Data migration method for physical damage of wafer storage unit caused by high-energy particles

Abstract

The invention relates to a data migration method for wafer storage unit physical damage caused by high-energy particles, which belongs to the technical field of data processing, and comprises the following steps: monitoring a storage unit in real time to judge whether wafer physical damage caused by high-energy particle impact exists or not; if the physical damage exists, determining a damage radiation area by taking a damage point as a center; data, meeting a preset migration condition, in the damaged radiation area serves as data to be migrated, the data to be migrated are migrated to a safe area of the storage unit, and the safe area at least meets the condition that the safe area is not overlapped with any damaged radiation area determined before the current time. The method has the advantage that the active data preservation effect on the damaged data is realized aiming at the physical damage of the wafer storage unit caused by the high-energy particles.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a data migration method for physical damage of wafer storage units caused by high-energy particles. BACKGROUND

[0002] With the continuous advancement of semiconductor manufacturing processes, the physical size of storage units continues to shrink, and the integration density per unit area is increasingly high. While this trend improves storage density and performance, it also makes storage chips, especially those used in aerospace, satellite communication, high-altitude computing, and other fields, more sensitive to space high-energy particles (such as cosmic rays, high-energy protons, etc.).

[0003] When high-energy particles bombard wafer storage units, the huge energy they carry will directly penetrate the unit structure, forming a small physical damage point (usually 10-100 nanometers in diameter). This damage point not only causes the hit storage unit to fail immediately, but more seriously, the surrounding area will experience long-term physical and chemical effects such as charge leakage, material defect expansion, electrochemical migration, or local oxidation, resulting in progressive data corrosion. This corrosion process is dynamic and slow (over hours to weeks), and it manifests as a center of damage points, with the surrounding normal storage units experiencing data errors or complete failure one bit at a time.

[0004] Existing technologies mainly rely on error correction code (ECC) technology to deal with soft errors of the storage (i.e., random bit flips, a kind of transient error). However, the error correction capability of ECC is limited (usually only a few bits of error can be corrected per 512 bytes of data). For the multi-bit error or even regional data collapse caused by the physical damage described above, the traditional ECC scheme will quickly fail due to the number of error bits exceeding its error correction capacity, ultimately resulting in permanent data loss. Therefore, there is an urgent need for a mechanism that can actively intervene before ECC fails after physical damage occurs to preserve data SUMMARY In order to address the physical damage of wafer storage units caused by high-energy particles and achieve the active data preservation effect of damaged data, the present application provides a data migration method for physical damage of wafer storage units caused by high-energy particles.

[0005] In a first aspect, the present application provides a data migration method for physical damage of wafer storage units caused by high-energy particles, comprising: real-time monitoring of the storage unit to determine whether there is physical damage of the wafer caused by high-energy particle impact; if there is physical damage, then determining the damage radiation area centered on the damage point; The data in the damage radiation region satisfying the preset migration condition is taken as to-be-migrated data, and the to-be-migrated data is migrated to a safe region of the storage unit, and the safe region at least satisfies: being mutually non-overlapped with any damage radiation region determined before the current time.

[0006] By adopting the technical scheme, the traditional ECC passive error correction mode is changed, after detecting physical damage, the data is actively migrated to a safe region before the data is completely disabled, the transition from "error correction" to "damage prevention" is realized, the long-term reliability of the data in a high radiation environment is greatly improved, through the "migration condition" judgment, the blind and unnecessary migration of invalid data or damaged data in the entire large region is avoided, and the valuable bandwidth, power consumption and storage space are saved.

[0007] Optionally, the preset migration condition at least includes: condition one: the to-be-migrated data is valid data; and condition two: the valid data is in an ECC correctable range. The to-be-migrated data is migrated to the safe region of the storage unit, and the safe region at least satisfies: being mutually non-overlapped with any damage radiation region determined before the current time. A migration queue is generated, all valid data in the damage radiation region is read, error correction state metadata is obtained through ECC error correction, and the valid data and the corresponding error correction state metadata are combined as a data packet and added to the migration queue. The data packets in the migration queue are read one by one, whether the corresponding data packet is in the ECC correctable range is determined based on the error correction state metadata, if yes, the data packet is migrated to the safe region of the storage unit, and if not, the data packet is discarded.

