Re: Getting at an address...

["Alan D. Brunelle" <Alan dot Brunelle at hp dot com>]

Frank Ch. Eigler wrote:
> "Alan D. Brunelle" <Alan.Brunelle@hp.com> writes:
>
>   
> > I am working on trying to track down a potential hot lock used within
> > the QLA device driver, and would like to sue STAP 
> >     
>
> Oh no!  Call in the lawyers!
>   
:-)
s/sue/use/

>   
> > [...] what I want is the address of
> > $sp->lun_queue->fclun->fcport->ha->hardware_lock
> >     
>
> OK.
>
>   
> > what follows "works" in that I think I get the address of
> > $sp->lun_queue->fclun->fcport->ha [...]
> >     
>
> Actually your function returns the address of a local copy of the
> pointer.
>   
I thought so, but was hoping otherwise! :-)
> But this should get you closer:
>
> # %{
> # #include <..../qla_def.h>  /* for scsi_qla_host */
> # %}
>   
I actually tried this, but qla_def.h isn't "released" onto 
/usr/include/linux - so I'd have to reference the source copy directly...

Anyways, this seems to work - had to add in a (long) cast in hwl_addr...
> #
> # function hwl_addr:long (arg:long)
> # %{
> #         THIS->__retvalue = & ((struct scsi_qla_host *) (THIS->arg))->hardware_lock;
> # %}
> # 
> # probe module("*").function("qla2x00_start_scsi")
> # [...]
> #         lockmap = hwl_addr($sp->lun_queue->fclun->fcport->ha)
> # [...]
>
>
> - FChE
>   
Thanks - this will help narrow this down quickly!
Alan