Re: [Bug-ddrescue] Tip: Improving recovery chances with Seagate "F3" models

[Scott Dwyer]

On 5/13/2014 12:14 AM, Franc Zabkar wrote:
> The fix could be incorporated into ddrescue, or it could be 
> implemented as a small helper application. However, it requires a 
> little knowledge of WD's Vendor Specific Commands, or access to 
> certain commercial tools.
>
> If someone would be willing to write this tool, I would be only too 
> willing to help, but only if the code were open source and freely 
> available to all. The procedure is extremely simple and the code would 
> be very small. In fact I could perform the required task with a small 
> MHDD batch routine.

If I can ever find the time again to get back to my ddrutility 
programming for more than just bug fixes and get caught up, I may likely 
take up things like this. I have some really good things I was working 
on, but have not been able to pursue them lately.

But I think I need to work on getting a "donate" button on my site (will 
be working on this very soon), because when I am finally ready to 
release the good stuff, I think at least one item might become a bit 
popular. Except for Mac users... Mac users will be SOL as the good stuff 
only works in Linux, sorry.

Maybe I am really trying to encourage myself with this to try to make 
some time, but here are a few teasers:

Did you know that some newer drives have the ability to change the error 
timer? This is the amount of time a drive will spend when a read error 
is encountered, which can easily take 2-4 seconds per error. On some 
drives it may be able to be changed to as low as 1/10th second which can 
turbo charge a recovery first pass, although it does increase the number 
of errors when used which can somewhat offset the benefit.

Ever heard of the elusive, controversial, and now twice obsolete read 
long command? Ever found any freeware that can actually perform it 
successfully? I haven't. How would one know if it was useful or just 
overrated if they did not have the ability to try it for themselves?

Did you know the Linux kernel does excessive retries when read errors 
are encountered? As ddrescue is universal it uses normal reads so it 
suffers from this, although there are usually a few less retries when 
using the direct option. It is possible to bypass this behavior for much 
increased error processing performance using special pass-through 
commands. I am sure no one would be interested in a patch for ddrescue 
that would allow it to process errors 5 times faster, would they?

I have said too much...

P.S.
I will not get into any conversations here about any of this, except 
maybe the ddrescue patch as it is directly related, but only to a point. 
This is Antonio's place for ddrescue and I am just responding to the 
post. If you wish to discuss anything else I have mentioned above, 
please post in the ddrutility discussion on sourceforge or email me 
directly.

Scott