Re: sata + vinum + Asus p4p800 = :(
On Monday, Oct 20, 2003, at 19:02 Europe/Budapest, Alexander Leidinger wrote: Greg 'groggy' Lehey schrieb: Provide a dump? Analyse the problem yourself? This *is* -CURRENT, after all. I can reproduce this here (same MB). I don't think it's a vinum problem, but vinum seems to be a good way to reproduce the bug. I set up a stripe over 2 SATA disks, newfs it, run "iozone -a" and BOOM. Well, my /etc dir is gone, and I have no other bootable device. Tomorrow my hardware guy will bring me a new be7s motherboard... Sorry guys, the rest is in your hands. -- jul ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: sata + vinum + Asus p4p800 = :(
Greg 'groggy' Lehey schrieb: Provide a dump? Analyse the problem yourself? This *is* -CURRENT, after all. I can reproduce this here (same MB). I don't think it's a vinum problem, but vinum seems to be a good way to reproduce the bug. I set up a stripe over 2 SATA disks, newfs it, run "iozone -a" and BOOM. Using just one disk without vinum doesn't result in a panic, but with a verbose boot I see a lot of spurious interrupt messages while running iozone on one disk. The panic is: panic: ata_dmasetup: transfer active on this disk. Backtrace: ata_dmastart ata_pci_dmastart ata_transaction ata_start g_disk_start g_io_schedule_down g_down_procbody fork_exit fork_trampoline Trying to write a dump results in a hard hang of the system. The backtrace is with todays kernel source: ata-lowlevel 1.20 ata-pci.c 1.69 ata-queue.c 1.11 atapi-cd.c 1.149 Verbose dmesg attached (booted with an oct 17 kernel, if it's not enough, I rebuild the kernel with a larger message buffer). Greg, I don't know if the following is ATA related: I'm able to fdisk and disklabel the disks without any prolems, and the label survives a reboot, but rebooting after setting up the stripe results in a lost configuration, only the names of the drives show up with "vinum l", everything else is "clean" (0 volumes, 0 plexes, 0 subdisks). The config is: ---snip--- drive SATA1 device /dev/ad1s1a drive SATA2 device /dev/ad2s1a volume space setupstate plex org striped 279k sd length 0 drive SATA1 sd length 0 drive SATA2 ---snip--- S/oren, one additional datapoint: With the oct 17 kernel I've seen the following output with a verbose boot before it hunged hard (I had DDB_UNATTENDED in the oct 17 kernel): - spurious interrupt messages for ata 0 and 2 - WARNING FLUSHCASHE for ad0 - WARNING WRITE_DMA recovered from missing int for ad1 - TIMEOUT WRITE_DMA retrying for ad2 - ata3 reset tp1 mask=03 ostat0=d0 ostat1=00 ---snip--- Feel free to use me as an testing-ape, I also can provide a root login to this machine if needed (just tell me and I grab the ssh key from freefall). Bye, Alexander. ngnt=0x00 (0 ns), maxlat=0x00 (0 ns) found-> vendor=0x8086, dev=0x2571, revid=0x02 bus=0, slot=1, func=0 class=06-04-00, hdrtype=0x01, mfdev=0 cmdreg=0x0107, statreg=0x00a0, cachelnsz=0 (dwords) lattimer=0x40 (1920 ns), mingnt=0x08 (2000 ns), maxlat=0x00 (0 ns) map[20]: type 4, range 32, base ef00, size 5, enabled pcib0: matched entry for 0.29.INTA (source \\_SB_.LNKA) pcib0: slot 29 INTA is routed to irq 10 found-> vendor=0x8086, dev=0x24d2, revid=0x02 bus=0, slot=29, func=0 class=0c-03-00, hdrtype=0x00, mfdev=1 cmdreg=0x0005, statreg=0x0280, cachelnsz=0 (dwords) lattimer=0x00 (0 ns), mingnt=0x00 (0 ns), maxlat=0x00 (0 ns) intpin=a, irq=10 map[20]: type 4, range 32, base ef20, size 5, enabled pcib0: matched entry for 0.29.INTB (source \\_SB_.LNKD) pcib0: slot 29 INTB is routed to irq 5 found-> vendor=0x8086, dev=0x24d4, revid=0x02 bus=0, slot=29, func=1 class=0c-03-00, hdrtype=0x00, mfdev=0 cmdreg=0x0005, statreg=0x0280, cachelnsz=0 (dwords) lattimer=0x00 (0 ns), mingnt=0x00 (0 ns), maxlat=0x00 (0 ns) intpin=b, irq=5 map[20]: type 4, range 32, base ef40, size 5, enabled pcib0: matched entry for 0.29.INTC (source \\_SB_.LNKC) pcib0: slot 29 INTC is routed to irq 5 found-> vendor=0x8086, dev=0x24d7, revid=0x02 bus=0, slot=29, func=2 class=0c-03-00, hdrtype=0x00, mfdev=0 cmdreg=0x0005, statreg=0x0280, cachelnsz=0 (dwords) lattimer=0x00 (0 ns), mingnt=0x00 (0 ns), maxlat=0x00 (0 ns) intpin=c, irq=5 map[20]: type 4, range 32, base ef80, size 5, enabled pcib0: matched entry for 0.29.INTA (source \\_SB_.LNKA) pcib0: slot 29 INTA is routed to irq 10 found-> vendor=0x8086, dev=0x24de, revid=0x02 bus=0, slot=29, func=3 class=0c-03-00, hdrtype=0x00, mfdev=0 cmdreg=0x0005, statreg=0x0280, cachelnsz=0 (dwords) lattimer=0x00 (0 ns), mingnt=0x00 (0 ns), maxlat=0x00 (0 ns) intpin=a, irq=10 map[10]: type 1, range 32, base febff800, size 10, enabled pcib0: matched entry for 0.29.INTD (source \\_SB_.LNKH) pcib0: slot 29 INTD is routed to irq 11 found-> vendor=0x8086, dev=0x24dd, revid=0x02 bus=0, slot=29, func=7 class=0c-03-20, hdrtype=0x00, mfdev=0 cmdreg=0x0106, statreg=0x0290, cachelnsz=0 (dwords) lattimer=0x00 (0 ns), mingnt=0x00 (0 ns), maxlat=0x00 (0 ns) intpin=d, irq=11 powerspec 2 supports D0 D3 current D0 found-> vendor=0x8086, dev=0x244e, revid=0xc2 bus=0, slot=30, func=0 class=06-04-00, hdrtype=0x01, mfdev=0 cmdreg=0x0107, statreg=0x0080, cachelnsz=0 (dwords) lattimer=0x00 (0 ns), mingnt=0x02 (500 ns), maxlat=0x00 (0 ns) found-> vendor=0x8086, dev=0x24d0, revid=0x02 bus=0, slot=31, func=0
Re: sata + vinum + Asus p4p800 = :(
On Wednesday, Oct 15, 2003, at 10:34 Europe/Budapest, Soren Schmidt wrote: First loose vinum from the mix, if we should have a chance finding the problem we need to lower the number of unknowns. Then tell us if and how it fails, dmesg from verbose booted system, backtrace etc, you know the drill if running -current. Lunchtime I went home just to reboot my box with no vinum config. Sadly it works seamlessly. -- jul ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: sata + vinum + Asus p4p800 = :(
It seems Balazs Nagy wrote: > Hi, > > I had a -CURRENT setting with an Abit BE7-S and two SATA disks with > vinum configuration. It worked very well until a power failure, and > the mainboard died. Yesterday I got a replacement mainboard, the only > type met the requirements (eg. two SATA ports) in the store: an Asus > P4P800. > > My only problem is with the disks. I can use all USB ports (8; what a > server could do with eight USB ports?), the 3C940 Gigabit Ethernet port > (I disabled the sound subsystem), and everything works until the first > fsck, when the kernel paniced. Here is the dmesg: > > GEOM: create disk ad0 dp=0xc6b4bb70 > ad0: 4028MB [8184/16/63] at ata0-master UDMA33 > acd0: CDROM at ata0-slave PIO4 > GEOM: create disk ad4 dp=0xc6b4b070 > ad4: 117246MB [238216/16/63] at ata2-master UDMA133 > GEOM: create disk ad6 dp=0xc6b4b170 > ad6: 117246MB [238216/16/63] at ata3-master UDMA133 > Mounting root from ufs:/dev/vinum/root > panic: ata_dmasetup: transfer active on this device! > > I did further investigation: I booted from ata0-master, and mounted > /dev/vinum/root as /mnt. A simple fsck -f -B /mnt killed the system. > I did the same with /dev/ad4s1a (this is the boot hack partition from > the handbook), then I switched off softupdates. No win. I tried to > boot with safe mode either, but it hung with page fault. > > What can I do? Besides, why my SATA interfaces are recognized as > UDMA133? First loose vinum from the mix, if we should have a chance finding the problem we need to lower the number of unknowns. Then tell us if and how it fails, dmesg from verbose booted system, backtrace etc, you know the drill if running -current. That your disks are said to be ATA133 is of no concern if they are SATA connected, no matter what DMA speed the disk is set to the interface will run full SATA speed. This is all just for backwards compat in case you wonder and yes I'd have liked a bit in there stating that it ran SATA speed, but thats simply not there.. -Søren ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: sata + vinum + Asus p4p800 = :(
