The defective drive is how to fix it. BAD-sectors. What is it with what it is eaten and how to get rid of them

Winchester is one of the most unreliable devices in the computer. In addition to sophisticated electronics, it contains continuously operating mechanical parts. Over time, they wear out, and various problems begin, the most common of which is the appearance of BAD-blocks. This is especially true for old hard disk models that can still be used (in particular, in enterprises where movies, games and other "heavy" content are not kept on computers) and which have already been fairly worn out. Many users are surprised, and they do not know what to do next. Therefore, this article was written. In it, we will consider all available ways of getting rid of these problems at home.

A bit of history

BAD-sectors ( from the English. - bad, unfit) is on any hard drives. No matter how carefully their discs were made, there are several places on each of them, the recording or reading of which is accompanied by errors. In addition, there are simply buggy parts of the surface that can eventually degenerate into defects, which is unacceptable for the user. Therefore, each drive after manufacturing in the factory, undergoes thorough testing, in the process of which the spoiled sectors are identified. They are marked as unusable and entered in a special table - defect sheet.

The very first hard drives had a defect sheet in the form of a paper sticker, into which the addresses of unstable sites were inscribed in the factory. These devices, which are a slightly modified copy of a regular floppy disk drive, could only work under their physical parameters: the number of tracks, sectors and heads indicated in their passport exactly coincided with their actual number. By purchasing such a device, the user read the sticker and himself entered the addresses of the killed sites in the FAT. After that, the operating system stopped noticing these defects, just like it does not notice the bad blocks on floppy disks, if they were removed by the Scandisk utility. Probably, in those distant times the term "bad block" appeared: the block was called cluster  - the minimum unit of logical disk space. At the physical level, the cluster consists of several sectors, and if one sector is damaged, the OS declares the whole cluster unsuitable. There were no other methods of concealing defects at that time. And when there were ways to hide certain sectors, people did not invent new concepts, and until now successfully continue to use the word "block".

It was not long before the manufacturers came up with a very interesting thing: if the user still marks the bad blocks as unnecessary, they reasoned, why not mark them directly at the factory? But how to do this, if there is no file system on the hard drive, and what will be unknown? That's when they came up with a tricky thing called " compiler": on the" pancakes "began to record a special table, which noted which sectors should be hidden from the user and which ones should be left to him.The translator became a kind of intermediate link connecting the physical system of" disks-heads "with the drive interface.

It was assumed that when turned on, the hard disk first reads its internal tables, hiding the defect addresses marked in them, and then allows itself to the BIOS, OS and application programs. And so that the user does not accidentally lose the translator during operation, he was placed in a special area of ​​the disk, inaccessible to normal programs. Only the controller could access it. This event produced a real revolution in the hard drive, and marked the emergence of a new generation of drives - with the service area.

In order to ensure that all screws of the same model, but with different number of defects, have the same capacity, each of them began to leave spare tracks - a reserve specially provided for leveling the capacity of the same type of drives to the standard declared value. He was placed at the end of the disk, near his center, and he too was not available to the user. Such hard drives when leaving the factory did not have not one visible bad sector. If new defects appeared during the operation, the user could make low-level formatting a universal utility from the motherboard's BIOS, and try to hide them. Sometimes, like on floppy disks, it was possible. But if the "evil spirits" were physical, it did not help: adding new defects to the table and rewriting the translator without special programs was impossible. Therefore bad blocks on many old screws (before 1995), it was necessary to hide all the same, in an outdated way - through FAT. And only firms Seagate, Maxtor and Western Digital have released utilities to hide defects with their replacement from the reserve.

Time passed and the screws changed even more. In an effort to increase the density of the recording, the developers began to apply various non-standard tricks: the plates were applied servo marks, designed to more accurately hit the heads on the tracks. There was a technology of zone-sectional recording (ZBR), the meaning of which was a different number of sectors on external and internal tracks. The drive of the heads has changed - instead of a stepping motor, the positioner has been used in the form of a moving coil. And the heads and disks themselves have changed so much that each company has developed its low-level format structure, sharpened only for their technology. This made it impossible to use universal utilities of low-level formatting due to the fact that the translator of such screws learned to hide the physical format of drives, translating it into a virtual one.

The number of cylinders, sectors and heads written on the hard drive body has ceased to correspond to its true values, and attempts to format such a screw with old utilities usually ended unsuccessfully: its controller rejected the standard ATA command 50h, or simply mimicked the formatting by filling the screw with zeros. This was specifically left for compatibility with older programs. For the same reason, the Low-Level Format procedure was excluded from the BIOS of modern motherboards. And to make such hard disks the present low-level formatting, it was necessary to bypass the translator, having received direct access to physical paths and heads. To do this, use a technology utility that launches a special microcode, written in the drive's ROM. The command to call this microcode is unique for each model, and refers to technological commands that are not disclosed by the firm. Often such formatting could not be done through the standard IDE interface: many HDD models of the 90's - Conner, Teac, etc., as well as all modern Seagate, require the connection of a separate connector to the terminal via the COM port.

As for the technology tools, they were never widely distributed and the regular user was not available. For a wide range of applications, "dummy programs" were written that perform pseudo-formatting via the interface: filling the disk with zeros to clear it from information. This can be seen even from the names of these utilities, which can be found on the websites of manufacturers of hard disks: wdclear, fjerase, zerofill etc. Naturally, there are no technological commands in these programs, and therefore they can be applied to any hard drives. Such utilities are often useful, helping to get rid of some types of BAD's, which we'll talk about a bit later.

Why did the manufacturers act so cruelly, depriving us of the opportunity to do the correct low-level formatting, and hide the defects ourselves? There is still no consensus on this issue, but the official response of most businesses sounds like this: "It's such a complicated and dangerous operation that an ordinary user can not be allowed to it, otherwise many hard drives will simply be killed." Therefore, low-level formatting can only be done at the factory, or in the company's service center. "

Let's try to figure out whether this is really so. And at the same time we will consider what is the present low-level formatting of modern hard drives, can we do it ourselves, and what is most important - do we need it?

