Second hard drive --> crash

Message

sheepdog · Post by **sheepdog** » Wed Sep 03, 2003 10:41 pm

After adding a second hard drive I find the system crashes when doing long disk-to-disk copies. I'm mystified so any help would be much appreciated.

The crashes occur with PartImage, tar, dar, and rdiff-backup. They occur even in single user mode with the root disk mounted read-only. Otherwise the disk can be mounted, read and short copying can be done without a problem.

While booting a message appears saying "No 80 connector cable attached." I have no idea what this means. The first hard drive is on channel 0 and at boot it is listed as ATA100. The new HD (Western Digital) is on IDE channel 1 and at boot it is listed as ATA33.

Thank you for reading.

Hardware details:

capabilities: smbios-2.2 dmi-2.2
*-core
description: Motherboard
product: AMD762-W83627
*-firmware
description: BIOS
vendor: Phoenix Technologies, LTD
version: 6.00 PG (08/26/2002)
size: 128KB
capacity: 192KB
capabilities: isa pci pnp apm upgrade shadowing escd cdboot bootselect socketedrom edd int13floppy360 int13floppy1200 int13floppy720 int13floppy2880 int5printscreen int9keyboard int14serial int17printer int10video acpi usb agp ls120boot zipboot
*-cpu:0
product: AMD Athlon(tm) MP 2200+
vendor: Advanced Micro Devices [AMD]
version: 6.8.0
slot: Socket 7
size: 1800MHz
clock: 133MHz
capabilities: fpu fpu_exception wp vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
*-cache:0
description: Synchronous Internal L1 write-back Cache
slot: Internal Cache
size: 128KB
capacity: 128KB
*-cache:1
description: Synchronous External L2 write-back Cache
slot: External Cache
size: 256KB
capacity: 256KB
*-cpu:1
product: AMD Athlon(tm) MP
vendor: Advanced Micro Devices [AMD]
version: 6.8.0
slot: Socket 7
size: 1800MHz
clock: 133MHz
capabilities: fpu fpu_exception wp vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow
*-cache:0
description: Synchronous Internal L1 write-back Cache
slot: Internal Cache
size: 128KB
capacity: 128KB
*-cache:1
description: Synchronous External L2 write-back Cache
slot: External Cache
size: 256KB
capacity: 256KB
*-memory
description: System Memory
slot: System board or motherboard
size: 2GB
capacity: 1536MB
*-bank:0
description: DIMM
slot: A0
size: 1GB
*-bank:1
description: DIMM
slot: A1
size: 1GB
*-bank:2
description: DIMM
slot: A2
*-pci
description: Host bridge
product: AMD-760 MP [IGD4-2P] System Controller
vendor: Advanced Micro Devices [AMD]
version: 11
clock: 66MHz
*-pci:0
description: PCI bridge (Normal decode)
product: AMD-760 MP [IGD4-2P] AGP Bridge
vendor: Advanced Micro Devices [AMD]
version: 00
clock: 66MHz
capabilities: normal_decode bus_master
*-display UNCLAIMED
description: VGA compatible controller (VGA)
product: NV18 [GeForce4 MX 440 AGP 8x]
vendor: nVidia Corporation
version: a2
size: 128MB
clock: 66MHz
capabilities: vga bus_master cap_list
configuration: irq=17
*-isa UNCLAIMED
description: ISA bridge
product: AMD-768 [Opus] ISA
vendor: Advanced Micro Devices [AMD]
version: 05
clock: 66MHz
capabilities: bus_master
*-ide
description: IDE interface
product: AMD-768 [Opus] IDE
vendor: Advanced Micro Devices [AMD]
version: 04
clock: 33MHz
capabilities: bus_master
configuration: driver=AMD
*-ide:0
description: Channel 0
logical name: ide0
clock: 33MHz
*-cdrom
product: SONY CD-RW CRX175E
logical name: /dev/hda
capacity: 1023GB
*-disk
description: ATA Disk
product: IC35L060AVVA07-0
logical name: /dev/hdb
version: VA3OA52A
serial: VNC302A3CE1HXA
size: 57GB
capacity: 57GB
capabilities: ata dma lba iordy smart security pm apm
configuration: apm=off mode=udma5 smart=on
*-ide:1
description: Channel 1
logical name: ide1
clock: 33MHz
*-disk
description: ATA Disk
product: WDC WD800JB-00CRA1
logical name: /dev/hdc
version: 17.07W17
serial: WD-WMA8E3114949
size: 74GB
capacity: 74GB
capabilities: ata dma lba iordy smart security pm
configuration: mode=udma2 smart=on
*-bridge UNCLAIMED
description: Bridge
product: AMD-768 [Opus] ACPI
vendor: Advanced Micro Devices [AMD]
version: 03
clock: 33MHz
*-pci:1
description: PCI bridge (Normal decode)
product: AMD-768 [Opus] PCI
vendor: Advanced Micro Devices [AMD]
version: 05
clock: 66MHz
capabilities: normal_decode bus_master
*-usb
description: USB Controller (OHCI)
product: AMD-768 [Opus] USB
vendor: Advanced Micro Devices [AMD]
version: 07
clock: 33MHz
capabilities: ohci bus_master
configuration: driver=usb-ohci irq=19
*-network
description: Ethernet controller
product: 3c905C-TX/TX-M [Tornado]
vendor: 3Com Corporation
version: 74
clock: 33MHz
capabilities: bus_master cap_list
configuration: driver=3c59x irq=16
*-multimedia
description: Multimedia audio controller
product: SB Audigy
vendor: Creative Labs
version: 03
clock: 33MHz
capabilities: bus_master cap_list
configuration: driver=EMU10K1_Audigy irq=19
*-input UNCLAIMED
description: Input device controller
product: SB Audigy MIDI/Game port
vendor: Creative Labs
version: 03
clock: 33MHz
capabilities: bus_master cap_list
*-firewire UNCLAIMED
description: FireWire (IEEE 1394) (OHCI)
product: SB Audigy FireWire Port
vendor: Creative Labs
version: 00
clock: 33MHz
capabilities: ohci bus_master cap_list
configuration: irq=16
*-scsi
description: SCSI host adapter emulation for IDE ATAPI devices
logical name: scsi0
capabilities: emulated
configuration: driver=ide-scsi
*-channel
description: Channel 0
*-cdrom
description: CD-R/CD-RW writer
product: CD-RW CRX175E
vendor: SONY
logical name: /dev/cdrom
version: 1.0j
capabilities: removable audio cd-r cd-rw

