dtrace discuss - Feb 2006 - Script for Stackdepth by Thread/LWP?

I''m interested in monitoring the amount of stack used by a
multi-threaded program.  I assume ''stackdepth'' built-in would
be useful...but not sure.  Been through demo''s, ToolKit, and
internals..but it''s just not clicking for me yet.

Not sure how to measure start/end of stack size dynamically...Anyone know how to
break this down?
This message posted from opensolaris.org

G''Day Mike,

On Wed, 15 Feb 2006, Mike Kuhnkey wrote:
> I''m interested in monitoring the amount of stack used by a
> multi-threaded program.  I assume ''stackdepth'' built-in
would be
> useful...but not sure.  Been through demo''s, ToolKit, and
internals..but
> it''s just not clicking for me yet.
>
> Not sure how to measure start/end of stack size dynamically...Anyone
> know how to break this down?
If you just want the stack size,

# dtrace -n ''on-cpu { @[execname] =
max(curthread->t_procp->p_stksize); }''
dtrace: description ''on-cpu '' matched 3 probes
^C

  svc.startd                                                     8192
  fsflush                                                        8192
  sched                                                          8192
  nscd                                                           8192
  bash                                                          12288
  svcs                                                          12288
  svc.configd                                                   16384
  gnome-vfs-daemon                                              16384
  dtrace                                                        16384
  miniserv.pl                                                   24576
  gnome-panel                                                   32768
  gnome-terminal                                                32768
  sshd                                                          57344
  nautilus                                                      73728
  snmpd                                                        348160

no worries,

Brendan

[Sydney, Australia]

On Fri, Feb 17, 2006 at 01:16:00AM +1100, Brendan Gregg
wrote:
> G''Day Mike,
> 
> On Wed, 15 Feb 2006, Mike Kuhnkey wrote:
> 
> > I''m interested in monitoring the amount of stack used by a
> > multi-threaded program.  I assume ''stackdepth''
built-in would be
> > useful...but not sure.  Been through demo''s, ToolKit, and
internals..but
> > it''s just not clicking for me yet.
> >
> > Not sure how to measure start/end of stack size dynamically...Anyone
> > know how to break this down?
> 
> If you just want the stack size,
> 
> # dtrace -n ''on-cpu { @[execname] =
max(curthread->t_procp->p_stksize); }''
> dtrace: description ''on-cpu '' matched 3 probes
> ^C
> 
>   svc.startd                                                     8192
>   fsflush                                                        8192
>   sched                                                          8192
>   nscd                                                           8192
>   bash                                                          12288
>   svcs                                                          12288
>   svc.configd                                                   16384
>   gnome-vfs-daemon                                              16384
>   dtrace                                                        16384
>   miniserv.pl                                                   24576
>   gnome-panel                                                   32768
>   gnome-terminal                                                32768
>   sshd                                                          57344
>   nautilus                                                      73728
>   snmpd                                                        348160
This doesn''t help with multi-threaded programs, however. 
You''re going
to need to do something like:

--- cut here ---
#!/usr/sbin/dtrace -s

this uintptr_t stkinfoptr;
this uintptr_t stkptr;

sched:::on-cpu, profile:::profile-997
{
	this->stkinfoptr = 0;
	this->stkptr = 0;
}

sched:::on-cpu, profile:::profile-997
/execname != "sched"/
{
	this->stkinfoptr = curthread->t_lwp->lwp_ustack;
	this->stkptr = (uintptr_t)0;
}

sched:::on-cpu, profile:::profile-997
/this->stkinfoptr != 0 && curpsinfo->pr_dmodel == PR_MODEL_ILP32/
{
	this->stkinfo32 = *(stack32_t *)copyin(this->stkinfoptr,
	    sizeof (stack32_t));
	this->stktop = (uintptr_t)this->stkinfo32.ss_sp +
	    this->stkinfo32.ss_size;
	this->stkptr = (uintptr_t)uregs[R_SP];
}

