Re: [PATCH] NMI trigger switch support for debugging

[Mikael Pettersson <mikpe@csd.uu.se>]

Paul Jackson writes:
 > Mikael Pettersson, replying to AKIYAMA Nobuyuki:
 > >  > +	if (!old_state == !unknown_nmi_panic)
 > >  > +		return 0;
 > > 
 > > This conditional looks terribly obscure.
 > 
 > Would the following variant seem clearer:
 > 
 > 	if (!!unknown_nmi_panic == !!old_state)
 > 		return 0;
 > 
 > Odd, I know.  For those of us familiar with the '!!' idiom, which
 > converts any value to its binary logical equivalent 0 (if zero) or
 > 1 (otherwise), this reads as:
 > 
 > 	if (the logical value of unknown_nmi_panic is unchanged)
 > 		return 0;

The !! idiom has the advantage of making it crystal clear
that the author is comparing boolean-normalised values.
The code I commented on was unusual enough that I couldn't
ignore the possibility of a bug.

In this case, I'd prefer the !! idiom, or moving the test into
the two state-changing code snippets below it.

/Mikael