diff options
author | David S. Miller <davem@davemloft.net> | 2019-12-11 23:46:13 +0300 |
---|---|---|
committer | David S. Miller <davem@davemloft.net> | 2019-12-11 23:46:13 +0300 |
commit | bea0f4a5115aaf6f59c6d2125f52ff149874b6d2 (patch) | |
tree | 409bb3dbcbad81f281ab7beeceb0da7883f4c232 /fs/pnode.c | |
parent | b58e9fd483482a075bf6cad988c1ef43022417ce (diff) | |
parent | 1cb89a14c80a87abe1f5bd47b1e7b3171b968f8c (diff) | |
download | linux-bea0f4a5115aaf6f59c6d2125f52ff149874b6d2.tar.xz |
Merge branch 'sfp-slow-to-probe-copper'
Russell King says:
====================
Add support for slow-to-probe-PHY copper SFP modules
This series, following on from the previous adding SFP+ copper support,
adds support for a range of Copper SFP modules, made by a variety of
companies, all of which have a Marvell 88E1111 PHY on them, but take
far longer than the Marvell spec'd 15ms to start communicating on the
I2C bus.
Researching the Champion One 1000SFPT module reveals that TX_DISABLE is
routed through a MAX1971 switching regulator and reset IC which adds a
175ms delay to releasing the 88E1111 reset.
It is not known whether other modules use a similar setup, but there
are a range of modules that are slow for the Marvell PHY to appear.
This patch series adds support for these modules by repeatedly trying
to probe the PHY for up to 600ms.
====================
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'fs/pnode.c')
0 files changed, 0 insertions, 0 deletions