kern/124127: [msk] watchdog timeout (missed Tx interrupts) -- recovering

[Andrew Bliznak]

The following reply was made to PR kern/124127; it has been noted by GNATS.

From: Andrew Bliznak <andriko.b at gmail.com>
To: pyunyh at gmail.com
Cc: bug-followup at freebsd.org
Subject: Re: kern/124127: [msk] watchdog timeout (missed Tx interrupts) -- 
	recovering
Date: Thu, 9 Jul 2009 14:48:12 +0300

 2009/7/8 Pyun YongHyeon <pyunyh at gmail.com>:
 > On Mon, Jul 06, 2009 at 11:00:09AM +0000, Andrew Bliznak wrote:
 >> The following reply was made to PR kern/124127; it has been noted by GNA=
 TS.
 >>
 >> From: Andrew Bliznak <andriko.b at gmail.com>
 >> To: bug-followup at FreeBSD.org, skylord at linkline.ru
 >> Cc:
 >> Subject: Re: kern/124127: [msk] watchdog timeout (missed Tx interrupts) =
 --
 >> =A0 =A0 =A0 recovering
 >> Date: Mon, 6 Jul 2009 13:20:12 +0300
 >>
 >> =A0Same problem with Asus P5BV-C/4L in production.
 >>
 >
 > Show me dmesg output (msk(4) and e1000phy(4) part) to see what
 > controller you have. Due to the diverse Yukon controllers and their
 > silicon bugs for each revision it's really hard to tell you're
 > seeing the same issue of this PR. Please show me more information.
 >
 
 mskc0: <Marvell Yukon 88E8056 Gigabit Ethernet> port 0xd800-0xd8ff mem
 0xfbefc000-0xfbefffff irq 18 at device 0.0 on pci5
 msk0: <Marvell Technology Group Ltd. Yukon EC Ultra Id 0xb4 Rev 0x03> on ms=
 kc0
 msk0: Ethernet address: 00:22:15:ef:02:06
 miibus0: <MII bus> on msk0
 e1000phy0: <Marvell 88E1149 Gigabit PHY> PHY 0 on miibus0
 e1000phy0:  10baseT, 10baseT-FDX, 100baseTX, 100baseTX-FDX, 1000baseTX-FDX,=
  auto
 mskc0: [FILTER]
 
 Huh, yes, it's different problem, which bug number for this?
 
 >> =A0I found desktop with Asus P5LD2 MB and use it for experimental setup.
 >> =A0Three host is same gig switch, msk0 configured with vlans and
 >> =A0ipforwarding
 >> =A0For tests I run iperf -c 192.168.4.21 -t 60000 -w 128k -P 4 between
 >> =A0hosts on different vlans.
 >> =A07.2-RELEASE kernel with if_msk* from head +patch. Need to specify
 >> =A0ifconfig -vlanhwtag =A0to run without input errors,
 >> =A0after 30 minutes card locked, =A0down/up interface not help, need to =
 reboot.
 >
 > Need to know which controller you have.
 >
 
 mskc0: <Marvell Yukon 88E8053 Gigabit Ethernet> port 0xc800-0xc8ff mem
 0xcbffc000-0xcbffffff irq 19 at device 0.0 on pci2
 mskc0: Reserved 0x4000 bytes for rid 0x10 type 3 at 0xcbffc000
 mskc0: MSI count : 2
 mskc0: attempting to allocate 2 MSI vectors (2 supported)
 mskc0: using IRQs 256-257 for MSI
 mskc0: RAM buffer size : 48KB
 mskc0: Port 0 : Rx Queue 32KB(0x00000000:0x00007fff)
 mskc0: Port 0 : Tx Queue 16KB(0x00008000:0x0000bfff)
 msk0: <Marvell Technology Group Ltd. Yukon EC Id 0xb6 Rev 0x02> on mskc0
 msk0: bpf attached
 msk0: Ethernet address: 00:17:31:bd:1a:64
 miibus0: <MII bus> on msk0
 mskc0: [MPSAFE]
 mskc0: [FILTER]
 mskc0: Uncorrectable PCI Express error
 msk0: link state changed to UP
 
 
 >> =A0Next I connect disk witch Ubuntu 9.04, setup vlans, test run ok,
 >> =A0little lover forwarding rate.
 >> =A0After, boot again freebsd, run test. Hmm, test run ok...
 >> =A0Reboot with stock 7.2-REL GENERIC kernel. Box running well over weeke=
 nd...
 >
 > This confuse me. So you don't see issues any more?
 >
 
 Yes.
 
 I instaled 8.0-BETA1-amd64 on this box, rerun tests, all ok.
 Also try clear CMOS from BIOS - no change. Next, I reflash BIOS from disket=
 te
 and problem returns.
 Build new kernel with patch-2.diff, now test run ok.
 Rebuild kernel without patch, and this time msk works.
 Look like very strange HW, but patch obviously helped.