On Wednesday, Oct 15, 2003, at 01:24 Europe/Budapest, Greg 'groggy' Lehey wrote: On Tuesday, 14 October 2003 at 18:46:44 +0200, Balazs Nagy wrote: What can I do? Provide a dump? Analyse the problem yourself? This *is* -CURRENT, after all. Well... There is a tricky problem... I have two panics: %fsck -t ufs -B /mnt panic: ata_dmasetup: transfer active on this device! syncing disks, buffers remaining... 1015 1015 1015 1015 1015 1015 1015 1015 1015 1015 1015 1015 1015 1015 1015 1015 1015 1015 1015 1015 giving up on 949 buffers Uptime: 1m32s ad4: TIMEOUT - WRITE_DMA retrying (2 retries left) ata2: resetting devices .. panic: free locked buf Uptime: 1m32s of course, no dump created. Besides, why my SATA interfaces are recognized as UDMA133? It sounds like this could be an issue with ATA compatibility issues with this mother board. You should be able to mount your root file system from the underlying UFS partition, thus disabling Vinum; at least that would help you track down the problem. It has ICH5R south bridge, with correct PCI ids. Film at 11, or when I go home to generate some crash dump. -- jul ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-current To unsubscribe, send any mail to "[EMAIL PROTECTED]"
Re: sata + vinum + Asus p4p800 = :(
On Tuesday, 14 October 2003 at 18:46:44 +0200, Balazs Nagy wrote: > Hi, > > I had a -CURRENT setting with an Abit BE7-S and two SATA disks with > vinum configuration. It worked very well until a power failure, and > the mainboard died. Yesterday I got a replacement mainboard, the only > type met the requirements (eg. two SATA ports) in the store: an Asus > P4P800. > > My only problem is with the disks. I can use all USB ports (8; what a > server could do with eight USB ports?), the 3C940 Gigabit Ethernet port > (I disabled the sound subsystem), and everything works until the first > fsck, when the kernel paniced. Here is the dmesg: > > GEOM: create disk ad0 dp=0xc6b4bb70 > ad0: 4028MB [8184/16/63] at ata0-master UDMA33 > acd0: CDROM at ata0-slave PIO4 > GEOM: create disk ad4 dp=0xc6b4b070 > ad4: 117246MB [238216/16/63] at ata2-master UDMA133 > GEOM: create disk ad6 dp=0xc6b4b170 > ad6: 117246MB [238216/16/63] at ata3-master UDMA133 > Mounting root from ufs:/dev/vinum/root > panic: ata_dmasetup: transfer active on this device! > > I did further investigation: I booted from ata0-master, and mounted > /dev/vinum/root as /mnt. A simple fsck -f -B /mnt killed the system. > I did the same with /dev/ad4s1a (this is the boot hack partition from > the handbook), then I switched off softupdates. No win. I tried to > boot with safe mode either, but it hung with page fault. > > What can I do? Provide a dump? Analyse the problem yourself? This *is* -CURRENT, after all. > Besides, why my SATA interfaces are recognized as UDMA133? It sounds like this could be an issue with ATA compatibility issues with this mother board. You should be able to mount your root file system from the underlying UFS partition, thus disabling Vinum; at least that would help you track down the problem. Greg -- See complete headers for address and phone numbers. pgp0.pgp Description: PGP signature