Hi,
we got a machine with a Nforce2 chipset and a SiI3112 SATA controller.
Two Maxtor disks are configured in software raid, level 1.
The machine is quite stable as long as I keep the raid degraded. If try
to syncronize the array the machines dies within 20 minutes. No message
given.
This has lead med to believe there is a problem with the two drives
sharing interrupt line 11. I guess it is impossible to force ide2 and
ide3 to use different interrupts as both devices are on the same chip.
I am running Linux 2.4.22-ac3.
Has anyone else seen this?
>From dmesg:
SiI3112 Serial ATA: IDE controller at PCI slot 01:0b.0
SiI3112 Serial ATA: chipset revision 2
SiI3112 Serial ATA: not 100% native mode: will probe irqs later
ide2: MMIO-DMA , BIOS settings: hde:pio, hdf:pio
ide3: MMIO-DMA , BIOS settings: hdg:pio, hdh:pio
hda: ASUS CD-S400, ATAPI CD/DVD-ROM drive
hde: Maxtor 6Y080M0, ATA DISK drive
blk: queue c046a7c8, I/O limit 4095Mb (mask 0xffffffff)
hdg: Maxtor 6Y080M0, ATA DISK drive
blk: queue c046ac3c, I/O limit 4095Mb (mask 0xffffffff)
ide0 at 0x1f0-0x1f7,0x3f6 on irq 14
ide2 at 0xf885e080-0xf885e087,0xf885e08a on irq 11
ide3 at 0xf885e0c0-0xf885e0c7,0xf885e0ca on irq 11
hde: attached ide-disk driver.
hde: host protected area => 1
hde: 160086528 sectors (81964 MB) w/7936KiB Cache, CHS=158816/16/63
hdg: attached ide-disk driver.
hdg: host protected area => 1
hdg: 160086528 sectors (81964 MB) w/7936KiB Cache, CHS=158816/16/63
--
Per Andreas Buer
On Llu, 2003-09-22 at 18:06, Per Andreas Buer wrote:
> we got a machine with a Nforce2 chipset and a SiI3112 SATA controller.
> Two Maxtor disks are configured in software raid, level 1.
>
> The machine is quite stable as long as I keep the raid degraded. If try
> to syncronize the array the machines dies within 20 minutes. No message
> given.
Just about every bug report I have about SI3112 now is on Nforce
chipsets. At the moment however I don't know what the magic connection
is.
On Mon, 2003-09-22 at 21:42, Allen Martin wrote:
> There are two issues I know about:
>
> 1) Earlier versions of the Asus BIOS would program incorrect timing in the
> nForce internal P2P bridge, causing failures with SI3112 under high disk
> activity. This is fixed in rev 1005 of their BIOS or later.
I am running the latest (1006, I think) ASUS (this is a ASUS A7
something Deluxe motherboard) BIOS revision.
> 2) PCI interrupts getting put into edge triggered mode when ACPI/APIC are
> enabled. Andrew de Quincey said this should be fixed in 2.4.22, but I
> haven't tested it myself (I have ACPI disabled on all my test systems). I
> did verify his patch on 2.6 when he first posted it, and it works.
I've tried disabeling APIC - it did not help.
I'll try 2.6.something tomorrow.
--
There are only 10 different kinds of people in the world,
those who understand binary, and those who don't.
> Just about every bug report I have about SI3112 now is on Nforce
> chipsets. At the moment however I don't know what the magic connection
> is.
>
There are two issues I know about:
1) Earlier versions of the Asus BIOS would program incorrect timing in the
nForce internal P2P bridge, causing failures with SI3112 under high disk
activity. This is fixed in rev 1005 of their BIOS or later.
2) PCI interrupts getting put into edge triggered mode when ACPI/APIC are
enabled. Andrew de Quincey said this should be fixed in 2.4.22, but I
haven't tested it myself (I have ACPI disabled on all my test systems). I
did verify his patch on 2.6 when he first posted it, and it works.
-Allen
On Mon, Sep 22, 2003 at 07:49:00PM +0100, Alan Cox wrote:
> Just about every bug report I have about SI3112 now is on Nforce
> chipsets. At the moment however I don't know what the magic connection
> is.
I don't know if this comes under the heading of "SiI3112 bugs", but
Adaptec AAR-1210SA still doesn't work for me (and I've seen one other
identical report on LKML in the last fortnight).
Any attempt to access the drive (Seagate Barracuda V) causes this
to happen:
Sep 22 20:56:00 src@vlad kernel: hda: dma_timer_expiry: dma status == 0x24
Sep 22 20:56:10 src@vlad kernel: hda: DMA interrupt recovery
Sep 22 20:56:10 src@vlad kernel: hda: lost interrupt
Data does eventually get to/from the drive, but so slowly as to be
unusable.
Hugo.
--
=== Hugo Mills: hugo@... carfax.org.uk | darksatanic.net | lug.org.uk ===
PGP key: 1C335860 from wwwkeys.eu.pgp.net or http://www.carfax.org.uk
--- In event of Last Trump, please form an orderly queue ---
and await judgement.
Dnia pon 22. wrze?nia 2003 21:49, Per Andreas Buer napisa?:
> On Mon, 2003-09-22 at 21:42, Allen Martin wrote:
> > There are two issues I know about:
> >
> > 1) Earlier versions of the Asus BIOS would program incorrect timing in
> > the nForce internal P2P bridge, causing failures with SI3112 under high
> > disk activity. This is fixed in rev 1005 of their BIOS or later.
>
> I am running the latest (1006, I think) ASUS (this is a ASUS A7
> something Deluxe motherboard) BIOS revision.
>
> > 2) PCI interrupts getting put into edge triggered mode when ACPI/APIC are
> > enabled. Andrew de Quincey said this should be fixed in 2.4.22, but I
> > haven't tested it myself (I have ACPI disabled on all my test systems).
> > I did verify his patch on 2.6 when he first posted it, and it works.
>
> I've tried disabeling APIC - it did not help.
>
> I'll try 2.6.something tomorrow.
Try to disable both ACPI and APIC, it helped in my case.
--
Witold Kr?cicki (adasi) adasi [at] culm.net
GPG key: 7AE20871
http://www.culm.net
> > I've tried disabeling APIC - it did not help.
> >
> > I'll try 2.6.something tomorrow.
> Try to disable both ACPI and APIC, it helped in my case.
You can look at /proc/interrupts and see if any PCI devices have interrupts
in edge triggered mode, if so your system will be unstable unless you
disable ACPI. If not, disabling ACPI / APIC probably won't help.
-Allen