This is the mail archive of the
mailing list for the eCos project.
RE : RE :Bug in CtrlC handling when sharing Ethernet btwRedboot/eCos
- From: "Arnaud Chataignier" <achataignier at neotion dot com>
- To: "'Gary Thomas'" <gary at mlbassoc dot com>
- Cc: "'Bart Veer'" <bartv at ecoscentric dot com>,"'eCos patches'" <ecos-patches at ecos dot sourceware dot org>
- Date: Tue, 2 Nov 2004 14:16:36 +0100
- Subject: RE : RE :Bug in CtrlC handling when sharing Ethernet btwRedboot/eCos
I agree the solution I propose is not elegant. It would be much better
to modify hal_ctrlc_check prototype, even if there would be a break in
the virtual vectors interface (consider some people may have some custom
drivers calling the old interface).
I am not familiar with the various macros defined in hal_if.h, and
unfortunately I don't have much time to understand them, but please feel
free to make the corresponding modifications if you can.
By the way, thank you for all the work you are doing on eCos/Redboot.
De : firstname.lastname@example.org
[mailto:email@example.com] De la part de Gary Thomas
Envoyé : mardi 2 novembre 2004 13:01
À : Arnaud Chataignier
Cc : 'Bart Veer'; eCos patches
Objet : Re: RE :Bug in CtrlC handling when sharing Ethernet
On Tue, 2004-11-02 at 03:19, Arnaud Chataignier wrote:
> The solution works. Here is the corresponding patch.
> To eCos maintainers : my opinion is that it should be included in CVS
> tree to provide an always working ctrlc when ethernet is shared
> between eCos app and Redboot/stubs. The drawback is the break always
> occuring in the function cyg_hal_user_break(), but it is still useful
> since the user can then navigate between the different threads with
> the 'thread awreness' functionality included in stubs/GDB. This
> drawback is unfortunately necessary since it's not possible with the
> current Ethernet driver architecture, to check Ethernet frames
> directly in the ISR, which would be necessary for a cleaner solution.
While I agree in principle with the problem, the proposed solution (a
shared global "flag") is not very elegant. A much better solution would
be to pass the context to HAL_CTRLC_CHECK (i.e. add a third parameter).
Since there are only 6 places this is used in the eCos codebase, this is
a pretty minor change.
Gary Thomas <firstname.lastname@example.org>