Re: [Bug-ddrescue] New features patch now available

[Florian Sedivy]

I believe the reverse pass of the copying phase just jumps and skips backwards, 
but still reads full clusters. That means that the blocks inside the cluster 
will be read forward. So performance should be similar to going forward. 

Generally with hard drives writing in groups of several 100 MB, and lots of 
predictive logic in reading from storage, and heads being even regularly parked 
for maintenance in some drives, I don't believe, anything ddrescue could 
possibly do can guarantee strictly linear head movement. All it can do is try 
to avoid unnecessary full transits from end to start. 

If you need strictly linear head movement, you'll have to use drive-specific 
hardware or software instead of (or in addition to) ddrescue. 

Florian

On 06.07.2014 at 02:26 Franc Zabkar wrote:

> A reverse pass effectively disables the drive's read-ahead caching. Data 
> recovery professionals will tell you that this can be (or is?) a good thing.
> 
> -Franc
> 
> 
> At 04:52 AM 6/07/14, Scott wrote:
> 
>> I have released a features patch for ddrescue that I have been working on. 
>> It is available on my ddrutility page on sourceforge. Its purpose is to add 
>> a few new options for more control over certain things.
>> 
>> http://sourceforge.net/projects/ddrutility/files/ddrescue%20patches/
>> 
>> New options are:
>> --no-reverse-pass          do not switch direction for each pass (for those 
>> that want it, as it really does not help any and is mostly useless in my 
>> opinion)