On 23/01/2013 00:48, Troy Kisky wrote:
On 11/10/2012 12:28 AM, Stefano Babic wrote:
On 23/10/2012 04:40, Troy Kisky wrote:
This series tries to separate the mii regsisters from the ethernet registers as suggested by Andy Fleming. Then, mx6qsabrelite is changed to find the phy address from the possibles 4-7.
The V3 series is very different from V2.
Troy Kisky (9): doc/README.fec_mxc: add documentation net: fec_mxc: delete CONFIG_FEC_MXC_MULTI net: fec_mxc: change fec_mii_setspeed parameter net: fec_mxc: have fecmxc_initialize call fecmxc_initialize_multi phy: add phy_find_by_mask/phy_connect_dev net: fec_mxc: use fec_set_dev_name to set name net: fec_mxc: only call phy_connect in fec_probe net: fec_mxc: get phydev before fec_probe mx6qsabrelite: search mii phy address 4-7
Hi Joe,
board/freescale/mx6qsabrelite/mx6qsabrelite.c | 24 +++- doc/README.fec_mxc | 27 ++++ drivers/net/fec_mxc.c | 164 ++++++++++++++----------- drivers/net/fec_mxc.h | 2 +- drivers/net/phy/phy.c | 128 +++++++++++-------- include/configs/m28evk.h | 1 - include/configs/mx28evk.h | 1 - include/configs/sc_sps_1.h | 1 - include/netdev.h | 7 ++ include/phy.h | 3 + 10 files changed, 227 insertions(+), 131 deletions(-) create mode 100644 doc/README.fec_mxc
Patches are assigned to me in patchworks. Do you have any issue with this patchset ? If it is ok for you, I will apply it to u-boot-imx.
Best regards, Stefano
Who should I pester to see some progress on this ?
Hi Troy,
you're right, this patchset stalls - but on the other side, I have not found any comment that should avoid that it can be merged.
If nobody complains, I propose I go on and I try to merge it into u-boot-imx. Here my first attempt:
as far as I can see, the patchset breaks mx6qsabreauto and mx35 boards. Generally speking, it breaks boards like flea3 that define CONFIG_PHYLIB in their config file.
undefined reference to `phy_connect_dev' drivers/net/libnet.o: In function `fecmxc_initialize_multi':
I think the patchset assumes that boards have its own board_eth_init(), making the required phy initialization, but most boards call only cpu_eth_init().
I have also a general question : it seems to me that the interface for PHY is set always to RMII - what about boards using MII ?
Best regards, Stefano Babic