This series refactors the EEPROM write logic in e1000 and e1000e drivers
to avoid processing uninitialized memory. Instead of looping over the
entire buffer, we now only perform endianness conversion on the boundary
words that were actually read from the hardware.

Patch 1: e1000: limit endianness conversion to boundary words
Patch 2: e1000e: limit endianness conversion to boundary words
---
v3:
 - Reverted to v1's "check-then-convert" logic in patch for e1000e: the
   return value of e1000_read_nvm() is now checked before performing
   le16_to_cpus().
 - Removed the redundant full-buffer loops in patch for e1000e that
   caused double endianness conversion in v2.

v2:
 - Moved these improvements to the 'net-next' tree.
 - Improved commit description for clarity.

 .../net/ethernet/intel/e1000/e1000_ethtool.c  | 11 +++++++----
 drivers/net/ethernet/intel/e1000e/ethtool.c   | 19 ++++++++++++-------
 2 files changed, 19 insertions(+), 11 deletions(-)

-- 
2.51.0

On Wed, 01. Apr 15:08, Agalakov Daniil wrote:
> This series refactors the EEPROM write logic in e1000 and e1000e drivers
> to avoid processing uninitialized memory. Instead of looping over the
> entire buffer, we now only perform endianness conversion on the boundary
> words that were actually read from the hardware.
> 
> Patch 1: e1000: limit endianness conversion to boundary words
> Patch 2: e1000e: limit endianness conversion to boundary words
> ---

Daniil, for future submissions, please post new versions of the patches
in a separate email-thread, not In-Reply-To.

https://docs.kernel.org/process/maintainer-netdev.html#resending-after-review
https://docs.kernel.org/process/maintainer-netdev.html#changes-requested