On 08/06/2013 04:50 AM, Mateusz Zalega wrote:
This commit postpones initialization of USB hardware until usb_board_init() is called by a command implementation (ie. do_dfu()) or a driver.
diff --git a/arch/arm/include/asm/arch-tegra/usb.h b/arch/arm/include/asm/arch-tegra/usb.h
/* Setup USB on the board */ -int board_usb_init(const void *blob); +int usb_process_devicetree(const void *blob);
This needs to be part of the previous patch (or part of that patch needs to be split out and put into this patch) so that this prototype change happens in the same patch that the implementation is renamed. Perhaps this can happen before the current patch 1?
diff --git a/board/nvidia/common/board.c b/board/nvidia/common/board.c
@@ -151,10 +152,6 @@ int board_init(void) # endif /* CONFIG_TEGRA_PMU */ #endif /* CONFIG_SYS_I2C_TEGRA */
-#ifdef CONFIG_USB_EHCI_TEGRA
- pin_mux_usb();
- board_usb_init(gd->fdt_blob);
-#endif
+#ifdef CONFIG_USB_EHCI_TEGRA +int board_usb_init(enum board_usb_init_type what_to_init) +{
- pin_mux_usb();
- usb_process_devicetree(gd->fdt_blob);
- return 0;
+} +#endif
Moving the pinmux initialization might be dangerous; pin_mux_usb() might prevent the pins used for USB having a mux function selected that conflicts with some other pins selected mux function. We really do need to initialize all of the pinmux early, and not defer it.