Solve : HDD Disaster?

1.	Solve : HDD Disaster?
Answer» Samsung 500GB 7200rpm SATA 3Gb/s My HDD crashed. Some Windows XP Pro files are missing and need to repair. I tried to repair using Win XP SP2 CD Installer, but FAILED. This HDD is partitioned to three (C, D, and E). I tried to connect it to other PC but only D & E are accessible. So I decided to reformat... First reformat using Win XP SP2 CD Installer, FAILED. Second reformat using Win 7 Ultimate 32-bit CD Installer, FAILED. THIRD reformat using Win XP SP2 CD Installer again, SUCCESSFUL. Then the same day, I just noticed that my system runs slow and the sound coming from my HDD is kinda different, like a strange clicking sound. So I decided to run Disk Error Checking. Reboot my computer. After Disk Checking, my computer won't load to Windows anymore. Some Windows XP Pro files are missing and need to repair again. (I forgot to note the exact error, but I remember when I search it in Google, result is something related to hardware problem.) Reboot again my computer but cannot detect anymore my HDD. My BIOS cannot detect also my HDD. Tried to change SATA cables but still cannot detect. Tried to connect it to 3 different desktop computers, but still BIOS cannot detect it. I just bought a new SATA HDD (Western Digital Caviar Blue) and everything are fine now. Are there any possible solution to fix my Samsung HDD?Yes, recycle it. That will help cut down on the electronic waste going into public garbage fills. Otherwise, any HDD that has obvious major errors is beyond home repair. The cost of hard dives has come down so much. Good quality Hard Drives that have been puller from commercial use are available at a fraction of the original price. Or, if you don't need huge capacity, brand new drives are under $40 for up to 250GBThere are tools, such as SpinRite that might be able to fix your drive long enough for you to get the data you need, but yes most likely recycle that drive. Drives are relatively cheap these days, and most are only under warranty for 3-5 years. Did the drive by chance give you an errors before hand? Such as failing S.M.A.R.T tests?Quote from: geek hoodlum on December 12, 2011, 07:11:37 PM the sound coming from my HDD is kinda different, like a strange clicking sound. Classic sign of disk failure. Quote from: geek hoodlum on December 12, 2011, 07:11:37 PM Are there any possible solution to fix my Samsung HDD? You cannot "repair" a failed drive. That is why having proper backups of important data is critical. Quote There are tools, such as SpinRite that might be able to fix your drive long enough for you to get the data you need, but yes most likely recycle that drive. Personally I avoid any products affiliated with it's author like the plague. But I guess even a stopped clock appears to be working twice a day.Avoid SpinRite. Edit: T there is FREE software out there to help you. But it is not magic. If your drive has major failures, you can not really fix it. The kink below provides information and some free items. If a drive is usable, the programs they have will do a good job to r recovery and repair. If the drive still works. http://www.easeus.com/ You can not repair a major crash with software. To illustrate. Even tho a flat tire is flat just on the bottom, if it has a big hole in the sidewall, putting more air in it does not provide a temporary fix. Same idea with a hard drive. The drive must be 90% usable before you and fix it. Quote from: Geek-9pm on December 13, 2011, 01:47:00 AM The drive must be 90% usable before you and fix it. Wha??? How would you know if it's 90% usable? Like... write random data to the drive and 90 times out of a 100 the data verifies correctly...? And the verification must only occur in the presence of me....? http://www.easeus.com/datarecoverywizard/free-data-recovery-software.htm#1 It claims it marks bad sectors before attempting data recovery... but you still wouldn't be able to recover the data on the bad sectors.Quote from: Transfusion on December 13, 2011, 06:14:34 AM Wha??? How would you know if it's 90% usable? If 90% of the drive clusters are not marked as "bad", I guess. IMO any drive that has bad sectors needs to have any and all important data copied off of it IMMEDIATELY. I speak from experience (and I would hope, common sense). They simply are no longer reliable. If you are lucky, the cause of the "bad sectors" was in fact due to a flakey PSU causing issues during the read cycle that were interpreted as a problem reading some unlucky portion of the disk, and can be "re-checked" and marked good. On the other hand, you cannot, to my understanding, force the drive to recheck those sectors that it remapped to a spare sector through it's firmware, only those marked by software. Which is sort of the issue; by the time you are able to see bad sectors, from fsck or chkdsk, the drive has already run out of sectors to remap itself and the I/O error has started to bubble up to the file-system level. Of course the topic isn't with regard to a drive with bad sectors, but rather one that is no longer detectable by the system. If one is lucky in that situation, they can simply swap the logic board with one from a working drive of the same brand and model. However, the changed acoustic properties of the drive seem to indicate that it is, to use the technical term, screwed.Quote from: Transfusion on December 13, 2011, 06:14:34 AM Wha??? How would you know if it's 90% usable? Like... write random data to the drive and 90 times