Re: [smartmontools-support] System will busy AND STUCK for seconds during executing "smartctl.exe -a /dev/csmiX, X"

[Christian F. <Chr...@t-...>]

Davis wrote:
> Yes, there are slightly visible and audible delays (<1 second) between  those Input/Output lines.
> The delay happens too fast to observe which command cause the delays exactly.
> It could be SMART READ ATTRIBUTE VALUES or SMART READ ATTRIBUTE THRESHOLDS from my observation, but I'm not pretty sure about this.
> I've updated to 5.42 final (2011-10-20 r3458), but the problem is still there.
>
> REPORT-IOCTL: Device=csmi1,3 Command=IDENTIFY DEVICE
>   Input:   FR=...., SC=0x01, LL=...., LM=...., LH=...., DEV=...., CMD=0xec IN
>   Output: ERR=0x00, SC=0x00, LL=0x00, LM=0x4f, LH=0xc2, DEV=0x00, STS=0x50
> REPORT-IOCTL: Device=csmi1,3 Command=IDENTIFY DEVICE returned 0
>
> REPORT-IOCTL: Device=csmi1,3 Command=SMART READ ATTRIBUTE VALUES
>   Input:   FR=0xd0, SC=0x01, LL=...., LM=0x4f, LH=0xc2, DEV=...., CMD=0xb0 IN
> =================Could be here==================
>   Output: ERR=0x00, SC=0x00, LL=0x00, LM=0x4f, LH=0xc2, DEV=0x00, STS=0x50
> REPORT-IOCTL: Device=csmi1,3 Command=SMART READ ATTRIBUTE VALUES returned 0
>
> REPORT-IOCTL: Device=csmi1,3 Command=SMART READ ATTRIBUTE THRESHOLDS
>   Input:   FR=0xd1, SC=0x01, LL=0x01, LM=0x4f, LH=0xc2, DEV=...., CMD=0xb0 IN
> ===================Or here=====================
>   Output: ERR=0x00, SC=0x00, LL=0x00, LM=0x4f, LH=0xc2, DEV=0x00, STS=0x50
> REPORT-IOCTL: Device=csmi1,3 Command=SMART READ ATTRIBUTE THRESHOLDS returned 0
>
>

Longer SMART commands apparently result in a system stuck. This is 
likely an Intel driver issue, so we cannot fix this. You might want 
report this to Intel support.


>> Is this only one specific system where this happens? MB type?
> I did more test yesterday, the same problem could be reprocuded in at least 4 different PC with the same set of disk.
>
> Tested MB:
> Intel P55, H55, Q67, and some other Intel chipset MB with ICH9R south bridge

Vendor? Any common special properties? Specific non-standard system 
software, PCI(e) hardware, ...?


> Tested Intel RST driver version:
> 10.8.0.1003 (latest version from Intel offical website)

I could not reproduce the problem with this version, even with a disk 
which has long SMART command durations.


> 11.0.0.1032 (leaked version from some forums)

Not tested yet. Please note that this is not a "leaked" version.
The same files are also available from Intel here:
http://downloadcenter.intel.com/Detail_Desc.aspx?DwnldID=20868
A normal driver search does not show this download because this version 
is only advertised for Intel MBs.


> All of them above could reproduce the delay problem with the same disk.

Which is this "same" disk?


> Or could it be disk specific problem?

Possibly: Disk which have longer ATA SMART command durations may result 
in this problem under some (still unknown) circumstances.

Please test if SATA interface resets occur in conjunction with system 
stucks. Run "smartctl -H -l sataphy ..." several times and observe 
whether counter with ID 0x000a (COMRESET) or other counters change:

SATA Phy Event Counters (GP Log 0x11)
ID      Size Value  Description
0x000a  2        6  Device-to-host register FISes sent due to a COMRESET
0x0001  2        0  Command failed due to ICRC error
... [more counters may follow]


Thanks,
Christian