisks: 0 (16 configured)
V raid State: down Plexes: 0 Size: 0 B
(juppie no panic!)
Not much of anything, in fact.
(Ok, after repairing 'striped' in x)
# vinum create x
Configuration summary
Drives: 0 (4 configured)
Volumes:2 (4 configured)
Plexes: 1 (8 configured)
Subdisks: 2 (16 configured)
V raid State: down Plexes: 1 Size:100 MB
P raid.p0 S State: down Subdisks: 2 Size:100 MB
S raid.p0.s0State: stalePO:0 B Size: 50 MB
S raid.p0.s1State: stalePO: 256 kB Size: 50 MB
(Hmm, stale sigh. The vinum(8) GOTCHAS 2. is not explicit about my
situation.
Well, I would have thought that the fact you have no drives would give
you something to think about.
this time, no `vinum resetconfig`, so:)
# vinum detach raid.p0
# vinum detach raid.p0.s0
# vinum detach raid.p0.s1
(Maybe this was too stupid and I get a deserved:)
..panic..
IdlePTD 3686400
initial pcb at 2f747c
panicstr: integer divide fault
panic messages:
---
Fatal trap 18: integer divide fault while in kernel mode
instruction pointer = 0x8:0xc0283b27
stack pointer = 0x10:0xc377cc64
frame pointer = 0x10:0xc377ccd8
code segment= base 0x0, limit 0xf, type 0x1b
= DPL 0, pres 1, def32 1, gran 1
processor eflags= interrupt enabled, resume, IOPL = 0
current process = 330 (vinum)
interrupt mask =
trap number = 18
panic: integer divide fault
syncing disks... 3 3 done
running gdb post mortem:
#0 boot (howto=256) at ../../kern/kern_shutdown.c:287
#1 0xc016e6f5 in panic (fmt=0xc02c9ea9 integer divide fault)
at ../../kern/kern_shutdown.c:448
#2 0xc025a8ce in trap_fatal (frame=0xc377cc28, eva=0)
at ../../i386/i386/trap.c:943
#3 0xc025a32c in trap (frame={tf_es = 16, tf_ds = 16, tf_edi = 0,
tf_esi = 102400, tf_ebp = -1015558952, tf_isp = -1015559088, tf_ebx = 0,
tf_edx = 0, tf_ecx = 102400, tf_eax = 1, tf_trapno = 18, tf_err = 0,
tf_eip = -1071105241, tf_cs = 8, tf_eflags = 66118, tf_esp = 0,
tf_ss = -1065135960}) at ../../i386/i386/trap.c:586
#4 0xc0283b27 in __qdivrem (uq=102400, vq=0, arq=0xc377ccf4)
at ../../libkern/qdivrem.c:100
#5 0xc028496f in __umoddi3 (a=102400, b=0) at ../../libkern/umoddi3.c:51
#6 0xc0842fe6 in ?? ()
#7 0xc0848dc4 in ?? ()
#8 0xc0848528 in ?? ()
#9 0xc01a0567 in spec_ioctl (ap=0xc377ce0c)
at ../../miscfs/specfs/spec_vnops.c:440
#10 0xc019fe79 in spec_vnoperate (ap=0xc377ce0c)
at ../../miscfs/specfs/spec_vnops.c:129
#11 0xc02309a9 in ufs_vnoperatespec (ap=0xc377ce0c)
at ../../ufs/ufs/ufs_vnops.c:2327
#12 0xc019aae5 in vn_ioctl (fp=0xc0811040, com=3223602776,
data=0xc377ced0 \001, p=0xc352b780) at vnode_if.h:395
#13 0xc017a114 in ioctl (p=0xc352b780, uap=0xc377cf84)
at ../../kern/sys_generic.c:564
#14 0xc025ab17 in syscall (frame={tf_es = 47, tf_ds = 47,
tf_edi = -1077945576, tf_esi = 1, tf_ebp = -1077945696,
tf_isp = -1015558188, tf_ebx = -1077945732, tf_edx = 0,
tf_ecx = 134781696, tf_eax = 54, tf_trapno = 12, tf_err = 2,
tf_eip = 134619728, tf_cs = 31, tf_eflags = 663, tf_esp = -1077945760,
tf_ss = 47}) at ../../i386/i386/trap.c:1101
#15 0xc024f20c in Xint0x80_syscall ()
#16 0x80483d1 in ?? ()
#17 0x8048295 in ?? ()
#18 0x80480ec in ?? ()
Anything else I can do or inspect?