[0008] By adopting the technical scheme, since the material properties of the peripheral region of the data hit by the high-energy particles will continuously deteriorate (such as oxide layer trap charge accumulation, leakage current increase, metal electromigration, etc.), the corresponding data corrosion is also progressive. If the traditional serial process of "reading - ECC checking to determine whether it is within the correction range - migration if it is within the correction range" is executed, when the Nth data page is determined to meet the condition and is preparing to be migrated, the N+1th data page may be corroding and the number of error bits is increasing. In this regard, the application proposes to use the migration queue as an intermediate temporary transition piece to realize parallel processing of the "reading" operation and the "determining whether it can be migrated to the safe area" operation. All valid data is read into the migration queue (data backup is realized to avoid continuous corrosion of the damaged points), and then each data in the migration queue is read to determine whether it can be corrected. That is, the "state" of the data is effectively frozen in the migration queue at the reading time. Even if the source storage unit is completely damaged immediately after the reading time, the migration queue still stores the correctable state copy of the source storage unit at the reading time, ensuring the success rate of migration.

[0009] Optionally, the migrating the data packet to the safe area of the storage unit comprises: According to the value level of all data stored in the data packet, the data of different value levels is stored in different storage positions in the safe area, and the data of high value level is stored in a position away from the safe area boundary by a distance greater than a preset distance, and the data of low value level is stored in a position away from the safe area boundary by a distance less than the preset distance; wherein the preset distance refers to a distance value calculated based on the material diffusion speed of the storage unit. If the current determined safe area and the historical period, the safe area storing data have a physical overlapping position, the data stored in the physical overlapping position in the historical period and the data to be stored in the physical overlapping position at present are taken as target data, the target data is encoded with error correction code to generate N data fragments; the plurality of data fragments are stored in different storage positions in the current safe area, so that any K uncorrupted data fragments can restore the target data; N>K.

[0010] By adopting the technical scheme, when data is written into the safe area, high-value data is consciously stored in the center position of the safe area, and the distribution dispersion degree of the high-value data is ensured, so that even if the spread range of a certain damage point accidentally reaches the edge of the safe area in the future, valuable buffer warning time can be provided for the high-value data in the core position. The operation of filling low-priority data in the physical edge of the safe area is considering that when the system runs for a long time and the available safe space is extremely tight, two safe areas are allowed to physically overlap, the physical overlapping position of the two safe areas is highly likely to be the low-value data stored by them respectively, and the low-value data (i.e. target data) located in the physical overlapping position is further encoded and stored in a dispersed manner, so that even if the data at the physical overlapping position is lost, the target data can be recovered by using the lost K data fragments, so as to reduce the risk of permanent loss of data at the physical overlapping position and improve the availability of physical storage space.

[0011] Optionally, the migrating the data packet to the safe area of the storage unit further includes: Before migrating the data packet to the safe area of the storage unit, performing data compression and deduplication processing on the data in the data packet to update the data packet.

[0012] By adopting the technical scheme, the overhead of the safe area and the data transmission amount of the migration process are greatly reduced, and the system cost and power consumption are directly reduced.

[0013] Optionally, the migrating the data packet to the safe area of the storage unit includes: Performing endogenous redundancy processing on the data packet to generate a reinforced data packet, and the reinforced data packet contains the payload of the original data packet before the endogenous redundancy processing and endogenous redundancy information, the endogenous redundancy information is configured to work with the payload to assist in error detection or error correction when the payload is damaged; so that the reinforced data packet has a higher probability of being within the ECC correctable range after being physically damaged compared with the original data packet.

[0014] By adopting the technical scheme, before the data packet is written into the safe area, the endogenous redundancy specially designed for it is injected, that is, the endogenous redundancy processing is implemented, and the purpose of the redundancy is to realize error detection and correction, and work with the main ECC to improve the success probability of the main ECC, thereby delaying the progress of the data packet being declared "uncorrectable".

[0015] Optionally, the endogenous redundancy processing on the data packet to generate a reinforced data packet includes: Splitting the data packet into a plurality of inner frames, calculating a local error correction code for each inner frame, appending the local error correction code after the corresponding inner frame to jointly form a reinforced data packet; Encoding the original data of the data packet using an anti-interference encoding algorithm to convert the original data into a code word sequence with inherent error correction capability; Randomizing the data packet and generating and storing mirror inverse code bits for key bits in the data bits thereof.