Preparing the hard drive in the factory

Before hiding the bad sectors at the plant, it is very important to identify all, even very small defects, as well as unstable patches that can eventually grow into bad blocks. After all, if this happens in the process of exploitation, the user can lose an important file, and the reputation of the company that released such an "unfinished" drive will be spoiled. Therefore, testing of hard drives before hiding defects takes a very long time, at least several hours, and is performed in a technological mode. This is done to exclude the time delays that inevitably arise during the operation of the translator, the transfer of data through the cache and interface logic. Therefore at the factory the surface is scanned only by physical parameters. Usually this is not an external program, but a special module in the hard disk ROM that works without the interface. The end result of such testing is the receipt of a defect sheet-an electronic list of unusable areas of disk space. It is stored in the service area of ​​the screw and stored there for the entire service life.

Modern hard disks have two main defect-sheets: one is filled in the factory during the manufacture of the drive and is called P-list (primary), and the second one is called G-list (from the word growing), and replenished during the operation of the screw, with the appearance of new defects . In addition, some screws also have a servo-defect sheet (servo marks applied to hard drive plates, too, sometimes have errors), and many modern models also contain a list of temporary (pending) defects. In it, the controller enters "suspicious" sectors from its point of view, for example, those that were not read the first time, or with errors.

Having received the defect sheet, they begin to hide the defects. There are several ways to hide them, each of which has its own characteristics. Theoretically, you can just reassign the addresses of the corrupted sectors to the reserve and take them from there, but this will cause the loss of the screw performance, as it, each time detecting a sector marked as unusable, will have to move the heads to a spare area that can be far from the defect site. If there are many reassigned sectors - the drive's performance will drop very much, since most of the time it will be spent on useless jerking heads. Moreover, the speed of screws with different number of defects will vary greatly, which of course, in mass production is unacceptable. This method of concealing defects was called " substitution method" or remap (from the English. sector map rebalancing).

Because of the many drawbacks inherent in a remap, this method is never used in the industrial manufacture of screws, and another algorithm is used: after identifying all the defects, the addresses of all serviceable sectors are rewritten anew, so that their numbers go in order. Bad sectors are simply ignored and do not participate in further work. The backup area also remains continuous and part of it is attached to the end of the working area - to equalize the volume. This method of hiding beats is more difficult to implement than a remap, but the result is worth the effort spent on it - with any number of faulty sectors, the drive does not slow down. This, the second main type of concealment of defects was called " sector pass method"(There are other algorithms for factory defect concealment, for example, by excluding an entire track, or by using a spare sector on each track, but they have drawbacks and therefore are not used in modern drives).

The process of address translation with missing defects was called "internal formatting". Internal - because the whole process occurs completely inside the hard disk, at physical addresses and without the participation of the interface. At this time, the screw is controlled by a firmware embedded in its ROM, which analyzes the defect sheet and controls the formatting. External commands can not be interrupted. When the formatting is completed, the firmware automatically recalculates the translator (or creates it again), and the screw becomes ready for use. After that, he, without a single bad block, comes from the factory to the buyer.

New technologies

Now it's clear why proprietary utilities do not do any operations related to direct access to the service area. After all, concealing defects by formatting is almost a complete repair cycle based on external parameters and associated with a clear understanding of each step. And it's enough to do something wrong to destroy the drive. Here is a simple example: the user decided to make a "real" low-level formatting by running a ROM subroutine in the production mode. The process usually lasts 10-60 minutes, but then there is a power failure or a banal hang - and the screw remains without the translator, because. just does not have time to recreate it. This means that to further work such a device will be unsuitable - it simply will not see either the OS or the BIOS.

It's frightening to even imagine how many drives can be killed in this way, out of mere curiosity or by mistake. Especially, if these utilities fall into the hands of teapots, running everything on their computers. Of course, the disk does not corrupt irrevocably, and by re-starting the formatting, you can bring it back to life. But the thinking of most users is designed so that, faced with problems (not identified in the BIOS corpse, instead of a screw), many panic, blaming all manufacturers. And they need unnecessary problems, of course they do not need it - it is much more important to get the hard drive to work out the warranty period. Therefore, several years ago, in the drives began to pawn the opportunity to "repair" the bad sites themselves - to do a remap. As it was said before, the remap did not find application for factory preparation of drives, but it turned out to be a very successful solution for hiding defects in domestic conditions.

Advantages of the remap before internal formatting - the lack of transfer of the screw in the technological mode, speed and safety for the drive. In addition, in many cases, you can do a remap without deleting the file system, and without the associated destruction of the data. This technology is called automatic defect reassignment  (automatic reassignment of defects), and the process itself - reassign. In this way remap  and reassign  - this is largely the same, although the term reassign is usually applied to a separate sector, and remap - to the entire disk.

The remap works as follows: if an error occurs while trying to access the sector, the smart controller understands that this sector is faulty and marks it as a BAD on the fly. Its address is immediately entered in the table of defects (G-list). Many screws have this happening so quickly that the user does not even notice the detection of a defect and its concealment. During operation, the drive constantly compares current sector addresses with addresses from the table and does not refer to defective sectors. Instead, he moves the heads to the spare area and reads the sector from there. Unfortunately, due to the time spent on long-distance positioning, such sectors will look like small gaps in the reading schedule. The same will happen when recording.

If an error occurs during normal operation of the OS, automatic remap occurs extremely rarely. This is due to the fact that, on most HDDs, reassign only works when recording. And many OSes before writing check the sector for integrity, and finding an error, refuses to write to it. Therefore, in most cases, for the production of a screw remap it is necessary to "ask" for this - to perform compulsory low-level sector overwriting, bypassing the standard OS and BIOS functions. This is done by a program that can access the hard drive directly through the ports of the IDE controller. If an error occurs during such an entry, the controller will automatically replace this sector from the reserve, and the BAD will disappear.

