Re: Trying Stapdyn Out

[Josh Stone <jistone at redhat dot com>]

On 08/20/2012 12:00 PM, sami wagiaalla wrote:
>>> Failed to create process /usr/bin/ls
>>> --SERIOUS-- #68: Failed to create process for /usr/bin/ls: Process
>>> exited during startup
>>> --FATAL-- #68: Dyninst was unable to create the specified process
>>> --FATAL-- #68: create process failed bootstrap
>>> WARNING: /usr/bin/stapdyn exited with status: 139
>>> Pass 5: run failed.  Try again with another '--vp 00001' option.
>>
>> I think I've seen something like this, and that exit 139 was a SIGFPE
>> somewhere in the dyninst library.  But for me, the problem went away as
>> soon as I did "debuginfo-install dyninst".  Of course that shouldn't be
>> required, and it certainly shouldn't crash with or without debuginfo,
>> but I haven't had a chance to dig into this issue yet.
>>
> 
> Figure it out. It turns out I had to 'setsebool deny_ptrace=0' :) I 
> should have been on the lookout for SELinux denials.

Ah yes, that will block dyninst quite well, as with any debugger. :)

But I'm still worried about that exit 139, since I believe that's still
a crash.  There are other ways that ptrace can fail (e.g. if the target
is already attached), so it ought to accept failure more gracefully.

You also need allow_execstack, which I think I mentioned before,
although I'm looking for ways to get around that.

Josh