Tek-Tips is the largest IT community on the Internet today!
Members share and learn making Tek-Tips Forums the best source of peer-reviewed technical information on the Internet!
-
Congratulations Chris Miller on being selected by the Tek-Tips community for having the most helpful posts in the forums last week. Way to Go!
What is signal 11 error? 2
- Thread starter Gesrix
- Start date
screwloose
MIS
#define SIGSEGV 11 /* (*) segmentation violation */
Signals are in /usr/include/sys/signal.h
Signals are in /usr/include/sys/signal.h
- Thread starter
- #3
hi screwloose,
thanks a lot for replying. can you tell me what it is all about? I mean, why it's happening. Actually, the case is bothering me now. Our system is behaving abnormally these few days. It's frequently shutting down. So, when I look at the log of our application. I always see this "Caught by Signal 11" message. Is it related to the memory, or any resources in our system?
Your response will be highly appreciated. thanks.
thanks a lot for replying. can you tell me what it is all about? I mean, why it's happening. Actually, the case is bothering me now. Our system is behaving abnormally these few days. It's frequently shutting down. So, when I look at the log of our application. I always see this "Caught by Signal 11" message. Is it related to the memory, or any resources in our system?
Your response will be highly appreciated. thanks.
MoshiachNow
IS-IT--Management
I would first run "diag" on the memory and check errpt for HW errors using "errpt" command.
If these show no HW issues - possibly you have some bug in the application you are running that causes this by bad memory handling.
Please post the output of "errpt -30" taken around the time of this error occurance.
Thanks
Long live king Moshiach !
If these show no HW issues - possibly you have some bug in the application you are running that causes this by bad memory handling.
Please post the output of "errpt -30" taken around the time of this error occurance.
Thanks
Long live king Moshiach !
-
2
- #5
hello,
Under AIX, the virtual memory is divided in segments.
When a 32 bit process is running, he can access 16 segments of 256 MB, which is 4 GB of memory. These segments have an id from 0x0 to 0xF, and some of them have specific purpose : segment 0 is where the kernel code is accessed from the process, so for example if a program fails and that the faulting adress begins with 0x0...., it means that some kernel code was executed when the process stopped.
When a process asks for some memory, it use the malloc function to request x MB of memory, and this memory is allocated from a pool of segments which is called the data area of the process. By default, a process can not access more than one segment, therefore if no special action is taken, a process can not ask more than 256 MB. If it tries to ask more memory, he will be killed by the kernel with a signal 11 (the other segments are mostly used to store shared libraries code and data).
AIX documentation is sometimes confusing about the maximum number of segments a process can have (8,10, 12 ?? not clear, it changes with every version), but if you want a process to be able to have a data area of (for example) 1 GB, before launching the process you type :
export LDR_CNTRL=MAXDATA=0x4000000
which means the processES you will launch after this command will have 4 segments of 256 MB for its data area. There are other ways than this environment variable to obtain the same result (compile a programm with the -bmaxdata option for example).
Note : when you add some segments to the data area, the same amount of segments will be unavailable for share libraries.
If you are interested in the subject, look in AIX documentaiton with keywords "user process model", or redbook sg24-5674 (C/C++ applications development under AIX, from
To summarize, your process will mostly receive signal 11 when it tries to allocate more memory then the environment allows him to have. My guess is there are probably several ways to make segment violations than allocating more memory that allowed, so my explanations are probably not complete.
regards,
Under AIX, the virtual memory is divided in segments.
When a 32 bit process is running, he can access 16 segments of 256 MB, which is 4 GB of memory. These segments have an id from 0x0 to 0xF, and some of them have specific purpose : segment 0 is where the kernel code is accessed from the process, so for example if a program fails and that the faulting adress begins with 0x0...., it means that some kernel code was executed when the process stopped.
When a process asks for some memory, it use the malloc function to request x MB of memory, and this memory is allocated from a pool of segments which is called the data area of the process. By default, a process can not access more than one segment, therefore if no special action is taken, a process can not ask more than 256 MB. If it tries to ask more memory, he will be killed by the kernel with a signal 11 (the other segments are mostly used to store shared libraries code and data).
AIX documentation is sometimes confusing about the maximum number of segments a process can have (8,10, 12 ?? not clear, it changes with every version), but if you want a process to be able to have a data area of (for example) 1 GB, before launching the process you type :
export LDR_CNTRL=MAXDATA=0x4000000
which means the processES you will launch after this command will have 4 segments of 256 MB for its data area. There are other ways than this environment variable to obtain the same result (compile a programm with the -bmaxdata option for example).
Note : when you add some segments to the data area, the same amount of segments will be unavailable for share libraries.
If you are interested in the subject, look in AIX documentaiton with keywords "user process model", or redbook sg24-5674 (C/C++ applications development under AIX, from
To summarize, your process will mostly receive signal 11 when it tries to allocate more memory then the environment allows him to have. My guess is there are probably several ways to make segment violations than allocating more memory that allowed, so my explanations are probably not complete.
regards,
- Thread starter
- #6
Hi guys,
Thanks a lot for all your replies. I'm leraning from you.
By the way, is there a tool I can use to diagnose it? I want to know exactly what process triggers this error. want to know if it is on the application side. There 200 users concurrently logging daily so I need to find out where we have the botteneck.
Thanks.
Thanks a lot for all your replies. I'm leraning from you.
By the way, is there a tool I can use to diagnose it? I want to know exactly what process triggers this error. want to know if it is on the application side. There 200 users concurrently logging daily so I need to find out where we have the botteneck.
Thanks.
screwloose
MIS
You can use trcrpt for diagnostics. Also you can use ldedit to change the maxdata for the XCOFF header.
- Thread starter
- #10
MoshiachNow
IS-IT--Management
Possibly it's a bug - I would also search an APAR dbase on:
or IBM support that shows severall entries on this error:
Long live king Moshiach !
or IBM support that shows severall entries on this error:
Long live king Moshiach !
- Thread starter
- #13
Hi letis
We are using ROSS Systems' General Ledger module.
This is an excerpt from the log file.
08/05 12:19:16.515 0 0 27120
/gemsrc/gembuild/v61/ref/gm_brk_unix.c:53: SEVERE-ERROR: Caught signal 11
08/05 13:06:49.642 67 0 75376
/gemsrc/gembuild/v61/ref/gem_broker.c:430: ERROR: Broker has ended: Error returned by GMIPC
We are using ROSS Systems' General Ledger module.
This is an excerpt from the log file.
08/05 12:19:16.515 0 0 27120
/gemsrc/gembuild/v61/ref/gm_brk_unix.c:53: SEVERE-ERROR: Caught signal 11
08/05 13:06:49.642 67 0 75376
/gemsrc/gembuild/v61/ref/gem_broker.c:430: ERROR: Broker has ended: Error returned by GMIPC
- Status
Similar threads
- Locked
- Question
- Locked
- Question