This principle is based on the work of most utilities of the so-called "low-level formatting" from the manufacturers. All of them, if desired, can be used for the screws of other companies (if such programs refuse to work with other drives, this is done for marketing reasons). And of course, the remap functions are present in many universal and free programs, the features of which we will consider later. In the meantime, there is still some theory.

The most common myth among users is the statement that for each screw you need your own, "special" program to hide defects, and that remap is low-level formatting. In fact, this is not so. Remap is just a form of recording information by standard means, and in most cases any remap utility can be applied to any screws. Remap does not external programs, but the hard disk controller. Only he decides to reassign the defective sectors. To spoil the drive "foreign" programs also can not, because technological commands are not used in them, and in normal mode the screw will never allow to do anything with itself, except for standard read-write operations. The only difference between proprietary utilities is the number of attempts to write / read / verify for different screws. In order for the controller to "believe" that there is a BAD to be hidden in the sector, some HDDs need only one cycle, and another - several.

Again about S.M.A.R.T.

Almost all hard drives that were released after the year 95 have a system of operational monitoring of their condition - S.M.A.R.T. (Self Monitoring And Reporting Technology). This technology allows you to assess at any time such important parameters of the drive as the number of hours worked, the number of errors that occurred during the process of reading / writing, and much more. The first hard drives equipped with this system (for example, WD AC21200) had a very imperfect SMART of four to six attributes. But soon the SMART-II standard was developed, and since its appearance in most drives there has appeared such a feature as internal diagnostics and self-control. This function is based on a series of self-contained internal tests that can be run by standard ATA commands, and is intended for in-depth monitoring of the mechanics of the drive, the surface of disks and many other parameters.

After running the tests, the drive automatically updates the readings in all SMART-attributes, according to its current state. Testing time can vary from a few seconds to 54 minutes. You can activate SMART tests, for example, with the MHDD program (the console command "smart test"). After running the tests, "strange" phenomena are possible, very similar to those that occur during the defragmenter operation: continuous burning of the HDD indicator and the sound of intense head movement. This is normal: the hard drive scans the surface to find defects. You just need to wait for a while until the self-test runs out and the screw calms down.

Later, the SMART-III specification appeared, in which there is not only a function for detecting surface defects, but also the possibility of their recovery on the fly and many other innovations. One of its varieties became the system of Data Lifeguard, used in Western Digital drives. Its essence is the following: if no calls are made to the screw, it starts scanning the surface independently, identifying unstable sectors, and, if detected, transfers the data to the spare area. Then makes him reassign. Thus, the data are saved even before a real BAD occurs on this site. Unlike SMART-monitoring, Data Lifeguard can not be disabled by external commands and works constantly. Therefore, "visible" BAD-blocks on modern Western Digital hard drives almost never appear.

To view the SMART-status of the hard disk, use programs called SMART-monitors. One of them is part of the HddUtil for DOS complex and is called smartudm. This program works with any hard drives and controllers. In addition, in the kit with it is a detailed documentation with a description of all the attributes. There are SMART monitors and for Windows 9x, for example, SiGuardian and SmartVision are very popular, but they may not work on some systems. This is explained by the fact that the programs work with the screw directly, through ports, and bus mastering drivers of some chipsets interfere with this. Owners of Windows XP should pay attention to the monitor SmartWiew www.upsystems.com.ua/ - the application works correctly in this system even on VIA chipsets.

There is some relationship between the SMART attributes and the state of the surface. Consider those of them that are directly related to bad blocks:

• Reallocated sector count and Reallocated event count: the number of sectors reassigned. These attributes show the number of sectors remapped by a remaps in the Grown defect sheet. For new screws they must necessarily be zero! If their value differs from zero, then it means that the hard disk was already in use, badges appeared on it, and a remape was made to it. Be careful when buying second-hand!
• Raw read error rate: the number of read errors. For many HDDs, they are always above zero, but if Value is within the norm (green zone), there is nothing to fear. These are "soft" errors, successfully corrected by the electronics of the drive and do not lead to distortion of the data. It is dangerous when this parameter drops sharply in a short time, passing into the yellow zone. This indicates serious problems in the drive, the possible appearance of badges in the near future, and that it's time to backup important data;
• Current Pending SectorThis attribute reflects the contents of the "temporary" defect sheet present on all modern drives, i.e. the current number of unstable sectors. These sectors could not be read from the first time. The raw value field of this attribute shows the total number of sectors that the drive currently considers as candidates for remap. If in the future some of these sectors will be read (or rewritten) successfully, then it is excluded from the list of applicants. A constant value of this attribute above zero indicates a problem in the drive.
• Uncorrectable Sector: shows the number of sectors, errors in which could not be corrected by the ECC-code. If its value is higher than zero, it means that the screw needs to do a remap: it is possible that during the data recording the OS will break into this sector and as a result some important information or system file will be corrupted. However, for some screws this attribute is not discarded for some reason after a remap, so it is not necessary to trust its indications.

Types of defects and the causes of their appearance

It's time to figure out why, in fact, there is such a nuisance as badds? To do this, consider the structure of the sector, in the form seen by the electronics of the screw "from within":

Fig. 1. Simplified structure of hard drive sector

As you can see from Figure 1, everything is much more complicated than it might seem at first glance, even with the help of a disk editor. The sector consists of an identifier header and a data area. The beginning of the sector is marked with a special byte-address marker (1). It serves to inform the controller that the sector is under the head. Then follow the cells that contain the unique sector address in the CHS (2) format and its checksum - to check the integrity of the recorded address (3). 512 bytes of user data are placed in a separate field (4), to which several dozens of bytes of redundant information are added during recording to correct reading errors with the ECC code (5). Next to the data, 4 bytes of the cyclic data checksum (CRC) are placed, which serves to verify the integrity of the user data, and to the error correction system when it is violated (6). For more reliable operation of the sector, there are bytes-gaps (7) at speed variations. Some HDDs have an additional byte after AM - in it the sector is marked as BAD.

