Re: Onboard RTL8111C not attached to by re(4)

On Nov 16, 2007 10:52 PM, Aryeh M. Friedman <aryeh.friedman_at_gmail.com> wrote:
>
> Something else must be going on because on my Mobo (MSI Neo-F
> [P35/IHC9]) I have no issue with this (and never have [except for some
> re(4) issues that have nothing to do with reconizing the nic]):
>
>
> re0_at_pci0:4:0:0: class=0x020000 card=0x360c1462 chip=0x816810ec
> rev=0x01 hdr=0x00
>     vendor     = 'Realtek Semiconductor'
>     device     = 'RTL8168/8111 PCI-E Gigabit Ethernet NIC'
>     class      = network
>     subclass = ethernet
>
>
> re0: <RealTek 8168/8111B PCIe Gigabit Ethernet> port 0xe800-0xe8ff mem
> 0xfcfff000-0xfcffffff irq 17 at device 0.0 on pci4
> re0: Using 2 MSI messages
> miibus0: <MII bus> on re0
> re0: Ethernet address: 00:19:db:b5:f8:0f
> re0: [FILTER]
> re0: [FILTER]
>
> FreeBSD monster 8.0-CURRENT FreeBSD 8.0-CURRENT #1: Thu Nov 15
> 19:17:50 EST 2007
> aryeh_at_monster:/usr/obj/FreeBSD/FreeBSD-current/src/sys/MONSTER  amd64
>
>
>


Your card is discovered as card=0x360c1462, mine as card=0xe0001458. I
think yours is really a 8168/8111B, while mine is a 8111C.


Accroding to the followling line in /sys/dev/re/if_re.c, re_attach(dev),
  hwrev = CSR_READ_4(sc, RL_TXCFG) & RL_TXCFG_HWREV;

and the definition in /sys/pci/if_rlreg.h,
  #define RL_TXCFG_HWREV          0x7CC00000


I traced a debug kernel and found that
  CSR_READ_4(sc, RL_TXCFG)
for my card is
  0x3f2000c0
so the masked hwrev should be
  0x3c000000


I added this line in /sys/pci/if_rlreg.h,
  #define RL_HWREV_8111C		0x3c000000
and these lines to the table in /sys/dev/re/if_re.c,
  { RT_VENDORID, RT_DEVICEID_8168, RL_HWREV_8111C,
    "RealTek 8111C Single-Chip PCIe Gigabit Ethernet" },
and these to the table in the same .c.
  { RL_HWREV_8111C, RL_8169, "8111C"},


After reboot, re(4) successfully recognized the chip. However, it
failed with the following error:

  re0: MII without any phy!


I traced the mii_phy_probe() function in /sys/dev/mii/mii.c, only to
find all the bmsr values read by MIIBUS_READREG was:
  0, 0xffff, 0, 0, ... (all 0's)

so that mii_phy_probe() did not find any PHY.


What's the problem?


-- 
Quan Qiu <jackqqpro_at_gmail.com>