topic Re: tusc output in Operating System - HP-UX

tusc output

Donald Thaler — Thu, 28 May 2009 00:49:55 GMT

oracle had me run the tusc to trace a process called genclntsh... this was the first error i saw in the output file:

[10527] mmap(NULL, 0, PROT_NONE, MAP_VARIABLE|0x200000, -1, 0) ERR#25 ENOTTY

any idea what this means ??

Re: tusc output

Dennis Handly — Thu, 28 May 2009 05:03:24 GMT

>this was the first error I saw in the output file:

You may want to list a window of system calls around your error.
Do you have the correct tusc for your OS version?
That length of 0, doesn't look right.

You may want to look for more errors.

Re: tusc output

Steven E. Protter — Thu, 28 May 2009 07:52:05 GMT

Shalom,

It would be helpful to see a little more data.

SEP

Re: tusc output

Donald Thaler — Thu, 28 May 2009 08:27:10 GMT

a little more data:
[10504] execve(0x9fffffffffffdf20, 0x9fffffffffffea80, 0x9fffffffffffea90) [entry]
argv[0] @ 0x9fffffffffffed12: "genclntsh"
[10504] execve("/u01/app/oracle/product/10.2.0.4/bin/genclntsh", 0x9fffffffffffea80, 0x9fffffffffffea90) = 0 [32-bit]
[10504] mmap(NULL, 8192, PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x777fc000
[10504] open("/usr/lib/hpux32/dld.so", O_RDONLY, 0) ...... = 4
[10504] read(4, "7fE L F 0102010101\0\0\0\0\0\0\0".., 1024) = 1024
[10504] mmap(NULL, 665520, PROT_READ|PROT_EXEC, MAP_SHARED|MAP_FILE|MAP_SHLIB, 4, 0) = 0xc0014000
[10504] sysconf(_SC_PAGE_SIZE) ........................... = 4096
[10504] mmap(NULL, 6712, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS|MAP_SHLIB, -1, 0) = 0x777fa000
[10504] mmap(0x777f7000, 11416, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_FILE|MAP_SHLIB, 4, 720896) = 0x777f7000
[10504] close(4) ......................................... = 0
[10504] mmap(NULL, 8192, PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x777f5000
[10504] sysconf(_SC_PAGE_SIZE) ........................... = 4096
[10504] stat("/usr/lib/hpux32/dpd", 0x7fffe730) .......... = 0
[10504] open("/usr/lib/hpux32/dpd", O_RDONLY, 0) ......... = 4
[10504] fcntl(4, F_SETFD, 0) ............................. = 0
[10504] mmap(NULL, 16384, PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE|MAP_ANONYMOUS, -1, 0) = 0x777f0000
[10504] getdents(4, 0x777f0028, 8192) .................... = 80
[10504] getdents(4, 0x777f0028, 8192) .................... = 0
[10504] close(4) ......................................... = 0
[10504] getuid() ......................................... = 304 (304)
[10504] getgid() ......................................... = 304 (304)
[10504] open("/bin/sh", O_RDONLY, 0) ..................... = 4
[10504] pread(4, "\0\0\002\0\0\0, \0\0\0\aH P \0\0".., 60, 404) = 60
[10504] close(4) ......................................... = 0
[10504] utssys(0x7fffe3b0, 0, 0) ......................... = 0
[10504] gettune("privileges_enabled", 0x7fffe370) ........ ERR#2 ENOENT
[10504] open("/usr/lib/hpux32/libc.so.1", O_RDONLY, 0) ... = 4
[10504] fstat(4, 0x7fffe680) ............................. = 0

and the tusc version i'm using is for hp-ux
obtained from http://hpux.connect.org.uk/hppd/hpux/Sysadmin/tusc-7.9/

is the err#2 noent an error ??
if so what does it mean...

the entire tusc output was too large for me to paste the whole thing here..

the tusc output is currently under review by oracle.. is there anything i could look for in the tusc output that would identify why the genclntsh process is not working ??

Re: tusc output

Donald Thaler — Thu, 28 May 2009 08:30:15 GMT

found an error but don't know how to interpret it:

[10527] write(2, "l d : C a n ' t f i n d l ".., 37) = 37
[10527] write(2, "\n", 1) ................................ = 1
[10527] write(2, "F a t a l e r r o r . ", 12) ......... = 12
[10527] write(2, "\n", 1) ................................ = 1
[10527] exit(1) .......................................... WIFEXITED(1)

Re: tusc output

Laurent Menase — Thu, 28 May 2009 08:44:54 GMT

the errno is somewhat weird for mmap()
all those params looks weird.
Are you using the right version of tusc?
Are you using some 3rd party kernel intrusive code which could change syscalls?

Re: tusc output

Laurent Menase — Thu, 28 May 2009 08:54:35 GMT

Typically it is the lines above your
found an error but don't know how to interpret it:

[10527] write(2, "l d : C a n ' t f i n d l ".., 37) = 37
[10527] write(2, "\n", 1) ................................ = 1
[10527] write(2, "F a t a l e r r o r . ", 12) ......... = 12
[10527] write(2, "\n", 1) ................................ = 1
[10527] exit(1) .......................................... WIFEXITED(1)
which are interesting.

Re: tusc output

Donald Thaler — Thu, 28 May 2009 11:15:32 GMT

what i've been able to discern is tha the file librt.so can't be found, i can see it searching thru the path directories. the file actually exists in /usr/lib/hpux64 but that director isn't being searched, which is probably the problem. i ran the tusc routine on my other server where the genclntsh routine works, and when i look at the tusc.out file i can see it looking for it in /usr/lib/hpux64 ?

updated the path statement so that the /usr/lib/hpux64 is first but that didn't help...

thanks for your efforts, does anyone have an idea how i can force it to look at /usr/lib/hpux64

again the tusc routine was downloaded from the hp site and i definitely selected the version for hp-ux 11.23...

Re: tusc output

Denver Osborn — Thu, 28 May 2009 16:26:17 GMT

Look at the output of chatr and ldd against the binary. chatr (change attributes) will show if SHLIB_PATH is enabled. If SHLIB_PATH is enable, set it or change it.

If you run ldd against the binary and it complains about a lib, it may not be in your SHLIB_PATH.

hope this helps,
-denver

Re: tusc output

rick jones — Thu, 28 May 2009 23:49:16 GMT

Something in /usr/lib/hpux64 would only be applicable to a 64-bit application. If the application were 32-bit then even if /usr/lib/hpux64 were in the search path it would not matter.

Re: tusc output

Dennis Handly — Fri, 29 May 2009 10:25:27 GMT

>a little more data:

I see no ENOTTY.

>is the err#2 noent an error? if so what does it mean.

It depends, it seems dld.so has an error recovery. See gettune(2):
The specified tunable parameter does not exist.

>the entire tusc output was too large for me to paste the whole thing here.

That's why I said "window".

>is there anything I could look for in the tusc output that would identify why the genclntsh process is not working?

Sure, errors like you found later.

>found an error but don't know how to interpret it:

You need to have tusc have larger write buffers.
tusc ... -w2 -b 200 ...

>the file actually exists in /usr/lib/hpux64 but that director isn't being searched

It should also exist in hpux32/.

>i can see it looking for it in /usr/lib/hpux64?

Does it find it?

>does anyone have an idea how I can force it to look at /usr/lib/hpux64

It should just do it. Make sure LD_LIBRARY_PATH has it.

Your genclntsh is 32 bit.