Slashdot Mirror
Recovering Secret HD Space
An anonymous reader writes "Just browsing hardocp.com and noticed a link to this article.
'The Inquirer has posted a method of getting massive amounts of hard drive space from your current drive. Supposedly by following the steps outlined, they have gotten 150GB from an 80GB EIDE drive, 510GB from a 200GB SATA drive and so on.' Could this be true? I'm not about to try with my hard drive." Needless to say, this might be a time to avoid the bleeding edge. (See Jeff Garzik's warning in the letters page linked from the Register article.)
This does sound suspect, but it reminds me of the trick you used to be able to do with 720 floppy disks - you could drill a hole where the hole on a 1.4MB disk would be and use it as a 1.4MB disk. Trouble was, it wouldn't retain data for very long, but it usually lasted for a day at least before the data degraded.
The old Linux IDE guy spoke of something like this a while back. Apparently the drive vendors got sick of stocking every drive model for warranty replacement, and implemented a scheme where they could "flash" a generic drive with a specific model number and capacity. Therefore it's possible that your "120GB" drive is really qualified for 160GB but was set that way for inventory reasons.
This was on the linux-kernel list a while back, too lazy too find it. (And it's possible I misunderstood -- Hedrick is a crackpot who is barely able to articulate what he is thinking.)
But really, has anyone ran over the data with a bunch of unique files to see if it's not just sharing tables and writing over itself on the respected sides?
Well thats what they advertise...
There are lots of internal sectors that are reserved for errors. There are builtin algorithms on the disk to diagnose and correct physical errors. You just don't notice them because the disk remaps those sectors transparently.
Hooray! I learned something in class for once!
I've done something similer in the past with a 40GB drive. I managed to get 67GB out of it. Worked fine and all the space was usable. The only problem was bad sectors, after only 2 weeks I had 15% of the dirve unusable, and after a month I couldn't even accsess it. So while it dose work it will quickly devistate the life expectince of the drive.
:)
On a side note a freand of mine tried this with his 20GB drive at around the same time, cranked it up to 32GB... Funny thing is it still fully works. Amazing isn't. Just don't try it at home
... that makes me want an article moderation capabilities to slashdot. I mean, how great would've it been to avoid seeing this at all because it had gotten (Score: -1, bullshit).
I mean tricking an OS into seeing the partition table twice hardly counts for doubling the actual drive capacity. Geeez.
Mmmm.. already dreaming of (Score: +4, top news) and (Score: -1, dupe)
1 Earth is warming, 2 It's us, 3 it's royally bad, 4 we need to take action NOW
It seems to work by deliberately corrupting your partition tabes by overlapping your patitions:
Partition a from 0 to 200 GB
Partition b from 1 to 200 GB etc.
You could probably get it to say almost any amount, but it wouldn't be usable space.
Some drives may have a little extra space but not 70 GB on a 80GB drive. No sane company is going to sell a 150 GB drive as an 80 GB because they pay as much to manufacture platters and heads no matter how they're used. The cost of the unused parts would come right out of their profits. Also, sometimes there is "unused space" used for the hard drive's bios, or for relocating data from bad sectors.
"The price good men pay for indifference to public affairs is to be ruled by evil men." -Plato
Case modder - okay
CPU overclocker - okay
Grapic card overclocker - okay
HD modder - ???
Actually there are guys that mod their harddrives.
Notice the less than clean working area with metal particles from the dremeling everywhere. This is less than wise, as the probability that foreign material will get in the drive and act like sandpaper is high. I certainly wouldn't put a modded drive like this in a production machine.
I think modding is great, but this is where I draw the line.
Take the cheese to sickbay, the doctor should see it as soon as possible - B'Elanna Torres, "Learning Curve"
Heh, I remember how Amigas used to have a more powerful FDD controller than PC's, meaning they could squeeze more on a disk; the space-optimized filesystems there let you squeeze almost 1MB onto a single DD floppy vs the already impressive default of 880k; and yup, you got nearly 2M from a HD floppy! ;)
Anyone wanting to try such amazing technology today can use a Catweasel, although I'm not sure if it supports anything more exotic than standard Mac/Amiga floppies.
(I post this here because maybe you've been around long enough to remember when ARC vs. ZIP vs. LZH vs. some others was a big deal.)
.BOB.
.BOB extension, but hey, as they say, there's one born every minute.
Back in the days of the "archive format wars" somebody made a program called NaBob that was pretty funny. It made archives that were so perfectly compressed that they approached singularity. That is, every archive turned out to be one byte long.
The various compression methods, it was said, were named after different types of quarks. So, as the files were compressed, it would report, "upping," "downing", "charming," "stranging," etc.
The file extension was
When you ran the uncompress process, all your files would be mysteriously "extracted" from the archive again. Amazing! It really stored all that data in a single byte!
