Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 30 Jun 2003 09:40:18 +0100
From:      "Duncan Barclay" <dmlb@dmlb.org>
To:        "Kenneth D. Merry" <ken@kdm.org>
Cc:        freebsd-current@freebsd.org
Subject:   PCI help needed - Re: Very alpha BroadCom 4401 driver
Message-ID:  <005b01c33ee4$455dd9f0$a7ac77c1@DJK1Comp>
References:  <000501c33e6f$e954da10$4bc8a8c0@orac> <20030629162638.A93448@panzer.kdm.org> <001801c33e8e$799ff4b0$4bc8a8c0@orac> <20030629170229.A93754@panzer.kdm.org>

next in thread | previous in thread | raw e-mail | index | archive | help
<CC'd to -current as this is a PCI memory mapping problem than I don't know
how to get around>

From: "Kenneth D. Merry" <ken@kdm.org>

> I sent dmesg and pciconf output last week...I've attached it again.

Thanks. The penny is starting to drop here. A night's sleep does wonders.

When you boot your machine it gets things right:
    bcm0: <Broadcom 10/100 Base-T Ethernet> mem 0xfaffe000-0xfaffffff irq 11
at device 0.0 on pci2
    pcib2: device bcm0 requested decoded memory range 0xfaffe000-0xfaffffff
    bcm0: Ethernet address: 00:0b:db:94:bf:42
    bcm0: core 0x806 revi 0x4 vend 0x4243 match 0x18000058 (type 0x0 base
0x18000000) inst 0x0
    bcm0: core 0x804 revi 0x2 vend 0x4243 match 0x18002058 (type 0x0 base
0x18002000) inst 0x0

When loaded as a module, the value for "match" is 0xffffffff, and this is
wrong.

> I just tried loading it as a module, and got the panic again.  One
> interesting thing is that is shows the MAC address as ff:ff:ff:ff:ff:ff.
>
> Another interesting thing is that if I insert a cardbus fxp card (the
> Broadcom chip is on board), the whole system locks up hard if the bcm
> driver is compiled into the kernel.  If the bcm driver isn't there, things
> work as expected.

All this points to something strange with memory mappings - it may be your
machine or the Broadcom chip set. On your machine, the chip is at a very
high address 0xfaffe000-0xfaffffff. For some reason, when used as a module,
this isn't being correctly mapped to the device driver - I have no idea why
not as this is wanderering into PCI voodoo.

Can anyone else suggest what's happening here? Ken has dmesg output
available.

> Ken

Duncan





Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?005b01c33ee4$455dd9f0$a7ac77c1>