bce issues still outstanding
Scott Long
scottl at samsco.org
Thu Oct 12 17:05:25 UTC 2006
Tom Judge wrote:
> Scott Long wrote:
>
>> Bill Moran wrote:
>>
>>> I've copied many of the people who I've been working with directly on
>>> this issue.
>>>
>>> Can anyone provide a status update on these problems? Discussion seems
>>> to have stopped since Oct 5.
>>>
>>> Any new patches to test?
>>>
>>
>> I'm actively working on fixing the driver right now.
>>
>> Scott
>>
>
> If there are any patches you want testing I have 3 very expensive
> paperweights sat around the office at the moment in the form of Dell
> PE2950's with twin adapters in them:
>
> bce0: <Broadcom NetXtreme II BCM5708 1000Base-T (B1), v0.9.6> mem
> 0xf4000000-0xf5ffffff irq 16 at device 0.0 on pci9
> bce0: ASIC ID 0x57081010; Revision (B1); PCI-X 64-bit 133MHz
> miibus0: <MII bus> on bce0
> brgphy0: <BCM5708C 10/100/1000baseTX PHY> on miibus0
> brgphy0: 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, 1000baseTX,
> 1000baseTX-FDX, auto
> bce0: Ethernet address: 00:13:72:60:35:8a
>
>
> bce1: <Broadcom NetXtreme II BCM5708 1000Base-T (B1), v0.9.6> mem
> 0xf8000000-0xf9ffffff irq 16 at device 0.0 on pci5
> bce1: ASIC ID 0x57081010; Revision (B1); PCI-X 64-bit 133MHz
> miibus1: <MII bus> on bce1
> brgphy1: <BCM5708C 10/100/1000baseTX PHY> on miibus1
> brgphy1: 10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, 1000baseTX,
> 1000baseTX-FDX, auto
> bce1: Ethernet address: 00:13:72:60:35:88
>
> The easiest way to cause the watchdog time out on these systems is to
> write a file larger than 50Mb to an NFS file system that is mounted over
> UDP (TCP doesn't cause the time out strangely). From the testing I have
> done the timeout is triggered when arround 49Mb has been copied to the
> NFS server. Perhaps this could suggest a bug in the udp packet checksum
> offload code?
>
> I have 2 systems available to test on right now, one running code from
> RELENG_6 just after beta2 was announced, and one cvsup'd from the uk
> mirror today.
>
> The kernels are compiled with:
>
> option INVARIANTS
> option INVARIANT_SUPPORT
>
> However this doesn't cause a kernel panic.
>
> Let me know if you need any more information.
>
> Tom
It's very easy to make UDP overrun the transmit ring in the bce driver.
This is my primary focus at the moment. After that, I need to look at
performance problems when using cvs.
Scott
More information about the freebsd-stable
mailing list