Re: [Bug-ddrescue] Linux drive access without BIOS detection

[Sam]

Seth,

> I thought recovery was not recoverable through software means 
> if the BIOS doesn't detect the drive, but it turns out not always
> true.

Correct - that is not _always_ true, although it usually is.  It all depends on 
the _exact_ behaviour of the faulty drive.

>  So he plugs the drive into sata port and boot up the computer
> and the drive starts clicking and I see that it not detected in the
> BIOS.

That means that *within the timeout period used by the BIOS for it to get a 
response to an Identify Device command*, the BIOS received no response. However 
the story with some drives which are failing, but not yet dead, is that when 
they are given some more time, they will eventually be able to finish loading 
their firmware from the platters and will then respond via the SATA interface.  
They sometimes just need more time...  For example, by the time the OS has 
started to boot and its drivers are then scanning for drives (which, for OS 
newer than DOS, is independent from whether the BIOS detected the drive or 
not), then the drive may now be responding on the SATA interface and so can be 
detected by the OS. Or the drive is still responding slowly even when Windows 
boots, but the Windows command timeout period is longer than that used by the 
BIOS i.e. the OS waits for longer than the BIOS was waiting for a response, and 
so Windows can detect the drive.

In some cases, the drive can remain faulty for longer and still be detected via 
what _appears_ to the OS, to be a "hot-plug" event when it eventually starts to 
respond via its SATA interface (which appears to be a hot-plug even though the 
drive was never unplugged).  The exact details and timing can depend on the 
drive's behaviour, the OS, its SATA driver, and the specific SATA controller.

>  After Windows 7 boots up, he goes to "Computer" and the drive is there.

Yes, that fits with what I've suggested - the drive probably just needed more 
time than the BIOS was allowing.

> They copy at 3 MB/s which I guess means it is doing it over PIO mode 0

More likely this slow I/O is the result of the internal problem which the drive 
has in the first place.  The drive is definitely failing, otherwise it would 
appear in the BIOS summary / disk setup screens.

> How does Windows access this drive without BIOS detection?

See above.  Of course to see the exact behaviour of that drive, I'd need to 
have it here, but I've seen enough similar results that the above is a likely 
reason, though not a 100% guarantee that this is always the situation.

> The disk won't show up as a /dev/sdX device in Linux.

The kernel boot messages may give you a clue what Linux is seeing. With failing 
drives, there is no guarantee that different OS and their drivers will behave 
in the way that you want. :-( I've also seen the reverse situation i.e. Linux 
recognising that a failing drive  is there, where Windows did not.

Good luck,

Sam