linux-kernel - [PATCH 0/3] Use new for_each macro to create hexdumps

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-ID: <20250113221721.362093-1-nnac123@linux.ibm.com>
Date: Mon, 13 Jan 2025 16:17:18 -0600
From: Nick Child <nnac123@...ux.ibm.com>
To: linux-kernel@...r.kernel.org
Cc: nick.child@....com, netdev@...r.kernel.org,
        Nick Child <nnac123@...ux.ibm.com>
Subject: [PATCH 0/3] Use new for_each macro to create hexdumps

Apologies, not sure what mailing list/tree to target. First 2 patches look
like *-next and last patch should go to net-next.

Currently, obtaining a hexdump can be done through one of the following:
 1. hex_dump_to_buffer - takes at most 32 bytes of a buffer and returns a 
     hexdump string representation
 2. print_hex_dump - prints output of hex_dump_to_buffer iteratively over
    a large buffer

There is no functionality for iterating over a large buffer and receiving
the string representation. It seems most users of hex_dump_to_buffer are
calling hex_dump_to_buffer within the body of a loop which iterates
through a buffer.

This patchset creates a for_each macro that accepts a buffer and fills
out an output string with the converted hexdump. This loops over the
buffer and takes care of incrementing pointers. Hopefully this makes
writing sequential calls to hex_dump_to_buffer more straightforward.

>From a users perspective there should be no difference in output.

The inspiration here was I wanted to use print_hex_dump in ibmvnic code
but I wanted to print through netdevice printing functions to maintain
formatting. Looking at other users of hex_dump_to_buffer it seems they had
similar intentions.

Side question:
  hex_dump_to_buffer automatically sets groupsize to 1 if user given
  groupsize is not a multiple of len. When printing large buffers this
  makes non-uniform output. For example, this is a 31 byte 8 groupsize
  buffer:
   ibmvnic 30000003 env3: 6c6774732e737561 2e6d62692e736261
   ibmvnic 30000003 env3: 63 6f 6d 00 03 00 05 65 6e 76 33 00 00 00 00
  Since the second line is only 15 bytes, the group size is set to 1. I
  have written a patch which keeps groupsize so output would be:
   ibmvnic 30000003 env3: 6c6774732e737561 2e6d62692e736261
   ibmvnic 30000003 env3: 636f6d0003000565 6e763300000000
  But since I am not sure if this would break some dependency for someone,
  and my justification for change is purely pedantic, I chose to omit
  that patch in this patchset. Let me know if there is any interest and
  I will send a different patchset for that. 

Thanks for your consideration/review.

Nick Child (3):
  hexdump: Implement macro for converting large buffers
  hexdump: Use for_each macro in print_hex_dump
  ibmvnic: Print data buffers with kernel API's

 drivers/net/ethernet/ibm/ibmvnic.c | 23 ++++++++++++++---------
 include/linux/printk.h             | 21 +++++++++++++++++++++
 lib/hexdump.c                      | 11 +++--------
 3 files changed, 38 insertions(+), 17 deletions(-)

-- 
2.47.1