On Tue, Apr 03, 2012 at 05:35:25PM -0400, Mike Frysinger wrote:
On Thursday 22 March 2012 15:09:14 Tom Rini wrote:
Unless a board defines CONFIG_SERIAL_MULTI we assume it is using an on-module UART somewhere. We leave the exact configuration per-board.
Start an environment section in omap3_common.h and default to allowing the environment to override console/ethaddr. omap3_mvblx retains its behavior via opt-out.
not a big deal, but it seems like the env settings slipped in after the fact
Yeah, I had to think a few times about how to split-up some of the serial/console related stuff.
--- a/include/configs/omap3_common.h +++ b/include/configs/omap3_common.h
+/*
- Default serial port configuration is an NS16550-compatible UART.
- Boards can override this as needed with CONFIG_SERIAL_MULTI
- */
+#ifndef CONFIG_SERIAL_MULTI +#define CONFIG_SYS_NS16550 +#define CONFIG_SYS_NS16550_SERIAL +#define CONFIG_SYS_NS16550_REG_SIZE (-4) +#define CONFIG_SYS_NS16550_CLK 48000000 +#define CONFIG_SYS_BAUDRATE_TABLE {4800, 9600, 19200, 38400, 57600,\
115200}+#endif
the baudrate table is fairly independent of the serial driver. might want to split that out into its own ifndef/define/endif check.
What I really wanted here was a test for "not using the SoM UART". Looking at the examples we have today, it's either the SoM and the above table but differing in which UART is physically exposed or it's SERIAL_MULTI and wants it's own baudrate table.
Part of me thinks I should whack at include/config_defaults.h and add something like #ifndef CONFIG_SYS_BAUDRATE_TABLE, standard rates, #endif. And same for CONFIG_SYS_PROMPT_HUSH_PS2 and probably a few others.