Hi,
Did you follow this sequence:
1. power down machine
2. install hardware
3. reboot machine
4. install drivers/software for expander
5. *no* reboot
6. crash
Vikram
Will Murnane wrote:
Oh, sorry. I forgot to give details. Here's the machine in question:
w...@will-fs:~$ uname -a
SunOS will-fs 5.11 snv_122 i86pc i386 i86pc
w...@will-fs:~$ isainfo
amd64 i386
w...@will-fs:~$ prtdiag
System Configuration: Intel Corporation S3210SH
BIOS Configuration: Intel Corporation
S3200X38.86B.00.00.0046.011420090950 01/14/2009
BMC Configuration: IPMI 2.0 (KCS: Keyboard Controller Style)
==== Processor Sockets ====================================
Version Location Tag
-------------------------------- --------------------------
Intel(R) Core(TM)2 Quad CPU Q6600 @ 2.40GHz Intel(R) Genuine processor
==== Memory Device Sockets ================================
Type Status Set Device Locator Bank Locator
----------- ------ --- ------------------- ----------------
DDR2 in use 0 DIMM_A1 CHAN A DIMM 1
DDR2 empty 0 DIMM_A2 CHAN A DIMM 2
DDR2 empty 0 DIMM_B1 CHAN B DIMM 1
DDR2 empty 0 DIMM_B2 CHAN B DIMM 2
==== On-Board Devices =====================================
PCIe x1 VGA embedded in ServerEngine(tm) Pilot-II
Intel 82541PI Ethernet Device
Intel 82566DM Ethernet Device
==== Upgradeable Slots ====================================
ID Status Type Description
--- --------- ---------------- ----------------------------
1 available PCI PCI SLOT 1 PCI 32/33
2 available PCI PCI SLOT 2 PCI 32/33
3 available PCI Express PCIe x4 SLOT 3
5 in use PCI Express PCIe x8 SLOT 5
6 in use PCI Express PCIe x16 SLOT 6 / FL RISER
Slot 3, which says "available" is in fact where the SAS card is.
When the box first crashed, I took out one stick of memory (there's
normally two, for a total of 4GB). The box hasn't crashed in about
three hours now. Once zpool scrub finishes I can pull out the disks
and put the other stick back in and see what happens.
Will
On Tue, Sep 15, 2009 at 21:12, Vikram Hegde <[email protected]> wrote:
Hi,
What build are you running ? Several bugs in this area have been fixed in
recent builds.
Vikram
Will Murnane wrote:
I recently purchased an HP-branded SAS expander, model number
468406-B21. Today I installed it, and got a panic ("genunix: [ID
655072 kern.notice]" at beginning of lines removed):
panic[cpu1]/thread=ffffff0004883c60:
Freeing a free IOMMU page: paddr=0x14eb7000
ffffff0004883100 rootnex:iommu_page_free+cb ()
ffffff0004883120 rootnex:iommu_free_page+15 ()
ffffff0004883190 rootnex:iommu_setup_level_table+a4 ()
ffffff00048831d0 rootnex:iommu_setup_page_table+e1 ()
ffffff0004883250 rootnex:iommu_map_page_range+6a ()
ffffff00048832a0 rootnex:iommu_map_dvma+50 ()
ffffff0004883360 rootnex:intel_iommu_map_sgl+22f ()
ffffff0004883400 rootnex:rootnex_coredma_bindhdl+11e ()
ffffff0004883440 rootnex:rootnex_dma_bindhdl+36 ()
ffffff00048834e0 genunix:ddi_dma_buf_bind_handle+117 ()
ffffff0004883540 scsi:scsi_dma_buf_bind_attr+48 ()
ffffff00048835d0 scsi:scsi_init_cache_pkt+2e1 ()
ffffff0004883650 scsi:scsi_init_pkt+5c ()
ffffff0004883730 sd:sd_setup_rw_pkt+12a ()