Well, you could RTFM, in this case vinum(4). It tells you in some
detail what to do if you have a panic. The stack trace you have there
doesn't help. If you give me (root) access to your machine, I'll take
a look at it.
Greg
--
See complete headers for address, home page and phone numbers
finger g...@lemis.com for
Re: panic: integer divide fault with vinum
On Sat, 10 Apr 1999, Greg Lehey wrote: ... You need to configure it. Use disklabel -e. No, they where configured. I have one ' d: 17848151 63 vinum' slice per drive. The only fault where the misspelling of 'stripe' versus 'striped' in the configfile! ... Not much of anything, in fact. (Ok, after repairing 'striped' in x) # vinum create x Configuration summary Drives: 0 (4 configured) Volumes:2 (4 configured) Plexes: 1 (8 configured) Subdisks: 2 (16 configured) V raid State: down Plexes: 1 Size:100 MB P raid.p0 S State: down Subdisks: 2 Size:100 MB S raid.p0.s0State: stalePO:0 B Size: 50 MB S raid.p0.s1State: stalePO: 256 kB Size: 50 MB (Hmm, stale sigh. The vinum(8) GOTCHAS 2. is not explicit about my situation. Well, I would have thought that the fact you have no drives would give you something to think about. Hmm, strange - see my above comment regarding the drive configuration. The same disks, the same configfile (but with proper spelled 'striped': # cat x drive d1 device /dev/da1d drive d2 device /dev/da2d volume raid plex org striped 256k sd length 50m drive d1 sd length 50m drive d2 # vinum create x Configuration summary Drives: 2 (4 configured) Volumes:1 (4 configured) Plexes: 1 (8 configured) Subdisks: 2 (16 configured) D d1State: up Device /dev/da1dAvail: 8664/8714 MB (99%) D d2State: up Device /dev/da2dAvail: 8664/8714 MB (99%) V raid State: up Plexes: 1 Size:100 MB P raid.p0 S State: up Subdisks: 2 Size:100 MB S raid.p0.s0State: up PO:0 B Size: 50 MB S raid.p0.s1State: up PO: 256 kB Size: 50 MB The panic occurs only, if I first use the incorrect config file with 'stripe', then the corrected one with 'striped' and then detaching the plex and the subdisks. BTW: It should'n panic in any case. ... Well, you could RTFM, in this case vinum(4). It tells you in some detail what to do if you have a panic. The stack trace you have there I haven't set up remote debugging. Sorry, its behind a firewall which I can't manipulate and it blocks telnet and ssh at the moment. But I'm shure you can reproduce the panic by following the steps I have taken. Bye! Michael Reifenberger Plaut Software GmbH, R/3 Basis To Unsubscribe: send mail to majord...@freebsd.org with unsubscribe freebsd-current in the body of the message
Re: panic: integer divide fault with vinum
[Format recovered--see http://www.lemis.com/email/email-format.html] On Saturday, 10 April 1999 at 3:37:47 +0200, Michael Reifenberger wrote: On Sat, 10 Apr 1999, Greg Lehey wrote: ... You need to configure it. Use disklabel -e. No, they where configured. I have one ' d: 17848151 63 vinum' slice per drive. It looks like you have them now, but you didn't then. The only fault where the misspelling of 'stripe' versus 'striped' in the configfile! That's not what your error messages are trying to tell you. (Ok, after repairing 'striped' in x) # vinum create x Configuration summary Drives: 0 (4 configured) Volumes:2 (4 configured) Plexes: 1 (8 configured) Subdisks: 2 (16 configured) V raid State: down Plexes: 1 Size:100 MB P raid.p0 S State: down Subdisks: 2 Size:100 MB S raid.p0.s0State: stalePO:0 B Size: 50 MB S raid.p0.s1State: stalePO: 256 kB Size: 50 MB (Hmm, stale sigh. The vinum(8) GOTCHAS 2. is not explicit about my situation. Well, I would have thought that the fact you have no drives would give you something to think about. Hmm, strange - see my above comment regarding the drive configuration. Nope, the error messages are clear. The same disks, the same configfile (but with proper spelled 'striped': # cat x drive d1 device /dev/da1d drive d2 device /dev/da2d volume raid