Re: [PATCH] nve(4) locking cleanup

On Thursday 17 November 2005 12:08 pm, Mike Tancsa wrote:
> At 11:35 AM 17/11/2005, John Baldwin wrote:
> >Hmm, there are a couple of changes such as calling pfnInit() in attach()
> > and calling pfnClearTXDesc() in nve_stop() that might be interesting. 
> > I've stuck those in my work tree and have made a seperate patch relative
> > to HEAD if folks want to try them out:
> >
> >http://www.FreeBSD.org/~jhb/patches/nve_dffixes.patch
>
> On a RELENG_6 kernel with the drivers from HEAD with the above patch
> applied, driver loaded as kld,
>
> the box does not seem to crash, but the nic does wedge
>
> nve0: link state changed to DOWN
> nve0: link state changed to UP
> nve0: link state changed to DOWN
> nve0: link state changed to UP
> nve0: device timeout (64)
> nve0: link state changed to DOWN
> nve0: link state changed to UP
> nve0: link state changed to DOWN
> nve0: link state changed to UP
>
> while generating a lot of 2 way traffic.
>
> This is a Tyan 939 board.
>
> Trying to mount root from ufs:/dev/da0s1a
> nve0: <NVIDIA nForce MCP9 Networking Adapter> port 0xe400-0xe407 mem
> 0xfebfc000-0xfebfcfff irq 23 at device 10.0 on pci0
> nve0: Ethernet address 00:e0:81:58:91:fa
> miibus2: <MII bus> on nve0
> ukphy0: <Generic IEEE 802.3u media interface> on miibus2
> ukphy0:  10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, 1000baseT,
> 1000baseT-FDX, auto
> nve0: Ethernet address: 00:e0:81:58:91:fa
> nve0: [GIANT-LOCKED]

Did it work before?  Is the new behavior better or worse than the unpatched 
driver?

-- 
John Baldwin <jhb_at_FreeBSD.org>  <><  http://www.FreeBSD.org/~jhb/
"Power Users Use the Power to Serve"  =  http://www.FreeBSD.org