out of a 100 the data verifies correctly...? Please do not take that literally to mean 90 % of the sectors have to be good. I meant that the drive has to have most of its ability. It must be more the 90%, more like 99.7% But if I said that, somebody will say "How do you know it is exactly 99.7%" I thought that by saying 90% it was understand to mean having most of its important ability. The intent was to use terms a layman would grasp. Drive failure is more that just X numbers of bad sectors. Drive failure can be divided up into about ten categories. Now are you going to ask me "Is it exactly ten." The failure of HDD is not due to PSU problems Quote It claims it marks bad sectors before attempting data recovery... but you still wouldn't be able to recover the data on the bad sectors. Wrong! Mechanical, physical damage to the head is either fatal or soon will be. The electronics in the drive can go sour and it can fail to read data. There is no cure for ICs that are at the end of life. They might recover after cooling, but they do not self-heal. All the data recovery tools attempt to make the best of a bad condition. . Bad sectors are due to defects in manufacturing. The defects are not apparent during early treating. In time the defect becomes worse and the drive system flags the spot as an area that needs to be moved. And yes, the bad sector can be read reliably and its data moved to another area. Whether or not the sector becomes worse does not matter nonce it has been flagged. Maybe it will get worse. maybe not. but it will no longer be used. The exact algorithm for this and the parameters are proprietary to the hardware / software makers. Don't even try to second guess. If they wanted you to knew, they would tell you. So I can't tell you either. But it works better that you may THINK it does. I did not want this to be a lecture non modern drive technology. The point is that the technology is very sophisticated and usually takes care of itself. The S.M.A.R.T. works good about half the time. Which is much better than nothing. If the drive only has some bad sectors, and no other problem, it is possible to restore the drive back to good heath and reset the monitoring system. But this is not something amateurs ought to do, except as a hobby. Formatting a drive with then right tools may restore a drive to almost new condition. Windows Check Disk does a fair job. The dive manufacturers have tools you can also use. But you have to read the instructions, not guess. More specifically, if a chuck of the oxide comes off and is n flying around inside the chamber, the drive is going to start failing very fast.Any attempt to read the whole drive will just shorten its life and reduce the opportunity to recover vital data. It depends. A few bad sectors are not a problem. But if severe damage has been done to the head or the surface, mapping bad sectors is a waste of time. Quote from: BC_Programmer on December 13, 2011, 01:42:05 AM You cannot "repair" a failed drive. That is why having proper backups of important data is critical. Personally I avoid any products affiliated with it's author like the plague. But I guess even a stopped clock appears to be working twice a day. I personally don't use it either, but I was making a point if he needed data their are software utilities that might help him.Quote from: Geek-9pm on December 13, 2011, 08:13:36 AM Bad sectors are due to defects in manufacturing. The defects are not apparent during early treating. In time the defect becomes worse and the drive system flags the spot as an area that needs to be moved. And yes, the bad sector can be read reliably and its data moved to another area. Whether or not the sector becomes worse does not matter nonce it has been flagged. Maybe it will get worse. maybe not. but it will no longer be used. The exact algorithm for this and the parameters are proprietary to the hardware / software makers. All Hard drives ship with defective sectors. The drive itself remaps them to spare sectors, at which point the original sector location's condition is irrelevant. However, the original sector can still be perfectly fine. a aberrant power fluctuation from the PSU could easily cause the drive logic to make an assumption about where it is reading that is no longer valid; or undervoltage of the read head during a critical decoding stage could cause the drive to read a CRC error. The tolerance of the drive for what is and is not an acceptable number of retries varies, but when you have a PSU that simply hooks the 5v rail to PWR_GOOD, problems like that are inevitable. Also, the algorithms used for Hard Drive sector remapping are rather well documented. P-Sector remapping and G-List remapping. I doubt HD manufacturers would use anything different, given that their behaviour in these situations (bad sector remapping) is altogether consistent. Quote The S.M.A.R.T. works good about half the time. Self-Monitoring And Reporting Technology is utterly useless for it's intended purpose. The purpose being, of course, to warn when a disk is failing. But it doesn't work. The various numbers that it keeps track of could easily be skewed as a result of environmental factors; when S.M.A.R.T sets off a warning, it basically means some number approached some other number. you can't easily get at the numbers to find out what set it off, and it doesn't matter which value hits the tolerance, the drive is simply reporting by most System BIOS's as a S.M.A.R.T failure on the drive. More to