Re: [Qemu-devel] [PATCH 1/2] hmp: fix info cpus for sparc targets

[Luiz Capitulino <lcapitulino@redhat.com>]

On Wed, 17 Oct 2012 08:06:13 +0200
Aurelien Jarno <aurelien@aurel32.net> wrote:

> On Tue, Oct 16, 2012 at 11:01:45PM -0300, Luiz Capitulino wrote:
> > On Wed, 17 Oct 2012 01:28:34 +0200
> > Aurelien Jarno <aurelien@aurel32.net> wrote:
> > 
> > > On sparc targets, info cpus returns this kind of output:
> > > 
> > > | info cpus
> > > | * CPU #0: pc=0x0000000000424d18pc=0x0000000000424d18npc=0x0000000000424d1c thread_id=19460
> > > 
> > > pc is printed twice, there is no space between pc, pc and npc. As npc
> > > implies pc, don't set has_pc on sparc, and add a space between pc and
> > > npc.
> > > 
> > > Cc: Luiz Capitulino <lcapitulino@redhat.com>
> > > Cc: Markus Armbruster <armbru@redhat.com>
> > > Cc: Blue Swirl <blauwirbel@gmail.com>
> > > Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
> > > ---
> > >  cpus.c |    2 --
> > >  hmp.c  |    2 +-
> > >  2 files changed, 1 insertion(+), 3 deletions(-)
> > > 
> > > diff --git a/cpus.c b/cpus.c
> > > index 750a76f..cbc8556 100644
> > > --- a/cpus.c
> > > +++ b/cpus.c
> > > @@ -1216,8 +1216,6 @@ CpuInfoList *qmp_query_cpus(Error **errp)
> > >          info->value->has_nip = true;
> > >          info->value->nip = env->nip;
> > >  #elif defined(TARGET_SPARC)
> > > -        info->value->has_pc = true;
> > > -        info->value->pc = env->pc;
> > >          info->value->has_npc = true;
> > >          info->value->npc = env->npc;
> > >  #elif defined(TARGET_MIPS)
> > > diff --git a/hmp.c b/hmp.c
> > > index 70bdec2..64b6ab5 100644
> > > --- a/hmp.c
> > > +++ b/hmp.c
> > > @@ -242,7 +242,7 @@ void hmp_info_cpus(Monitor *mon)
> > >              monitor_printf(mon, "nip=0x%016" PRIx64, cpu->value->nip);
> > >          }
> > >          if (cpu->value->has_npc) {
> > > -            monitor_printf(mon, "pc=0x%016" PRIx64, cpu->value->pc);
> > > +            monitor_printf(mon, "pc=0x%016 " PRIx64, cpu->value->pc);
> > 
> > This patch changes qmp_query_cpus() to never set pc, so why do you print it?
> 
> Well the print part of the code isn't change beside adding a space at
> the end. With the current code, has_npc implies has_pc, that's why I
> have removed the latter from qmp_query_cpus() in the sparc case.

My point is that you dropped the pc assignment along:

-        info->value->pc = env->pc;

So, value->pc will be zeroed.

> If you prefer to change this behavior and not have has_npc implying
> has_pc, I can send another patch.

If pc is going to contain relevant info, then we have to set has_pc even
if has_npc implies it. If your point is that pc is irrelevant, then we have
to set has_pc to false and drop its printing from HMP.

> 
> > >              monitor_printf(mon, "npc=0x%016" PRIx64, cpu->value->npc);
> > >          }
> > >          if (cpu->value->has_PC) {
> > 
> > 
>