ffffff00048837a0 sd:sd_initpkt_for_buf+ad ()
ffffff0004883810 sd:sd_start_cmds+197 ()
ffffff0004883860 sd:sd_core_iostart+184 ()
ffffff00048838d0 sd:sd_mapblockaddr_iostart+302 ()
ffffff0004883910 sd:sd_xbuf_strategy+50 ()
ffffff0004883960 sd:xbuf_iostart+1e5 ()
ffffff00048839a0 sd:ddi_xbuf_qstrategy+d3 ()
ffffff00048839d0 sd:sdstrategy+10b ()
ffffff0004883a00 genunix:bdev_strategy+75 ()
ffffff0004883a30 genunix:ldi_strategy+59 ()
ffffff0004883a70 zfs:vdev_disk_io_start+d0 ()
ffffff0004883ab0 zfs:zio_vdev_io_start+17d ()
ffffff0004883ae0 zfs:zio_execute+a0 ()
ffffff0004883b00 zfs:zio_nowait+42 ()
ffffff0004883b40 zfs:vdev_queue_io_done+9c ()
ffffff0004883b70 zfs:zio_vdev_io_done+62 ()
ffffff0004883ba0 zfs:zio_execute+a0 ()
ffffff0004883c40 genunix:taskq_thread+1b7 ()
ffffff0004883c50 unix:thread_start+8 ()
On the next boot, I saw this set of messages three times:
WARNING: bios issue: rmrr is not in reserved memory range
WARNING: rmrr overlap with physmem [0x24400000 - 0x7fcf0000] for
pci8086,34d0
WARNING: rmrr overlap with physmem [0x7fd96000 - 0x7fdfd000] for
pci8086,34d0
and this, several times:
NOTICE: IRQ21 is being shared by drivers with different interrupt levels.
This may result in reduced system performance.
I found a suggestion for showing the interrupt map:
# echo ::interrupts -d | mdb -k
IRQ Vect IPL Bus Trg Type CPU Share APIC/INT# Driver Name(s)
1 0x41 5 ISA Edg Fixed 3 1 0x0/0x1 i8042#1
3 0xb1 12 ISA Edg Fixed 3 1 0x0/0x3 asy#3
4 0xb0 12 ISA Edg Fixed 2 1 0x0/0x4 asy#2
6 0x44 5 ISA Edg Fixed 1 1 0x0/0x6 fdc#1
9 0x82 9 PCI Lvl Fixed 1 1 0x0/0x9 acpi_wrapper_isr
12 0x42 5 ISA Edg Fixed 0 1 0x0/0xc i8042#1
14 0x40 5 ISA Edg Fixed 2 1 0x0/0xe ata#2
16 0x84 9 PCI Lvl Fixed 3 1 0x0/0x10 nvidia#1
17 0x85 9 PCI Lvl Fixed 0 1 0x0/0x11 ehci#2
18 0x87 9 PCI Lvl Fixed 2 3 0x0/0x12 e1000g#3, uhci#10,
uhci#6
19 0x89 9 PCI Lvl Fixed 0 1 0x0/0x13 uhci#9
21 0x88 9 PCI Lvl Fixed 3 1 0x0/0x15 pci-ide#2
23 0x86 9 PCI Lvl Fixed 1 2 0x0/0x17 uhci#8, ehci#3
24 0x83 7 PCI Edg MSI 1 1 - pcieb#2
25 0x30 4 PCI Edg MSI 2 1 - pcieb#3
26 0x60 6 PCI Edg MSI 1 1 - e1000g#0
27 0x43 5 PCI Edg MSI 3 1 - mpt#0
128 0x81 8 Edg IPI all 1 - iommu_intr_handler
160 0xa0 0 Edg IPI all 0 - poke_cpu
208 0xd0 14 Edg IPI all 1 - kcpc_hw_overflow_intr
209 0xd1 14 Edg IPI all 1 - cbe_fire
210 0xd3 14 Edg IPI all 1 - cbe_fire
240 0xe0 15 Edg IPI all 1 - xc_serv
241 0xe1 15 Edg IPI all 1 - apic_error_intr
... which seems to indicate that IRQ21 is in fact not shared.
Any suggestions where I should start debugging this? From what I can
find, the rmrr issue is a should-never-happen case; should I raise a
bug with Intel? This didn't happen before adding the SAS expander, so
I'm reluctant to point fingers at the onboard ethernet controller
(whose PCI ID that is). Maybe swapping cards around will help; I'll
try that in a bit.
Will
_______________________________________________
indiana-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/indiana-discuss
_______________________________________________
indiana-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/indiana-discuss
