Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > comp.os.linux.hardware > #2935 > unrolled thread

IRQ storm from non existing device ?

Back to article view | Back to comp.os.linux.hardware

Contents

  IRQ storm from non existing device ? marc.lefranc@univ-lille1.fr - 2015-12-18 03:36 -0800
    Re: IRQ storm from non existing device ? J G Miller <miller@yoyo.ORG> - 2015-12-18 16:42 +0000
      Re: IRQ storm from non existing device ? marc.lefranc@univ-lille1.fr - 2015-12-18 09:18 -0800
        Re: IRQ storm from non existing device ? J G Miller <miller@yoyo.ORG> - 2015-12-18 21:46 +0000
          Re: IRQ storm from non existing device ? marc.lefranc@univ-lille1.fr - 2015-12-21 02:57 -0800
            Re: IRQ storm from non existing device ? J G Miller <miller@yoyo.ORG> - 2015-12-21 15:50 +0000

#2935 — IRQ storm from non existing device ?

Hi all,

on my new Dell system configured with Debian 8, several instances ksoftirqd keep consuming CPU time (more than 2 minutes together 30 minutes after booting) and I get all these error messages from the kernel:

output of dmesg|grep ata3:

[    0.829175] ata3: SATA max UDMA/133 abar m2048@0xfb13d000 port 0xfb13d100 irq 67
[    3.048081] ata3: COMRESET failed (errno=-32)
[    3.048163] ata3: reset failed (errno=-32), retrying in 8 secs
[   10.828076] ata3: limiting SATA link speed to 1.5 Gbps
[   13.092062] ata3: COMRESET failed (errno=-32)
[   13.092132] ata3: reset failed (errno=-32), retrying in 8 secs
[   23.044102] ata3: SATA link down (SStatus 1 SControl 310)
[   23.053782] ata3: exception Emask 0x10 SAct 0x0 SErr 0x4000000 action 0xe frozen
[   23.053858] ata3: irq_stat 0x00000040, connection status changed
[   23.053927] ata3: SError: { DevExch }
[   23.053993] ata3: hard resetting link
[   25.380040] ata3: SATA link down (SStatus 1 SControl 300)
[   25.380055] ata3: EH complete
... ad libitum...

Now it turns out that there is no device on ata 3. I have a CDROM on ata4, a SSD on ata7 and a hard drive on ata8 ??? You can see that there is no device reporting on ata 3 at boot, in contrast with, say, ata4 :

[    0.829178] ata4: SATA max UDMA/133 abar m2048@0xfb13d000 port 0xfb13d180 irq 67
[    1.151003] ata4: SATA link up 1.5 Gbps (SStatus 113 SControl 300)
[    1.156553] ata4.00: ATAPI: PLDS DVD+/-RW DH-16AES, DD11, max UDMA/133
[    1.157578] ata4.00: configured for UDMA/133

So what can be causing the problem? A faulty controller? Motherboard problem? How can disable it in the mean time since it seems to impact the CPU performance of the system?

Thanks in advance,
Marc

[toc] | [next] | [standalone]

#2936

On Friday, December 18th, 2015, at 03:36:26h -0800, Marc Lefranc observed:

> [    0.829175] ata3: SATA max UDMA/133 abar m2048@0xfb13d000 port 0xfb13d100 irq 67
> [    3.048081] ata3: COMRESET failed (errno=-32)
> [    3.048163] ata3: reset failed (errno=-32), retrying in 8 secs
> [   10.828076] ata3: limiting SATA link speed to 1.5 Gbps
> [   13.092062] ata3: COMRESET failed (errno=-32)
> [   13.092132] ata3: reset failed (errno=-32), retrying in 8 secs
> [   23.044102] ata3: SATA link down (SStatus 1 SControl 310)

These messages typicall indicate that whatever you have connected to ata3,
a SATA hard disk, is starting to fail and that you need to ensure you have
a backup of your files on it, and be actively seeking out a replacement.

Or the SATA cable is faulty and/or needs reseating, but this is much, much
less probable than the disk its-self starting to fail.

