Bug #536

Savecore broken?

Added by wa1ter almost 8 years ago. Updated over 4 years ago.

Status:ClosedStart date:
Priority:NormalDue date:
Assignee:-% Done:

0%

Category:-
Target version:-

Description

I'm running the latest -DEVEL world/kernel built with gcc41. I'm
seeing a kernel panic when trying mount_smbfs, so I've been trying
to get a crashdump to send you.

The problem is this:
# savecore /var/crash
savecore: /kernel: _dumpmag not in namelist

Anyone else seeing this?

History

#1 Updated by bastyaelvtars almost 8 years ago

On Wed, 24 Jan 2007 06:04:37 -0800
walt <> wrote:

Me.
--
Gergo Szakal <>
University Of Szeged, HU
Faculty Of General Medicine

/* Please do not CC me with replies, thank you. */

#2 Updated by corecode almost 8 years ago

fixed

#3 Updated by wa1ter almost 8 years ago

On Wed, 24 Jan 2007, Simon 'corecode' Schubert wrote:

Thanks. Now, do you want the crashdump I finally got? :o)
The panic is 100% reproducible on my end.

#4 Updated by dillon almost 8 years ago

:> Simon 'corecode' Schubert <> added the comment:
:>
:> fixed
:
:Thanks. Now, do you want the crashdump I finally got? :o)
:The panic is 100% reproducible on my end.

Sure. I am putting in the journaling fixes and branching though, it
can't be held off any longer, so any bug fix(es) will have to be
committed to both branches.

-Matt
Matthew Dillon
<>

#5 Updated by wa1ter almost 8 years ago

On Thu, 25 Jan 2007, Matthew Dillon wrote:

Okay, the files are in my crash directory on leaf. I also gave you my
libiconv, libmchain, and smbfs kernel modules because they have to be
loaded in order to use mount_smbfs. For future reference, do you really
need those kernel modules to do your debugging?

Thanks.

#6 Updated by corecode almost 8 years ago

always. they are part of the kernel and equally important.

cheers
simon

#7 Updated by dillon over 7 years ago

:Okay, the files are in my crash directory on leaf. I also gave you my
:libiconv, libmchain, and smbfs kernel modules because they have to be
:loaded in order to use mount_smbfs. For future reference, do you really
:need those kernel modules to do your debugging?
:
:Thanks.

Ok, here is the backtrace:

at /usr/src/sys/platform/pc32/i386/trap.c:1085
#8 0xc02b9748 in trap_pfault (frame=0xcefecab8, usermode=0, eva=6)
at /usr/src/sys/platform/pc32/i386/trap.c:991
#9 0xc02b9c68 in trap (frame=0xcefecab8)
at /usr/src/sys/platform/pc32/i386/trap.c:674
#10 0xc02abc36 in calltrap ()
at /usr/src/sys/platform/pc32/i386/exception.s:783
#11 0xc0290de7 in strcmp (s1=0xcefecb28 "xlat",
s2=0x6 <Address 0x6 out of bounds>) at /usr/src/sys/libkern/strcmp.c:48
#12 0xced388f3 in iconv_sysctl_add (oidp=0xced392c0, arg1=0x0, arg2=0,
req=0xcefecbf0) at /usr/src/sys/libiconv/iconv.c:146
#13 0xc019a5fa in sysctl_root (oidp=Variable "oidp" is not available.
) at /usr/src/sys/kern/kern_sysctl.c:1194
#14 0xc019a70d in userland_sysctl (name=0xcefecc78, namelen=3, old=0xbfbfef9c,
oldlenp=0xbfbfef98, inkernel=0, new=0xbfbfefa0, newlen=108,
---Type <return> to continue, or q <return> to quit---q

I think it may be bounds checking on the iconv_add_in structure.

Please try this patch. If it doesn't work we will need to add more
debugging there to figure out why it is failing.

-Matt
Matthew Dillon
<>

Index: libiconv/iconv.c
===================================================================
RCS file: /cvs/src/sys/libiconv/iconv.c,v
retrieving revision 1.5
diff -u -r1.5 iconv.c
--- libiconv/iconv.c 5 Sep 2006 00:55:46 -0000 1.5
+++ libiconv/iconv.c 26 Jan 2007 18:50:41 -0000
@@ -354,6 +354,11 @@
return EINVAL;
if (din.ia_datalen > ICONV_CSMAXDATALEN)
return EINVAL;
+
+ din.ia_converter[ICONV_CNVNMAXLEN-1] = 0;
+ din.ia_to[ICONV_CSNMAXLEN-1] = 0;
+ din.ia_from[ICONV_CSNMAXLEN-1] = 0;
+
if (iconv_lookupconv(din.ia_converter, &dcp) != 0)
return EINVAL;
error = iconv_register_cspair(din.ia_to, din.ia_from, dcp, NULL, &csp);

#8 Updated by wa1ter over 7 years ago

Matthew Dillon wrote:

Sorry, the panic is exactly the same. :o(

#9 Updated by swildner almost 6 years ago

Walt,

do you still have this smbfs problem or can we close this issue?

#10 Updated by tuxillo almost 5 years ago

Hi,

iconv was a bit reworked recently and the dumping system has been refactored. Is
this still happening? Please, provide some feedback.

Cheers,
Antonio Huete

Also available in: Atom PDF