[smartmontools-support] smartctl -a results in ... failed command: SMART ... resetting link ...

[Andreas B. <bec...@cs...>]

The two Sandforce based SSDs SuperTalent FTM24CT25H and ADATA S599 256GB
(just posted smartctl -x output to smartmontools-database@, smartctl -a
output from one disk is attached here) cause a link reset on smartctl -a
(but not smartctl -x). I used the 5.40 release.

There is a longer pause during output, probably while waiting for the
link to reset:

...
Error SMART Error Log Read failed: Input/output error
Smartctl: SMART Error Log Read Failed
*** longer pause here ***
Error SMART Error Self-Test Log Read failed: Input/output error
Smartctl: SMART Self Test Log Read Failed
SMART Selective self-test log data structure revision number 1
...

in dmesg I find this:

[60194.792078] ata2.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x6
frozen
[60194.792118] ata2.00: failed command: SMART
[60194.792155] ata2.00: cmd b0/d5:01:06:4f:c2/00:00:00:00:00/00 tag 0
pio 512 in
[60194.792159]          res 40/00:00:00:4f:c2/00:00:00:00:00/00 Emask
0x4 (timeout)
[60194.792218] ata2.00: status: { DRDY }
[60194.792250] ata2: hard resetting link
[60195.112043] ata2: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
[60195.200597] ata2.00: configured for UDMA/133
[60195.200658] ata2: EH complete

The bad thing about these resets is that at some point the kernel
downgrades the link to SATA 1.5 Gb/s.

I currently can't test the SSDs in a different machine/kernel.

* This could be a smartctl bug, issuing a "wrong" command that is no
longer supported with the extended logs ... but by analyzing some
feature flags provided by the disk this would have been obvious.
* This could be a firmware bug ... so lets hope there will be an update
soon.
* This could be a kernel problem, I'm using 2.6.36 currently.


Andreas