the point, S.M.A.R.T, aside from having the number of firmware-remapped sectors as one of it's magic numbers, doesn't help prevent bad sectors, or disk failures; prime example being the Samsung drive of the Original Poster which by all appearances has failed. Quote If the drive only has some bad sectors, and no other problem, it is possible to restore the drive back to good heath and reset the monitoring system. But this is not something amateurs ought to do, except as a hobby. If you want to reset smart data, you either don't get what smart is about, or you have bad intentions- such as trying to reset the S.M.A.R.T data for values like "Hours Run" and "Spin-ups" in ORDER to sell the drive as "new". And resetting the values for remapped sectors only changes the value for the S.M.A.R.T reporting... it doesn't actually make changes to the P-List and G-List and remove the mappings. Either way, the problem with S.M.A.R.T is that the tolerances are set differently for all different drives, and oftentimes they are given utterly stupid values. Some Maxtor Drives report "impending Drive failure" (fail SMART) simply by being spun-up a set number of times. This despite there being no evidence to suggest that the number of times a drive has spun up is in any way related to the frequency with which a drive fails. Quote Formatting a drive with then right tools may restore a drive to almost new condition. Formatting a drive can never return a drive to "almost new" condition. Even if you get some special tool to reset the S.M.A.R.T values and reformat the drive- like say, flashing a newer firmware, which can sometimes reset accumulated S.M.A.R.T values as a side-effect- it will merely mean that S.M.A.R.T reporting for that drive is even more useless because the values it actually stores in no way reflect the drive. Remapped sectors will still exist, but unless the drive has to remap more after the values are reset, it will still report the value as 0; and either way the value will be off by whatever number of remapped sectors there were before. Quote Windows Check Disk does a fair job. Of doing what? Windows Chkdsk checks the File-System structure, and can perform a surface scan which attempts to locate bad sectors. If chkdsk is reporting bad sectors at all- the drive is basically screwed. chkdsk can report that there aren't any bad sectors when in fact there are because the drive will have remapped them. This is invisible to any software and the only software-visible way to figure if there are remapped sectors would be to investigate the S.M.A.R.T fields. Of course if you reset that data then even that won't work, and either way software cannot determine what sectors are remapped and which ones aren't, it's just a count. A chkdsk surface scan can make the drive remap bad sectors itself, though, when they are accessed. (which is a good thing, better to remap bad sectors earlier rather than later) Quote mapping bad sectors is a waste of time. Tell that the the manufacturers that have been incorporating P-List and G-List sector remapping within their firmware and ASIC's for almost 20 years.To all reading this. BC knows programming. He knows very little about hard drive technology, other than what he reads in comic books. Quote All Hard drives ship with defective sectors. false Quote Also, the algorithms used for Hard Drive sector remapping are rather well documented false Quote aberrant power fluctuation from the PSU could easily cause the drive logic false. It is extremely rare. Quote Of doing what? Windows Chkdsk checks the File-System structure, and can perform a surface scan which attempts to locate bad sectors. If chkdsk is reporting bad sectors at all- the drive is false. How do you think you know what it does? Geek should get a warning for the amount of rubbish he posts. His above effort is nonsense. Each of his points answering a remark of BC_P's is a complete howler. Quote from: Geek-9pm on December 13, 2011, 12:03:01 PM To all reading this. BC knows programming. He knows very little about hard drive technology, other than what he reads in comic books. Funny. Quote Quote All Hard drives ship with defective sectors. false This is common knowledge. It has always been true. With MFM/RLL drives, the defective sectors would actually be listed On the drive, since that was before Sector remapping was put into use (since it was Before IDE and the common usage of ASIC chips, making it so the circuitry for such functionality would take up too much real estate). Once ASIC chips became cheaper and IDE was conceived and established, they looked to the problem of factory defects and how they can help. The solution was to simply PACKAGE the drives with spare sectors, and have it remap them, in a way that is invisible to the Operating System. The number of pre-existing bad sectors per Megabyte has decreased by a LARGE margin thanks to new manufacturing techniques, but hard drive size has also skyrocketed so for the most part there will be at least a half dozen or so bad blocks that need to be remapped. Manufacturers will sometimes run the hard disks through QA (to remap the existing defects) and reset the S.M.A.R.T fields so that it shows none remapped. This isn't that dishonest though since they typically will decrement the tolerance value for the field as well. This is why sometimes you can get two of the exact same model hard drive that have different "Max allowed" values for their S.M.A.R.T fields. Essentially they allow the remapping and