From patchwork Fri May 29 00:24:57 2020 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Miquel Raynal X-Patchwork-Id: 11577319 Return-Path: Received: from mail.kernel.org (pdx-korg-mail-1.web.codeaurora.org [172.30.200.123]) by pdx-korg-patchwork-2.web.codeaurora.org (Postfix) with ESMTP id 2FF79913 for ; Fri, 29 May 2020 00:25:44 +0000 (UTC) Received: from bombadil.infradead.org (bombadil.infradead.org [198.137.202.133]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by mail.kernel.org (Postfix) with ESMTPS id D26FB207F9 for ; Fri, 29 May 2020 00:25:43 +0000 (UTC) Authentication-Results: mail.kernel.org; dkim=pass (2048-bit key) header.d=lists.infradead.org header.i=@lists.infradead.org header.b="VzJcPRe9" DMARC-Filter: OpenDMARC Filter v1.3.2 mail.kernel.org D26FB207F9 Authentication-Results: mail.kernel.org; dmarc=none (p=none dis=none) header.from=bootlin.com Authentication-Results: mail.kernel.org; spf=none smtp.mailfrom=linux-arm-kernel-bounces+patchwork-linux-arm=patchwork.kernel.org@lists.infradead.org DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20170209; h=Sender: Content-Transfer-Encoding:Content-Type:Cc:List-Subscribe:List-Help:List-Post: List-Archive:List-Unsubscribe:List-Id:MIME-Version:Message-Id:Date:Subject:To :From:Reply-To:Content-ID:Content-Description:Resent-Date:Resent-From: Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID:In-Reply-To:References: List-Owner; bh=lGZ3jzTKVLz8oEl+RJrdFyhHK5/+W3Tdm6pVdCu/xIs=; b=VzJcPRe9HE8bJy qSJdUvaZ7f0ANNK8mcTsV6ElOAboAPnp1MKNnGpBwJG1OjzTkTf/E7LuSgGXs9DfLCg87Vs9bEm9N ObKb767rcx076KbbDL0vryKZiMehlXSqiaJ3EfPDy3xtIHWil00FdhO/ooFMijRT3YT22Y9FnzoYm zdXRFVAtfD8cKpIaReJz5XSMpP6dPCrmzyp/xKG62e3TeL95wwId+nO69sIcWZiBf8BW5OwnBD1H5 b5z3D11qD8+vIDMGFknmTVO+Zc53OlvYG5TpABu6WlVr2OUs9cQzHAMSmswib3n0kqRmjqJ1/nhwQ Lz5ph5RHhl45c4FYZybA==; Received: from localhost ([127.0.0.1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.92.3 #3 (Red Hat Linux)) id 1jeSq0-0005z6-Nv; Fri, 29 May 2020 00:25:28 +0000 Received: from relay12.mail.gandi.net ([217.70.178.232]) by bombadil.infradead.org with esmtps (Exim 4.92.3 #3 (Red Hat Linux)) id 1jeSpw-0005xu-GL; Fri, 29 May 2020 00:25:27 +0000 Received: from localhost.localdomain (unknown [91.224.148.103]) (Authenticated sender: miquel.raynal@bootlin.com) by relay12.mail.gandi.net (Postfix) with ESMTPSA id 60654200002; Fri, 29 May 2020 00:25:18 +0000 (UTC) From: Miquel Raynal To: Richard Weinberger , Vignesh Raghavendra , Tudor Ambarus , , Rob Herring , Mark Rutland , Subject: [PATCH v7 00/20] Introduce the generic ECC engine abstraction Date: Fri, 29 May 2020 02:24:57 +0200 Message-Id: <20200529002517.3546-1-miquel.raynal@bootlin.com> X-Mailer: git-send-email 2.20.1 MIME-Version: 1.0 X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.8.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20200528_172524_821884_3DD5747E X-CRM114-Status: GOOD ( 27.36 ) X-Spam-Score: -0.7 (/) X-Spam-Report: SpamAssassin version 3.4.4 on bombadil.infradead.org summary: Content analysis details: (-0.7 points) pts rule name description ---- ---------------------- -------------------------------------------------- -0.7 RCVD_IN_DNSWL_LOW RBL: Sender listed at https://www.dnswl.org/, low trust [217.70.178.232 listed in list.dnswl.org] -0.0 RCVD_IN_MSPIKE_H2 RBL: Average reputation (+2) [217.70.178.232 listed in wl.mailspike.net] -0.0 SPF_PASS SPF: sender matches SPF record 0.0 SPF_HELO_NONE SPF: HELO does not publish an SPF Record X-BeenThere: linux-arm-kernel@lists.infradead.org X-Mailman-Version: 2.1.29 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Cc: Julien Su , Weijie Gao , Paul Cercueil , Boris Brezillon , Thomas Petazzoni , Miquel Raynal , Mason Yang , Chuanhong Guo , linux-arm-kernel@lists.infradead.org Sender: "linux-arm-kernel" Errors-To: linux-arm-kernel-bounces+patchwork-linux-arm=patchwork.kernel.org@lists.infradead.org As of today, only raw NAND controllers used to feature an integrated ECC engine and so controller drivers always embedded some code to enable/disable the correction. This statement is no longer correct as SPI-NAND devices might not embed an on-die ECC engine and must make use of an external ECC engine. We figured there are three possible situations for (generic) NAND devices: either the engine is 'on-die' (most of the SPI-NANDs, a few raw NANDs), or the engine is part of the host controller (most raw NANDs), or the engine may be external (SPI controllers might feature an hardware ECC engine, otherwise software correction can also be used). To solve this situation, I already proposed the creation of an ECC engine abstraction (still work in progress) and this is the first step for that. The logic in this series is: 1/ Use the generic NAND core for all NAND devices (raw and SPI). 2/ Create the ECC engine interface in drivers/mtd/nand/ecc.c 3/ Move code in driver/mtd/nand/ecc.c Later, I will repost: 4/ Make both software engines (Hamming and BCH) generic, move them in the ecc/ directory, clean them a bit and instantiate ECC engines. Write raw NAND helpers to use these two new engines. 5/ Isolate SPI-NAND on-die ECC engine in its own driver. 6/ Make use from the SPI-NAND layer of all the ECC engines listed above (on user request, people can now make use of soft BCH if they don't have an ECC-engine). 7/ Extend the logic to hardware external engines. A proposal of a driver for Macronix external ECC engine will follow in another series. This work is now almost ready, the next steps will be: 1/ Migrate the raw NAND core to make a proper use of these ECC engines. 2/ Deprecate in the raw NAND subsystem the interfaces used until now (I expect we should get rid of a lot of boilerplate). Thanks, Miquèl Changes in v7 ============= * Collected tags. * Fixed subjects s/OOB placement/ECC bytes placement/. * Moved a tab change into space to the appropriate patch. * Moved the removal of the nand_ecc_mode public enumumeration to the right patch. * Created helpers to retrieve the OOB layouts. * Moved the "flags" entry of nand_ecc_props to the commit using it. * Changed the select MTD_NAND_CORE into a depends on and moved it to the right patch too. * Split the patch moving raw NAND bits into the ECC framework. * Dropped an extra helper: nanddev_get_flash_node(). * Updated all the ECC framework introductory comment, following Boris' suggestions. Changes in v6 ============= * Rebased on top of nand/next. * Dropped the new nand-ecc-provider property. Instead, we will use nand-ecc-engine and boolean properties like nand-no-ecc-engine / nand-use-soft-ecc-engine (will be added in another series). * Moved most of the bindings to the next series. * Used platform_dev_put() instead of of_dev_put() on Rob's advice. * Renamed objects: ondie -> on_die_hw. * Renamed objects: hw -> on_host_hw. * Used engine_type everywhere instead of provider. * Created a rawnand compatibility layer to avoid moving deprecated code due to the rawnand core history in the generic (and clean) NAND layer. * Enhanced the nand_ecc_algo enumeration. Changes in v5 ============= * Rebased on top of nand/next * Avoided a fallthrough situation in commit: mtd: rawnand: Separate the ECC engine type and the OOB placement * Fixed an of_dev_put() build issue due to a missing dummy helper. * Extracted a