On 11 April 2017 at 08:13, Tom Rini <tr...@konsulko.com> wrote:
> On Tue, Apr 11, 2017 at 08:03:07AM -0600, Simon Glass wrote:
>> Hi Tom,
>>
>> On 11 April 2017 at 08:00, Tom Rini <tr...@konsulko.com> wrote:
>> > On Tue, Apr 11, 2017 at 07:56:06AM -0600, Simon Glass wrote:
>> >> On 10 April 2017 at 00:53, Vignesh R <vigne...@ti.com> wrote:
>> >> > Add support for %p, %pa[p], %pM, %pm and %pI4 formats to tiny-printf.
>> >> > %pM and %pI4 are widely used by SPL networking stack and is required if
>> >> > networking support is desired in SPL.
>> >> > %p, %pa and %pap are mostly used by debug prints and hence supported
>> >> > only when DEBUG is enabled.
>> > [snip]
>> >> > +static void pointer(struct printf_info *info, const char *fmt, void
>> >> > *ptr)
>> >> > +{
>> >> > +#ifdef DEBUG
>> >>
>> >> What is the #ifdef DEBUG for? It may not be enabled globally so I
>> >> don't think you can do this. You probably need this code always.
>> >
>> > So, %p/%pa/%pa[p] are used in debug() prints, which only matter when
>> > DEBUG is set. Doing it this way means we globally bloat by (I think I
>> > snipped out..) 25 bytes? instead of ~250 bytes. And since we're in
>> > tiny-printf, which we use when every byte counts, I'm happier about only
>> > bloating by 25 bytes here.
>>
>> What I mean is that typically DEBUG is enabled file by file. So if I
>> want to output something I need to enable DEBUG in this file as well?
>> That seems confusing to me. Perhaps it needs another CONFIG option?
>
> I usually end up whacking DEBUG into common.h myself, so I hadn't
> thought about it that way. Long-term, yeah, we should think about how
> to handle debug stuff as there's times you want everything on and times
> you just want a little bit on, and we're talking about the we-need-space
> case here too.
OK. Would be a nice little project for someone :-)
Reviewed-by: Simon Glass <s...@chromium.org>
_______________________________________________
U-Boot mailing list
U-Boot@lists.denx.de
https://lists.denx.de/listinfo/u-boot
