From mboxrd@z Thu Jan 1 00:00:00 1970 From: broonie@opensource.wolfsonmicro.com (Mark Brown) Date: Thu, 25 Oct 2012 11:27:56 +0100 Subject: [PATCH] [RFC] pinctrl: mvebu: reset pins to an UNKNOWN state on startup In-Reply-To: References: <1351106281-31288-1-git-send-email-thomas.petazzoni@free-electrons.com> Message-ID: <20121025102756.GJ18814@opensource.wolfsonmicro.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org On Thu, Oct 25, 2012 at 08:46:38AM +0200, Linus Walleij wrote: > Now I don't know which kernel senior being it was that told me > never to screw around with the defaults from the boot loader > if not really needed. It is better if the driver reads the hardware > to figure out what state it's in and move on from there. This depends on the platform quite a bit - there's two schools of thought on what bootloaders should do, and typically platforms decide on a per platform basis which they'll use: 1. The bootloader does enough to load the kernel and nothing more, the kernel should ignore anything the bootloader did. Some people prefer this as it places minimal reliance on the bootloader which may be of uncertain quality and minimises the need to upgrade the bootloader which is often highly risky. 2. The bootloader does all the pin setup and so on. This provides a place for board specific stuff and avoids the kernel having to know about lots of device variants. For example AIUI OMAP tends towards the second view (partly due to number of device variants) but for example Samsung platforms typically use the first method (their pinmux is pretty simple). -------------- next part -------------- A non-text attachment was scrubbed... Name: signature.asc Type: application/pgp-signature Size: 836 bytes Desc: Digital signature URL: