diff options
author | David S. Miller <davem@davemloft.net> | 2018-06-26 22:53:29 +0900 |
---|---|---|
committer | David S. Miller <davem@davemloft.net> | 2018-06-26 22:53:29 +0900 |
commit | 094bdaddf129417ee5e5d885d2fe76a5adfada0f (patch) | |
tree | 21bdad75629bdd61a9f65a455f7d729a487f6183 /net | |
parent | 0ee1f4734967af8321ecebaf9c74221ace34f2d5 (diff) | |
parent | 9343ac87f2a4e09bf6e27b5f31e72e9e3a82abff (diff) | |
download | op-kernel-dev-094bdaddf129417ee5e5d885d2fe76a5adfada0f.zip op-kernel-dev-094bdaddf129417ee5e5d885d2fe76a5adfada0f.tar.gz |
Merge branch 'lan78xx-minor-fixes'
Dave Stevenson says:
====================
lan78xx minor fixes
This is a small set of patches for the Microchip LAN78xx chip,
as used in the Raspberry Pi 3B+.
The main debug/discussion was on
https://github.com/raspberrypi/linux/issues/2458
Initial symptoms were that VLANs were very unreliable.
A couple of things were found:
- firstly that the hardware timeout value set failed to
take into account the VLAN tag, so a full MTU packet
would be timed out.
- second was that regular checksum failures were being
reported. Disabling checksum offload confirmed that
the checksums were valid, and further experimentation
identified that it was only if the VLAN tags were being
passed through to the kernel that there were issues.
The hardware supports VLAN filtering and tag stripping,
therefore those have been implemented (much of the work
was already done), and the driver drops back to s/w
checksums should the choice be made not to use the h/w
VLAN stripping.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net')
0 files changed, 0 insertions, 0 deletions