Skip to contentSkip to conversation

Undecoded Interrupt with SiL3112 IDE?

DISCUSSION Johannes Deisenhofer 2 participants 3 messages Apr 2, 2004 – Apr 4, 2004

lore
Share & tools
lore
#0 lore raw

Hi,

I have a recent problem with my sil3112 onboard SATA adapter.

Starting after a while (sometimes after hours of uptime), i get this every few seconds:

----- snip -----
irq 18: nobody cared!

Call Trace:
 [<c0108f03>] __report_bad_irq+0x33/0x90
 [<c0108fe0>] note_interrupt+0x50/0x80
 [<c01091e9>] do_IRQ+0xa9/0x130
 [<c0105030>] default_idle+0x0/0x30
 [<c010795c>] common_interrupt+0x18/0x20
 [<c0105030>] default_idle+0x0/0x30
 [<c0105053>] default_idle+0x23/0x30
 [<c01050de>] cpu_idle+0x2e/0x40
 [<c0103055>] _stext+0x55/0x60
 [<c04e66f5>] start_kernel+0x155/0x160

handlers:
[<c02ae9f0>] (ide_intr+0x0/0x180)
[<c02ae9f0>] (ide_intr+0x0/0x180)
Disabling IRQ #18
----- snip -----

This started recently (without changes in hardware or software) and drags down my machine quite a bit. No hangs / data losses, however.

- There is always an interrupt storm (about 100000 IRQ in ca. 1 sec) on IRQ 18 when this message is logged.
- kernel 2.6.5-rc2
- siimage driver (not libata)
- two SATA drives on adapter, ST3120026AS and WDC WD1200JD-00FYB0
- Asus A7N8X board (nforce2 chipset), latest bios
- There is only the onboard SATA adapter on this IRQ. I've pulled the PCI card physically sharing the same IRQ line.
- Same problem with kernel 2.4, although it handles it less gracefully (system
freezes for some time).
- Disabling ACPI doesn't change a thing (IRQ #11 will be disabled, then)
- System has otherwise been stable.
- After reboot, problem will disappear for a while

I suspect some unhandled error condition of the sil chip. After disconnecting and reseating both SATA connectors, problems disappeared for two days. Coincidence?

Anything I can test before I go and buy new cables?

From lspci -v -xxxx

01:0b.0 RAID bus controller: CMD Technology Inc Silicon Image SiI 3112 SATARaid Controller (rev 01)

        Subsystem: CMD Technology Inc: Unknown device 6112
        Flags: bus master, 66Mhz, medium devsel, latency 32, IRQ 18
        I/O ports at 9800 [size=8]
        I/O ports at 9c00 [size=4]
        I/O ports at a000 [size=8]
        I/O ports at a400 [size=4]
        I/O ports at a800 [size=16]
        Memory at de005000 (32-bit, non-prefetchable) [size=512]
        Expansion ROM at <unassigned> [disabled] [size=512K]
        Capabilities: [60] Power Management version 2

00: 95 10 12 31 07 00 b0 02 01 00 04 01 01 20 00 00 10: 01 98 00 00 01 9c 00 00 01 a0 00 00 01 a4 00 00 20: 01 a8 00 00 00 50 00 de 00 00 00 00 95 10 12 61 30: 00 00 00 00 60 00 00 00 00 00 00 00 0b 01 00 00 40: 02 00 00 00 00 82 08 ba 00 00 00 00 00 00 00 00 50: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 60: 01 00 22 06 00 40 00 64 00 00 00 00 00 00 00 00 70: 00 00 20 00 00 e0 d5 37 00 00 20 00 00 c0 d5 37 80: 03 00 00 00 03 00 00 00 00 00 10 00 da a9 50 7e 90: 00 fc 01 01 0f ff 00 00 00 00 00 18 00 00 00 00
a0: 01 60 8a 32 8a 32 dd 62 c1 10 92 43 01 40 09 40
b0: 01 60 8a 32 8a 32 dd 62 c1 10 92 43 02 40 09 40
c0: 84 01 00 00 13 01 00 00 00 00 00 00 00 00 00 00
d0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
e0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
f0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

P.S.: I'm not on the linux-kernel list, but I read the archives

Jo

#1 lore raw

siimage driver has buggy interrupt handling - have seen similar behaviour. It appears to be unmaintained. Recommend using libata instead.

Justin

Quoted 92 lines · Johannes Deisenhofer

Hi,

I have a recent problem with my sil3112 onboard SATA adapter.

Starting after a while (sometimes after hours of uptime), i get this every few seconds:

----- snip -----
irq 18: nobody cared!

