Instability in re driver?

[Dennis Glatting]

On Thu, 2014-02-20 at 10:27 -0600, Dan Nelson wrote:
> In the last episode (Feb 19), Brett Glass said:
> > We've been experiencing occasional crashes on heavily loaded machines that
> > have Realtek gigabit Ethernet ports built into their motherboards.  This
> > evening, at a time of peak usage, one of the machines showed the log
> > messages
> > 
> > Feb 19 18:44:14 <kern.crit> server kernel: re0: watchdog timeout
> > Feb 19 18:44:14 <kern.notice> server kernel: re0: link state changed to DOWN
> > Feb 19 18:44:18 <kern.notice> server kernel: re0: link state changed to UP
> > 
> > even though we did not take the port down or up. A few minutes later, the
> > entire machine locked up solid and required a power cycle.
> > 
> > Here are the boot time messages from the same server (as I rebooted it
> > following the crash):
> > 
> > Feb 19 19:08:22 <kern.crit> server kernel: re0: <RealTek 8168/8111 B/C/CP/D/DP/E/F PCIe Gigabit Ethernet> port 0xd800-0xd8ff mem 0xfeadf000-0xfeadffff,0xfdefc000-0xfdefffff irq 16 at device 0.0 on pci1
> > Feb 19 19:08:22 <kern.crit> server kernel: re0: Using 1 MSI-X message
> > Feb 19 19:08:22 <kern.crit> server kernel: re0: Chip rev. 0x28000000
> > Feb 19 19:08:22 <kern.crit> server kernel: re0: MAC rev. 0x00000000
> > Feb 19 19:08:22 <kern.crit> server kernel: miibus0: <MII bus> on re0
> > Feb 19 19:08:22 <kern.crit> server kernel: rgephy0: <RTL8169S/8110S/8211 1000BASE-T media interface> PHY 1 on miibus0
> > Feb 19 19:08:22 <kern.crit> server kernel: rgephy0:  none, 10baseT, 10baseT-FDX, 10baseT-FDX-flow, 100baseTX, 100baseTX-FDX, 100baseTX-FDX-flow, 1000baseT, 1000baseT-master, 1000baseT-FDX, 1000baseT-FDX-master, 1000baseT-FDX-flow, 1000baseT-FDX-flow-master, auto, auto-flow
> > 
> > This problem is occurring on several servers built with the same
> > motherboard/chipset.  Are there known bugs, hangs, or memory leaks that
> > might have led to this condition?  Are there recent fixes which address
> > it?  (The machines that are crashing are all running the latest security
> > patch level of FreeBSD 9.1-RELEASE.)
> 
> I have had a similar persistent problem with a re0 interface in one of my
> machines, where medium UDP traffic sometimes causes a problem with the clock
> timer.  Packets stop flowing, calls to nanosleep() never return, etc.  I
> have watchdogd enabled, so the system ends up resetting on its own.  Without
> that, it would just sit in a half-frozen state indefinitely.  The machine is
> an old Dell Studio 540s, and I've seen the problem on FreeBSD 7, 8 and 9,
> i386 and amd64.  I assumed it was just a problem with my motherboard since
> nobody else has ever reported a similar problem, and it only resets once or
> twice a month, so it doesn't really affect me.  Neither disabling MSI nor
> using a different timecounter seems to help.
> 

I have the same problem on a new ASUS A78M-A MB. Transfer is really fast
(300+ GB files) then poof. Generally, I stay away from RealTek,
replacing my network interfaces with Intel, but on A78M-A I don't much
choice.



> re0: <RealTek 8168/8111 B/C/CP/D/DP/E/F/G PCIe Gigabit Ethernet> port 0xe800-0xe8ff mem 0xfebff000-0xfebfffff,0xfdff0000-0xfdffffff irq 17 at device 0.0 on pci3
> re0: MSI count : 1
> re0: MSI-X count : 2
> re0: attempting to allocate 1 MSI-X vectors (2 supported)
> msi: routing MSI-X IRQ 259 to local APIC 0 vector 58
> re0: using IRQ 259 for MSI-X
> re0: Using 1 MSI-X message
> re0: Chip rev. 0x3c000000
> re0: MAC rev. 0x00400000
> miibus0: <MII bus> on re0
> rgephy0: <RTL8169S/8110S/8211 1000BASE-T media interface> PHY 1 on miibus0
> rgephy0: OUI 0x00e04c, model 0x0011, rev. 2
> rgephy0:  none, 10baseT, 10baseT-FDX, 10baseT-FDX-flow, 100baseTX, 100baseTX-FDX, 100baseTX-FDX-flow, 1000baseT, 1000baseT-master, 1000baseT-FDX, 1000baseT-FDX-master, 1000baseT-FDX-flow, 1000baseT-FDX-flow-master, auto, auto-flow
> 
>