Bug#653192: Intel X-25M SSD destroyed

[Jonathan Nieder <jrnieder@gmail.com>]

tags 653192 + unreproducible
severity 653192 important
quit

Hi Steven,

Steven Chamberlain wrote:

> Today I tried booting 3.2~rc4-1~experimental.1, which seemed fine until
> after a few minutes various programs in a graphical desktop environment
> started to hang, apparently when trying to load anything from disk that
> was not already in cache.
>
> Ctrl+Alt+F1-6 wasn't working.  After a SysRq-S,U,B forced reboot, one of
> my SSDs (sdb) now seems to be toast.  I'm only able to read the first
> 16KiB of data from it, which still contains the original intact
> partition table and embedded GRUB.  The rest is neither readable nor
> writable, with these errors:
>
>> [  749.427493] end_request: I/O error, dev sdb, sector 32
>> [  749.427501] __ratelimit: 2619 callbacks suppressed
>> [  749.427508] Buffer I/O error on device sdb, logical block 4
>> [  757.362967] sd 4:0:0:0: [sdb] Unhandled error code
>> [  757.362979] sd 4:0:0:0: [sdb] Result: hostbyte=DID_BAD_TARGET driverbyte=DRIVER_OK

No matter how poorly written the kernel is, this shouldn't be possible
unless a problem occured while flashing the disk's firmware or there
is a hardware or firmware bug.  Based on the above description, I
doubt the former was occuring.  Please contact the hard drive vendor
for information on diagnosis and recovery and get back to us.

> I don't imagine my bug report is of much help fixing the issue, but I at
> least wanted to warn anyone before trying it.

Thanks.  I believe v3.2-rc4 has been pretty well tested with common
SSDs by now but perhaps the realtime kernel hasn't.

Hope that helps,
Jonathan