Linux kernel stable tree
Go to file
Jacob Keller e7fdf5dddc lib: packing: fix QUIRK_MSB_ON_THE_RIGHT behavior
The QUIRK_MSB_ON_THE_RIGHT quirk is intended to modify pack() and unpack()
so that the most significant bit of each byte in the packed layout is on
the right.

The way the quirk is currently implemented is broken whenever the packing
code packs or unpacks any value that is not exactly a full byte.

The broken behavior can occur when packing any values smaller than one
byte, when packing any value that is not exactly a whole number of bytes,
or when the packing is not aligned to a byte boundary.

This quirk is documented in the following way:

  1. Normally (no quirks), we would do it like this:

  ::

    63 62 61 60 59 58 57 56 55 54 53 52 51 50 49 48 47 46 45 44 43 42 41 40 39 38 37 36 35 34 33 32
    7                       6                       5                        4
    31 30 29 28 27 26 25 24 23 22 21 20 19 18 17 16 15 14 13 12 11 10  9  8  7  6  5  4  3  2  1  0
    3                       2                       1                        0

  <snip>

  2. If QUIRK_MSB_ON_THE_RIGHT is set, we do it like this:

  ::

    56 57 58 59 60 61 62 63 48 49 50 51 52 53 54 55 40 41 42 43 44 45 46 47 32 33 34 35 36 37 38 39
    7                       6                        5                       4
    24 25 26 27 28 29 30 31 16 17 18 19 20 21 22 23  8  9 10 11 12 13 14 15  0  1  2  3  4  5  6  7
    3                       2                        1                       0

  That is, QUIRK_MSB_ON_THE_RIGHT does not affect byte positioning, but
  inverts bit offsets inside a byte.

Essentially, the mapping for physical bit offsets should be reserved for a
given byte within the payload. This reversal should be fixed to the bytes
in the packing layout.

The logic to implement this quirk is handled within the
adjust_for_msb_right_quirk() function. This function does not work properly
when dealing with the bytes that contain only a partial amount of data.

In particular, consider trying to pack or unpack the range 53-44. We should
always be mapping the bits from the logical ordering to their physical
ordering in the same way, regardless of what sequence of bits we are
unpacking.

This, we should grab the following logical bits:

  Logical: 55 54 53 52 51 50 49 48 47 45 44 43 42 41 40 39
                  ^  ^  ^  ^  ^  ^  ^  ^  ^

And pack them into the physical bits:

   Physical: 48 49 50 51 52 53 54 55 40 41 42 43 44 45 46 47
    Logical: 48 49 50 51 52 53                   44 45 46 47
              ^  ^  ^  ^  ^  ^                    ^  ^  ^  ^

The current logic in adjust_for_msb_right_quirk is broken. I believe it is
intending to map according to the following:

  Physical: 48 49 50 51 52 53 54 55 40 41 42 43 44 45 46 47
   Logical:       48 49 50 51 52 53 44 45 46 47
                   ^  ^  ^  ^  ^  ^  ^  ^  ^  ^

That is, it tries to keep the bits at the start and end of a packing
together. This is wrong, as it makes the packing change what bit is being
mapped to what based on which bits you're currently packing or unpacking.

Worse, the actual calculations within adjust_for_msb_right_quirk don't make
sense.

Consider the case when packing the last byte of an unaligned packing. It
might have a start bit of 7 and an end bit of 5. This would have a width of
3 bits. The new_start_bit will be calculated as the width - the box_end_bit
- 1. This will underflow and produce a negative value, which will
ultimate result in generating a new box_mask of all 0s.

For any other values, the result of the calculations of the
new_box_end_bit, new_box_start_bit, and the new box_mask will result in the
exact same values for the box_end_bit, box_start_bit, and box_mask. This
makes the calculations completely irrelevant.

If box_end_bit is 0, and box_start_bit is 7, then the entire function of
adjust_for_msb_right_quirk will boil down to just:

    *to_write = bitrev8(*to_write)

The other adjustments are attempting (incorrectly) to keep the bits in the
same place but just reversed. This is not the right behavior even if
implemented correctly, as it leaves the mapping dependent on the bit values
being packed or unpacked.

Remove adjust_for_msb_right_quirk() and just use bitrev8 to reverse the
byte order when interacting with the packed data.

In particular, for packing, we need to reverse both the box_mask and the
physical value being packed. This is done after shifting the value by
box_end_bit so that the reversed mapping is always aligned to the physical
buffer byte boundary. The box_mask is reversed as we're about to use it to
clear any stale bits in the physical buffer at this block.

For unpacking, we need to reverse the contents of the physical buffer
*before* masking with the box_mask. This is critical, as the box_mask is a
logical mask of the bit layout before handling the QUIRK_MSB_ON_THE_RIGHT.

Add several new tests which cover this behavior. These tests will fail
without the fix and pass afterwards. Note that no current drivers make use
of QUIRK_MSB_ON_THE_RIGHT. I suspect this is why there have been no reports
of this inconsistency before.

Signed-off-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Link: https://patch.msgid.link/20241002-packing-kunit-tests-and-split-pack-unpack-v2-8-8373e551eae3@intel.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
2024-10-03 15:32:04 -07:00
arch move asm/unaligned.h to linux/unaligned.h 2024-10-02 17:23:23 -04:00
block move asm/unaligned.h to linux/unaligned.h 2024-10-02 17:23:23 -04:00
certs sign-file,extract-cert: use pkcs11 provider for OPENSSL MAJOR >= 3 2024-09-20 19:52:48 +03:00
crypto move asm/unaligned.h to linux/unaligned.h 2024-10-02 17:23:23 -04:00
Documentation lib: packing: adjust definitions and implementation for arbitrary buffer lengths 2024-10-03 15:32:03 -07:00
drivers Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net 2024-10-03 10:05:55 -07:00
fs three small ksmbd server fixes 2024-10-03 09:38:16 -07:00
include lib: packing: duplicate pack() and unpack() implementations 2024-10-03 15:32:04 -07:00
init Rust changes for v6.12 2024-09-25 10:25:40 -07:00
io_uring for-6.12/io_uring-20240922 2024-09-24 11:11:38 -07:00
ipc struct fd layout change (and conversion to accessor helpers) 2024-09-23 09:35:36 -07:00
kernel Getting rid of asm/unaligned.h includes 2024-10-02 16:42:28 -07:00
lib lib: packing: fix QUIRK_MSB_ON_THE_RIGHT behavior 2024-10-03 15:32:04 -07:00
LICENSES LICENSES: add 0BSD license text 2024-09-01 20:43:24 -07:00
mm 19 hotfixes. 13 are cc:stable. 2024-09-27 10:27:22 -07:00
net Including fixes from ieee802154, bluetooth and netfilter. 2024-10-03 09:44:00 -07:00
rust Rust changes for v6.12 2024-09-25 10:25:40 -07:00
samples [tree-wide] finally take no_llseek out 2024-09-27 08:18:43 -07:00
scripts Reduce Coccinelle choices in string_choices.cocci 2024-09-28 21:33:11 +02:00
security move asm/unaligned.h to linux/unaligned.h 2024-10-02 17:23:23 -04:00
sound move asm/unaligned.h to linux/unaligned.h 2024-10-02 17:23:23 -04:00
tools Merge git://git.kernel.org/pub/scm/linux/kernel/git/netdev/net 2024-10-03 10:05:55 -07:00
usr initramfs: shorten cmd_initfs in usr/Makefile 2024-07-16 01:07:52 +09:00
virt x86: 2024-09-28 09:20:14 -07:00
.clang-format clang-format: Update with v6.11-rc1's for_each macro list 2024-08-02 13:20:31 +02:00
.cocciconfig scripts: add Linux .cocciconfig for coccinelle 2016-07-22 12:13:39 +02:00
.editorconfig .editorconfig: remove trim_trailing_whitespace option 2024-06-13 16:47:52 +02:00
.get_maintainer.ignore Add Jeff Kirsher to .get_maintainer.ignore 2024-03-08 11:36:54 +00:00
.gitattributes .gitattributes: set diff driver for Rust source code files 2023-05-31 17:48:25 +02:00
.gitignore Kbuild updates for v6.12 2024-09-24 13:02:06 -07:00
.mailmap Summary 2024-09-24 11:08:40 -07:00
.rustfmt.toml rust: add .rustfmt.toml 2022-09-28 09:02:20 +02:00
COPYING COPYING: state that all contributions really are covered by this file 2020-02-10 13:32:20 -08:00
CREDITS MAINTAINERS: Mark powerpc spufs as orphaned 2024-08-19 21:27:56 +10:00
Kbuild Kbuild updates for v6.1 2022-10-10 12:00:45 -07:00
Kconfig kbuild: ensure full rebuild when the compiler is updated 2020-05-12 13:28:33 +09:00
MAINTAINERS lib: packing: add KUnit tests adapted from selftests 2024-10-03 15:32:04 -07:00
Makefile Linux 6.12-rc1 2024-09-29 15:06:19 -07:00
README README: Fix spelling 2024-03-18 03:36:32 -06:00

Linux kernel
============

There are several guides for kernel developers and users. These guides can
be rendered in a number of formats, like HTML and PDF. Please read
Documentation/admin-guide/README.rst first.

In order to build the documentation, use ``make htmldocs`` or
``make pdfdocs``.  The formatted documentation can also be read online at:

    https://www.kernel.org/doc/html/latest/

There are various text files in the Documentation/ subdirectory,
several of them using the reStructuredText markup notation.

Please read the Documentation/process/changes.rst file, as it contains the
requirements for building and running the kernel, and information about
the problems which may result by upgrading your kernel.