On 7 October 2017 at 17:42, Marcel Holtmann <marcel@xxxxxxxxxxxx> wrote: > Hi Ian, > >> The current Kconfig for serdev is not compatible when adding ACPI support as it does not work when built as a module as it requires config SERIAL_DEV_CTRL_TTYPORT to be set. This patch makes serdev compiled into the kernel if selected so that config SERIAL_DEV_CTRL_TTYPORT can be correctly set if requiring ACPI support. >> --- >> drivers/tty/serdev/Kconfig | 5 +++-- >> 1 file changed, 3 insertions(+), 2 deletions(-) >> >> diff --git a/drivers/tty/serdev/Kconfig b/drivers/tty/serdev/Kconfig >> index cdc6b820cf93..a9fb09a9c105 100644 >> --- a/drivers/tty/serdev/Kconfig >> +++ b/drivers/tty/serdev/Kconfig >> @@ -2,7 +2,8 @@ >> # Serial bus device driver configuration >> # >> menuconfig SERIAL_DEV_BUS >> - tristate "Serial device bus" >> + bool "Serial device bus" >> + default y >> help >> Core support for devices connected via a serial port. >> >> @@ -11,6 +12,6 @@ if SERIAL_DEV_BUS >> config SERIAL_DEV_CTRL_TTYPORT >> bool "Serial device TTY port controller" >> depends on TTY >> - depends on SERIAL_DEV_BUS != m >> + default y > > actually we made hci_nokia.c be buildable as separate module. We might need to do the same for hci_bcm.c to avoid forcing SERIAL_DEV_BUS into a bool. Or you need to explain the reason behind this change a bit better. > > Regards > > Marcel > Hi Marcel, For serdev to support for devices connected via a serial port such as a tty (for example an ACPI enumerated BT driver) its code hooks into TTY code meaning serdev can no longer be compiled as a module for this functionality to work. With CONFIG_SERIAL_DEV_BUS originally being defined as a tristate a provision was made when compiling serdev as a module to return ENODEV when trying to register a tty port. This provision still exists with CONFIG_SERIAL_DEV_BUS defined as a bool because CONFIG_SERIAL_DEV_CTRL_TTYPORT actually controls this. My patch also aligns CONFIG_SERIAL_DEV_BUS (serdev) which provides "Core serial port support" to CONFIG_SERIAL_NONSTANDARD or "Non-standard serial port support" as this seems logical comparison. CONFIG_SERIAL_NONSTANDARD is a bool similar to how I've made CONFIG_SERIAL_DEV_BUS a bool. Whilst it may be preferable to make hci_bcm.c buildable as separate module like hci_nokia.c I don't relate the 'modularity' of bluetooth drivers with the 'modularity' of a serial device bus as I see the latter more similar to PCI support or ACPI support. Because of the link and/or cross-over between serial port support and tty support I believe serdev should be built-in rather than modularized if and when it is required. If this better explains the reason for the patch I can resubmit with an updated changelog together with a signoff which I accidentally during editing. Regards, Ian -- To unsubscribe from this list: send the line "unsubscribe linux-acpi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
