Spinrite 6.0 - Division Overflow Error Last Updated : 01 December 2011
I have an external (USB2.0) SATA2 WD 1TB Green FAT32 drive being used for photo back-up (non-business). With very little warning the drive began to have RW erros, so I pulled from the external enclosure so that I could run chkcdsk /f & /r options, fotunately my mobo (Asrock 4Core Dual SATA2) supports SATA2 drives. Initially this basic recovery worked, but the drive kept getting RW errors - bad. (Note: only minor SMART flags, BIOS has always reported SMART as GOOD for this HDD) Also at one point I had to reset the HDD dirty flag (fsutil dirty query n:) as it was dramatically slowing the XP load. And at the third chkdsk attempt this executable decided enough was enough and refused to run on the drive anymore. Bad news, as although the BIOS reported the drive as fine DOS could not see it and XP considered it unformatted.
I also used the WD Diagnostics software, can't recall whether it was before or after running the chkdsk routines. Anyway whilst the drive could be loaded and would always report the "Quick" SMART results, the software would crash when any other "tests" were tried.
I have alwyas known about Gibson research and Spinrite, so here was the time to test it. Online ordering and boot CD set-up quite self-explanatory. Loaded FreeDOS and spinrite without problems, software could see all the drives (1x IDE and 2x SATA2) however after completing the first "read" run (~40% used data, 1x bad sector and rest empty space)on the suspect drive any further analysis stopped due to the "Division Overflow Error". Same error no matter whether level 2 or 4 of Spinrite analysis was tried.
Emailed errors back to GRC who immediately responded that it was possibly a problem with the SATA interface, and was it possible under the BIOS to run the drive under "Legacy IDE". Unfortunately my BIOS had no option for this, but I tested changing all other BIOS options but still had exactly the same error. At that point I was considering getting addition SATA 2 IDE hardware, or possibly a PCI interface card that could run SATA as IDE. However this would add another layer of hardware (and possible problems) so I wanted some advise from GRC before doing this.
GRC support said before trying any hardware changes try running software from MS-DOS boot disk rather than FreeDOS. Used XP SP3 to create a boot floppy and copied both Spinrite and a copy of SP5 onto the disk. Exactly the same error happened when I ran Spinrite, and also got the "Division Overflow Error" when I tried SP5 (could still try this version as HDD was formatted in FAT 32).
Obviously a bit stressed at this point, but I decided to try and isolate the error by running Spinrite on the other 40GB IDE and 40GB SATA2 drives connected in the system. Of course both drives ran the Level 1 diagnosis fine. I then tried the Level 1 diagnosis on the suspect drive expecting it to crash. However it didn't, and after completing 18% I cancelled this test, reset it to Level 2 and tried again.
After approx. 10 hours the data recovery routine completed and the data was readable again on the drive. I am currently copying data to networked back-up HDD - the transfer under XP is *very slow* but I'm not concerned as long as it works.
So why did the same Level 2 diagnostics now work on the drive?
The only difference from initial failures to working was that I ran the Level 1 diagnostics on the other drives and then the suspect drive. What I did notice was that when I re-ran the Level 2 test, it immediately jumped to 17% - which was *exactly* the point where I cancelled the Level 1 scan. Couldn't be just a coincidence could it? So anyway if you are having the "Division Overflow Error" try running LEvel 1 analysis first - maybe it will help. I have reported this to GRC, if they can shed any further light on this I will update.