diff options
author | Petr Machata <petrm@nvidia.com> | 2022-01-18 14:57:34 +0100 |
---|---|---|
committer | Josh Boyer <jwboyer@kernel.org> | 2022-01-18 09:49:45 -0500 |
commit | 4954931df6542157865f5a44454caee68a7c4924 (patch) | |
tree | de7e43c5232be1ed87ea9e53468829b72d22337d /intel/ibt-hw-37.7.10-fw-1.0.2.3.d.bseq | |
parent | ea056f9be2f8d1535db05ffe758e2a279e3e4047 (diff) | |
download | linux-firmware-4954931df6542157865f5a44454caee68a7c4924.tar.gz |
Mellanox: Add new mlxsw_spectrum firmware xx.2010.1232
Add latest verified version of Nvidia Spectrum-family switch firmware,
for Spectrum (13.2010.1232), Spectrum-2 (29.2010.1232) and Spectrum-3
(30.2010.1232).
The following bugs were fixed:
- On Spectrum-3, when cut-through forwarding large frames from a 1Gbps port
to a 100Gbps one, packets would get corrupted.
- Several issues in handling of MPLS on Spectrum-2.
- An issue that caused fast reboot to take longer than 30s.
- An issue with bulk update of FDB where roamed FDB entries were not
updated.
- Maximum MTU reported for Quantum chips was 4222 bytes instead of 4096.
- Packets wrongly discarded due to egress STP filter when VxLAN is
configured.
- On Spectrum-2, on 1Gbps or 10Gbps ports, when congestion occurs, packets
would very rarely get stuck in the chip and cause switch to hang.
- On Spectrum-3, when ECMP has many next hops based on VLAN interfaces
packets may rarely get a wrong VLAN tag and be dropped.
- On Spectrum-3, effective speed of a 400Gbps port would only be 380Gbps
with 1500-byte packets.
- On a switch with an external PHY, attempts to configure PFC are no longer
bounced by FW as unsupported.
- An issue in handling of QinQ packets with EtherType 8100 on Spectrum-3.
- On SN4800, an issue in PCIe gen3 link negotiation instability.
Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: Josh Boyer <jwboyer@kernel.org>
Diffstat (limited to 'intel/ibt-hw-37.7.10-fw-1.0.2.3.d.bseq')
0 files changed, 0 insertions, 0 deletions