Hi Jeff,

Apache segfaults right way on the first attempt to
connect to the database.

Note that the oracle library works correctly
from /usr/bin/perl - the fault occurs only when run
within Apache.

Also, a coworker has verified that, on Fedora Core 6,
this fault does not occur, even within Apache.

I agree that it is strange for this library to call
_dl_make_stack_executable, and i don't think that
it does.  objdump --dynamic-syms on the oracle
library does not show any reference to _make_stack_executable.

So it could be that the stack trace we see has been corrupted.
I don't know what sort of stack frame magic ld-linux may employ,
or if gdb can interpret a 'trampoline' correctly, so i'm uncertain on
this point.

I thought that ld-linux.so may have thought it necessary
to make this call. Apache originally had the elf header

   $readelf -l /usr/sbin/apache | grep STACK
   GNU_STACK      0x000000 0x00000000 0x00000000 0x00000 0x00000 RW  0x4

whereas libclntsh.so lacks the GNU_STACK header.
So, that seemed like a reason why ld-linux might call
_dl_make_stack_executable.

However, editing /usr/sbin/apache to change the
GNU_STACK permissions to RWE, so that the stack
would be executable from the start, did not change
the fault.

My Ubuntu system's glibc was built on 2006-10-10:

   $ /lib/libc-2.4.so
   ...
   Compiled on a Linux 2.6.17-10 system on 2006-10-10.

As this is not the most recent build, I will try to update to
the January build to see if the fault is still present.

-SEan

On Jan 24, 2007, at 6:38 AM, Jeff Bailey wrote:

> It's hard to tell without the code in hand, but my best guess is that
> the oracle library is handing in a bad pointer to some function.
>
> I'm a bit surprised to see a call right from kpummfpg into
> _dl_make_stack_executable.  Does it segfault reasonably quickly, or  
> does
> it have to run for a while?
>
> -- 
> ld-linux.so.2 segfaults in _dl_make_stack_executable
> https://launchpad.net/bugs/81232

-- 
ld-linux.so.2 segfaults in _dl_make_stack_executable
https://launchpad.net/bugs/81232

-- 
ubuntu-bugs mailing list
ubuntu-bugs@lists.ubuntu.com
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to