[0016] By adopting the above technical solutions, a plurality of inner frames are obtained by splitting the data packet, and a local error correction code is set to reduce the burden of the main ECC (i.e., the ECC protecting the entire data packet), so that the number of errors it faces is smaller, thereby significantly improving the success rate of error correction. The data packet is encoded using an anti-interference encoding algorithm, which greatly increases the damage intensity or range required to cause a complete data frame to fail, thereby effectively delaying the overall damage progress of the data packet. The randomization operation effectively delays the impact of some types of progressive corrosion (such as charge leakage) on data damage.

[0017] Optionally, the method further comprises: Whenever a damage radiation area or a safe area is determined, the relative positions of all damage radiation areas and safe areas determined before the current time are updated in a preset storage unit global map in real time, and the storage unit global map is displayed for monitoring personnel to know; wherein the storage unit global map is used to digitally represent the physical layout of the safe area and the damage radiation area on the wafer storage unit.

[0018] By adopting the above technical solutions, the microscopic and invisible internal physical state of the wafer storage unit is converted into a macroscopic visible map, realizing visual judgment of the fault, and using visual interaction to enable the monitoring personnel to intuitively see the damage hotspots, the remaining capacity of the safe area, and the overall healthy area of the storage unit, providing a basis for the monitoring personnel to make decisions later.

[0019] In a second aspect, the present application provides a data migration system for physical damage of wafer storage units caused by high-energy particles, comprising, A physical damage real-time monitoring module for real-time monitoring of the storage unit to determine whether there is a physical damage of the wafer caused by high-energy particle impact; A damage radiation area determination module for determining a damage radiation area centered on a damage point if there is a physical damage; The damage area data migration module is configured to: take data in the damage radiation area that meets a preset migration condition as to-be-migrated data, and migrate the to-be-migrated data to a safe area of the storage unit, wherein the safe area at least meets: being mutually non-overlapped with any damage radiation area determined before a current time.

[0020] In a third aspect, the present application provides a data migration device for physical damage of a wafer storage unit caused by high-energy particles, comprising a storage and a processor, wherein the storage stores a computer program capable of being loaded and executed by the processor to perform the method according to any one of the first aspect.

[0021] In a fourth aspect, the present application provides a computer readable storage medium storing a computer program capable of being loaded and executed by a processor to perform the method according to any one of the first aspect.

[0022] In summary, the present application has the following beneficial technical effects: The present application changes the traditional ECC passive error correction mode, and actively migrates data to a safe area before the data is completely disabled after detecting physical damage, realizes the leap from "error correction" to "damage prevention", greatly improves the long-term reliability of data in a high radiation environment, and avoids blind and unnecessary migration of invalid data or damaged data in the entire large area through "migration condition" judgment, thereby saving valuable bandwidth, power consumption and storage space. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can also be obtained by those skilled in the art without creative labor.

[0024] Figure 1 is a flowchart of the data migration method for physical damage of a wafer storage unit caused by high-energy particles disclosed by the embodiments of the present application.

[0025] Figure 2 is a structural block diagram of the data migration system for physical damage of a wafer storage unit caused by high-energy particles disclosed by the embodiments of the present application.

[0026] Label explanation: 201, physical damage real-time monitoring module; 202, damage radiation area determination module; 203, damage area data migration module. DETAILED DESCRIPTION

[0027] The following will be described in detail in combination with the accompanying drawings. Figures 1-2 The present application will be further described in detail.

[0028] Embodiments of the present application disclose a data migration method for physical damage of wafer storage units caused by high-energy particles (hereinafter referred to as a data migration method), and the execution subject of the data migration method is a data migration system for physical damage of wafer storage units caused by high-energy particles (hereinafter referred to as a data migration system). The following will be combined with Figure 1 The data migration system specifically executes the specific execution process of the data migration method.

[0029] S101, real-time monitoring of the storage unit is performed to determine whether there is physical damage of the wafer caused by high-energy particle impact; S102, if there is physical damage, a damage radiation area is determined with the damage point as the center; S103, data in the damage radiation area that meets a preset migration condition is taken as to-be-migrated data, and the to-be-migrated data is migrated to a safe area of the storage unit, and the safe area at least meets: any damage radiation area determined before the current time does not overlap with each other.

