[one-users] Generating core dump using SIGQUIT

Jeroen Nijhof jeroen.nijhof at sara.nl
Thu Jul 22 02:41:48 PDT 2010


Hi,

The crash is caused by the function xmlrpc_array_get_item from 
libxmlrpc. This function has not become deprecated for nothing, instead 
they should use xmlrpc_array_read_item(). I though it was fixed in 
xmlrpc-c version 1.16.07.

There is a debian experimental package available for this version:
http://packages.debian.org/source/experimental/xmlrpc-c

I'm not sure about ubuntu though.

Regards,
  Jeroen Nijhof

On 07/22/2010 11:31 AM, Shashank Rachamalla wrote:
> Hi
>
> Consider the following piece of code from Nebula.cc
>
>   // -----------------------------------------------------------
>   // Block all signals before creating any Nebula thread
>   // -----------------------------------------------------------
>
>   sigfillset(&mask);
>
>   pthread_sigmask(SIG_BLOCK, &mask, NULL);
>
>
> I think the above line is blocking signals which can cause core dump. is
> this true ?. if yes, is this desired ?
>
>   ABRT      6   core
>      FPE      8   core
>         ILL      4   core
>     QUIT      3   core
>   SEGV    11   core
>   TRAP      5   core
>
> However, I could get the dump after commenting the above line in code
> and rebuilding binaries. I had to set the following env variable for
> console stack trace.
>
> export MALLOC_CHECK_=3
>
> oneadmin at vx64p:~$ *** glibc detected *** /usr/bin/oned: free(): invalid
> pointer: 0x0000000000805ea0 ***
> ======= Backtrace: =========
> /lib/libc.so.6[0x7f4abe4bc2f6]
> /usr/lib/libxmlrpc.so.3(
> xmlrpc_array_get_item+0x32)[0x7f4abf58d8f2]
> /usr/lib/libxmlrpc.so.3[0x7f4abf58c9c1]
> /usr/lib/libxmlrpc.so.3(xmlrpc_parse_value_va+0x35)[0x7f4abf58cd15]
> /usr/lib/libxmlrpc.so.3(xmlrpc_parse_value+0x83)[0x7f4abf58cda3]
> /usr/lib/libxmlrpc_server.so.3(xmlrpc_dispatchCall+0x16e)[0x7f4abfdd888e]
> /usr/lib/libxmlrpc_server.so.3(xmlrpc_registry_process_call+0xec)[0x7f4abfdd89bc]
> /usr/lib/libxmlrpc_server_abyss.so.3[0x7f4abffdeaa0]
> /usr/lib/libxmlrpc_abyss.so.3[0x7f4abfbcf9f2]
> /usr/lib/libxmlrpc_abyss.so.3[0x7f4abfbcfae9]
> /usr/lib/libxmlrpc_abyss.so.3[0x7f4abfbcbfd7]
> /usr/lib/libxmlrpc_abyss.so.3[0x7f4abfbd21f8]
> /lib/libpthread.so.0[0x7f4abf9afa04]
> /lib/libc.so.6(clone+0x6d)[0x7f4abe525d4d]
>
> Note that we are using Java API to talk to opennebula.  Also, we are
> using libxmlrpc with version 1.06.27-1ubuntu6.1
>
> please throw some light on this issue.
>
> On 21 July 2010 13:07, Shashank Rachamalla
> <shashank.rachamalla at hexagrid.com
> <mailto:shashank.rachamalla at hexagrid.com>> wrote:
>
>     Hi
>
>     Are there any additional steps one should follow while compiling
>     opennebula on a x86_64 machine ?
>
>     I am making debian files from the compiled binaries and using them
>     on other x86_64 machines. should i keep anything in mind while doing
>     this ?
>
>     I could have used the debian installer distributed by opennebula but
>     i am waiting for 1.4.2 version. current version is 1.4.0.
>
>
>     On 20 July 2010 21:42, Shashank Rachamalla
>     <shashank.rachamalla at hexagrid.com
>     <mailto:shashank.rachamalla at hexagrid.com>> wrote:
>
>         Hi
>
>         tried setting core file size to a max value. still no luck.
>
>         also any ideas on this general ip protection errors. i am
>         getting these errors even after switching down apparmor on
>         ubuntu lucid.
>
>
>         On 20 July 2010 17:20, Manish Sapariya <manish at gslab.com
>         <mailto:manish at gslab.com>> wrote:
>
>             Hi Shashank,
>             Try setting core size to big value, typically its 0.
>
>             # ulimit -a
>             core file size          (blocks, -c) 0
>
>             shows you limits of various size. You can set core
>             file size using
>
>             #ulimit -c 999999999999
>
>             and using the same shell start oned and then try to
>             generate the core dump.
>
>             Hope this helps.
>
>             Thanks and Regards,
>             Manish
>
>             Office Phone : 46711053
>             Extension    : #1053
>
>
>             On 7/20/2010 3:00 PM, Shashank Rachamalla wrote:
>
>                 Hi
>
>                 I have set ONE_MAD_DEBUG=1 in /etc/one/defaultrc and started
>                 opennebula.  After that, I have sent SIGQUIT ( kill -3 )
>                 to oned
>                 process. I couldn't see any core dump generated. I have
>                 compiled
>                 opennebula from source on x86_64 arch but it has crashed
>                 a couple of
>                 times yesterday with the following messages in syslog:
>
>                 Jul 19 20:01:30 vx64p kernel: [418164.847674]
>                 oned[24466] general
>                 protection ip:7f50848d6954 sp:7f508676bb00 error:0 in
>                 libxmlrpc_util.so.3.6.15[7f50848d5000+2000]
>
>                 (and also)
>
>                 Jul 20 00:59:29 vx64p kernel: [363056.672265]
>                 oned[14290] general
>                 protection ip:7fb753aecc19 sp:7fb756824b00 error:0 in
>                 libc-2.10.1.so <http://libc-2.10.1.so>
>                 <http://libc-2.10.1.so>[7fb753a72000+166000]
>
>
>                 I have checked out "general protection fault" on wiki (
>                 http://en.wikipedia.org/wiki/General_protection_fault ).
>                 However, I am
>                 not able to determine if is a problem with selinux (
>                 apparmor on ubuntu
>                 ) or an issue with 64 bit architecture. Hence, I was
>                 testing the core
>                 dump functionality in order to ensure that the dump gets
>                 created when it
>                 actually crashes. Please let me know if I am missing
>                 anything here.
>
>                 --
>                 Regards,
>                 Shashank Rachamalla
>
>                 --
>                 This message has been scanned for viruses and
>                 dangerous content by *MailScanner*
>                 <http://www.mailscanner.info/>, and is
>                 believed to be clean.
>
>
>
>                 _______________________________________________
>                 Users mailing list
>                 Users at lists.opennebula.org
>                 <mailto:Users at lists.opennebula.org>
>                 http://lists.opennebula.org/listinfo.cgi/users-opennebula.org
>
>
>
>
>         --
>         Regards,
>         Shashank Rachamalla
>
>
>
>
>     --
>     Regards,
>     Shashank Rachamalla
>
>
>
>
> --
> Regards,
> Shashank Rachamalla




More information about the Users mailing list