pretend that the drive came with fewer spare sectors. Quote Quote Also, the algorithms used for Hard Drive sector remapping are rather well documented false http://www.mjm.co.uk/sectorremapping.html Quote Quote aberrant power fluctuation from the PSU could easily cause the drive logic Quote false. It is extremely rare. So... err.. "That doesn't happen. It's rare" How is that not a contradiction of terms? And you are seriously suggesting that the hard drive, despite running on electrical power from the power supply both the 5v for the logic board and the 12 volts for the spindle motor, is somehow immune to abberations from bad power supplies or power supplies that continue to send Power_Good to the machine even when the power is not good? Not to mention that power fluctuations on the logic board can cause bit decay in the same manner that insufficient power to System memory will; and if that happens to occur while the logic board is reading or writing data, a CRC error can easily occur as it finds that the data it has in memory has changed. Anecdotally, I have two hard drives that in my older machine were working quite well- until I added a third Hard drive to the configuration. Drives would disappear at random intervals, and very shortly after changing the configuration they all started to fail the S.M.A.R.T diag the BIOS performed at boot-up, and I would have to chkdsk /r at least once a week to fix hard freezes. Once I replaced the hard drive, all these issues went away. I was able to chkdsk /b and re-evaluate bad clusters at the file-system level and Suddenly the system decided they were no longer bad. I wasn't able to "unmap" the P or G-Lists, obviously. Best I would have been able to achieve was resetting the SMART data. I imagine hard drive manufacturers have internal software they might have that can deal with those particulars. But I don't think any are available to the public. Quote Quote Windows Chkdsk checks the File-System structure, and can perform a surface scan which attempts to locate bad sectors How do you think you know what it does? I know it works on the file system level. chkdsk, like format, calls into fmifs.dll (Format Manager for Installable File Systems, presumably) It exports a number of functions, such as Chkdsk, ChkdskEx, Format and FormatEx; this can be confirmed quite simply with a quick dumpbin /exports. FMIFS itself, some might assume, is where you might think at this point that FMIFS is where the file system on-disk structure knowledge is located. That is not the case. FMIFS, as it's name suggests, is more or less a stub program for a set of installable File-Systems. For Most windows installations, this means FAT and NTFS. The logic for FMIFS to deal with these two filesystems is contained in ufat.dll and untfs.dll, located in the windows system32 directory. FMIFS itself, as can be viewed using dumpbin /disasm on it, mostly consists of epilog code, prolog code, and setting up and acquiring handles to DLLs and Function addresses (the installable file system drivers are enumerated from the registry). Appropriate calls to FMIFS.dll's various functions is essentially taking the string argument passed to it's function, taking U on the front and .DLL on the end, calling LoadLibrary() using the created string, and then attempting to use GetProcAddress() to acquire pointers to the function it will call, either the chkdsk or format functions, as needed,and there is error handling, of course. This is all easily confirmable by viewing the dissassembly, even if it is hardly an easy parse. the installable file system drivers- UFAT and UNTFS in this case- are the portions that actually "know" about the On-disk FAT and NTFS layout, so that the system can perform consistency checking or formatting operations. UNTFS and UFAT (and I believe UXFAT, or something similar for the ExFat System, presumably) are the DLLs that actually read/write from the disk during chkdsk and format operations. They in fact open the device directly and read/write clusters themselves; this, again, is quite easily confirmed via any number of Application tracing tools. Since UNTFS and UFAT deal only at the file-system level, and all reads/writes done during a chkdsk are done by those dlls, it stands to reason that the disk check is at the file-system level. your dismissal of the topic was so quick, you left it vacant of any actual information. Anyway, the obvious purpose of this heirarchy, which is well-documented in several SysInternals Articles, blog entries, and the like, is to make it easy for MS to add new file system types to the system without having to write custom versions of the file-system management tools. They simply write a DLL for the new file system type and make the management tools (specifically, FMIFS.DLL, and even explorer for it's format dialog) aware of the new format name. This can be seen in action with the installable Device Driver for ExFat on Windows XP systems, which make use of this capability. Add to this the fact that chkdskx, a sysinternals tool that exactly duplicates the functionality of chkdsk (quite literally, the two are indistinguishable) is Open Source, I think it's a fair bet to say, particularly based on your attempt at ad hominem that I "know programming" that I, and really anybody, can know how it works by perusing that source. By the way, the source can be found here. The actual source was ripped from SysInternals sites back when MS acquired them, but there are quite a few mirror sites.