The structure of the low-level format differs greatly among different drive models, and is determined by the type of controller used, its microprogram and the ingenuity of the developers.

While the structure of the format is not broken, the hard drive is working properly, clearly fulfilling its duties - storing information. But it's worth interfering with the evil forces - and depending on the type of destruction, they manifest themselves as BAD-sectors of different severity.

Defects can be divided into two large groups: physical and logical. Let's look at each of them in detail.

Physical defects

Surface defects. Occur with mechanical damage to the magnetic coating inside the sector space, for example, due to scratches caused by dust, aging pancakes or careless handling of the hard drive. Such a sector should be marked as unfit and expelled from circulation.

Servo errors. All modern drives use a system called the voice coil to move the heads, which, unlike the stepping motor of old screws, does not have any discreteness of movement. To accurately hit the heads on the tracks in the screws, a feedback system is used, which is guided by special magnetic servos placed on the disk. Servo marks are on each side of each disc. They are located evenly along all the tracks, and strictly radially, like spokes in the wheel, forming a servo format. It does not belong to the lower-level format and is not shown in the figure, but it is absolutely present in all modern hard disks, and plays an important role. Servo drives stabilize the engine rotation speed and keep the head on a given track, regardless of external influences and thermal deformation of the elements.

However, during the operation of the screw, some servo marks may be destroyed. If there are too many servo marks, this place will begin to malfunction when accessing the information track: the head, instead of taking the position it wants and reading the data, will start to rush from side to side. It will look like a fat and particularly insolent BAD, or even as their group. Their presence is often accompanied by a knocking of heads, hovering of the drive and the inability to fix it with the usual utilities. Elimination of such defects is possible only by special programs, by disabling defective tracks, and sometimes the entire disk surface. For these purposes, some drives have a servo-object sheet that stores information about bad servos. Unlike the P and G-sheets, the servo-leaf sheet is used not by the translator, but by the whole microprogram of the screw. To sectors with defective servo marks, access is blocked even by physical parameters, which avoids knocking and disruption when accessing them. Self-winchester can not restore the servo format, this is done only at the factory.

Hardware BAD Sectors. Occur due to a malfunction of the mechanics or electronics of the drive. Such problems include: breakage of heads, displacement of disks or bent shaft as a result of impact, dusting of the hermetic zone, as well as various glitches in the work of electronics. Errors of this type are usually catastrophic and can not be corrected programmatically.

Logical defects

These errors do not arise from damage to the surface, but from violations of the logic of the sector. They can be divided into irreparable and irreparable. Logical defects have the same external manifestations as physical ones, and they can be distinguished only indirectly, based on the results of various tests.

Correctible logical defects (soft-bad): appear if the sector checksum does not match the checksum of the data written to it. For example, due to interference or power failure during recording, when the screw has already written data to the sector, and the checksum was not recorded (Figure 1). At subsequent reading of such an "unfinished" sector there will be a failure: the hard drive first reads the data field, then computes their checksum and compares the received data with the recorded data. If they do not match, the drive controller decides that an error has occurred and will make several attempts to reread the sector. If this also does not help (and it does not help, since the checksum is known to be incorrect), then, using code redundancy, it will try to correct the error, and if this does not work out, the screw will give an error to the external device. On the operating system side, this will look like BAD. Some hard drives had an increased propensity to form soft-bads due to errors in the firmware - under certain conditions, the checksums were not calculated correctly; in others this was due to defects in mechanics.

The operating system or the BIOS can not fix the logical defect by itself, because before they write to the sector, they check it for integrity, run into an error and refuse to write. At the same time, the screw controller can not correct this mistake either: it tries in vain to read this sector from the second, at the third attempt, and when it does not work out, he tries to help himself with all his might, adjusting the reading channel and the servo system. At the same time, a heart-rending screech is heard. This scratch does not produce "heads on the surface," as many people would think, but only a coil of the positioner, because of the specific shape of the current flowing through it, and it is absolutely safe. The address of the unread sector falls into the temporary defect list, changing the value of the Current Pending Sector attribute in SMART, and is stored in it. Remapa does not happen when reading.

And only compulsory low-level overwriting of this sector by a special program bypassing the BIOS results in automatic recalculation and overwriting of the checksum, that is, the BAD-block disappears without a trace. You can rewrite it with a disk editor that can work with the screw directly through the ports, but usually "overwrite" the entire disk, filling its sectors with zeros. Utilities that do this are freely distributed by drive manufacturers, and are often incorrectly called "low-level formatting programs." In fact, these are simple "obnutili", which does not in the least prevent them from delivering the screw from the bad: when successful recording soft-badges disappear, and if unsuccessful - the bad is considered physical, and autoremap occurs.

Incorrect Logical Errors. These are errors in the internal format of the hard drive, leading to the same effect as surface defects. Occur when the sector headers are destroyed, for example, due to the action of a strong magnetic field on the screw. But unlike physical defects, they can be corrected programmatically. And they are called incorrigible only because for their correction it is necessary to make "correct" low-level formatting, which is difficult for ordinary users due to the lack of specialized utilities. Therefore, in everyday life such a sector is turned off the same way as a physical one - with the help of a remap. Currently, more and more screws are produced using ID-less technology (sectors without headers), so this type of error is no longer so relevant.

"Adaptive" pads. Despite the fact that the screws are very accurate devices, when mass production of them inevitably there is a spread of the parameters of mechanics, radio components, magnetic coatings and heads. The old drives did not hurt, but modern screws with their huge recording density, slight deviations in the size of parts or in the amplitudes of signals, can lead to deterioration of the product's properties, the appearance of errors, up to the complete loss of its performance. Therefore, all modern drives in the production process are individually tuned, in the process of which such parameters of electric signals are selected, under which the device is working better. This setting is carried out by the program ROM during technological scanning of the surface. In this case, so-called adaptive variables are generated, which contain information about the features of a particular air-block. Adaptives are saved on pancakes in the service area, and sometimes in flash memory on the controller board.

