[Bug kprobes/22147] stopped to probe do_arch_ctrl_64

dsmith at redhat dot com sourceware-bugzilla@sourceware.org
Tue Sep 19 16:00:00 GMT 2017


https://sourceware.org/bugzilla/show_bug.cgi?id=22147

--- Comment #3 from David Smith <dsmith at redhat dot com> ---
(In reply to David Smith from comment #2)

Here's something else about this probe:

> ====
> probe kernel.function("do_arch_prctl_64").call 
> {
>  if ( pid() == target() ) {
>     if ( ulonglong_arg(2) ==  0x1002 ) {
>       ++cnt_fs_base_set
>       printf ("do_arch_prctl_64 ARCH_SET_FS fun %s option %x arg %x\n",
> ppfunc(), ulonglong_arg(2), ulonglong_arg(3) )
>      printf( " tid %d current fs base %x \n", tid(), my_get_fs_base() )
> 
>     }
>  }
> }
> ====

In the upstream kernel source, that function is defined like this:

====
long do_arch_prctl_64(struct task_struct *task, int option, unsigned long
arg2);
====

You are treating argument 2 as a ulonglong (8 bytes on x86_64), where it is
really an int (4 bytes on x86_64). So, you are probably getting incorrect
values for argument 2. So, you should use 'int_arg(2)', not 'ulonglong_arg(2)'.
If you aren't getting incorrect values, you are getting lucky.

(Or course argument 3 isn't really a ulonglong, but since a ulonglong takes the
same number of bytes as a long on x86_64 you should be OK there.)

-- 
You are receiving this mail because:
You are the assignee for the bug.


More information about the Systemtap mailing list