Cannot get dbx to work correctly with a running process


 
Thread Tools Search this Thread
Top Forums Programming Cannot get dbx to work correctly with a running process
# 1  
Old 02-25-2011
Cannot get dbx to work correctly with a running process

Hi everyone,

I've been struggling with this for a few weeks now. I'm trying to debug a running process with dbx on an AIX box.

The command I'm using is 'dbx -a <pid> core'
There is a function I can perform in my application that crashes this process, but it does not show up as crashed in dbx. However, if I do 'ps -ef | grep <process>' it shows a new PID, so I know something is happening. The output I get when I first start dbx is,

Quote:
Waiting to attach to process 573546 ...
Successfully attached to dps643.
Type 'help' for help.
reading symbolic information ...
stopped in msgrcv at 0x900000000247918 ($t1)
0x900000000247918 (msgrcv+0xd8) e8410028 ld r2,0x28(r1)
(dbx)
If anyone can give me a hand, I'd appreciate it.

---------- Post updated at 12:29 PM ---------- Previous update was at 12:09 PM ----------

Another thing I just noticed is, the PID doesn't change until I actually type 'quit' in dbx. The service appears to crash (application locks up - then displays connection lost message), but the PID doesn't change until I exit my dbx session. Don't know if that matters.
Login or Register to Ask a Question

Previous Thread | Next Thread

10 More Discussions You Might Find Interesting

1. Shell Programming and Scripting

Why does this awk script not work correctly?

I have a large database with English on the left hand side and Indic words on the left hand. It so happens that since the Indic words have been entered by hand, there are duplicates in the entries. The structure is as under: English headword=Indic gloss,Indic gloss A small sample will... (6 Replies)
Discussion started by: gimley
6 Replies

2. Shell Programming and Scripting

Work with setsid to open a session and close it correctly

I would like to create the following script: run a python script with setsid python may or may not fail with exception check if all of the group processes were terminated correctly if not, kill the remaining processes How can I do that? Thanks a lot (3 Replies)
Discussion started by: ASF Studio
3 Replies

3. Gentoo

LDAP-Auth does not work correctly with systemd

Hi, since the upgrade to Gnome 3.6 (now i have 3.8) the authentication over LDAP stops working. The whole machine does not start anymore. The machine boot, but no gdm and no X. I can login, with root, but then the tty hangs. When i look at ttyF12 i see a lot of systemd service the runs random,... (1 Reply)
Discussion started by: darktux
1 Replies

4. SCO

Set NIC correctly , but the network does not work

I'm trying to virtualize an instance of SCO Open Server 5.0.2c in VirtualBox (called VM- A) , I can not configure the network (NIC). The NIC I'm using is PCnet -FAST III (Am79C973 ) (this NIC works with VirtualBox + SCO 5.0.5M) When I add from ' Add new LAN adapter' I detects the NIC... (2 Replies)
Discussion started by: flako
2 Replies

5. UNIX for Advanced & Expert Users

Libvirt does not work correctly anymore on my gentoo

Hi, Since a year my libvirtd does not work anymore on my Gentoodesktop. In the meantime a used virtualbox. But I would like to have back libvirt. The problem was after libvirt should not only work with root privileges. I deinstalled all things with libvirt an kvm. I removed all things from /var... (4 Replies)
Discussion started by: darktux
4 Replies

6. Programming

AIX dbx - attaching to a process that is crashing

Hey everyone, I'm trying to attach to a process that is crashing so I can debug the source code. I've tried this: dbx -a PID stop at "file.cpp":line# However, nothing ever breaks. The service crashes and restarts, but I never see it hitting the code. I've tried to use a core file, but... (2 Replies)
Discussion started by: ctote
2 Replies

7. Programming

dbx - attach to process, break when crash

Hey everyone, I have a process that is crashing, and I'd like to have some way to see where it crashes. Is this possible? (2 Replies)
Discussion started by: ctote
2 Replies

8. Shell Programming and Scripting

Script not running correctly

Hi all, My script below was working fine till I changed the LOGFILE path (which originally points to a file within the same folder of the script for testing) Only the event.log file path is changed, the rest are still within the current folder. Before the change, running the script works.... (1 Reply)
Discussion started by: phixsius
1 Replies

9. Programming

question with running dbx

If your program unfortunately takes a LONG time to finish running in dbx and you find yourself hitting Ctrl+C or Ctrl+Z to stop the running, does it produce any adverse effects at all ? The reason I am asking is because my program is in an infinite loop state (which leads to the huge delay in... (1 Reply)
Discussion started by: JamesGoh
1 Replies