If the adapters are destroyed during the operation of the screw (this can happen as a result of errors in the screw itself, static electricity, or because of poor quality power), the consequences can be unpredictable: from the banal pile of heads to the complete inoperability of the device, with the failure to go on readiness for interface. "Adaptive" pads differ from usual ones in that they are "floating": today they are, and tomorrow they can disappear and appear in a completely different place. Remapping such a screw is useless - ghost-defects will appear again and again. And thus the disk surface can be in a faultless condition! Adaptive heads are treated with a selfscan run, an internal testing program similar to that used in the factory for making screws. At the same time, new adaptations are created, and the screw returns to its normal state. This is done in the conditions of branded service centers.

Outlined defects

These are areas of the surface on which a pronounced defect has not yet formed, but problems with the reading speed are already apparent. This happens because the sector is not read by the controller the first time, and the screw is forced to do several turns of the disk, trying to read it without errors. If you still can read the data, the screw will not tell the operating system anything, and the error will go unnoticed until a real BAD block appears on this place. As a rule, it immediately turns out that it was in this place that a very important file was stored, in a single copy, and it is impossible to save it. Therefore, you need to periodically test the disks. This can be done by Scandisk or Norton Disk Doctor in the surface testing mode, but better - a special utility that works independently of the file system and is able to detect emerging BAD sectors, measuring the read time of each sector.

Practice

Each firm that produces hard drives usually develops a special software for diagnostics and maintenance of its drives, placing it on the Web for free use. Sometimes these utilities already contain an operating system (usually one of the DOS versions), such as the Sea Tool from Seagate or Drive Fitness Test from IBM (now Hitachi). And sometimes it's just an executable file that you need to run from DOS yourself, like Maxtor (already owned by Seagate) or Fujitsu. This software allows you to test the drive for errors and, if possible, fix them. Among the methods of fixing, you can often find disk cleaning functions (filling it with zeros with destruction of all information), as well as hiding defects with the remap method. But we will not consider branded utilities - as we learned, these programs do quite standard things: writing zeros and checking the surface. So let's pay attention to some very good alternative programs.

So, we have such a funny thing, which is "bad hard". Or we want to insure ourselves against "surprises" and check it while it is working. For this, first of all, the program MHDD. Anyone who has an ancient screw up to 8.4 GB, especially old Western Digital, is recommended to have a DOS program on the farm.

First of all, you need to prepare a diagnostic software and create a boot disk with MS-DOS. You can sacrifice a Windows 9x boot floppy by removing all files from it except io.sys, msdos.sys and command.com. On the vacant space we write the executable file of the MHDD program: mhdd2743.exe and the configuration file mhdd.cfg. Since there is still plenty of free space on the floppy disk, write to it SMART-monitor smartudm.exe, and some file manager, for example Volkov Commander. It will be used to view the contents of the program performance reports. For convenience, all the files are placed in the root directory of the floppy disk. Alternatively, you can not create a floppy disk at all or use it for DOS boot only, and start all programs directly from the main hard drive by connecting the scanned screw to another channel. It is not necessary to write programs on CD to run them from there - the disk must be open for writing, as the programs will create work logs on it, and failing at the same time, will simply malfunction. After a close look at the description of MHDD and SMARTUDM, you can proceed with the execution. First, let's look at the SMART-information of our drive (in the future it will have to do more than once).

We booted from our floppy disk, and if the investigated screw hangs on the primary IDE channel, type: a: smartudm on the command line, and if on the secondary one - a: smartudm 1. If there are more than two hard disks in the system, the figure can be greater than 1. Before us, a table appears that characterizes the state of the drive (Figure 2).

Fig.2. SMART hard disk status graph for quick evaluation of its status

Each line of the table is one of the parameters of the current state of the screw. Opposite to each of them, in the column "Indicator", there is a scale divided into three color zones. As the drive wears out, the length of the indicators decreases, as more and more green squares in their right-hand part are extinguished. There are yellow and red. When all the green squares in any indicator disappear, it means that the screw has exhausted its resource or is defective. It is recommended that you save important data, because at any time the screw can die completely. If there was only a red square - the screw is already in an emergency condition, and further storage of files is not suitable.

The inscription "T.E.C. not detected" means that the current state of the screw is in order. If it is not, a warning will be issued, highlighted in red. Having looked at the color chart, you can quickly evaluate which SMART-attribute caused such discontent of the program. In the case of a large number of badges, they will most likely be the highest (Raw Read Error Rate). But this information is approximate, but we need absolute values ​​of attributes, so we press and see about such a picture (Figure 3):

Fig.3. Advanced SMART-status of the hard disk (exact values ​​of attributes)

This mode of the SMART-monitor is the main one, and we will monitor the status of the drive on it in case of any further actions. For example, after looking at the value of attribute 5 (Reallocated Sector Count), we will see the contents of the user defect list and will be able to judge whether the defects were hidden. When you press the key, the current SMART-log is saved to a file. By pressing the key, you can exit the program in DOS. Some of the attributes are updated by the controller "on the fly", several times per minute, so to get the most reliable result, the screw needs to be tested, what we are doing now.

We leave the SMART-monitor, and run the MHDD program by entering the name of its executable file on the command line. After boot, you should immediately press a key combination - the program will scan the bus and show the list of drives connected to the system. Select the one you want to check by entering the desired digit from 1 to 10 into the console (Figure 4). Then press to initialize the selected screw.

Fig.4. Initializing the screw by command

After these actions, the drive will provide information about its volume, the maximum supported DMA mode and much more. The MHDD program sees the screws entirely, completely uninterested in their partitioning and the type of file systems. It will see all the IDE screws, regardless of whether they are detected in the BIOS or not. Even if the motherboard does not support large-capacity drives, the program will still see them at full capacity, if only the hard disks were in good order. If this happens, you can start checking the surface.

