Re: call for sk(4) testers

On Sat 14 Jan 06 21:02, Pyun YongHyeon <pyunyh_at_gmail.com> wrote:
> On Sat, Jan 14, 2006 at 12:06:48AM -0700, Joshua Tinnin wrote:
>  > On Thu 12 Jan 06 04:32, Pyun YongHyeon <pyunyh_at_gmail.com> wrote:
>  > > I'd like to hear any success/failure reports. If you have a NIC
>  > > that is supported by sk(4) please give it a try and let me know
>  > > the result.
>  >
>  > So far so good, running 6.0-RELEASE-p2 on an i386 machine. The
>  > sk(4) NIC is an onboard Marvell 88E8001 Yukon Gigabit Ethernet on
>  > an Asus A7N8X-E Deluxe motherboard:
>
> Thank you for your time and testing.
> Did you notice a increase in throughput?  While testing stock sk(4)
> I saw numerous "discard frame w/o leading ethernet header..."
> messages under high packet rates. You should not see these messages
> anymore in new driver.

To be honest with you I cannot detect an increase in throughput, but I 
haven't measured it, so it may be better. I can probably try to test 
that some time tomorrow. However, the biggest problem I had with this 
NIC and sk(4) is that I would get watchdog timeouts; the NIC would not 
go completely down, as it happens with some people, but it would go 
down for a few seconds and come back up by itself. That hasn't happened 
with your driver, at least not yet. It typically happens at least once 
a day, so this is a good sign that it hasn't happened so far.

>  > # uname -a
>  > FreeBSD smogmonster.local 6.0-RELEASE-p2 FreeBSD 6.0-RELEASE-p2
>  > #0: Fri Jan 13 23:13:57 MST 2006
>  > krinklyfig_at_smogmonster.local:/usr/obj/usr/src/sys/MYKERNEL60  i386
>  >
>  > # dmesg | grep sk
>  > skc0: <Marvell Gigabit Ethernet> port 0x9000-0x90ff mem
>  > 0xec000000-0xec003fff irq 17 at device 4.0 on pci1
>  > skc0: Marvell Yukon Lite Gigabit Ethernet rev. A3(0x7)
>  > sk0: <Marvell Semiconductor, Inc. Yukon> on skc0
>  > sk0: Ethernet address: 00:0e:a6:70:56:75
>  >
>  > As for dual port, I also have:
>  >
>  > miibus0: <MII bus> on sk0
>  > e1000phy0: <Marvell 88E1000 Gigabit PHY> on miibus0
>  > e1000phy0:  10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX,
>  > 1000baseTX-FDX, auto
>  >
>  > ... although I haven't used that one, yet. I can try it if you
>  > like.
>
> Yes. I'd like to know it has locking bugs on dual port handling.
> I'd also like to know sk(4) works on second generation chips from
> Marvell.(e.g. Yukon-II XL/EC/FE)

OK, I'll give it a shot ...

- jt