Le Monday 14 March 2016 à 07:52 -0700, Peter Hurley a écrit : > On 03/14/2016 02:42 AM, Jean Delvare wrote: > > Hi Peter, > > > > On Fri, 11 Mar 2016 06:43:59 -0800, Peter Hurley wrote: > >> On 03/11/2016 06:31 AM, Jean Delvare wrote: > >>> Le Thursday 10 March 2016 à 14:19 -0800, Peter Hurley a écrit : > >>>> On 03/08/2016 05:30 AM, Jean Delvare wrote: > >>>>> Kconfig option SERIAL_8250_RT288X seems to be only relevant on a > >>>>> limited number of platforms, so do not present it on other > >>>>> architectures, unless build-testing. > >>>>> > >>>>> Signed-off-by: Jean Delvare <jdelvare@xxxxxxx> > >>>>> Acked-by: John Crispin <blogic@xxxxxxxxxxx> > >>>>> Cc: Mans Rullgard <mans@xxxxxxxxx> > >>>>> Cc: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx> > >>>>> Cc: Jiri Slaby <jslaby@xxxxxxxx> > >>>>> --- > >>>>> Changes since v1: > >>>>> * Add ARCH_TANGO to the list of platforms, as Måns Rullgård tells me > >>>>> it has this hardware too. > >>>>> > >>>>> drivers/tty/serial/8250/Kconfig | 1 + > >>>>> 1 file changed, 1 insertion(+) > >>>>> > >>>>> --- linux-4.5-rc7.orig/drivers/tty/serial/8250/Kconfig 2016-03-08 09:18:45.610936903 +0100 > >>>>> +++ linux-4.5-rc7/drivers/tty/serial/8250/Kconfig 2016-03-08 14:26:14.565737750 +0100 > >>>>> @@ -295,6 +295,7 @@ config SERIAL_8250_EM > >>>>> config SERIAL_8250_RT288X > >>>>> bool "Ralink RT288x/RT305x/RT3662/RT3883 serial port support" > >>>>> depends on SERIAL_8250 > >>>>> + depends on MIPS || ARCH_TANGO || COMPILE_TEST > >>>> > >>>> This seems pretty speculative. > >>> > >>> This is based on the Kconfig plus information I was given meanwhile. > >>> This may be incremental but not speculative. > >> > >> Well, that's not the only way to configure a serial port. > >> It can be done from userspace with TIOCSSERIAL ioctl. > > > > I don't follow you. Aren't we talking about hardware-specific > > characteristics? That can't be changed by software at run-time. > > Yes, it can. > > ioctl(TIOCSSERIAL) can change the base port address, irq, iotype (including > to UPIO_AU which is the port type relevant to this discussion). OK, it can be done. Question is, does it make any sense? And an even better question: do people actually do that? I don't care much about what's possible. By "it's possible" we could drop all Kconfig dependencies. > >>>> I'm concerned that patch v1 was only for MIPS, and now is already > >>>> needed on at least one ARM platform. > >>> > >>> This is still a short list, and this is why I think it should be added. > >> > >> So what happens when it turns out that this breaks some > >> platform and we find out 5 or even 10 versions from now? > > > > We can always revisit if/when the list gets too long. This isn't carved > > in stone. > > -stable doesn't take build patches. > > How are you going to make this list longer? Stable takes one-line fixes, and cares in particular about regressions. That being said... All hardware dependency statements I add have this "|| COMPILE_TEST" happened, so you can always actually enable the driver if you really have to, without touching the code. -- Jean Delvare SUSE L3 Support -- To unsubscribe from this list: send the line "unsubscribe linux-serial" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
