[Bug-ddrescue] New passthrough patch released

[Scott Dwyer]

I have released a new passthrough patch to work with ddrescue-1.19 on my 
site for ddrutility:

http://sourceforge.net/projects/ddrutility/files/ddrescue%20patches/passthrough%20patch/

There are a few improvements over the previous release.

First, since it does only work and compile on Linux, it now works with 
the new configure --enable-linux option. Meaning that if you just do a 
normal configure on a patched ddrescue, the passthrough options will not 
be available, and it should still compile on a non-linux system. This 
was tested on a Mac OSX system, and a patched ddrescue compiled and 
operated normally.

Second, there are some improvements in the error handling of the 
passthrough commands. And it also now detects the maximum cluster size 
that can be used with the passthrough commands on the drive.

Third, and maybe the best new improvement, when using the 
--ata-passthrough option no sector is tried twice. This is possible 
because a failed ATA command will return the first bad sector of the 
attempted read. The patch will now mark that sector as bad, and then 
make one attempt to read the sectors up to the error. This can speed up 
a recovery, as normally the linux kernel returns the whole block as 
failed and then ddrescue has to read sector by sector to find any bad 
sectors. And by making the attempt to read the data up to the error 
while still on the copy pass can also increase the amount of good data 
recovered on the copy passes.

I have done some extensive testing of this patch, including byte for 
byte comparisons of recovered data on a couple different drives. While 
that is no guarantee that it is perfect under every possible condition, 
I am confident enough in it that I just used it to perform a recovery of 
a friends 750GB drive, along with use of the latest release of 
ddrutility (ddru_ntfsbitmap and ddru_ntfsfindbad). I can say that the 
recovery looks very successful!

Scott