Re: [Qemu-devel] [Qemu-ppc] [PATCH 4/4] ppc: remove excessive logging

[Joakim Tjernlund]

Riku Voipio <address@hidden> wrote on 2014/07/16 09:55:50:

> From: Riku Voipio <address@hidden>
> To: Joakim Tjernlund <address@hidden>, 
> Cc: Alexander Graf <address@hidden>, Peter Maydell 
<address@hidden>, "address@hidden" <address@hidden>, 
QEMU Developers <address@hidden>
> Date: 2014/07/16 09:55
> Subject: Re: [Qemu-devel] [Qemu-ppc] [PATCH 4/4] ppc: remove excessive 
logging
> 
> On Sat, Jul 12, 2014 at 04:06:09PM +0200, Joakim Tjernlund wrote:
> > Alexander Graf <address@hidden> wrote on 2014/07/12 12:41:05:
> > > 
> > > On 12.07.14 12:40, Peter Maydell wrote:
> > > > On 12 July 2014 10:39, Alexander Graf <address@hidden> wrote:
> > > >> On 12.07.14 10:58, Peter Maydell wrote:
> > > >>> On 12 July 2014 01:39, Alexander Graf <address@hidden> wrote:
> > > >>>> What do the other platforms do on illegal instructions during 
user 
> > mode?
> > > >>>> Any way we can get consistency across the board?
> > > >>> Mostly it looks like they just silently generate the SIGILL.
> > > >>> Consistency has never been our strong point :-)
> > > >>
> > > >> That means this patch brings things towards consistency? It's 
> > certainly good
> > > >> for me then :)
> > > > No, this just removes one use of this logging. If you
> > > > wanted consistency we should remove all of them...
> 
> > > Agreed :)
> 
> > So can I infer from this discussion that you will apply the patch?
> 
> I think Peter and Alex suggest that the EXCP_DUMP() loggings should be
> removed from all cases in PPC code where TARGET_SIGILL is risen. Your
> patch removes just once case, and that would make PPC code become
> internally inconsistent where some SIGILLs are logged and others aren't.

Something like that. This is one step in that direction. We(or I cannot) 
do
the consistency for all cases/arches at once. With the patch we become
one step closer to the Linux kernel so I don't see why not apply it.

> 
> Even more dramatically, we remove the whole EXCP_DUMP and users since 
none
> of the other archs output anything for SIGFPE/SIGSEGV either. After all,
> the Linux kernel (ppc or others) doesn't output anything either. And
> it is the behaviour of linux we try to match.

hmm, not clear to me what you mean here

 Jocke