sched:::on-cpu, profile:::profile-997
/this->stkinfoptr != 0 && curpsinfo->pr_dmodel == PR_MODEL_LP64/
{
	this->stkinfo = *(stack_t *)copyin(this->stkinfoptr,
	    sizeof (stack_t));
	this->stktop = (uintptr_t)this->stkinfo.ss_sp +
	    this->stkinfo.ss_size;
	this->stkptr = (uintptr_t)uregs[R_SP];
}

sched:::on-cpu, profile:::profile-997
/this->stkptr != 0/
{
	@a[execname] = quantize(this->stktop - this->stkptr);
}

dtrace:::ERROR
{
	@b[execname] = count();
}

dtrace:::END
{
	printa(@a);
	printf("\nErrors:\n");
	printa("    %@d %s\n", @b);
}
--- cut here ---

Cheers,
- jonathan

-- 
Jonathan Adams, Solaris Kernel Development

G''Day Jonathan,

On Thu, 16 Feb 2006, Jonathan Adams wrote:
> On Fri, Feb 17, 2006 at 01:16:00AM +1100, Brendan Gregg wrote:
> > G''Day Mike,
> >
> > On Wed, 15 Feb 2006, Mike Kuhnkey wrote:
> >
> > > I''m interested in monitoring the amount of stack used by
a
> > > multi-threaded program.  I assume ''stackdepth''
built-in would be
> > > useful...but not sure.  Been through demo''s, ToolKit,
and internals..but
> > > it''s just not clicking for me yet.
> > >
> > > Not sure how to measure start/end of stack size
dynamically...Anyone
> > > know how to break this down?
> >
> > If you just want the stack size,
> >
> > # dtrace -n ''on-cpu { @[execname] =
max(curthread->t_procp->p_stksize); }''
> > dtrace: description ''on-cpu '' matched 3 probes
> > ^C
[...]
>
> This doesn''t help with multi-threaded programs, however.
Nope, but tracing them is much harder.
> You''re going to need to do something like:
bah - stop making it look easy! :-)

When I realised I''d have to look into the stack structures themselves,
I
imagined it to be much more convoluted. Very cool script!

Brendan

> --- cut here ---
> #!/usr/sbin/dtrace -s
>
> this uintptr_t stkinfoptr;
> this uintptr_t stkptr;
>
> sched:::on-cpu, profile:::profile-997
> {
[...]

On Sat, Feb 18, 2006 at 10:01:35AM +1100, Brendan Gregg
wrote:
> G''Day Jonathan,
> 
> On Thu, 16 Feb 2006, Jonathan Adams wrote:
> 
> > On Fri, Feb 17, 2006 at 01:16:00AM +1100, Brendan Gregg wrote:
> > > G''Day Mike,
> > >
> > > On Wed, 15 Feb 2006, Mike Kuhnkey wrote:
> > >
> > > > I''m interested in monitoring the amount of stack
used by a
> > > > multi-threaded program.  I assume
''stackdepth'' built-in would be
> > > > useful...but not sure.  Been through demo''s,
ToolKit, and internals..but
> > > > it''s just not clicking for me yet.
> > > >
> > > > Not sure how to measure start/end of stack size
dynamically...Anyone
> > > > know how to break this down?
> > >
> > > If you just want the stack size,
> > >
> > > # dtrace -n ''on-cpu { @[execname] =
max(curthread->t_procp->p_stksize); }''
> > > dtrace: description ''on-cpu '' matched 3 probes
> > > ^C
> [...]
> >
> > This doesn''t help with multi-threaded programs, however.
> 
> Nope, but tracing them is much harder.
> 
> > You''re going to need to do something like:
> 
> bah - stop making it look easy! :-)
> 
> When I realised I''d have to look into the stack structures
themselves, I
> imagined it to be much more convoluted. Very cool script!
Adam''s work in early Solaris 10 to add a "ustack" structure
simplifies
it significantly.  Without that, it would be almost impossible.

Cheers,
- jonathan

-- 
Jonathan Adams, Solaris Kernel Development