Answer»

Samsung 500GB 7200rpm SATA 3Gb/s

My HDD crashed. Some Windows XP Pro files are missing and need to repair. I tried to repair using Win XP SP2 CD Installer, but FAILED. This HDD is partitioned to three (C, D, and E). I tried to connect it to other PC but only D & E are accessible. So I decided to reformat...
First reformat using Win XP SP2 CD Installer, FAILED.
Second reformat using Win 7 Ultimate 32-bit CD Installer, FAILED.
THIRD reformat using Win XP SP2 CD Installer again, SUCCESSFUL.

Then the same day, I just noticed that my system runs slow and the sound coming from my HDD is kinda different, like a strange clicking sound. So I decided to run Disk Error Checking. Reboot my computer. After Disk Checking, my computer won't load to Windows anymore. Some Windows XP Pro files are missing and need to repair again. (I forgot to note the exact error, but I remember when I search it in Google, result is something related to hardware problem.) Reboot again my computer but cannot detect anymore my HDD. My BIOS cannot detect also my HDD. Tried to change SATA cables but still cannot detect. Tried to connect it to 3 different desktop computers, but still BIOS cannot detect it.

I just bought a new SATA HDD (Western Digital Caviar Blue) and everything are fine now. Are there any possible solution to fix my Samsung HDD?Yes, recycle it.
That will help cut down on the electronic waste going into public garbage fills. Otherwise, any HDD that has obvious major errors is beyond home repair. The cost of hard dives has come down so much.

Good quality Hard Drives that have been puller from commercial use are available at a fraction of the original price. Or, if you don't need huge capacity, brand new drives are under $40 for up to 250GBThere are tools, such as SpinRite that might be able to fix your drive long enough for you to get the data you need, but yes most likely recycle that drive.

Drives are relatively cheap these days, and most are only under warranty for 3-5 years.

Did the drive by chance give you an errors before hand? Such as failing S.M.A.R.T tests?Quote from: geek hoodlum on December 12, 2011, 07:11:37 PM

the sound coming from my HDD is kinda different, like a strange clicking sound.

Classic sign of disk failure.
Quote from: geek hoodlum on December 12, 2011, 07:11:37 PM

Are there any possible solution to fix my Samsung HDD?

You cannot "repair" a failed drive. That is why having proper backups of important data is critical.

Quote

There are tools, such as SpinRite that might be able to fix your drive long enough for you to get the data you need, but yes most likely recycle that drive.

Personally I avoid any products affiliated with it's author like the plague. But I guess even a stopped clock appears to be working twice a day.Avoid SpinRite.

Edit:
T there is FREE software out there to help you. But it is not magic. If your drive has major failures, you can not really fix it.
The kink below provides information and some free items. If a drive is usable, the programs they have will do a good job to r recovery and repair. If the drive still works.
http://www.easeus.com/
You can not repair a major crash with software. To illustrate. Even tho a flat tire is flat just on the bottom, if it has a big hole in the sidewall, putting more air in it does not provide a temporary fix. Same idea with a hard drive. The drive must be 90% usable before you and fix it.
Quote from: Geek-9pm on December 13, 2011, 01:47:00 AM

The drive must be 90% usable before you and fix it.

Wha??? How would you know if it's 90% usable? Like... write random data to the drive and 90 times out of a 100 the data verifies correctly...? And the verification must only occur in the presence of me....?

http://www.easeus.com/datarecoverywizard/free-data-recovery-software.htm#1
It claims it marks bad sectors before attempting data recovery... but you still wouldn't be able to recover the data on the bad sectors.Quote from: Transfusion on December 13, 2011, 06:14:34 AM

Wha??? How would you know if it's 90% usable?

If 90% of the drive clusters are not marked as "bad", I guess.

IMO any drive that has bad sectors needs to have any and all important data copied off of it IMMEDIATELY. I speak from experience (and I would hope, common sense). They simply are no longer reliable. If you are lucky, the cause of the "bad sectors" was in fact due to a flakey PSU causing issues during the read cycle that were interpreted as a problem reading some unlucky portion of the disk, and can be "re-checked" and marked good. On the other hand, you cannot, to my understanding, force the drive to recheck those sectors that it remapped to a spare sector through it's firmware, only those marked by software. Which is sort of the issue; by the time you are able to see bad sectors, from fsck or chkdsk, the drive has already run out of sectors to remap itself and the I/O error has started to bubble up to the file-system level.

Of course the topic isn't with regard to a drive with bad sectors, but rather one that is no longer detectable by the system. If one is lucky in that situation, they can simply swap the logic board with one from a working drive of the same brand and model. However, the changed acoustic properties of the drive seem to indicate that it is, to use the technical term, screwed.Quote from: Transfusion on December 13, 2011, 06:14:34 AM

Wha??? How would you know if it's 90% usable? Like... write random data to the drive and 90 times out of a 100 the data verifies correctly...?

