lkml.org 
[lkml]   [2004]   [Mar]   [3]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [Kgdb-bugreport] [PATCH] Kill kgdb_serial
On Wed, Mar 03, 2004 at 04:51:06PM +0100, Pavel Machek wrote:
> Hi!
>
> > > > More precisely:
> > > > http://lkml.org/lkml/2004/2/11/224
> > >
> > > Well, that just says Andrew does not care too much. I think that
> > > having both serial and ethernet support *is* good idea after all... I
> > > have few machines here, some of them do not have serial, and some of
> > > them do not have supported ethernet. It would be nice to use same
> > > kernel on all of them. Also distribution wants to have "debugging
> > > kernel", but does _not_ want to have 10 of them.
> >
> > But unless I'm missing something, supporting eth or 8250 at all times
> > doesn't work right now anyhow, as eth if available will always take over.
>
> Well, that can be fixed. [Probably if kgdbeth= is not passed, ethernet
> interface should not take over. So user selects which one should be
> used by either passing kgdbeth or kgdb8250. That means that 8250
> should not be initialized until user passes kgdb8250=... not sure how
> you'll like that].

At this point, I'm going to give up on killing kgdb_serial, and pass
along some comments from David Woodhouse on IRC as well (I was talking
about this issue, and the init/main.c change):
(Tartarus == me, dwmw2 == David Woodhouse)

<Tartarus> dwmw2, the problem is how do you deal with all of the
possibilities of i/o (8250, kgdboe, or other serial) and do you allow
for passing 'gdb' on the command line to result in kgdb not being dropped
into? You can always break in later on of course
<dwmw2> parse command line early for 'gdb=' argument specifying which
i/o device to use. init kgdb core early. init each i/o device as early
as possible for that i/o device. Start the selected i/o device as soon
as it becomes available.
<dwmw2> just like console could, if we looked for console= a little bit
earlier. (forget all the earlyconsole shite, it's not necessary)
<dwmw2> Tartarrus, do the __early_setup() thing to replace __setup() for
selected args. We can use that for console= too.
<dwmw2> since 'console=' on the command line _already_ remembers its
arguments, and starts to use the offending device as soon as it gets
registered with register_console().
<Tartarus> dwmw2, __early_setup() ?
<dwmw2> See __setup("gdb=", gdb_setup_func);
<dwmw2> Replace with __early_setup(...)
<Tartarus> where is __early_setup ?
<dwmw2> before we normally parse the command line
<dwmw2> in my head

So perhaps someone can take these ideas and fix both problems... :)
(I've got some other stuff I need to work on today).

--
Tom Rini
http://gate.crashing.org/~trini/
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 14:01    [W:0.087 / U:0.108 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site