plex org striped 256k sd length 50m drive d1 sd length 50m drive d2 # vinum create x Configuration summary Drives: 2 (4 configured) This isn't what you had last time Volumes:1 (4 configured) Plexes: 1 (8 configured) Subdisks: 2 (16 configured) D d1State: up Device /dev/da1dAvail: 8664/8714 MB (99%) D d2State: up Device /dev/da2dAvail: 8664/8714 MB (99%) V raid State: up Plexes: 1 Size:100 MB P raid.p0 S State: up Subdisks: 2 Size:100 MB S raid.p0.s0State: up PO:0 B Size: 50 MB S raid.p0.s1State: up PO: 256 kB Size: 50 MB The panic occurs only, if I first use the incorrect config file with 'stripe', then the corrected one with 'striped' and then detaching the plex and the subdisks. BTW: It should'n panic in any case. Of course not. Well, you could RTFM, in this case vinum(4). It tells you in some detail what to do if you have a panic. The stack trace you have there I haven't set up remote debugging. Sorry, its behind a firewall which I can't manipulate and it blocks telnet and ssh at the moment. Well, then at least you could do what I explain in vinum(4). But I'm shure you can reproduce the panic by following the steps I have taken. Possibly. I'll try it. Greg -- See complete headers for address, home page and phone numbers finger g...@lemis.com for PGP public key To Unsubscribe: send mail to majord...@freebsd.org with unsubscribe freebsd-current in the body of the message
Re: panic: integer divide fault with vinum
On Sat, 10 Apr 1999, Greg Lehey wrote: ... It looks like you have them now, but you didn't then. Strange, I'm shure I haven't touched them. The only fault where the misspelling of 'stripe' versus 'striped' in the configfile! That's not what your error messages are trying to tell you. That seems right. ... S raid.p0.s0State: stalePO:0 B Size: 50 MB S raid.p0.s1State: stalePO: 256 kB Size: 50 MB (Hmm, stale sigh. The vinum(8) GOTCHAS 2. is not explicit about my situation. ... The panic occurs only, if I first use the incorrect config file with 'stripe', then the corrected one with 'striped' and then detaching the plex and the subdisks. This point is still valid. I just reproduced the panic again. This time the disks showed up and the errormessages where slightly different. They are not important for the panic. I did: )# vinum l Configuration summary Drives: 2 (4 configured) Volumes:2 (4 configured) Plexes: 1 (8 configured) Subdisks: 2 (16 configured) D d1State: up Device /dev/da1dAvail: 8664/8714 MB (99%) D d2State: up Device /dev/da2dAvail: 8664/8714 MB (99%) V raid State: down Plexes: 1 Size:100 MB P raid.p0 S State: down Subdisks: 2 Size:100 MB S raid.p0.s0State: stalePO:0 B Size: 50 MB S raid.p0.s1State: stalePO: 256 kB Size: 50 MB # vinum detach raid.p0 # vinum detach raid.p0.s0 # vinum detach raid.p0.s1 ..panic.. It seems to be important to remove the last stale subdisk. Bye! Michael Reifenberger Plaut Software GmbH, R/3 Basis To Unsubscribe: send mail to majord...@freebsd.org with unsubscribe freebsd-current in the body of the message
Re: panic: integer divide fault with vinum
On Saturday, 10 April 1999 at 4:05:10 +0200, Michael Reifenberger wrote: On Sat, 10 Apr 1999, Greg Lehey wrote: The panic occurs only, if I first use the incorrect config file with 'stripe', then the corrected one with 'striped' and then detaching the plex and the subdisks. This point is still valid. In fact, it's not necessary. It happens any time you detach the last subdisk from a striped or RAID-5 plex. # vinum detach raid.p0 # vinum detach raid.p0.s0 # vinum detach raid.p0.s1 ..panic.. It seems to be important to remove the last stale subdisk. Correct. What happened to the stack traces? Don't bother, though, I've reproduced it here. Expect a fix in a couple of hours. Greg -- See complete headers for address, home page and phone numbers finger g...@lemis.com for PGP public key To Unsubscribe: send mail to majord...@freebsd.org with unsubscribe freebsd-current in the body of the message
