Re: [Qemu-devel] [PATCH v2 1/9] log: Add new IOMMU type

["Edgar E. Iglesias" <edgar.iglesias@gmail.com>]

On Mon, Sep 26, 2016 at 08:54:52AM +0200, Auger Eric wrote:
> Hi Edgar, Prem,
> 
> On 25/09/2016 16:58, Edgar E. Iglesias wrote:
> > On Tue, Sep 13, 2016 at 01:53:39AM +0530, Prem Mallappa wrote:
> >> On Fri, Sep 9, 2016 at 9:06 PM, Auger Eric <eric.auger@redhat.com> wrote:
> >>
> >>> Hi Prem,
> >>>
> >>> Missing commit message
> >>>
> >>>> Signed-off-by: Prem Mallappa <prem.mallappa@broadcom.com>
> >>>> ---
> >>>>  include/qemu/log.h | 1 +
> >>>>  util/log.c         | 2 ++
> >>>>  2 files changed, 3 insertions(+)
> >>>>
> >>>> diff --git a/include/qemu/log.h b/include/qemu/log.h
> >>>> index 234fa81..3dd2131 100644
> >>>> --- a/include/qemu/log.h
> >>>> +++ b/include/qemu/log.h
> >>>> @@ -42,6 +42,7 @@ static inline bool qemu_log_separate(void)
> >>>>  #define CPU_LOG_TB_NOCHAIN (1 << 13)
> >>>>  #define CPU_LOG_PAGE       (1 << 14)
> >>>>  #define LOG_TRACE          (1 << 15)
> >>>> +#define CPU_LOG_IOMMU      (1 << 16)
> >>> why is it prefixed with CPU_ ?
> >>> besides all arm gic devices seem to use LOG_GUEST_ERROR. what is the
> >>> rationale behind introducing a new enum?
> >>>
> >>
> >> Will change this to LOG_GUEST_ERROR, if others on the list are okay.
> > 
> > Hi,
> > 
> > LOG_GUEST_ERROR is used for cases when the guest programs things in bad
> > way. E.g sets up a register in an invalid manner or writes to regs that
> > don't exist.
> > 
> > In this case we're logging information for valid translation steps, I
> > would prefer if we could use something else than LOG_GUEST_ERROR.
> > An IOMMU logging class sounds good to me.
> 
> Thank you for the clarification; so indeed LOG_GUEST_ERROR which was my
> suggestion is not a good idea. With respect to that patch I was also
> wondering whether the CPU_ prefix was relevant.

Right, the CPU_ prefix should probably be dropped if we use a new IOMMU log mask.

Cheers,
Edgar