Re: [PATCH] 2.5.21 kill warnings 4/19

[Andrew Morton]

Thunder from the hill wrote:
> 
> Hi,
> 
> On Mon, 10 Jun 2002, Andrew Morton wrote:
> > __func__ does *not* work on egcs-1.1.2 and so cannot be used in Linux.
> >
> > `struct blah = { .open = driver_open };' *does* work in egcs-1.1.2
> > and is OK to use.
> 
> Gee. I guess we need a special host gcc to support our needs - on
> sparc(|64). We might have a patch that easily renames it in the sources of
> egcs...

?

> Anyway, what is the problem about new gcc on old sparc? It works at least
> on my sparc64, I can't complain.

I don't know - we need to find a davem to answer that question.

egcs-1.1.2 has a problem with the percpu infrastructure - it
fails to put the right things in the right sections.  The
workaround for that is to ensure that the percpu data structures
are always initialised at the definition site:

	/* Some compilers disobey section attribute on statics when not
	   initialized -- RR */
	static struct tasklet_head tasklet_vec __per_cpu_data = { NULL };
	static struct tasklet_head tasklet_hi_vec __per_cpu_data = { NULL };

Note the (otherwise unneeded) `= { NULL }'.  The result of getting
this wrong is a nasty and subtly-dead kernel.

It's not a very happy state of affairs.   I'll be using
2.91.66 on x86 from now on, so any problems (in the stuff
which I build and test) will be picked up.

-