diff options
| author | Florian Fainelli <f.fainelli@gmail.com> | 2014-08-11 14:50:45 -0700 |
|---|---|---|
| committer | David S. Miller <davem@davemloft.net> | 2014-08-11 15:10:11 -0700 |
| commit | 98bb7399d29a53a58f497409f98acb3bc0434ded (patch) | |
| tree | debcce8f9747a00e4d46d4c90847952d1ed74360 /drivers/crypto | |
| parent | c677ba8b3c47650358572091ed8a6af50bfca877 (diff) | |
| download | linux-98bb7399d29a53a58f497409f98acb3bc0434ded.tar.bz2 | |
net: bcmgenet: correctly resume adapter from Wake-on-LAN
In case we configured the adapter to be a wake up source from
Wake-on-LAN, but we never actually woke up using Wake-on-LAN, we will
leave the adapter in MagicPacket matching mode, which prevents any other
type of packets from reaching the RX engine. Fix this by calling
bcmgenet_power_up() with GENET_POWER_WOL_MAGIC to restore the adapter
configuration in bcmgenet_resume().
The second problem we had was an imbalanced clock disabling in
bcmgenet_wol_resume(), the Wake-on-LAN slow clock is only enabled in
bcmgenet_suspend() if we configured Wake-on-LAN, yet we unconditionally
disabled the clock in bcmgenet_wol_resume().
Fixes: 8c90db72f926 ("net: bcmgenet: suspend and resume from Wake-on-LAN")
Signed-off-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'drivers/crypto')
0 files changed, 0 insertions, 0 deletions