[0030] In implementation, the data migration system is used to monitor whether physical damage occurs by using ECC engine alarm analysis. Specifically, the error correction count and distribution of the ECC unit are monitored, and when the error correction times of one or more adjacent physical pages (Pages) or storage blocks (Blocks) sharply and abnormally increase in a short time (such as the number of error correction times in a specified time period exceeds a preset threshold value / the error correction time growth rate is higher than a preset growth rate), especially when the error bits present an aggregated distribution on the physical space (such as there are at least three error bits that meet: the distance between two adjacent error bits is less than a preset distance value), rather than a random distribution, it can be preliminarily determined that the region may have physical damage.

[0031] Exemplarily, the system (dedicated hardware circuit or firmware algorithm in the storage controller) will map each error bit to the physical layout coordinate system of the memory (based on the Page address, Wordline, Bitline and other information of the error bit), and a preset algorithm (such as a density-based DBSCAN or a distance-based clustering algorithm) will find an error cluster that is abnormally dense on the physical space: a cluster composed of dozens or even hundreds of error bits that are closely adjacent within microns. Once an error cluster is identified, the system can calculate the physical coordinates of the geometric center or the density center of the error cluster (such as located near Die 0, Plane 1, Block 200, Wordline 50), which is taken as the damage point.

[0032] After the damage point is located, a two-dimensional damage radiation area (e.g., a circular area with the damage point as the center and a preset length as the radius) can be established. Since the defect caused by the damage point will spread in the physical area over time, the damage radiation area should be a dynamic real-time changing area. To this end, the application further proposes that, on the basis of the initially established damage radiation area, the ECC clustering analysis is periodically used every specified monitoring period T (e.g., T = 10 minutes) to determine whether an error bit appears in the ring-shaped buffer zone (e.g., a ring-shaped buffer zone with a radius of the radius of the currently newly drawn damage radiation area + a specified distance (e.g., 0.5 μm)) defined on the periphery of the newly drawn damage radiation area and the number of error bits exceeds a preset threshold. If yes, the buffer zone is taken as a part of the currently newly drawn damage radiation area, and the damage radiation area is updated (i.e., the area amplification is realized). If no, the current damage radiation area remains unchanged.

[0033] The migration condition can include the following conditions: Condition one: only migrate the cells storing valid data Condition two: read and ECC check all valid data pages in the area. For those data pages whose current ECC can still be corrected (i.e., the number of error bits does not exceed the threshold), it is determined that the migration condition is met.

[0034] For all data pages meeting the migration condition, the storage controller starts the internal data migration process, sets different migration priorities according to the importance of the data, and sequentially performs data migration according to the migration priorities, reads data (which can be correctly repaired by relying on ECC at this time) from the source physical address in the radiation area, and then writes the data page to the target address in the pre-allocated safe backup storage block (i.e., the safe area), and satisfies that the safe area does not overlap with any damage radiation area before the current time and the ring-shaped buffer zone in the physical space.

[0035] After the migration is completed, the controller updates the address mapping table (Flash Translation Layer, FTL), points the logical address of the system to the new target physical address, and marks the original block at the source address as a bad block or unusable to prevent subsequent data from being written to the damage radiation area again.

[0036] Optionally, S103 specifically further includes the following steps: S1031, generating a migration queue, reading all valid data in the damage radiation area, obtaining error correction state metadata through ECC error correction, and merging the valid data and the corresponding error correction state metadata as a data packet and adding it to the migration queue; S1032, read the data packets in the migration queue one by one, determine whether the corresponding data packet is within the ECC correctable range based on the error correction state metadata, if yes, migrate the data packet to the safe area of the storage unit, if not, discard it.

[0037] In implementation, when judging the migration condition, the application proposes to open a separate and safe storage area (SRAM or protected DRAM) in the controller as a migration queue, which is physically independent of the vulnerable wafer storage unit. Then, the controller as a "producer" reads out all valid data in the damage radiation area and performs error correction by the ECC engine to generate a data packet containing the original data (i.e. the read valid data) and metadata (including the number of error bits and the error correction state), and sequentially stores the data packet into the migration queue. At the same time, another independent process or hardware unit as a "consumer" sequentially (the data packet is stored at the tail and taken out at the head) takes out the data packet from the migration queue, checks the metadata in the data packet, and if the metadata shows "successful reading and the number of error bits is within the ECC tolerance", it means that the valid data is within the ECC correctable range, and it can be stored in the safe area. If the metadata shows "reading failure or the number of error bits exceeds the ECC tolerance", it means that at the moment when the "producer" reads the data, it has been damaged to an extent that cannot be recovered, at this time, the data packet is directly discarded without migration operation, and its original physical address is marked as completely damaged.