Call Trace:
  [<c0108f03>] __report_bad_irq+0x33/0x90
  [<c0108fe0>] note_interrupt+0x50/0x80
  [<c01091e9>] do_IRQ+0xa9/0x130
  [<c0105030>] default_idle+0x0/0x30
  [<c010795c>] common_interrupt+0x18/0x20
  [<c0105030>] default_idle+0x0/0x30
  [<c0105053>] default_idle+0x23/0x30
  [<c01050de>] cpu_idle+0x2e/0x40
  [<c0103055>] _stext+0x55/0x60
  [<c04e66f5>] start_kernel+0x155/0x160

handlers:
[<c02ae9f0>] (ide_intr+0x0/0x180)
[<c02ae9f0>] (ide_intr+0x0/0x180)
Disabling IRQ #18
----- snip -----

This started recently (without changes in hardware or software) and drags down my machine quite a bit. No hangs / data losses, however.

- There is always an interrupt storm (about 100000 IRQ in ca. 1 sec) on IRQ 18 when this message is logged.
- kernel 2.6.5-rc2
- siimage driver (not libata)
- two SATA drives on adapter, ST3120026AS and WDC WD1200JD-00FYB0
- Asus A7N8X board (nforce2 chipset), latest bios
- There is only the onboard SATA adapter on this IRQ. I've pulled the PCI card physically sharing the same IRQ line.
- Same problem with kernel 2.4, although it handles it less gracefully (system freezes for some time).
- Disabling ACPI doesn't change a thing (IRQ #11 will be disabled, then)
- System has otherwise been stable.
- After reboot, problem will disappear for a while

I suspect some unhandled error condition of the sil chip. After disconnecting and reseating both SATA connectors, problems disappeared for two days. Coincidence?

Anything I can test before I go and buy new cables?

From lspci -v -xxxx

01:0b.0 RAID bus controller: CMD Technology Inc Silicon Image SiI 3112 SATARaid Controller (rev 01)

         Subsystem: CMD Technology Inc: Unknown device 6112
         Flags: bus master, 66Mhz, medium devsel, latency 32, IRQ 18
         I/O ports at 9800 [size=8]
         I/O ports at 9c00 [size=4]
         I/O ports at a000 [size=8]
         I/O ports at a400 [size=4]
         I/O ports at a800 [size=16]
         Memory at de005000 (32-bit, non-prefetchable) [size=512]
         Expansion ROM at <unassigned> [disabled] [size=512K]
         Capabilities: [60] Power Management version 2

00: 95 10 12 31 07 00 b0 02 01 00 04 01 01 20 00 00 10: 01 98 00 00 01 9c 00 00 01 a0 00 00 01 a4 00 00 20: 01 a8 00 00 00 50 00 de 00 00 00 00 95 10 12 61 30: 00 00 00 00 60 00 00 00 00 00 00 00 0b 01 00 00 40: 02 00 00 00 00 82 08 ba 00 00 00 00 00 00 00 00 50: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 60: 01 00 22 06 00 40 00 64 00 00 00 00 00 00 00 00 70: 00 00 20 00 00 e0 d5 37 00 00 20 00 00 c0 d5 37 80: 03 00 00 00 03 00 00 00 00 00 10 00 da a9 50 7e 90: 00 fc 01 01 0f ff 00 00 00 00 00 18 00 00 00 00
a0: 01 60 8a 32 8a 32 dd 62 c1 10 92 43 01 40 09 40
b0: 01 60 8a 32 8a 32 dd 62 c1 10 92 43 02 40 09 40
c0: 84 01 00 00 13 01 00 00 00 00 00 00 00 00 00 00
d0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
e0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
f0: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00

P.S.: I'm not on the linux-kernel list, but I read the archives

Jo

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

#2 lore raw
Quoted 4 lines · Justin Cormack

siimage driver has buggy interrupt handling - have seen similar behaviour. It appears to be unmaintained. Recommend using libata instead.

I tried libata from 2.6.5-rc3. Previous versions are considered broken by the author, especially for error handling.

However, I had far worse problems:

Apr  3 10:18:10 urmel kernel:  <3>ata1: DMA timeout, stat 0x0
Apr  3 10:18:10 urmel kernel: ATA: abnormal status 0x58 on port 0xF8854087
Apr  3 10:18:10 urmel kernel: scsi0: ERROR on channel 0, id 0, lun 0, CDB: Read (10) 00 00 07 65 3f 00 00 c8 00
Apr  3 10:18:10 urmel kernel: Current sda: sense key Medium Error
Apr  3 10:18:10 urmel kernel: Additional sense: Unrecovered read error - auto reallocate failed
Apr  3 10:18:10 urmel kernel: end_request: I/O error, dev sda, sector 484671
Apr  3 10:18:10 urmel kernel: ATA: abnormal status 0x58 on port 0xF8854087
Apr  3 10:18:10 urmel last message repeated 2 times
Apr  3 10:19:47 urmel PAM_pwdb[3086]: (login) session opened for user root by (uid=0)

This probably was a hardware problem. A new SATA cable seems to have fixed it.
Can't explain the 'medium error'.
SMART status of the drive is ok. No bad sectors according to SMART, none reallocated.

Jo