ale(4) not attaching after ACPI resume

[Sascha Klauder]

Hi all,

 I'm trying to get suspend/resume functionality to work on my
Acer Aspire One D250.  I'm running 8.2-STABLE i386 2011-02-25.

 Suspend was doing fine, but it would freeze on resume.  By
trial and error I've been able to pinpoint the problematic
driver as ale(4).  The machine will correctly resume once I un-
load if_ale before suspending.

 I wouldn't mind that as a workaround, however, it fails to
reattach after a suspend/resume cycle:

pci0: driver added
pci1: driver added
found-> vendor=0x1969, dev=0x1026, revid=0xb0
        domain=0, bus=1, slot=0, func=0
        class=02-00-00, hdrtype=0x00, mfdev=0
        cmdreg=0x0407, statreg=0x0010, cachelnsz=0 (dwords)
        lattimer=0x00 (0 ns), mingnt=0x00 (0 ns), maxlat=0x00 (0 ns)
        intpin=a, irq=16
        powerspec 2  supports D0 D3  current D3
        MSI supports 1 message, 64 bit
pci0:1:0:0: reprobing on driver added
pci0:1:0:0: Transition from D3 to D0
ale0: <Atheros AR8121/AR8113/AR8114 PCIe Ethernet> port 0x3000-0x307f mem 0x95200000-0x9523ffff irq 16 at device 0.0 on pci1
pcib1: ale0 requested memory range 0x95200000-0x9523ffff: good
ale0: phy write timeout : 29
ale0: phy write timeout : 30
ale0: phy write timeout : 29
ale0: phy write timeout : 30
ale0: phy write timeout : 29
ale0: phy write timeout : 30
ale0: phy write timeout : 29
ale0: phy write timeout : 29
ale0: phy write timeout : 29
ale0: phy write timeout : 29
ale0: master reset timeout!
ale0: reset timeout(0xffffffff)!
ale0: PCI device revision : 0x00b0
ale0: Chip id/revision : 0xffff
ale0: chip revision : 0xffff, 4294967295 Tx FIFO 4294967295 Rx FIFO -- not initialized?
device_attach: ale0 attach returned 6

 It does attach and work correctly even if loaded after booting
(provided there's been no suspend/resume cycle).  Repeated loading
and unloading works as well.

 Anything obvious that I could try?  Verbose dmesg is available
here: http://arwen.shopkeeper.de/~sascha/acer/dmesg.verbose

Cheers,
-sascha