[0038] Optionally, S1032 further includes the following steps: Before migrating the data packet to the safe area of the storage unit, the data in the data packet is compressed and deleted to update the data packet.

[0039] According to the value level of all data stored in the data packet, the data of different value levels is stored in different storage positions in the safe area, and it satisfies that the data of high value level is stored in a position with a distance greater than a preset distance from the boundary of the safe area, and the data of low value level is stored in a position with a distance less than the preset distance from the boundary of the safe area; wherein the preset distance refers to a distance value calculated based on the material diffusion speed of the storage unit. If the current determined safe area and the historical period, the safe area storing data have a physical overlapping position, the data stored in the physical overlapping position in the historical period and the data to be stored in the physical overlapping position at present are taken as target data, the target data is subjected to error correction code encoding to generate N data fragments; the plurality of data fragments are stored in different storage positions in the current safe area, so that any K undamaged data fragments can restore the target data; N>K.

[0040] In implementation, for the data packet meeting the preset migration condition and needing to be migrated to the safe area, the data in the data packet is first compressed (such as using LZ4, Zstd or a self-defined algorithm to perform real-time lossless compression on the data page), and a repeated data deletion process (such as using a hash calculation unit built in the controller to calculate a unique hash value for the data packet data, and each time a hash value is calculated, it is queried from a preset duplicate checking table (such as a content addressable memory (CAM) or a hash table) whether the current hash value has been stored in the duplicate checking table, if not, the current hash value is stored in the duplicate checking table, the duplicate checking table is updated, and the data packet is defined as a data packet to be written to the safe area, and then the subsequent data packet is continuously calculated and queried, if it already exists, it means that there is a data packet to be written with the same data content before the current time, at this time, the current data packet is directly discarded, and the duplicate checking is realized.

[0041] After the compression and repeated data deletion are completed, for all the data packets to be written which are not discarded, the value level of each data packet is defined (such as L1: core metadata, L2: code, L3: user data; L4: cache data), and for the selected safe area, it can be regarded as a micro geographical space, according to the physical material of the safe area after being damaged, the data damage diffusion speed and the preset theoretical diffusion time (such as 10 minutes) to calculate the preset distance, the preset distance = diffusion speed * preset theoretical diffusion time.

[0042] According to the preset distance, the safe area is divided into a center part and an edge part, such as a center circular area with the center of the safe area as the center and the preset distance as the radius as the center part, and the remaining annular part in the safe area located outside the center part.

[0043] After the high-value data (L1, L2) is split into several data blocks, it is stored in the center part of the safe area, and it is ensured that in the center part, the distance between adjacent data blocks corresponding to the high-value data is maximized to realize the decentralized distribution.

[0044] After the writing of the completely high-value data (L1, L2) is completed, before the writing of the low-value data (L3, L4) is completed, it is determined whether there is a safe area which has a physical overlapping position with the currently determined safe area in all the safe areas determined before the current time. It should be noted that the present application defaults that whenever a safe area is generated, the safe area and the safe area in the historical period can only have two situations: 1. There is no physical overlapping position; 2. Only the edge parts of each other have a physical overlapping position to form a physical overlapping position.

[0045] If there is a security region in the history period (hereinafter referred to as a history region) that physically overlaps with the current security region, the data stored in the history region at the physically overlapping position is taken as first data, and the data to be stored in the current security region at the physically overlapping position is taken as second data. The first data and the second data are input into a correction code encoder to generate N data fragments, such as F1, F2, F3, and F4. The data fragments are stored in different positions in the edge part of the current security region (which can be the physically overlapping position or a non-physically overlapping position in the edge part), and at least K (K=2) data fragments are ensured not to be stored in the physically overlapping position. The N data fragments satisfy that adjacent data fragments maintain a safe distance (not less than a specified distance) from each other.

