Hi, > > I wouldn't mind printing the pointer for %p[mMI], but %pa prints the > > *content* of the pointer which is really confusing. I.e. in > > pinctrl-single.c the reg value pairs are printed like > > > > dev_dbg(dev, "reg/val %pa/0x%08x\n", ®, val); > > > > with reg being a pointer to a physical address. So with tiny_printf > > the address of reg (which is a pointer to the stack) is printed in > > this case. > > > > I don't think we can print %p without putting more logic into the > > decoding. I think the culprit here is the fallthrough to %x, which > > then leads to the confusing behavior shown above. IMHO if we want to > > avoid that, we'd have to make %p entirely unsupported. > > > > diff --git a/lib/tiny-printf.c b/lib/tiny-printf.c > > index faf55d7f327..8147ffa2c1b 100644 > > --- a/lib/tiny-printf.c > > +++ b/lib/tiny-printf.c > > @@ -269,21 +269,18 @@ static int _vprintf(struct printf_info *info, const > > char *fmt, > > va_list va) > > div_out(info, &num, div); > > } > > break; > > +#if CONFIG_IS_ENABLED(NET) || CONFIG_IS_ENABLED(NET_LWIP) || _DEBUG > > What if we fine-tune tinyprinf via config here? > For example SPL_USE_TINY_PRINTF_POINTER_SUPPORT and > select it by NET or NET_LWIP. If someone needs it, > the pointer output can be enabled, otherwise '?' for > unsupported is output.
Yeah I had a similar idea, but I'm not sure if yet another config symbol is worth it. That's up to the maintainer to decide :) In any case, we have a different behavior to what is printed right now, as we drop the fallthrough to %x. Tom? Simon? -michael
signature.asc
Description: PGP signature
