Re: [Gems-users] Ruby magic instructions interference: Warning: in fn void magic_instruction_callback

[Dan Gibson <degibson@xxxxxxxx>]

The shift has been a subject of much confusion internally and among our 
users, as some versions of ruby use shifted magic numbers and others do 
not -- since the magic number is provided by a sethi instruction, the 
original intent was to have "sethi 4 $g0" correspond to magic number 4, 
for readability.
Nonetheless, the end result is the same. 6000 >> 16 is still 0, so Ruby 
is still treating the value as unhandled, which is, inevitably, what you 
probably want.
Regards,
Dan


Thomas De Schampheleire wrote:
> Hi,
>
> Thanks for the quick reply. See below:
>
> On 5/16/07, Dan Gibson <degibson@xxxxxxxx> wrote:
>   
> > Ruby handles assumes magic instructions are directives passed to Ruby to
> > perform various tasks (eg. begin_transaction, etc.). Its actually a
> > *good* thing that you're seeing warnings -- its because Ruby doesn't
> > know how to interpret the magic number, and therefore isn't doing anything.
> >
> > You should have a look at commands.C to see what values Ruby associates
> > with various actions. So long as you don't use Ruby's magic numbers, you
> > will not incur any Ruby-strangeness due to magic calls.
> >
> > Since you're using magic number 0, you can safely ignore these warnings.
> > You can trivially modify commands.C to recognize "0" as a null command
> > and not print the warning at all.
> >     
> Well, actually I'm using magic numbers in the range of 1000 upto 6000
> or something. However, I noticed that in the magic ruby code in
> commands.C, there is a right shift:
>   val = val >> 16;
> which effectively makes my magic numbers appear to be zero.
>
> Is there a particular reason why this shift takes place?
> Due to the shift, it seems that Ruby only uses magic numbers above
> 2^16 = 65 536.
>
> Thanks, thomas
>
>   
> > Regards,
> > Dan
> >
> > Thomas De Schampheleire wrote:
> >     
> > > Hi,
> > >
> > > I am using Simics 2.2.19 and Ruby release 1.2.
> > > I have added some magic instructions to the OpenSolaris kernel
> > > (UltraSPARC architecture), which are handled by my own callback
> > > function. This works as I expect.
> > >
> > > However, when adding Ruby to the simulation, I get warnings on every
> > > magic instruction:
> > >
> > > cpu0 [ cycle 56876262505L ]: [ magic 1021 ] disp_getwork()
> > > Warning: in fn void magic_instruction_callback(void*, void*,
> > > integer_t) in simics/commands.C:158: val is 0
> > > Warning: in fn void magic_instruction_callback(void*, void*,
> > > integer_t) in simics/commands.C:158: val is 0
> > > Warning: in fn void magic_instruction_callback(void*, void*,
> > > integer_t) in simics/commands.C:159: Unexpected magic call
> > > Warning: in fn void magic_instruction_callback(void*, void*,
> > > integer_t) in simics/commands.C:159: Unexpected magic call
> > >
> > > The first line is the line that is printed out by my own handler, and
> > > the subsequent four lines seem to come from Ruby code.
> > >
> > > Can I avoid having these warnings by using other numbers for my magic
> > > instructions? (which range of magic instructions is used by Ruby?)
> > > Can my magic instructions interfere with Ruby?
> > > Can I possibly turn off these warnings?
> > >
> > > Thank you,
> > > Thomas
> > > _______________________________________________
> > > Gems-users mailing list
> > > Gems-users@xxxxxxxxxxx
> > > https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> > > Use Google to search the GEMS Users mailing list by adding "site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.
> > >
> > >
> > >       
> > --
> > http://www.cs.wisc.edu/~gibson [esc]:wq!
> >
> > _______________________________________________
> > Gems-users mailing list
> > Gems-users@xxxxxxxxxxx
> > https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> > Use Google to search the GEMS Users mailing list by adding "site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.
> >
> >
> >     
> _______________________________________________
> Gems-users mailing list
> Gems-users@xxxxxxxxxxx
> https://lists.cs.wisc.edu/mailman/listinfo/gems-users
> Use Google to search the GEMS Users mailing list by adding "site:https://lists.cs.wisc.edu/archive/gems-users/"; to your search.
>
>   
--
http://www.cs.wisc.edu/~gibson [esc]:wq!