Of course, all it was really doing was setting the hidden file bit on all your files and creating a one-byte file with the
That program always cracked me up, so I just thought I'd share.
Breakfast served all day!
There were also programs that just "deleted" the file and strored the cluster numbers in the "compressed" file. Too bad if you happen to defrag or something else in the meantime.
the faq of comp.compression has a lot of really wired stuff...
HI O WISE PRINCE. WHT TOOK U SO DAM LONG?
The hard drives in many 60 hour ReplayTV's are actually 80 gigabytes (approx 1gig/hr), and can be reformatted as such. The formatting was reduced for marketing reasons.
I'll second that. Wish I had a mod point for you. I don't trust 3.5s any more at all. USB flash drives for quick mobile storage, CD-Rs for anything bigger or more long term. Even the CD-Rs don't last well anymore. Now all those 3.5s come with those stupid little plastic sliders instead of the sturdy old metal ones. Constantly I find those things coming off and getting jammed in the drives at work. And the plastic is so cheap and flimsy they are almost a real "floppy" disk again.
Of course, it doesn't help that now it's not just the computer geeks using these things and a bunch of stupid college kids are storing all of their term papers on these crappy things. Then they run around with them jammed in their back pocket or backpack until crushed, bent, or otherwise destroyed.
My job involves me helping people use the computer, but I'm about to put a sign up that help with college work will cost extra.
From my own page:
/dev/hdX
When I created my Linux filesystems with mke2fs, I didn't know there was an -m option. This option specifies how many percent of your disk Linux will "steal" so that root can use it to fix your system when the disk is full. This defaults to 5%, which for a disk used to store files is obviously 5% too many. So for all your non-system disks at least, simply correct the file system with tune2fs:
tune2fs -m 0
Et voila. The disk is 5% bigger as if by magic. For a 120GB drive this gives you an extra 6GB. Hey, you never know when you might need it. Also, if you do this on your system disk, don't say I didn't warn ya.
That's probably because the Amiga floppy controller wrote track-at-once, rather than secton-at-once but without either the controller or the trackdisk.device verifying that the entire track had been written correctly. Hence, if you updated a single sector on a track, the entire track would be re-written, and the "unmodified" tracks may get corrupted in the process.
There was a nice hack called TrackSalve which hacked the trackdisk.device so that it performed an automatic verify of tracks after writing. ISTR equivalent functionality may have been incorporated into trackdisk.device in 2.04/3.0+ Kickstarts, but before I started using TrackSalve, I used to frequently end up with corrupted diskette bitmaps (probably the most-rewritten track on an Amiga floppy).
Another, probably less significant factor is that the Amiga disk hardware wrote tracks with no gaps between sectors in order to get that extra 160KBytes. If a PC disk controller encountered an error in the inter-sector gaps, I doubt it would cause it many problems, but for Amigas, it increases the probability that an error will occur in an occupied cell of the disk.
--
"to some magic 1.22 meg format that mysteriously made my floppies faster"
No magic at all. I used the shit out of that program. It was called fdformat and even came with Pascal source code! scheweet There were two little parameters called Xnnn and Ynnn that did sector sliding.
From the fdformat docs... These options can be used to enhance the performance of your disk up to 100%. This is a bit difficult to explain. Imagine a standard 360 kB disk. It has 9 sectors on each track numbered 1 to 9. Normally the sectors on all tracks ordered "1 2 3 4 5 6 7 8 9". With sector sliding of 1 you order "1 2 3 4 5 6 7 8 9" on track 0, "9 1 2 3 4 5 6 7 8" on track 1, "8 9 1 2 3 4 5 6 7" of track 2 and so on. You can easily imagine, that it takes a little time, when your diskette drive head steps from one track to another. But your diskette continues rotating. Without sector sliding your diskette is positioned to sector 2 or 3 on the next track, when the stepping is done. It needs nearly a full revolution until sector 1 of the next track can be read. With sector sliding of 1 or 2 your diskette is positioned exactly on sector 1, when it starts reading again.
This little bit of magic was somewhat drive-specific, since some drives were faster than others, you needed to use different sliding numbers, but all in all, it's a very cool hack.
LOAD "SIG",8,1
LOADING...
READY.
RUN
Sorry, but this is complete bullshit.
Yes. I call it corrupting your partition table. ; )
Years ago, when an 800MB drive was "big", a friend of mine tried to convince myself and a group of IT staff friends, that he could get around BIOS limits of a particular DEC workstation, through some tricky settings of the geometry in the BIOS. LBA was not big in those days and MS OS were still using the BIOS for disk access beyond the boot process.
Anyway, my friend managed to "trick" the BIOS into seeing 800MB (previously 504MB).
So, in an attempt to prove him wrong, I then proceeded to format the drive. MS-DOS format claimed it was formatting the drive as 800MB, but this did not deter me. I knew that MS-DOS was simply fooled into thinking that 800MB was actually addressable on that particular (504MB through BIOS limited) machine.