fido · Post by **fido** » Wed Sep 03, 2003 11:20 pm

Well, I would suggest spending a few dollars at your local computer shack and picking up an 80-pin ide cable. It sounds like that might be worth looking into.

Nermal · Post by **Nermal** » Wed Sep 03, 2003 11:35 pm

an 80 pin IDE cable is basically an IDE cable with twice as many wires

Only 40 are connected though, the other 40 are inbetween each data wire as spacers to prevent interference when operating at speeds higher than UDMA 33.

To be honest, I would get an 80 pin cable, as running another hdd at speeds higher than UDMA 33 on a 40 way cable is likely to cause corruption and nasty stuff to happen

prince · Post by **prince** » Thu Sep 04, 2003 6:43 pm

Funny, I noticed the same problem after adding a second harddrive (same model as the first) which is much more quiet and wanted to mirror my system using dd. Browsing forums and playing with some kernel parameters brought up

Code: Select all

[*] Local APIC support on uniprocessors

fscked it up for me. You may want to try disabeling that. As I use the same ide-driver (amd74xx) as you do, it may point out a problem using that driver and doing fast transfers over the bus while having that APIC-support enabled. A dirty dirty workaround that worked for me is running some numbercrunching like Seti@home in the background, so traffic doesn't become too fast, but I guess you have to believe in it

I run 2.4.20-rc6 gentoo-sources, but using 2.4.22 vanilla-sources results in the same unstable behaviour for me. Maybe anyone complaining about "random lockups" should try and disable APIC especially running those kernel-versions.

