Re: [Xen-devel] [PATCH] xen/arm: traps: Add missing 0x in bad_trap

[Ian Campbell <Ian.Campbell@xxxxxxxxxx>]

On Fri, 2014-05-02 at 15:08 +0100, Jan Beulich wrote:
> >>> On 02.05.14 at 15:25, <julien.grall@xxxxxxxxxx> wrote:
> > On 05/02/2014 02:21 PM, Ian Campbell wrote:
> >> On Fri, 2014-05-02 at 14:19 +0100, Julien Grall wrote:
> >>> On 05/02/2014 01:56 PM, Ian Campbell wrote:
> >>>> On Thu, 2014-04-10 at 12:44 +0100, Julien Grall wrote:
> >>>>> The syndrome value is printed in hexadecimal. Prefix it by 0x for less
> >>>>> confusion.
> >>>>>
> >>>>> Signed-off-by: Julien Grall <julien.grall@xxxxxxxxxx>
> >>>>
> >>>> I've acked + applied this. Irrespective of any apparent move to prefer %
> >>>> #x this is an improvement.
> >>>
> >>> Thanks. So what was the conclusion about % vs #x. Which one shall we use?
> >> 
> >> I'm perfectly fine with 0x%x for ARM code. I think Andy's comment about
> >> "Xen moving towards" was overstating things somewhat.
> > 
> > Ok. FIY, Jan Beulich asked me to change some 0x%x into #x in common code
> > when I sent patch a couple of month ago.
> 
> I can only repeat that: The format string is one byte shorter with
> the %#x form, and since this accumulates and we shouldn't be
> making the hypervisor image bigger than it needs to be, I think we
> should prefer that form.

We'd need > 2000 of these before there was an even chance of pushing us
into the next page. Is this really the lowest hanging fruit in terms of
hypervisor size?

>  But of course - as everywhere - the
> maintainer(s) of a particular piece of code has/have the final say.

TBH it's purely cosmetic, for values which one expects to see in hex I
prefer to see 0x0 to 0, especially in e.g. a list of 0 and non-0 items.

Ian.


_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel