[Crash-utility] [PATCH 0/2] dis: Introduce the -f option

[Dave Anderson]

----- Original Message -----
> 
> 
> ----- Original Message -----
> > On Tue 2015-08-11 11:47 -0400, Dave Anderson wrote:
> > [ ... ]
> > > The bug is caused by this part of the 2/2 patch:
> > > 
> > >   @@ -1655,8 +1660,15 @@ cmd_dis(void)
> > >   
> > >                           extract_hex(buf2, &curaddr, ':', TRUE);
> > >   
> > >   +                       if (forward) {
> > >   +                               if (curaddr != target)
> > >   +                                       continue;
> > >   +                               else
> > >   +                                       forward = FALSE;
> > >   +                       }
> > >   +
> > >                           if (!reverse)
> > >   -                               if ((req->flags & GNU_FUNCTION_ONLY) &&
> > >   +                               if (!count_entered &&
> > >                                       (curaddr >= req->addr2))
> > >                                           break;
> > > 
> > > req->addr2 is 0 in this case, so it always breaks/fails.
> > 
> > Indeed; req->addr2 is only relevant in the case of: reverse, forward
> > or GNU_FUNCTION_ONLY. Where a single instruction is specified, we should
> > not evaluate here. As such the following change should resolve this bug:
> > 
> 
> > --- a/kernel.c
> > +++ b/kernel.c
> > @@ -1669,7 +1669,7 @@ cmd_dis(void)
> >                         }
> >  
> >                         if (!reverse)
> > -                               if (!count_entered &&
> > +                               if (!count_entered && req->addr2 &&
> >                                     (curaddr >= req->addr2))
> >                                         break;
> >  
> > 
> > --
> 
> OK good.  I'm going to run some sanity-checks on most of the other architectures.
> I don't foresee any problems, and hopefully I should get this checked in today.
> 
> Thanks,
>   Dave

Hi Aaron,

The testing went OK as expected, although I did make a couple minor additions.
In help.c, I added "f" to the SYNOPSIS line, and added an example to the help
page.  And in kernel.c, I made the -f and -r options mutually exclusive.  

Queued for crash-7.1.3:

  https://github.com/crash-utility/crash/commit/4935c333a6cd1e7a8949eda53319b8550640cfc0

Nice cleanup of cmd_dis() and a nifty addition to the toolbox!

Thanks,
  Dave