Please do not take that literally to mean 90 % of the sectors have to be good. I meant that the drive has to have most of its ability. It must be more the 90%, more like 99.7% But if I said that, somebody will say "How do you know it is exactly 99.7%"

I thought that by saying 90% it was understand to mean having most of its important ability. The intent was to use terms a layman would grasp. Drive failure is more that just X numbers of bad sectors. Drive failure can be divided up into about ten categories. Now are you going to ask me "Is it exactly ten."

The failure of HDD is not due to PSU problems

Quote

It claims it marks bad sectors before attempting data recovery... but you still wouldn't be able to recover the data on the bad sectors.

Wrong!

Mechanical, physical damage to the head is either fatal or soon will be. The electronics in the drive can go sour and it can fail to read data. There is no cure for ICs that are at the end of life. They might recover after cooling, but they do not self-heal.

All the data recovery tools attempt to make the best of a bad condition. . Bad sectors are due to defects in manufacturing. The defects are not apparent during early treating. In time the defect becomes worse and the drive system flags the spot as an area that needs to be moved. And yes, the bad sector can be read reliably and its data moved to another area. Whether or not the sector becomes worse does not matter nonce it has been flagged. Maybe it will get worse. maybe not. but it will no longer be used. The exact algorithm for this and the parameters are proprietary to the hardware / software makers. Don't even try to second guess. If they wanted you to knew, they would tell you. So I can't tell you either. But it works better that you may THINK it does.

I did not want this to be a lecture non modern drive technology. The point is that the technology is very sophisticated and usually takes care of itself. The S.M.A.R.T. works good about half the time. Which is much better than nothing. If the drive only has some bad sectors, and no other problem, it is possible to restore the drive back to good heath and reset the monitoring system. But this is not something amateurs ought to do, except as a hobby.

Formatting a drive with then right tools may restore a drive to almost new condition. Windows Check Disk does a fair job. The dive manufacturers have tools you can also use. But you have to read the instructions, not guess.

More specifically, if a chuck of the oxide comes off and is n flying around inside the chamber, the drive is going to start failing very fast.Any attempt to read the whole drive will just shorten its life and reduce the opportunity to recover vital data. It depends. A few bad sectors are not a problem. But if severe damage has been done to the head or the surface, mapping bad sectors is a waste of time.
Quote from: BC_Programmer on December 13, 2011, 01:42:05 AM

You cannot "repair" a failed drive. That is why having proper backups of important data is critical.

Personally I avoid any products affiliated with it's author like the plague. But I guess even a stopped clock appears to be working twice a day.

I personally don't use it either, but I was making a point if he needed data their are software utilities that might help him.Quote from: Geek-9pm on December 13, 2011, 08:13:36 AM

Bad sectors are due to defects in manufacturing. The defects are not apparent during early treating. In time the defect becomes worse and the drive system flags the spot as an area that needs to be moved. And yes, the bad sector can be read reliably and its data moved to another area. Whether or not the sector becomes worse does not matter nonce it has been flagged. Maybe it will get worse. maybe not. but it will no longer be used. The exact algorithm for this and the parameters are proprietary to the hardware / software makers.

All Hard drives ship with defective sectors. The drive itself remaps them to spare sectors, at which point the original sector location's condition is irrelevant. However, the original sector can still be perfectly fine. a aberrant power fluctuation from the PSU could easily cause the drive logic to make an assumption about where it is reading that is no longer valid; or undervoltage of the read head during a critical decoding stage could cause the drive to read a CRC error. The tolerance of the drive for what is and is not an acceptable number of retries varies, but when you have a PSU that simply hooks the 5v rail to PWR_GOOD, problems like that are inevitable.

Also, the algorithms used for Hard Drive sector remapping are rather well documented. P-Sector remapping and G-List remapping. I doubt HD manufacturers would use anything different, given that their behaviour in these situations (bad sector remapping) is altogether consistent.

Quote

The S.M.A.R.T. works good about half the time.

Self-Monitoring And Reporting Technology is utterly useless for it's intended purpose. The purpose being, of course, to warn when a disk is failing. But it doesn't work. The various numbers that it keeps track of could easily be skewed as a result of environmental factors; when S.M.A.R.T sets off a warning, it basically means some number approached some other number. you can't easily get at the numbers to find out what set it off, and it doesn't matter which value hits the tolerance, the drive is simply reporting by most System BIOS's as a S.M.A.R.T failure on the drive. More to the point, S.M.A.R.T, aside from having the number of firmware-remapped sectors as one of it's magic numbers, doesn't help prevent bad sectors, or disk failures; prime example being the Samsung drive of the Original Poster which by all appearances has failed.

