Re: [PATCH v5 12/20] vsprintf: add new `%pA` format specifier

[Petr Mladek <pmladek@suse.com>]

On Fri 2022-03-18 16:07:31, Andy Shevchenko wrote:
> On Thu, Mar 17, 2022 at 07:10:00PM +0100, Miguel Ojeda wrote:
> > From: Gary Guo <gary@garyguo.net>
> > 
> > This patch adds a format specifier `%pA` to `vsprintf` which formats
> > a pointer as `core::fmt::Arguments`. Doing so allows us to directly
> > format to the internal buffer of `printf`, so we do not have to use
> > a temporary buffer on the stack to pre-assemble the message on
> > the Rust side.
> > 
> > This specifier is intended only to be used from Rust and not for C, so
> > `checkpatch.pl` is intentionally unchanged to catch any misuse.
> 
> ...
> 
> > +	case 'A':
> > +		if (!IS_ENABLED(CONFIG_RUST)) {
> > +			WARN_ONCE(1, "Please remove %%pA from non-Rust code\n");
> > +			return error_string(buf, end, "(%pA?)", spec);
> > +		}
> 
> I'm wondering if the Big Scary Banner as trace_printk() does would be better
> (in case we can tell that %pA is used in the code when RUST=n).

Good question!

The advantage of WARN_ONCE() is that it shows the stack so that it is
easier to locate the caller.

On the other hand, WARN_ONCE() is a bit misused here. It should be
used only in situations that might be potentially fatal. It might
even cause panic() with "panic_on_warn" kernel parameter.

Well, I am not sure if it is worth huge effort. WARN_ONCE() is
practical in this case because of the backtrace. We could always
create something better if people hit it more frequently and
it causes real life problems.

Best Regards,
Petr