Giter Site home page Giter Site logo

yafut's Introduction

GitHub Workflow Status License

Yafut (Yet Another File UTility)

Overview

Yafut is a basic file copying utility for the Linux operating system that employs Yaffs Direct Interface (YDI) and Linux Memory Technology Devices (MTD) ioctls to interact with Yaffs file systems from userspace. It enables copying files from/to Yaffs file systems even if the kernel does not have native support for the Yaffs file system compiled in. Yafut also has limited support for copying files from/to Yaffs file system images stored in regular files.

Requirements

Build Requirements

  • C compiler (C99-compliant; Yafut is tested with GCC & Clang)
  • CMake 3.16+ (with a CMake-supported build tool, e.g. GNU Make, Ninja)
  • Yaffs source code (included as a Git submodule)
  • C standard library header files
  • Linux kernel header files

Runtime Requirements

  • Linux kernel with MTD support enabled
    • Linux kernel 6.1 or newer for NAND flash
    • Linux kernel 3.2 or newer for NOR flash
  • glibc or musl libc (other C standard libraries should also work, but have not been tested)

Building

cmake -B builddir
cmake --build builddir

Usage Examples

To copy a file called foo from the MTD partition represented by the character device /dev/mtd1 to a local file called bar, run:

yafut -d /dev/mtd1 -r -i foo -o bar

(Run cat /proc/mtd to list all available MTD partitions.)

To copy a local file called baz to a file called qux on the MTD partition represented by the character device /dev/mtd2, run:

yafut -d /dev/mtd2 -w -i baz -o qux

Pass - to -i/-o to read from/write to stdin/stdout. For example, the first command above could also be written as:

yafut -d /dev/mtd1 -r -i foo -o - > bar

while the second command above could also be written as:

cat baz | yafut -d /dev/mtd2 -w -i - -o qux

Run yafut -h for further usage instructions.

FAQ

Which flash memory types does this tool work with?

Yafut supports both NAND flash and NOR flash. However, while most parameters for an existing Yaffs file system stored on NAND flash can be autodetected, the Yaffs layout used on NOR flash can be fairly arbitrary and therefore Yaffs parameters for such a file system will likely need to be provided manually (see the next question below).

Do I need to manually set any Yaffs parameters?

Yafut tries to make reasonable assumptions about most Yaffs parameters to use for a given MTD, so that its use remains as simple as possible. However, not all Yaffs parameters can be discovered deterministically, in which case the educated guess attempted by Yafut may turn out to be wrong and specific values for certain Yaffs options will need to be forced by the user via command-line options.

The Yaffs parameters that can currently be controlled using command-line options are:

  • Chunk size (-C) and block size (-B). Yafut uses autodetected MTD parameters for these parameters by default, but they can be forced to specific values if necessary.

  • Use of inband tags (-T). Yafut assumes that inband tags are only necessary if the MTD does not have enough available bytes in the OOB area to store a Yaffs2 tag structure (whether that includes ECC data for tags or not depends on whether the -E option is used, see below). However, some file systems may use inband tags despite being stored on an MTD that does have enough available space in the OOB area to fit a Yaffs2 tag structure. For such file systems, the -T command-line option can be used to force use of inband tags.

  • Use of ECC for tags (-E). Yafut assumes that Yaffs2 tag structures include ECC data for tags by default. If a given file system does not use ECC for tags, the -E command-line option can be used to tell Yafut to act accordingly.

  • Use of Yaffs2 checkpoints (-P to disable). By default, Yaffs2 code stores a snapshot of its runtime state (called a "checkpoint") in the file system when the latter gets unmounted or sync()'d. Using these checkpoints speeds up file system mounting, but occupies extra storage space. If desired, reading and writing checkpoints can be disabled using the -P command-line option.

  • Writing Yaffs2 summaries (-S to disable). By default, Yaffs2 code stores a so-called summary at the end of each block. Using these summaries speeds up file system scanning, but occupies extra storage space. If desired, writing summaries can be disabled using the -S command-line option.

  • Byte order (-L to force little-endian byte order, -M to force big-endian byte order). By default, Yafut assumes that the byte order used on the MTD is the same as the byte order used by the host CPU. This can be overridden if necessary, allowing little-endian hosts to operate on big-endian file systems and vice versa.

Which Yaffs file system versions does this tool work with?

