Re: [PATCH] tracing user_events: Simplify user_event_parse_field() parsing

From: Steven Rostedt
Date: Mon Jan 08 2024 - 17:22:10 EST


On Mon, 8 Jan 2024 17:13:12 -0500
Steven Rostedt <rostedt@xxxxxxxxxxx> wrote:

> On Mon, 8 Jan 2024 21:47:44 +0000
> Beau Belgrave <beaub@xxxxxxxxxxxxxxxxxxx> wrote:
>
> > > - len = str_has_prefix(field, "__rel_loc ");
> > > - if (len)
> > > - goto skip_next;
> > > + if (!(len = str_has_prefix(field, "__data_loc unsigned ")) &&
> > > + !(len = str_has_prefix(field, "__data_loc ")) &&
> > > + !(len = str_has_prefix(field, "__rel_loc unsigned ")) &&
> > > + !(len = str_has_prefix(field, "__rel_loc "))) {
> > > + goto parse;
> > > + }
> >
> > This now triggers a checkpatch error:
> > ERROR: do not use assignment in if condition
>
> What a horrible message.
>
> > #1184: FILE: kernel/trace/trace_events_user.c:1184:
> > + if (!(len = str_has_prefix(field, "__data_loc unsigned ")) &&
> >
> > I personally prefer to keep these files fully checkpatch clean.
>
> I've stopped using checkpatch years ago because I disagreed with so much it :-p
> (Including this message)

Note that checkpatch is a guideline and not a rule. The general rule is, if
the code looks worse when applying the checkpatch rule, don't do it.

- Steve