[toc] | [prev] | [next] | [standalone]

#2937

Le vendredi 18 décembre 2015 17:44:41 UTC+1, J G Miller a écrit :
> On Friday, December 18th, 2015, at 03:36:26h -0800, Marc Lefranc observed:
> 
> > [    0.829175] ata3: SATA max UDMA/133 abar m2048@0xfb13d000 port 0xfb13d100 irq 67
> > [    3.048081] ata3: COMRESET failed (errno=-32)
> > [    3.048163] ata3: reset failed (errno=-32), retrying in 8 secs
> > [   10.828076] ata3: limiting SATA link speed to 1.5 Gbps
> > [   13.092062] ata3: COMRESET failed (errno=-32)
> > [   13.092132] ata3: reset failed (errno=-32), retrying in 8 secs
> > [   23.044102] ata3: SATA link down (SStatus 1 SControl 310)
> 
> These messages typicall indicate that whatever you have connected to ata3,
> a SATA hard disk, is starting to fail and that you need to ensure you have
> a backup of your files on it, and be actively seeking out a replacement.
> 
> Or the SATA cable is faulty and/or needs reseating, but this is much, much
> less probable than the disk its-self starting to fail.

Thanks for the answer. This is indeed what I would diagnose if there was something connected to ata3. However, as I wrote in my original message, there is nothing connected to ata3 ???

[toc] | [prev] | [next] | [standalone]

#2938

Le venredi, 18 décembre 2015, a 09:18:37h -0800, Marc Lefranc a écrit:

> However, as I wrote in my original message, there is nothing
> connected to ata3 ???

Oops sorry I missed that.

Looks like it is either a controller problem or possible a kernel bug.

Which motherboard brand and model and year of production are you using?

Which kernel version are you using?

If you are using a significantly old kernel, there is a possibility of there
being some bugs in libata causing this to happen.

> How can disable it in the mean time?

Imeediately after power up, go into the BIOS and you may be able to
disable this ATA port.

[toc] | [prev] | [next] | [standalone]

#2939

> Looks like it is either a controller problem or possible a kernel bug.
> 
> Which motherboard brand and model and year of production are you using?

This is a DELL Precision T5810 system, manufactured in july 2015. I am not sure about the exact motherboard model but can try to look inside if needed.
 
> Which kernel version are you using?

This is the kernel from Debian 8.2

Linux platon 3.16.0-4-amd64 #1 SMP Debian 3.16.7-ckt11-1+deb8u6 (2015-11-09) x86_64 GNU/Linux

> > How can disable it in the mean time?
> 
> Imeediately after power up, go into the BIOS and you may be able to
> disable this ATA port.

Indeed, disabling in the BIOS the drives that are not hooked to an actual drives make the messages disappear. ksoftirqd is still busy although less, but this may be the network card.

[toc] | [prev] | [next] | [standalone]

#2940

Le lundi, 21 décembre, 2015, á 02:57:31h -0800, Marc Lefranc a ecrit:

> This is a DELL Precision T5810 system, manufactured in july 2015.

With something as new as that, you should not be having a problem of
hardware failing.

> This is the kernel from Debian 8.2
> 
> Linux platon 3.16.0-4-amd64 #1 SMP Debian 3.16.7-ckt11-1+deb8u6
> (2015-11-09) x86_64 GNU/Linux

So a fairly new kernel then.

All I can suggest is you do some web searching for this issue related
to this version of the kernel and perhaps also seeking help from the
kernel mailing list if you cannot find anything relevant.

> Indeed, disabling in the BIOS the drives that are not hooked to an
> actual drives make the messages disappear.

Some temporary respite at least!

One thing you could try if you have a spare SATA hard disk, is to
connect that up to the problem ATA controller and see if when something
is connected it shews up correctly identified in the BIOS (if not that
could point to faulty hardware) and what messages if any appear concerning
the connection to the disk which could perhaps be used for better 
diagnosing a kernel libata problem.

[toc] | [prev] | [standalone]

Back to top | Article view | comp.os.linux.hardware

csiph-web