patch that deserved to be merged quickly. * Fixed a few issues reported by robots. Changes in v4 ============= * Rebased on top of a recent kernel version. * Added Boris' reviewed-by. * Added Maxime's Acked-by tag. * Added the missing of_device.h header to ecc.c. * Corrected a 'minimum' comparison by using min_t. * Updated the new Macronix raw NAND controller driver by using the new (ECC related) function names. * Fixed a function call in ndfc.c. * Update brcmnand.c file to fit new enumerations and structures (due to recent Kamal's changes). * Force sm_ftl to depends on the Hamming engine, because by just selecting it the ECC code would be embedded in the NAND core and the NAND core might not be compiled in with sm_ftl. * Fixed a structure field name that I previously added in davinci platform data. * Moved the oob_first placement scheme to Davinci driver. Removed any occurence of it out of the driver (unused). * Simplify structure names as proposed by Boris. * Change enumeration/string names about ECC engine providers/placements. * Change the logic in the of_get_nand_ecc_* helpers to ensure backward compatibility. * Use enums intead of unsigned integers in the core when referring to ECC engine type, placement and algorithm. * Add nand-ecc-placement DT property. * Deprecate hw_syndrome. * Deprecate nand-ecc-mode in favor of nand-ecc-provider. * Fixed a typo in the Macronix ECC driver, where I made a copy/paste error which I haven't spotted because it is located in a macro only compiled when building the driver as a module (name of the of_ids was prefixed marvell_nfc instead of mxic_ecc). * Simplified the ECC engine API by dropping the useless oobbuf parameter. Instead, ECC engine drivers are supposed to provide a spare OOB buffer if none is provided. Updated the three existing engines. * Fixed BCH software engine with the help of Mason from Macronix. * Added a mechanism called "tweaking req" to change the SPI-NAND requests and ensure they always contain the right amount of data/OOB needed for the ECC engine to work properly. Changes in v3 ============= * Added Boris' Reviewed-by tags. * Added a kernel doc header on the nand_page_io_req enumeration. * Added support for HW engines. * Droped the patch clarifying the value of the first entry in enumerations (which is always 0). * Rename the nand_ecc_conf structure as nand_ecc_props because the _conf suffix implies that it is possible to edit it, while in some cases (eg. on-die ECC) there is nothing to tweak. * Smoother introduction of the ECC engine abstraction. * Renamed the ECC engine module nand_ecc_engine.ko. * Moved all the ECC files into drivers/mtd/nand/. Forgot the ecc/ subdirectory. * Added a new series to drop the ECC mode enumeration wich mixes the provider (none, hw, sw, on-die) and the OOB placement (first, syndrome). * Various typos fixed. * Added a few patches to fix bugs found in SPI-NAND/mtdchar.c. * Introduced the external hardware ECC engine boilerplate. Changes in v2 ============= * SPDX license identifiers for soft BCH and Hamming: the license macro was right, "GPL" means "GPLv2 or higher", so do not change this portion. Also update the commit messages to fit the actual change. * Do not compile-in the NAND core by default, do it only for raw NAND. Remove the dependencies on CONFIG_MTD in a different patch. Also, keep an extra level of hierarchy in Kconfig for the NAND bits by adding a menu instead of a config. * Moved the standard OOB layouts in the ecc/engine.c driver instead of in the NAND core. * Used the nand_ecc_ prefix in most of the engines functions instead of just ecc_, which is now reserved for bare