Quote

If the drive only has some bad sectors, and no other problem, it is possible to restore the drive back to good heath and reset the monitoring system. But this is not something amateurs ought to do, except as a hobby.

If you want to reset smart data, you either don't get what smart is about, or you have bad intentions- such as trying to reset the S.M.A.R.T data for values like "Hours Run" and "Spin-ups" in ORDER to sell the drive as "new". And resetting the values for remapped sectors only changes the value for the S.M.A.R.T reporting... it doesn't actually make changes to the P-List and G-List and remove the mappings. Either way, the problem with S.M.A.R.T is that the tolerances are set differently for all different drives, and oftentimes they are given utterly stupid values. Some Maxtor Drives report "impending Drive failure" (fail SMART) simply by being spun-up a set number of times. This despite there being no evidence to suggest that the number of times a drive has spun up is in any way related to the frequency with which a drive fails.

Quote

Formatting a drive with then right tools may restore a drive to almost new condition.

Formatting a drive can never return a drive to "almost new" condition. Even if you get some special tool to reset the S.M.A.R.T values and reformat the drive- like say, flashing a newer firmware, which can sometimes reset accumulated S.M.A.R.T values as a side-effect- it will merely mean that S.M.A.R.T reporting for that drive is even more useless because the values it actually stores in no way reflect the drive. Remapped sectors will still exist, but unless the drive has to remap more after the values are reset, it will still report the value as 0; and either way the value will be off by whatever number of remapped sectors there were before.
Quote

Windows Check Disk does a fair job.

Of doing what? Windows Chkdsk checks the File-System structure, and can perform a surface scan which attempts to locate bad sectors. If chkdsk is reporting bad sectors at all- the drive is basically screwed. chkdsk can report that there aren't any bad sectors when in fact there are because the drive will have remapped them. This is invisible to any software and the only software-visible way to figure if there are remapped sectors would be to investigate the S.M.A.R.T fields. Of course if you reset that data then even that won't work, and either way software cannot determine what sectors are remapped and which ones aren't, it's just a count. A chkdsk surface scan can make the drive remap bad sectors itself, though, when they are accessed. (which is a good thing, better to remap bad sectors earlier rather than later)

Quote

mapping bad sectors is a waste of time.

Tell that the the manufacturers that have been incorporating P-List and G-List sector remapping within their firmware and ASIC's for almost 20 years.To all reading this. BC knows programming. He knows very little about hard drive technology, other than what he reads in comic books.

Quote

All Hard drives ship with defective sectors.

false
Quote

Also, the algorithms used for Hard Drive sector remapping are rather well documented

false
Quote

aberrant power fluctuation from the PSU could easily cause the drive logic

false. It is extremely rare.
Quote

Of doing what? Windows Chkdsk checks the File-System structure, and can perform a surface scan which attempts to locate bad sectors. If chkdsk is reporting bad sectors at all- the drive is

false. How do you think you know what it does?
Geek should get a warning for the amount of rubbish he posts. His above effort is nonsense. Each of his points answering a remark of BC_P's is a complete howler.
Quote from: Geek-9pm on December 13, 2011, 12:03:01 PM

To all reading this. BC knows programming. He knows very little about hard drive technology, other than what he reads in comic books.

Funny.

Quote

Quote
All Hard drives ship with defective sectors.
false

This is common knowledge. It has always been true. With MFM/RLL drives, the defective sectors would actually be listed On the drive, since that was before Sector remapping was put into use (since it was Before IDE and the common usage of ASIC chips, making it so the circuitry for such functionality would take up too much real estate). Once ASIC chips became cheaper and IDE was conceived and established, they looked to the problem of factory defects and how they can help. The solution was to simply PACKAGE the drives with spare sectors, and have it remap them, in a way that is invisible to the Operating System. The number of pre-existing bad sectors per Megabyte has decreased by a LARGE margin thanks to new manufacturing techniques, but hard drive size has also skyrocketed so for the most part there will be at least a half dozen or so bad blocks that need to be remapped. Manufacturers will sometimes run the hard disks through QA (to remap the existing defects) and reset the S.M.A.R.T fields so that it shows none remapped. This isn't that dishonest though since they typically will decrement the tolerance value for the field as well. This is why sometimes you can get two of the exact same model hard drive that have different "Max allowed" values for their S.M.A.R.T fields. Essentially they allow the remapping and pretend that the drive came with fewer spare sectors.

