On Mon, 2012-10-22 at 05:15 -0700, Joe Perches wrote:
> waOn Mon, 2012-10-22 at 10:37 +0100, David Howells wrote:
> > Joe Perches <joe@perches.com> wrote:
> > 
> > > On Sat, 2012-10-20 at 23:33 +0100, Ken O'Brien wrote:
> > > ...
> > > > -		pr_err("binder: %d: binder_alloc_buf failed to "
> > > > +		pr_err("binder: %d: binder_alloc_buf failed to " \
> > > >  		       "map pages in userspace, no vma\n", proc->pid);
> > > ...
> > > Nice try, but the "right" way to do this is to coalesce formats like:
> > > 
> > > 		pr_err("binder: %d: binder_alloc_buf failed to map pages in userspace, no vma\n",
> > > 		       proc->pid);
> > 
> > Surely the right way is to fix _checkpatch_ to see ["]\n\s*["] and merge the
> > lines before checking them?  We shouldn't have to fix up the source in either
> > of these fashions just because checkpatch is broken.
> 
> checkpatch is stupid, but it's not broken here.
> 
> Formats should be coalesced into a single line to make
> grepping the source for dmesg output easier.

But MODULE_PARM_DESC() multi-line strings are ok, right?

  grep - 'MODULE_PARM_DESC.*"$' drivers

shows it's in widespread use.

> More likely checkpatch could be changed to bleat yet
> another warning like
> 
> "line continuation used outside macro definition"