helpers. Get rid of the __ecc prefix. * In the sunxi NAND controller driver: moved the ECC structure from sunxi_nfc to sunxi_nand_chip as the ECC engine is per-chip and not per controller. * Software Hamming ECC engine is only enabled by default if raw NAND is also enabled. NDFC now selects the software Hamming ECC engine (instead of depending on it). * Mention in software BCH and Hamming Kconfig entries that booting from NAND is very likely to fail if the user selects these symbols as modules. * Added Boris Reviewed-by tag on the SPI-NAND typo fixing patch. * Renamed the "mode" into a "provider" entry in the ECC configuration structures. * Moved the "total" entry of the ECC configuration directly in the context structure (should probably not be public but let's keep it as is for now). * Split the generic ECC engine introduction into smaller patches to do some renaming aside. * Drop the "maximize" entry in the ECC engine configuration structure, keep using a flag like before. * Canceled the move of the SPI-NAND specific ECC engine out of the core file. * Amended the root ECC structures to have three nand_ecc_conf structures: one for the defaults, one for the chip requirements, one for the user desires. * Created a *ondie_engine pointer in the nand_ecc structure to save the on-die ECC engine, if any. For instance, saving a reference to this engine is done by the SPI-NAND core. * Dropped the SPI-NAND flag that was used to distinguish between NAND flavors from the NAND core, it should not be needed anymore. * Added an helper in the NAND core to put a reference on an ECC engine. This will be used by the hardware engines only. * Renamed the files ecc/sw-{bch,hamming}.c and their headers include/linux/mtd/nand-ecc-sw-{bch,hamming}-engine.h. * Created a MTD_NAND_ECC invisible Kconfig symbol. * Added plenty of missing EXPORT_SYMBOL{,_GPL}(). * Minor modifications so that everything still compiles even when modules and built-in drivers are mixed in Kconfig in the whole NAND directory. Miquel Raynal (20): dt-bindings: mtd: Document nand-ecc-placement mtd: rawnand: Create a new enumeration to describe ECC bytes placement mtd: rawnand: Separate the ECC engine type and the ECC byte placement mtd: rawnand: Create a helper to retrieve the ECC placement mtd: rawnand: Add a kernel doc to the ECC algorithm enumeration mtd: rawnand: Rename the ECC algorithm enumeration items mtd: rawnand: Create a new enumeration to describe properly ECC types mtd: rawnand: Use the new ECC engine type enumeration mtd: nand: Move nand_device forward declaration to the top mtd: nand: Add an extra level in the Kconfig hierarchy mtd: nand: Drop useless 'depends on' in Kconfig mtd: nand: Add a NAND page I/O request type mtd: nand: Rename a core structure mtd: nand: Add more parameters to the nand_ecc_props structure mtd: nand: Introduce the ECC engine abstraction mtd: nand: Convert generic NAND bits to use the ECC framework mtd: rawnand: Hide the generic OOB layout objects behind helpers mtd: rawnand: Write a compatibility layer mtd: rawnand: Move generic OOB layouts to the ECC framework mtd: rawnand: Move the user input parsing bits to the ECC framework .../bindings/mtd/nand-controller.yaml | 10 + arch/arm/mach-davinci/board-da830-evm.c | 2 +- arch/arm/mach-davinci/board-da850-evm.c | 2 +- arch/arm/mach-davinci/board-dm355-evm.c | 2 +- arch/arm/mach-davinci/board-dm355-leopard.c | 3 +- arch/arm/mach-davinci/board-dm365-evm.c | 2 +- arch/arm/mach-davinci/board-dm644x-evm.c | 2 +- arch/arm/mach-davinci/board-dm646x-evm.c | 2 +- arch/arm/mach-davinci/board-mityomapl138.c | 2 +- arch/arm/mach-davinci/board-neuros-osd2.c | 2 +- arch/arm/mach-davinci/board-omapl138-hawk.c | 2 +- arch/arm/mach-s3c24xx/common-smdk.c | 2 +- arch/arm/mach-s3c24xx/mach-anubis.c | 2 +- arch/arm/mach-s3c24xx/mach-at2440evb.c | 2 +- arch/arm/mach-s3c24xx/mach-bast.c | 2 +- arch/arm/mach-s3c24xx/mach-gta02.c | 2 +- arch/arm/mach-s3c24xx/mach-jive.c | 2 +- arch/arm/mach-s3c24xx/mach-mini2440.c | 2 +- arch/arm/mach-s3c24xx/mach-osiris.c | 2 +- arch/arm/mach-s3c24xx/mach-qt2410.c | 2 +- arch/arm/mach-s3c24xx/mach-rx1950.c | 2 +- arch/arm/mach-s3c24xx/mach-rx3715.c | 2 +- arch/arm/mach-s3c24xx/mach-vstms.c | 2 +- arch/arm/mach-s3c64xx/mach-hmt.c | 2 +- arch/arm/mach-s3c64xx/mach-mini6410.c | 2 +- arch/arm/mach-s3c64xx/mach-real6410.c | 2 +- drivers/mtd/nand/Kconfig | 13 + drivers/mtd/nand/Makefile | 2 + drivers/mtd/nand/ecc.c | 471 +++++++++++++++ drivers/mtd/nand/onenand/Kconfig | 1 - drivers/mtd/nand/raw/Kconfig | 2 +- drivers/mtd/nand/raw/ams-delta.c | 4 +- drivers/mtd/nand/raw/arasan-nand-controller.c | 16 +- drivers/mtd/nand/raw/atmel/nand-controller.c | 30 +- drivers/mtd/nand/raw/au1550nd.c | 4 +- .../mtd/nand/raw/bcm47xxnflash/ops_bcm4706.c | 3 +- drivers/mtd/nand/raw/brcmnand/brcmnand.c | 27 +- .../mtd/nand/raw/cadence-nand-controller.c | 4 +- drivers/mtd/nand/raw/cafe_nand.c | 3 +- drivers/mtd/nand/raw/cs553x_nand.c | 2 +- drivers/mtd/nand/raw/davinci_nand.c | 38 +- drivers/mtd/nand/raw/denali.c | 6 +- drivers/mtd/nand/raw/diskonchip.c | 3 +- drivers/mtd/nand/raw/fsl_elbc_nand.c | 20 +- drivers/mtd/nand/raw/fsl_ifc_nand.c | 12 +- drivers/mtd/nand/raw/fsl_upm.c | 4 +- drivers/mtd/nand/raw/fsmc_nand.c | 14 +- drivers/mtd/nand/raw/gpio.c | 4 +- drivers/mtd/nand/raw/gpmi-nand/gpmi-nand.c | 14 +- drivers/mtd/nand/raw/hisi504_nand.c | 6 +- .../mtd/nand/raw/ingenic/ingenic_nand_drv.c | 20 +- drivers/mtd/nand/raw/lpc32xx_mlc.c | 2 +- drivers/mtd/nand/raw/lpc32xx_slc.c | 3 +- drivers/mtd/nand/raw/marvell_nand.c | 34 +- drivers/mtd/nand/raw/meson_nand.c | 2 +- drivers/mtd/nand/raw/mpc5121_nfc.c | 4 +- drivers/mtd/nand/raw/mtk_nand.c | 10 +- drivers/mtd/nand/raw/mxc_nand.c | 25 +- drivers/mtd/nand/raw/nand_base.c | 565 +++++++----------- drivers/mtd/nand/raw/nand_esmt.c | 11 +- drivers/mtd/nand/raw/nand_hynix.c | 41 +- drivers/mtd/nand/raw/nand_jedec.c | 4 +- drivers/mtd/nand/raw/nand_micron.c | 20 +- drivers/mtd/nand/raw/nand_onfi.c | 8 +- drivers/mtd/nand/raw/nand_samsung.c | 19 +- drivers/mtd/nand/raw/nand_toshiba.c | 16 +- drivers/mtd/nand/raw/nandsim.c | 8 +- drivers/mtd/nand/raw/ndfc.c | 2 +- drivers/mtd/nand/raw/omap2.c | 22 +- drivers/mtd/nand/raw/orion_nand.c | 4 +- drivers/mtd/nand/raw/pasemi_nand.c | 4 +- drivers/mtd/nand/raw/plat_nand.c | 4 +- drivers/mtd/nand/raw/qcom_nandc.c | 2 +- drivers/mtd/nand/raw/r852.c | 3 +- drivers/mtd/nand/raw/s3c2410.c | 20 +- drivers/mtd/nand/raw/sh_flctl.c | 6 +- drivers/mtd/nand/raw/sharpsl.c | 2 +- drivers/mtd/nand/raw/socrates_nand.c | 5 +- drivers/mtd/nand/raw/stm32_fmc2_nand.c | 9 +- drivers/mtd/nand/raw/sunxi_nand.c | 26 +- drivers/mtd/nand/raw/tango_nand.c | 4 +- drivers/mtd/nand/raw/tegra_nand.c | 34 +- drivers/mtd/nand/raw/tmio_nand.c | 2 +- drivers/mtd/nand/raw/txx9ndfmc.c | 2 +- drivers/mtd/nand/raw/vf610_nfc.c | 6 +- drivers/mtd/nand/raw/xway_nand.c | 4 +- drivers/mtd/nand/spi/core.c | 12 +- drivers/mtd/nand/spi/macronix.c | 6 +- drivers/mtd/nand/spi/toshiba.c | 6 +- include/linux/mtd/nand.h | 164 ++++- include/linux/mtd/rawnand.h | 34 +- include/linux/mtd/spinand.h | 2 +- include/linux/platform_data/mtd-davinci.h | 9 +- .../linux/platform_data/mtd-nand-s3c2410.h | 2 +- 94 files changed, 1221 insertions(+), 731 deletions(-) create mode 100644 drivers/mtd/nand/ecc.c