Re: [patch] kgdb light, v6

[Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>]

On Sunday 10 February 2008, Ingo Molnar wrote:

[...]

> > > +__setup("kgdboc=", kgdboc_option_setup);
> > 
> > no need for obsolete __setup, we have module_param_call() below
> 
> it's needed for bzImage kernels. I just tested it and without __setup() 
> no init sequence is run and KGDB is not activated.

weird, should work with "kgdboc.kgdboc=" parameter

[...]

> > > +#ifndef KGDB_MAX_BREAKPOINTS
> > > +# define KGDB_MAX_BREAKPOINTS	1000
> > > +#endif
> > > +
> > > +#define KGDB_HW_BREAKPOINT	1
> > 
> > unused
> 
> hm, both KGDB_MAX_BREAKPOINTS and KGDB_HW_BREAKPOINT are used.

my bad

[...]

> > if kgdb_isremovedbreak() helper is moved before kgdb_set_sw_break() 
> > and converted to return 'i' on success and '-1' on failure then it can 
> > be used instead the above for () loop
> 
> dunno - that would complicate arch/x86/kernel/kgdb.c's use of 
> kgdb_isremovedbreak() and looks a bit complex. If you feel strongly 

the whole difference w.r.t. arch/x86/kernel/kgdb.c should be:

-	if (exception == 3 && kgdb_isremovedbreak(regs->ip - 1)) {
+	if (exception == 3 && kgdb_isremovedbreak(regs->ip - 1) >= 0) {

> about it, could you send a patch?

well, maybe in the future if nobody fixes it :)

[ added as low-prio to my TODO... ]

Thanks for fixing all the other stuff.

Bart