SIGSEGV, SEGV_ACCERR - Invalid Permissions for object in OCIEnvCreate

[Pinkerik]

When running application, it crashes inside oracle library when calling OCIEnvCreate

But, when I tryed to simulate it, by just calling it alone, it works fine, even tryed to run it inside thread (still ok)

I have literally spent half a day just googling and trying every suggestion I found on forums, but I don't know what to try next..

Just to mention, it works completely fine on other platforms

Problem is reproducible every time, and every time it fails at same point

at call to:

OCIEnvCreate(reinterpret_cast<OCIEnv**>(&oci_env_hndl_), OCI_THREADED, NULL, NULL, NULL, NULL, 0, NULL);

where oci_env_hndl_ is NULL at time being called (checked through gdb)

I'm desperate here, and open to any suggestions

Thx.

[Dennis Handly]

>even tried to run it inside thread

You could have a thread stack overflow.

What does this show?

(gdb) disas $pc-16*12 $pc+16*4

(gdb) set $sp_save = $sp

(gdb) frame 42

(gdb) p $sp - $sp_save

[Pinkerik]

Had to rerun the program, because firewall cut of my ssh connection,

but same crash at the same place,

just posting new backtrace again

[Dennis Handly · Accepted solution]

>(gdb) p $sp - $sp_save $1 = 335024

Yes,  you have a thread stack overflow.  You need something about twice the above value.

0x9fffffffbbf03c10:0 <skgfqio+0x00>: alloc r36=ar.pfs,0,42,8,0

0x9fffffffbbf03c10:1 <skgfqio+0x01>: mov r9=0x45730  << large frame!

0x9fffffffbbf03c30:0 <skgfqio+0x20>: sub r12=r12,r9

0x9fffffffbbf03c60:1 <skgfqio+0x51>: adds r39=16,r12

0x9fffffffbbf03c70:1 <skgfqio+0x61>: st4 [r39]=r0,8  <<< Storing into a local shouldn't give signal

[Pinkerik]

I even tryed to increase thread stack trace through 

pthread_attr_setstacksize

Old thread stack size: 262144
New thread stack size: 8388608

But that still didn't help

[Dennis Handly]

>I even tried to increase thread stack trace through pthread_attr_setstacksize

(Should I assume you know how to use that function and pass the attributes to pthread_create?)

>(gdb) p $sp - $sp_save $1 = 335024

If this calculation didn't change, then you didn't change the stacksize.

[Pinkerik]

I'm so sorry about my last post,

I have a realy bad habit of not reading documentation till the end...

But it works great now

Thank you so much for your help

[Sandeep_HCL]

Hi,

I am facing same issue and try to implement your solution. But, couldn't understand it. Could your please explain your solution in more elaborate manner and it could be greate if you explain it with some example? Quick response is highly appriociated and thanks in advance.

Regards,

Sandeep

[Dennis Handly]

>I am facing same issue and try to implement your solution.

Do you have debug output to prove this?

I.e. what is your thread stack size and what is your usage when it aborts?