Quote

Quote
Also, the algorithms used for Hard Drive sector remapping are rather well documented
false

http://www.mjm.co.uk/sectorremapping.html

Quote

Quote
aberrant power fluctuation from the PSU could easily cause the drive logic
Quote
false. It is extremely rare.

So... err.. "That doesn't happen. It's rare" How is that not a contradiction of terms?

And you are seriously suggesting that the hard drive, despite running on electrical power from the power supply both the 5v for the logic board and the 12 volts for the spindle motor, is somehow immune to abberations from bad power supplies or power supplies that continue to send Power_Good to the machine even when the power is not good? Not to mention that power fluctuations on the logic board can cause bit decay in the same manner that insufficient power to System memory will; and if that happens to occur while the logic board is reading or writing data, a CRC error can easily occur as it finds that the data it has in memory has changed.

Anecdotally, I have two hard drives that in my older machine were working quite well- until I added a third Hard drive to the configuration. Drives would disappear at random intervals, and very shortly after changing the configuration they all started to fail the S.M.A.R.T diag the BIOS performed at boot-up, and I would have to chkdsk /r at least once a week to fix hard freezes.

Once I replaced the hard drive, all these issues went away. I was able to chkdsk /b and re-evaluate bad clusters at the file-system level and Suddenly the system decided they were no longer bad. I wasn't able to "unmap" the P or G-Lists, obviously. Best I would have been able to achieve was resetting the SMART data. I imagine hard drive manufacturers have internal software they might have that can deal with those particulars. But I don't think any are available to the public.

Quote

Quote
Windows Chkdsk checks the File-System structure, and can perform a surface scan which attempts to locate bad sectors
How do you think you know what it does?

I know it works on the file system level. chkdsk, like format, calls into fmifs.dll (Format Manager for Installable File Systems, presumably) It exports a number of functions, such as Chkdsk, ChkdskEx, Format and FormatEx; this can be confirmed quite simply with a quick dumpbin /exports.

FMIFS itself, some might assume, is where you might think at this point that FMIFS is where the file system on-disk structure knowledge is located. That is not the case. FMIFS, as it's name suggests, is more or less a stub program for a set of installable File-Systems. For Most windows installations, this means FAT and NTFS. The logic for FMIFS to deal with these two filesystems is contained in ufat.dll and untfs.dll, located in the windows system32 directory. FMIFS itself, as can be viewed using dumpbin /disasm on it, mostly consists of epilog code, prolog code, and setting up and acquiring handles to DLLs and Function addresses (the installable file system drivers are enumerated from the registry). Appropriate calls to FMIFS.dll's various functions is essentially taking the string argument passed to it's function, taking U on the front and .DLL on the end, calling LoadLibrary() using the created string, and then attempting to use GetProcAddress() to acquire pointers to the function it will call, either the chkdsk or format functions, as needed,and there is error handling, of course. This is all easily confirmable by viewing the dissassembly, even if it is hardly an easy parse. the installable file system drivers- UFAT and UNTFS in this case- are the portions that actually "know" about the On-disk FAT and NTFS layout, so that the system can perform consistency checking or formatting operations.

UNTFS and UFAT (and I believe UXFAT, or something similar for the ExFat System, presumably) are the DLLs that actually read/write from the disk during chkdsk and format operations. They in fact open the device directly and read/write clusters themselves; this, again, is quite easily confirmed via any number of Application tracing tools. Since UNTFS and UFAT deal only at the file-system level, and all reads/writes done during a chkdsk are done by those dlls, it stands to reason that the disk check is at the file-system level. your dismissal of the topic was so quick, you left it vacant of any actual information.

Anyway, the obvious purpose of this heirarchy, which is well-documented in several SysInternals Articles, blog entries, and the like, is to make it easy for MS to add new file system types to the system without having to write custom versions of the file-system management tools. They simply write a DLL for the new file system type and make the management tools (specifically, FMIFS.DLL, and even explorer for it's format dialog) aware of the new format name. This can be seen in action with the installable Device Driver for ExFat on Windows XP systems, which make use of this capability.

Add to this the fact that chkdskx, a sysinternals tool that exactly duplicates the functionality of chkdsk (quite literally, the two are indistinguishable) is Open Source, I think it's a fair bet to say, particularly based on your attempt at ad hominem that I "know programming" that I, and really anybody, can know how it works by perusing that source.

By the way, the source can be found here. The actual source was ripped from SysInternals sites back when MS acquired them, but there are quite a few mirror sites.

Solve : HDD Disaster?

Discussion

No Comment Found

Related InterviewSolutions

Reply to Comment