Well, it's probably not the kernel, it's your HDD!
Yes, It tries to read the sectors over and over again.
And if there's ONE successful "read", it'll remap it
to a "save" place. But until this "happends" it'll try and try and fail... retry... retry... fail... (until the HDD has enough ;) ).
You can check if this is your problem with smartmontools (see "smartctl" for further details)

add -e 3 to ddrescue to reduce retry count

The problem I had in this situation: due to the number of read errors the IDE driver decided that UDMA mode is not good for me (which it should do if there is a cabling problem or hardware incompatibility) and switched into PIO mode. Yes, the ultra-slow ISA-bus-speed wordwise transfer mode. And I could not switch back. 2-3MB/s. 24h for 160GB.

Thanks for your help, folks.
As your replies came in, the job completed. After 90+ hours of running, I got 3.6Gb off the 10Gb disk.
The testdisk utility isn't being helpful, either. It won't let me set a partition table on it.

So I thought I'd try your advice and shut down and restarted the box.
That was a mistake. The disk is no longer detected by the bios or linux.
I let it all cool down and tried again. Nope.

So, at the end of the week, nothing.

Thanks again for your thoughts,

dad