For this we press, and in the top line of the appeared menu we put the parameter (by default there is CHS). Switching between CHS and LBA modes is done with the space bar. Then we press the second time. The gray rectangles will run across the screen. It will take 10-30 minutes, and it is absolutely safe for the information stored on the screw, since this only reads the sectors. Here is what the author of the program wrote about this regime in the documentation for it:

"When you perform a surface test on the right, a window will appear on the right, the first line of this window will display the current surface speed, the last one - two percentages .The first value shows the percentage of the current test running in the specified interval, and the second shows how far the heads left 0 cylinders and came to the last one.In the process of testing the surface, one square is 255 sectors (when testing in LBA mode), or the number of sectors in the HDD parameter string (usually, 63 when testing in CHS mode). the more time it takes for the drive to read this block of sectors.If the color squares have gone - it means that the drive did not fit into the allotted time interval for the work.Color squares display abnormal state of the surface (but without BADs). The lower the color of the menu, the more time it takes for the drive to read this hard-to-read area. Red color is a sign that BAD block was almost formed at this place. A question mark appears when the maximum waiting time for a standby is exceeded. That is, when [?] Appears, it can be assumed that the drive is "suspended" in this place and that there is either a serious surface defect or the magnetic head unit (MBG) is faulty. Anything below the question mark is errors (BAD block). If they appear in the testing process, then there are physical defects on the surface. "

In the presence of BAD-blocks, instead of the squares usually appear icons [x], obviously symbolizing the crosses. If the surface is OK and without color squares, and all SMART attributes are in the green zone, you can breathe freely: the screw is still working.

If MHDD showed that there are defects on the surface, and the screw "hangs" or emits scraping sounds, then there is a problem. But we will not think at once of the bad: after all the badges can be logical (soft-bad), so first we'll arrange the drive "brainwashing" - perform a low-level record of zeros in all sectors. (Note: In this case, all information will be destroyed, so we copy important data to another disk). The MHDD program has two commands for zeroing disks: erase  and aerase. We will use the first one, since it works faster.

We initialize the screw by pressing the key (this procedure should be done before any actions), and enter the ERASE command into the console. Be very careful when choosing a drive, otherwise you can mistake your working screw by mistake: the data is lost irrevocably, and even in the FSB they will not be restored! The cleaning procedure is rather slow, taking several tens of minutes. But in the future, after a little understanding of the program, you can erase the disk selectively by entering the initial and final sector number before starting the procedure. This is very convenient if the pads are closer to the end of the disc, and its beginning is flawless.

After cleaning, run the surface test again (by pressing twice or using the console command SCAN). In this case, the screw controller must recount the vital SMART attributes, which will make its SMART-status more reliable. If the bads are no more, the screw can be considered repaired. We leave MHDD, start our SMART-monitor and look at the value of the attribute Reallocated Sector Count. If it did not increase after cleaning, and the defects disappeared, then they were logical. If increased - they were physical, and the controller made a successful remap of these sectors. If, on the contrary, the badges remain, and the value of the Raw Read Error Rate attribute has fallen catastrophically - everything is much more complicated, and the screw has serious damages. We will try to treat it further - to do a remap.

You may already have noticed that when you press a key once in MHDD, a menu appears that contains additional scanning parameters (Figure 5)

Fig.5. Scan and remapse settings

Among these parameters is the remap function. By default, it is off, but by placing the cursor on it and pressing the "space", you can turn it on (Remap: ON). In this mode, MHDD will try to cure the defective sector, in every possible way showing the controller that there is a BAD there and it needs to be hidden. At the same time near each successfully hidden sector there is a blue square or an inscription. After the elimination of all the bads, you need to run the surface test again, exit the MHDD and start the SMART monitor again, making sure that the value of the Reallocated Sector Count has increased. This means that the remap was successful, without errors, and the defects were actually replaced from the reserve.

If you for some reason do not want to lose information from a corrupted hard drive, for example, it has nowhere to store - do not despair. You can try not to do a full zeroing, and go directly to the remap using MHDD. Information from the screw is not erased, except, perhaps, the one that was in the bad (but it still does not return). When the heads are found, the program will apply the same measures to them as in the case of zeroing - a low-level record, and therefore, even if the heads are logical, they can most likely be corrected. The exact result depends on the implementation of the microcode of a specific drive model. But if this does not help, and the defects will not disappear - you will still have to do the nullification, just in case. In some cases, only the use of the "aerase" command can help (it sets the screw to a different algorithm, but it works slower).

The MHDD program is constantly being supplemented and improved. Therefore, after visiting its official website, you can download its latest version.

It may happen that even after all the operations performed, the badges remain, and SMART will show that the remap does not happen. The reasons can be several:

• the screw is very old, and its controller does not support the Automatic Defect Reassignment function. For example, the screws of the company generally can not be repaved. It is treated only by special technological utilities;
• the screw can be filled with G-list, and there is no more room for new defects. This is clearly seen in SMART on the collapse of the attribute of the Reallocated Sector Count. Such a screw is repaired in workshop conditions, by transferring all the remaps to the P-list and subsequent low-level formatting;
• the procedure for auto-replacement of defects was disabled in the screw itself. The proprietary utilities of some drives allow you to do this, and they can also be re-enabled. This is a fairly rare case;
• a special defect that can not be repaired can appear on the screw. For example, if the sector header in which the sector is marked as BAD is physically destroyed, or severely damaged servo marks. Such a screw is repaired only in the technological mode, from a good specialist;
• the controller could not believe that the sector was really defective, since he could still read / write it, even if not at the first attempt. In this case, the remap will not occur. No concealed defect program ever writes directly to defect sheets. This can only be done by the controller itself, based on its observations. In order for him to "believe" that there is a bad in the given place, and hide it, sometimes it takes him quite a long time to explain this, in every possible way showing the problematic sector - repeatedly making a record / reading, before the error occurs. Therefore, utilities for remap will never hide the emerging defect. In order for this to happen, you need only "real" BAD. This "incredulity" is made specially: in fact, each moved sector worsens the drive's parameters, reducing its performance. And to hide 666 heads remapom most likely does not work out - the size of the user defect-sheet is limited, and at a certain value (from tens to several hundred sectors, depending on the specific model), the screw blanks SMART, informing that it's time for it repair or landfill.