Yaffs code that Yafut builds upon supports both Yaffs1 and Yaffs2 file systems. Yafut assumes that NAND devices with 512-byte pages use Yaffs1 while those with 1024-byte or larger pages use Yaffs2. All NOR devices are assumed to use Yaffs2 by default. While the autodetected Yaffs layout can be tweaked using the -C and -B command-line options (see above), there is currently no way to override the chunk size threshold used for autodetecting the Yaffs file system version used.

What's the deal with the Linux kernel version requirements?

Linux kernel version 6.1 is the first one that implements all the MTD ioctls that Yafut needs to do its job on NAND flash - specifically, it is the first kernel version that supports the MEMREAD ioctl, which enables userspace applications to use the Linux kernel's OOB autoplacement mechanism while reading data from NAND devices (see also the next question).

While Yafut can be built on hosts on which the Linux kernel headers do not include the definition of the MEMREAD ioctl, it needs Linux kernel version 6.1+ to actually work on NAND flash. (An older, custom kernel version with a backported implementation of the MEMREAD ioctl can also be used).

As Yafut does not need the MEMREAD ioctl to handle NOR flash, only Linux kernel 3.2 or newer is necessary for Yafut to work on that type of flash memory.

Do I need to know the OOB (spare) area layout of my NAND device?

No. Here is why:

  • Yaffs2: Yafut relies on the Linux kernel's autoplacement mechanism for reading/writing Yaffs2 metadata from/to the OOB area of the NAND device. This enables Yafut to work transparently with any NAND device supported by the Linux kernel, without requiring the user to provide the OOB layout details, which simplifies Yafut's source code and usage. However, the downside of this approach is that if one wanted to read data from an MTD partition that uses a different OOB layout than the one assumed by the Linux kernel, there is currently no way to make Yafut do that.

  • Yaffs1: Yaffs1 code assumes that each page of the NAND device has a 16-byte spare area. Yaffs assumes full control over the layout of that area. For that reason, Yaffs1 metadata is written to the NAND device in raw mode (i.e. without using the Linux kernel's autoplacement mechanism for writing to the OOB area) and the concept of customizing the OOB layout does not really apply for Yaffs1.

Is this tool also able to work with Yaffs image files?

Yes, to an extent. The argument passed via the -d command-line option can be a path to either an MTD character device representing NAND/NOR flash memory or a regular file containing a Yaffs file system image. However, there is currently no support for working with file system images that include OOB data (e.g. NAND flash dumps). Like for NOR flash, the Yaffs layout used for a given file system image can be fairly arbitrary and therefore Yaffs parameters for such a file system will likely need to be provided manually (see above).

Troubleshooting & Debugging

If Yafut fails to do what you expect it to and the error messages it reports back do not directly indicate the failure reason (or are downright confusing), it might be useful to run it again with the -v command-line option, which enables verbose output that includes potentially useful information, e.g. the executed ioctl() system calls and their return codes. Verbose output is printed to stderr. If err=-25 (Inappropriate ioctl for device) lines are present in the verbose output, it means that the running kernel does not support the ioctl() commands that Yafut relies on. Specifying -v twice also causes (truncated) hex dumps of the data passed around to be included in the output.

Yaffs tracing (implemented in Yaffs code itself) can be enabled by setting the YAFFS_TRACE_MASK environment variable to the desired value. For example, to enable full Yaffs tracing, run Yafut like this:

YAFFS_TRACE_MASK=0xffffffff yafut [...options-here...]

See the yaffs_trace.h file in the Yaffs source tree for what each bit in the tracing mask means. Note that Yaffs tracing code prints its messages to stdout, not stderr, so setting YAFFS_TRACE_MASK while copying the contents of a file from an MTD to stdout (-r -o -) is not a good idea. Setting YAFFS_TRACE_MASK can be combined with -v, but since the former prints to stdout while the latter prints to stderr, use something like this to collect both types of tracing information to a common location:

YAFFS_TRACE_MASK=0xffffffff yafut [...options-here...] >trace 2>&1

License

Yafut source code is released under GNU General Public License (GPL), version 2.

yafut's People

Contributors

kempniu avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    ๐Ÿ–– Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. ๐Ÿ“Š๐Ÿ“ˆ๐ŸŽ‰

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google โค๏ธ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.