Hope that helps!

sheepdog · Post by **sheepdog** » Fri Sep 05, 2003 12:22 am

Thanks for all the input.

I changed the cable and now the HD comes up with ATA100. And I examined my kernel settings and noticed that I had the kernel set for a Pentium when I am using an Athlon MP. Changing this eliminated the ACPI parameter. Unfortunately, none of this has improved the situation. Same problems continue.

Any further ideas?

prince · Post by **prince** » Fri Sep 05, 2003 1:33 am

Hello there...

sheepdog wrote:Changing this eliminated the ACPI parameter.

Maybe that was just a typo of yours but ACPI does not equal APIC. Roughly speaking the first one is about your power supply, power button, fans, temperature, etc. while the second one is about controlling system resources. The option

Code: Select all

[*] Local APIC support on uniprocessors

can be found in the menu "Processor type and features" and I was suggesting to disable it. I hope I didn't get you wrong but that APIC != ACPI used to confuse me und I just hope I'm not the only one who got confused

fishhead · Post by **fishhead** » Fri Sep 05, 2003 2:46 am

I'm not sure about this 100%, but you might want to make sure your power supply can handle the extra load of the second hard drive. I believe an overtaxed power supply can lead to failures, but I'd look into software issues (like your kernel config) first.

sheepdog · Post by **sheepdog** » Fri Sep 05, 2003 2:47 am

APIC != ACPI

Yes, I had been confused. Thank you for pointing this out.

However, I have been over and over the kernel parameters in both menuconfig and xconfig and this parameter is not settable in "Processor types" or in "General Setup". It looks as though it should be settable in General Setup since if I edit the .config file I see

Code: Select all

#
# General setup
#
CONFIG_NET=y
CONFIG_X86_IO_APIC=y
CONFIG_X86_LOCAL_APIC=y
CONFIG_PCI=y
# CONFIG_PCI_GOBIOS is not set
# CONFIG_PCI_GODIRECT is not set
CONFIG_PCI_GOANY=y
CONFIG_PCI_BIOS=y
CONFIG_PCI_DIRECT=y
CONFIG_ISA=y
CONFIG_PCI_NAMES=y
# CONFIG_EISA is not set
# CONFIG_MCA is not set
CONFIG_HOTPLUG=y

But in menuconfig this option does not appear at all and is not settable.

I'm reluctant to simply edit the change in the .config file directly. Should I try this?

prince · Post by **prince** » Fri Sep 05, 2003 8:31 pm

sheepdog wrote:I'm reluctant to simply edit the change in the .config file directly. Should I try this?

Trying won't hurt, just backup your old kernel

If you disable LOCAL_APIC you also have to say no to IO_APIC. I don't know if there are any other dependencies but i guess there aren't. Please correct me if I'm wrong. So this is how it should look like:

Code: Select all

CONFIG_X86_IO_APIC=n
CONFIG_X86_LOCAL_APIC=n

sheepdog · Post by **sheepdog** » Sat Sep 06, 2003 10:43 pm

I tried editing the .config file as you indicated. There is, in fact, one other location where APIC is referenced but I left that "y". Then, of course, I did

Code: Select all

make oldconfig

, as is necessary when manually modifying the .config. After building and installing the new kernel I tested it and it failed just as did the kernel before it.

I went back and looked at the APIC values in my .config file and they were all "y"?! So I modified them again, confirmed that they were set to "n" and did

Code: Select all

make oldconfig

again. Now they were back to "y". So the make oldconfig forces these to "y", I guess because I have specified SMP, since I have a dual processor machine.

From my reading it would indicate that it isn't a good idea to turn off APIC in an SMP machine. At least performance is severely degraded. At worse multithreaded programs may not work.

Anyone know to whom I might report this problem as a bug?

Second hard drive --> crash

Second hard drive --> crash

so far...

Re: so far...

oops!

Re: oops!

Cannot turn off APIC