But do not rush to throw such a drive. If it is relatively modern, and does not have a crowded defect sheet (attribute 5 is normal), there is still hope for a remap. Just need to try to apply to it another program that has more cycles to write to the defective sector. These programs include HDD Utility for DOS. It works a little differently than MHDD: in it, the surface and remapse test functions are separated, and the remap is made based on the protocol created by the scan. Therefore, we run the test first by going through the chain: - -, and then we move on to the - - item (Figure 6). Before this, it is desirable to read the description of this program, since it is very detailed and written in Russian. Disadvantages of Hdd Utility - a misunderstanding of drives with a volume of more than 8.4 GB and the failure to work with some models (the latter is due to the limitation of the free version). But this is not so important - "hard-to-screw" screws usually have a small capacity - usually they are different models of Western Digital with a capacity of 0.65-6.4 GB. For large screws, you can use the HddSpeed ​​v.2.4 program, it also has a Try to repair / reallocate found defects function and a Russian description (Figure 6).

Fig.6. HDD Utility. The process of hiding the defective sectors

Fig. 7: Remap with HddSpeed

It is impossible to estimate the real state of the drive according to the schedule received through its interface. This is due to the fact that the interface is inevitably delayed, since the controller of the screw, in addition to data transfer, performs many other operations: conversion of physical addresses to LBA, defect management, recording of internal SMART logs, data verification and calculation of their checksums, cache strategy management, thermal calibration, etc. Therefore, this method is suitable only for an approximate evaluation of the screw, the detection of gross errors, and is used only in everyday life. This is well understood by the authors of the testing programs, pointing out the inability to use their results as any evidence. The most reliable are the tests under pure DOS. In multitask environments, the situation is worse, since any background process distorts time intervals, which prevents a correct estimate of the state of the drive.

Alternative methods of concealing defects

As already mentioned above, the remap has a disadvantage, manifested in the form of jerks heads in the reserve area. In this case, the screw can click during operation, and the graph will show the dips. This can greatly complicate, for example, working with streaming video. This is particularly pronounced when the remaps are located at the beginning of the disk: the heads travel the maximum path, and the delay in their movement is very high. Therefore, in some cases, a remap may not be appropriate, and instead of it the best choice is to hide defects by means of the file system. For example, the usual high-level formatting format.com, Scandisk or Norton Disk Doctor. Only to decide on this step is necessary immediately after checking the surface, without trying to make the screw remap. Otherwise, if it is successfully carried out, you can not return the pads back and clear the defect table. Remap is a one-time procedure, and if the screw controller transfers sector addresses to a reserve, it will be impossible to return them back.

Another alternative to the remap is circumcision of the space at the end of the disk using HPA (Host Protected Area) technology, which is available for all modern screws. In this case, the screw will be detected in the BIOS to a smaller volume, and all the troubles, if they are located at the end, will remain "overboard" and become invisible. This method should be applied to drives that have many badges at the end of the disk (unfortunately, this happens rarely). At any time, the screw can return its full capacity and, accordingly, the badges too. You can do this with the MHDD program (HPA and NHPA console commands). If the screw is old, and does not support HPA, then you can create a separate logical partition, not only at the end, but also anywhere else on the disk, and arrange it so that a large group of blades is in it. This is done by the Fdisk program. Such a partition can be clogged with unnecessary files, or you can not format it at all, giving it the status of "non-dos" (then it will become invisible to the system).

But the best way to rid the screw from badges, especially if there are a lot of them, or they can not be repaved - is a bench repair from a qualified specialist. With the help of special equipment and utilities, you can perform a complete repair cycle, similar to the one that the screw goes through at the factory: correct low-level formatting, cleaning the screw from remaps, restoring the service information, and much more. After such repair, the screw will be indistinguishable from the new one, it will have an even schedule, and most importantly - this screw will have a safety margin for several years ahead.

Contrary to popular belief, remap and low-level formatting are not a universal tool for solving any problems. If the screw has a serious hardware malfunction, then these actions will not only not cure the patient, but can also cause harm to him, finishing finally. For example, if the screw monotonously knocks the heads when turned on and does not want to be detected in the BIOS, or breaks into a knock when copying files - do not torment it with software, they will not help. Such behavior is usually associated with physical breakage of heads, destroyed servo or controller malfunction. This screw needs not formatting, but repair from a competent specialist.

Features, glitches and prevention

Not all screws are damaged due to careless handling. Sometimes the cause of their glitches are errors made by the developers themselves. Some of them have irreparable consequences, since they can physically damage the magnetic surface. So it was, for example, in 1996 with hard disks Quantum ST. Due to an error in the microcode, these screws were unpacked heads a little earlier than the pancakes were gaining the required speed. As a result, the head scratched the surface, which led to a huge number of bad blocks and a fast drive failure. But this did not happen during normal operation, but only when the screw left the sleep mode, so for many this glitch remained unnoticed. And only after the permutation of the OS, if you forgot to turn off the "reduced power", the screw began to pour. This disease was so massive that it was called "grandmother's alarm clock" in the people - because of the characteristic metallic sound that issued the screw when the hooves were discarded. After each "awakening", the screw received a new portion of the badges, and attempts to do a remap were only helped as long as there was enough space in the defect table. Therefore, to save the survivors of the screws, Quantum released a patch. Unfortunately, it was already late - almost all the screws of this series were extinct earlier than in a year.

Similar problems were with the old Western Digital hard drives in 1995, but the badges appeared at the end of the disk. Quite often there is such a glitch: the screw simply ceases to be detected in the BIOS. The reason is the error of the programmer who wrote the firmware, as a result of which the screw itself spoils the service area: due to overflow of internal logs, mashing of neighboring areas occurs without which the HDD refuses to work. As a rule, this is preceded by any malfunctions, for example, the appearance of BAD-blocks or unsuccessful overclocking by a bus. This was the case with the IBM DTLA series: the error was hidden in SMART, and if it was enabled, the screw was dying. Similar problems were with Seagate, Fujitsu, and many others. Therefore, you need to monitor the release of updates for your hard drive, and regularly "overhaul" them. Unlike the firmware BIOS of motherboards, it is necessary to do this - if the firm released the firmware, it's no accident: perhaps, a serious bug has been found, the elimination of which will save troubles in the future.