The format completed fine! But I was still not detered. I said, "ok, now we start to fill this drive up...".
I started copying a large directory over and over to fill the drive. When we approached about 500MB... "Seek error: sector not found.". The drive no longer booted either.
What had happened, was that we managed to force the BIOS to accept geometry values which it could not fully address. Most Significant Bits which MS-DOS would send, would never get seen by the drive, since the BIOS could not go beyond a certain address width. So while formatting, MS-DOS would be sending write commands which would be honored by the drive, but the BIOS would be passively stripping some of the highest MSB's out of shere lack of support of them.
The end effect, was that at the 504MB point, the drive head would be about 504MB's in to the 800MB, then at 505MB, the address would go back to zero and the head would come back to the start! That first sector would be formatted again, the drive would report success, and MS-DOS format would think nothing of it. When it got to "800MB", it would have all appeared to format ok to MS-DOS.
The end result was an 800MB drive, with a partition table which that BIOS was never going to be able to fully service, even though MS-DOS format "saw the proof" that all was fine. ; ) When someone tried to copy data to the next "safe" sector beyond what the BIOS could address, what they were actually doing was writing back over the beginning of the disk! Corrupting the partition table.
; )
I was delighted, because everyone else was on my friends side, even though one of my buddies also had a background in electronics and should have known what I was talking about. Anyway, modern drives DO have secret areas set aside for remapping of bad sectors (to give you the consumer the perception of zero bad sectors and all the space you legally purchased), but this space is way smaller than what these jokers are claiming and it is normally not user accessible.
So, save yourself the hassle of wondering in a few months time, why your drive has "crashed". You might not remember the "magic" that you did to your drive.
War crimes, torture, lies, illegal spying... Would someone give Bush a blowjob, already, so he can be impeached?
As I understand it, and I'm certainly not a EE or chip designer, it's a matter of semiconductors can be driven "hot" which makes them work but shortens their life. Now Intel wants to run the voltage at 1.4 V to keep their heat dissipation at about 80 W. Imagine a P4 that has everything perfectly finished (if it were an engine it would be ported and polished). Now imagine one that all works but there are some thin gates, and doping wasn't quite as uniform. If you wanted it to run at 3 Ghz you might be able to if you crank the voltage up to 1.6 V and dissipate 110 W (note that these numbers are only guesses) Intel doesn't want to do that so they just mark it at 2.4 GHz.
The arbitrage, and reason for such excitement in overclocking, is that most of the time Intel's manufacturing is too good. It makes too many uniform pieces that qualify for 3 GHz. The company likes to sell a few processors at high prices at the cutting edge, most processors at a sweet spot (~$200), and the remander as budget processors. To meet the economic demand, they take over qualified processors and mark them (most of them are multiplier locked as well) at for lower speeds. Over clockers take the chance that they bought a "relabed" processor not a "binned" processor. The success of a large group of overclockers is an indication of how well the manufacturing process is at delivering things at good tolerances. If you recall the Barton launch over clocking was a much dicier prospect, or further back an old Cyrix chip, because the processors were more likely to not qualify at higher speeds.
Almost all manufactured goods are built to tolerances rather than exact specs. Go grab a precision instrument and check some. The tolerances allow for much lower cost, and are usually developed as a balance between cost reduction and usefulness. There is a whole branch of manufacturing statistics that has developed tools for deciding when a process is out of tolerance. (The stats aren't too tough the tools make it easy to check on the fly even if you have little or no stat's training).
Degaussing scares the bad magnetism out of the monitor and fills it with good karma.
Not always is their goal to make a profit, but rather market share...
The best example of this is the Celeron 300A debacle for Intel. Switch back to those days of yore for a moment...
Intel introduced the Celeron line to help blunt AMD's advance into the low end post-Pentium I market. One problem: The Celeron 233 and 266 with NO L2 cache suck so much ass nobody wanted them, but they couldn't just change over the production line to a new Celeron design at the drop of a hat. What to do, Andy? Easy. That production line in Malaysia that's pumping out the Deschutes 450 PIIs to the rescue! So Intel took a whack of those chips, gave them a lower L2 cache, dropped their "rated" bus speed to 66MHz and branded them Celeron 300As. Which is why pretty much every Malaysian Celeron 300A runs just fine at 450 MHz with the stock Intel cooler, no adjustment required.
Intel actually lost money doing it, but they didn't lose the low end market. But the damage the current batch of crap they call a Celeron is doing to their reputation down there seems to indicate they will lose it soon...
Do they actually phisically install a head on the side of the platter that is not being used for drives such as your 10GB and 30GB examples? It wouldn't seem completely unreasonable to build them all the same and just disable the extra head in the firmware.