]> www.infradead.org Git - users/jedix/linux-maple.git/commit
net: phy: fixed_phy: handle link-down case
authorStas Sergeev <stsp@list.ru>
Tue, 21 Jul 2015 00:49:56 +0000 (17:49 -0700)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Sat, 3 Oct 2015 11:49:17 +0000 (13:49 +0200)
commit282117acdfac7b5d621b18f002a9751d7f583daf
tree4bd7bfc94416895078297bded5746d82b5047d32
parent90eb52c94901af7f78f9751f8ffff67fbb0bf741
net: phy: fixed_phy: handle link-down case

[ Upstream 868a4215be9a6d80548ccb74763b883dc99d32a2 in net-next tree,
  will be pushed to Linus very soon. ]

fixed_phy_register() currently hardcodes the fixed PHY link to 1, and
expects to find a "speed" parameter to provide correct information
towards the fixed PHY consumer.

In a subsequent change, where we allow "managed" (e.g: (RS)GMII in-band
status auto-negotiation) fixed PHYs, none of these parameters can be
provided since they will be auto-negotiated, hence, we just provide a
zero-initialized fixed_phy_status to fixed_phy_register() which makes it
fail when we call fixed_phy_update_regs() since status.speed = 0 which
makes us hit the "default" label and error out.

Without this change, we would also see potentially inconsistent
speed/duplex parameters for fixed PHYs when the link is DOWN.

CC: netdev@vger.kernel.org
CC: linux-kernel@vger.kernel.org
Signed-off-by: Stas Sergeev <stsp@users.sourceforge.net>
[florian: add more background to why this is correct and desirable]
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/net/phy/fixed_phy.c