From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from mail.bootlin.com ([62.4.15.54]) by merlin.infradead.org with esmtp (Exim 4.90_1 #2 (Red Hat Linux)) id 1gHPMc-0001Lb-Rf for linux-mtd@lists.infradead.org; Tue, 30 Oct 2018 08:27:04 +0000 Date: Tue, 30 Oct 2018 09:26:05 +0100 From: Boris Brezillon To: "Grandbois, Brett" Cc: Marek Vasut , David Woodhouse , Brian Norris , Richard Weinberger , "linux-mtd@lists.infradead.org" , "linux-kernel@vger.kernel.org" Subject: Re: [PATCH v2] mtd: spi-nor: Add support for SPI boot flash access for AMD Family 16h Message-ID: <20181030092605.38851906@bbrezillon> In-Reply-To: References: <20181016005726.26813-1-brett.grandbois@opengear.com> <20181027173945.241e87b2@bbrezillon> MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit List-Id: Linux MTD discussion mailing list List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , On Mon, 29 Oct 2018 23:15:42 +0000 "Grandbois, Brett" wrote: > On 28/10/18 1:39 am, Boris Brezillon wrote: > > Hi Brett, > > > > On Tue, 16 Oct 2018 00:57:41 +0000 > > "Grandbois, Brett" wrote: > > > >> Add support to expose the SPI boot flash on AMD Family 16h CPUs as > >> a standard mtd device to give userspace BIOS updaters greater > >> feature support. The BIOS and Kernel Developer's Guide refers to > >> this as the 'SPI ROM' controller and so the driver follows that > >> naming convention for consistency. > >> > > We're currently trying to convert spi-nor controller drivers to the > > spi-mem interface [1]. Can you look at this new interface and tell > > me if you'd be able to implement it? If that's not possible, then > > I'd prefer to have this driver implement the mtd_info interface > > directly. > > So from going over the spi-mem interface it looks like the intent is > for these sorts of devices to be a standard spi_controller with only > mem_ops defined and the transfer/_one/_one_message left as NULL? Is > that correct? Yes > That's a bit of a pivot from how it's currently done > (it's conceptually similar to the intel-spi-pci driver so I was > following that) Yes, and that's exactly what I'd like to avoid. intel-spi-pci will probably be the trickiest conversion, so I'd like to avoid having another one ;-). > but I should be able to rework it to the new > interface. This then lives under drivers/spi and thus should be > submitted to linux-spi? Actually, if the controller is only ever connected to the same SPI NOR chip (no need for advanced detection scheme) and does not support support Octo/Quad/Dual modes (or any other advanced features), you'll be better off implementing mtd->_read/_write/_erase() directly (the driver would then live in drivers/mtd/devices/).