10. SuSE

vsft doesn't work correctly

I install vsftpd server on 2 SUSE 10.2 servers. The first works perfectly, but the second doesn't work how I expect. The second works only over local network and doesn't over internet. The vsftpd.conf and ../xinetd.d/vsftpd are the same in 2 servers. The only different was when I threw to log in... (1 Reply)
Discussion started by: zhivko.neychev
1 Replies
Login or Register to Ask a Question
dumpsys(8)						      System Manager's Manual							dumpsys(8)

NAME
dumpsys - Copies a snapshot of memory to a dump file SYNOPSIS
/sbin/dumpsys [-fisuz] [-r num] directory OPTIONS
Perform a full core dump -- the default is a partial dump. Ignore filesystem space limit warning -- copy the dump even if there is insuf- ficient filesystem space to save it. Only the portion of the dump that fits in the space available is copied. Set the expected dump com- pression ratio, defaulting to 0.5. A lower number means a better compression ratio is expected. Print the expected size of a full and partial dump file -- no dump is taken. Produce a non-compressed dump. Disable contiguous zero suppression. DESCRIPTION
The dumpsys command allows you to save a snapshot of the system memory to a dump file. There are times when system memory requires analy- sis but it may not be possible to halt the system and take a normal crash dump. Many problems can be resolved by taking a snapshot of the system memory while the system is running. The dumpsys command performs this function after it determines that there is enough file system space to save a core dump (see the following for information about minfree). Note that the system is running while dumpsys takes a snapshot of memory. This means that memory may be changing as it is copied. As a result, analysis of the resulting dump may show inconsistencies such as incomplete linked lists and partially zeroed pages. These are features caused by the transitory state of memory, caused by the working system. For this reason, some system problems cannot be detected by dumpsys and you must halt the system to take a normal crash dump. The dumpsys command writes information in directory. By default, directory is /var/adm/crash. The dump contains the contents of a portion of physical memory (or all of physical memory in the case of a full core dump) at the time of the command execution. The dumpsys command saves this information in the file vmzcore.n, or vmcore.n if compression is supressed. The command also copies the kernel executable image, usually /vmunix, to the vmunix.n file. You can then analyze the vmzcore.n and vmunix.n files (See the Kernel Debugging manual for information about analyzing core dump files.) The variable n indicates the number of the core file. For the first file, dumpsys creates the files vmunix.0 and vmcore.0. It then creates a file named directory/bounds and initializes the file with the value 1. For each succeeding dump, the dumpsys command uses the value in the directory/bounds file and then increments that value. By default dumpsys produces specially compressed dump files. The compression scheme used is not as powerful as compress or gzip but has the unusual feature that any byte in the file can be extracted without decompressing more than about 40K (typically less), even if the dump is very large. Tools such as dbx, ladebug, and kdbx, are able to read the compressed core files. The expand_dump utility is provided to con- vert compressed dumps into non-compressed dumps if you want to use an analysis tool that does not understand the compressed format. A traditional non-compressed dump can be generated using the -u switch. In this case the dump file will be named vmcore.n instead of vmz- core.n. It is possible that you may run dumpsys on a kernel that is not recent enough to support compressed dumps. If this happens, dumpsys will display a warning that the kernel is too old, but will then produce a non-compressed dump. The message is suppressed if -u is used. Con- versely, if you try to use older versions of dumpsys with a newer kernel a non-compressed dump is created without a message. The older dumpsys version will not recognize the -u or -r options. The text file directory/minfree specifies the minimum number of kilobytes that must be left on the filesystem containing directory after dumpsys copies the dump. By default, this file does not exist, indicating that the minimum is set to zero. To specify a minimum, create the file and store the number of kilobytes you want reserved in it.You can override the minimum check of directory/minfree using the -i option. The -s option displays the approximate number of disk blocks that full and partial dumps will require. The exact size can not be deter- mined ahead of time for many reasons, such as: By default, dumpsys optimizes disk space requirements by suppressing the writing of contigu- ous zeroes. System use of dynamic memory (malloc/free) changes while the system is in use. If the dump is to be compressed, the ratio by which it will be compressed is not known. The -z option disables contiguous zero suppression. A considerable amount of memory consists of contiguous zeros, that do not need to be written to disk. The dumpsys command optimizes disk space by default, but optimization of disk space causes longer execution times. If you specify the -z option, the run time can be 25% faster, although you will require more disk space. Note that if the dump is to be com- pressed, zero supression is not usedm therefore the -z option is meaningless. With the exception of the -s option, execution of dumpsys requires root (superuser) access rights. EXIT STATUS
Success -- dump taken General error -- dump failed Insufficient file system space -- dump failed FILES
Specifies the number of the next dump Specifies the minimum number of kilobytes to be left after dump files are written SEE ALSO
Commands: dbx(1), expand_dump(8), savecore(8) System Administration dumpsys(8)