Among many users there is still a rumor that some screws die from "improper" low-level formatting, for example, a program built into the BIOS of motherboards. While it was not possible to find sufficient evidence for this, however, there was a model of a screw with a hole in the microcode capable of leading to a similar effect. This is the Fujitsu TAU series (circa 1996), which incorrectly processes the ATA command 50h: it is the BIOS that produces the universal formatting, and this command is included in many programs for a la HddSpeed. So do not tempt fate by formatting these screws with little-known utilities or from the BIOS.

Many old screws with incorrect formatting acquired an uneven reading schedule. You can fix it by performing a disk zeroing in MHDD.

Another version of the program that can only be used by manufacturers is DMA mode switches: the changes between UDMA33 / 66/100 are a change in the part of the microcode of the screw, so trying to use someone else's utility can lead to damage to the firmware, and consequently to glitches with unpredictable consequences.

That's all. Let's hope that this material has helped you. But remember: any number of BAD-blocks on the screw is an excuse for applying for a guarantee. And the inability to remove them without degrading the characteristics of the disc is an occasion for the exchange of the device. And if you manage to convince the seller of this - consider that the concealment of BAD-sectors was 100% successful. Just do not forget about prevention and, perhaps, you will not need to hide anything.

Run our program and see the following window:

Select the "regeneration" menu item "start process under windows" (start the process from under windows). To start scanning bad sectors or bad blocks, we first need to "explain" to the program what exactly we want to do.

In the next window, we need to select the hard drive for scanning. In our case, it is one, select it and click on the "start process".

We continue. In the next window we will be asked to specify the option of scanning the disk. I advise you to immediately select the first option "scan and repair" (scan and restore). Simply enter the number "1" from the keyboard, as shown in the screenshot.

And in the last window before scanning the bad sectors ourselves, we are "asked" from which sector to begin scanning? I advise you to leave the number "0". This will ensure that the entire disk is scanned.

Press the "Enter" key and start scanning to the bad blocks. We'll look at the whole process on an example of a disk that contains bad sectors. Pay attention to the screenshot below, on it we see the progress of scanning (white bar) and on it - the three sectors found by the program, bad sectors.

Let's take a look at this screenshot in more detail: in the upper right hand corner we see the time and time remaining before the completion of the process from the start of the blockade block scan. If there are bad sectors on the disk, the program marks them with the English letter "B" and immediately tries to "cure" it. If it succeeds, then the letter "B" appears on the letter "R", which indicates the successful "restoration" of the bad block. In the lower left corner we see statistics on the scanned megabytes, as well as the number of "B" and "cured" "R" bad sectors found.

Upon completion of the scan, we will see the following window:

Three program sectors identified by the program are indicated here, and on the right - statistics already familiar to us, in which it is said that all the found bad blocks are eliminated.

Also consider the fact that when detecting bad sectors it is highly desirable (immediately or - after a few days) to start the test procedure again. The fact is that (with a serious disc defect) bad blocks can reappear and their number can only increase.

This is what the following inscriptions tell us in the screenshot above: "4 new bad sectors appear" and "18 bad sectors appear" are the newly appeared bad areas on the hard drive detected during the second scan. Such a disk can still be used very successfully for some time as an additional one and it is possible to store various (not very necessary) information and temporary files on it. But here's how a reliable data carrier or - a system drive it does not fit us unambiguously!

Actually, the whole simple testing process, I described you :) The very logic of what happens "behind the scenes" of the work of any program to restore bad sectors, we have discussed in the previous article, which is called "".

In addition to what has been said, I would like to mention another very useful feature of the "HDD Regenerator" program. It can burn its boot image to a CD.

Why is this necessary? Imagine the situation: you have problems with the hard drive (God forbid! :)) and the operating system because of this simply does not load. How do we run our program to scan bad sectors of the hard drive? In this case, the function of creating a bootable version of the program comes to our aid.

Let's examine this possibility. At the very beginning after starting the program in the menu "regeneration" select the item "create bootable CD / DVD" (create bootable CD or DVD disc).

In the next window, select our recording device installed in the system.

Press the "OK" button, insert a blank disc into the device and go to the last window just before the disc is recorded. Here we are asked to choose the recording speed. Select and press the "Burn CD" button (burn CD).

After the end of the recording, take our (now boot disk) with the program "HDD Regrenerator", insert it into the computer on which we want to test for bad sectors. We put it on the boot from CD and see the menu in which the program shows us the hard disks of the computer it finds.

As you can see, we have them - two. We choose (for example) the second one (we enter the number "2" from the keyboard) and press "enter". Then we see the following window.

There are several options for scanning the hard drive for bad sectors:

1. Scan, but do not fix the found bad blocks
2. Scan with correction of such sectors
3. View information about the program itself

We enter the number "2" from the keyboard (choose the second option). We see here such a window.

Here we indicate that we will scan immediately with the recovery of the bad sectors. Press the number "1", then - "enter" and the already started familiar testing process will start.

Also keep in mind the following point: a poor power supply (failures caused) or the use of various adapters may be the reason that the recovery program will signal the detection of a large number of bad sectors.

There were such cases in my practice. The SATA hard drive was connected via the molex to sata adapter:

The diagnostic program found a lot of bad blocks on it, but as soon as we put the corresponding one (which had Sata power connectors), the problem disappeared. So hard to remember - any adapters are - forced evil and if you can not do without them - get rid of them immediately!

That's all I wanted to tell you today about how to search and fix bad sectors on the disk. At the end of the article, as agreed, I give a link to the program "". Download, use.