[0046] Optionally, S1032 further includes the following steps: The data packet is subjected to endogenous redundancy processing to generate a reinforced data packet, and the reinforced data packet contains the payload of the original data packet before the endogenous redundancy processing and endogenous redundancy information configured to work with the payload to assist in error detection or error correction when the payload is damaged; so that the reinforced data packet has a higher probability of being within the ECC correctable range after being physically damaged compared to the original data packet. The endogenous redundancy processing of the data packet to generate the reinforced data packet specifically includes the following steps: The data packet is divided into multiple inner frames, a local error correction code is calculated for each inner frame, and the local error correction code is attached after the corresponding inner frame to collectively form the reinforced data packet. The original data of the data packet is encoded using an anti-interference encoding algorithm to convert the original data into a code word sequence with inherent error correction capability. The data packet is subjected to randomization operation, and mirror inverse code bits are generated and stored for key bits in the data bits.

[0047] In implementation, the present application proposes three schemes for realizing endogenous redundancy processing: Scheme one: custom inner frame ECC. Specifically, the original data to be migrated (e.g. 4K G) is divided into multiple smaller inner frames (e.g. 64 bytes per frame), a very strong local ECC check code (e.g. a 16 byte CRC or a BCH code) is calculated for each inner frame. The original data frame + local ECC is packed, then protected by the whole Page level main ECC, and finally written into the secure area. When the data packet is damaged, it may only affect a few inner frames. When reading the data packet, the local ECC of the inner frame can be used to attempt to repair the error of the frame. If the repair is successful, the number of data error bits submitted to the whole Page level main ECC is greatly reduced or even zero. This is equivalent to decomposing a big problem (multiple bit errors in the whole Page) into several small problems (several individual frame errors), and solving them locally first, which significantly improves the probability of successful error correction of the main ECC, thereby delaying the progress of declaring the data packet "uncorrectable". If the repair is not successful, the local ECC will mark the inner frame as an error, but the original data of the damaged frame will still be retained. All inner frames (including the marked damaged frames and the intact frames) are assembled and submitted to the next level, the main ECC (e.g. LDPC code) that protects the whole data packet for decoding. The main ECC algorithm can use the check relationship of other intact frames in the data packet to attempt to correct the error in the damaged frame.

[0048] Scheme two: interference-resistant coding. Specifically, interference-resistant coding algorithms are used, such as using Hamming code to encode the data packet data into longer code words with error correction capability. For example, 4-bit data is encoded into a 7-bit Hamming code word. When writing the data packet, ensure that the same inner frame or each bit of the same Hamming code word is physically stored separately (e.g. not on the same Bitline or in the same floating gate transistor group), which greatly increases the damage intensity or range required to cause a complete data frame to fail, thereby effectively delaying the overall damage progress of the data packet Scheme three: data randomization and mirroring for protection. Specifically, before writing, the data packet is randomized (scrambled) to avoid long strings of 0s or 1s, which are more susceptible to certain types of progressive corrosion (e.g. charge leakage). For the most critical data bits (e.g. the highest bits of the pointer), write their inverse code (mirror bits) and store them in a physically slightly distant location. In this way, randomization makes the charge level more evenly distributed, which may slow down the electrochemical migration of specific patterns. When reading, if the values of the original code bit and the mirror bit are no longer complementary, an error can be immediately detected. Combined with other information (such as the main ECC), it may even be possible to infer which bit is more likely to be correct, thereby discovering and possibly correcting it in advance before the number of error bits exceeds the ECC capacity.

[0049] Optionally, the data migration method further comprises the following steps: Whenever it is determined that a damage radiation area is generated or a safe area is generated, the relative positions of all damage radiation areas and safe areas determined before the current time in the preset storage unit global map are updated in real time, and the storage unit global map is displayed for the monitoring personnel to know; wherein, the storage unit global map is used to digitally represent the physical layout of the safe area and the damage radiation area on the wafer storage unit.

[0050] In implementation, the storage unit global map refers to a digital twin model of the column maintained in the controller memory or a dedicated cache, which is a multi-dimensional matrix whose coordinates correspond to the physical structure (Die, Plane, Block, Page) of the wafer storage unit. The damage radiation area, the historical safe area and the current safe area can be marked and distinguished by different colors in the storage unit global map, so as to visually display the hot spot of damage occurrence, the remaining capacity of the safe area and the overall healthy area of the storage medium to the detection personnel.

[0051] The embodiment of the application also discloses a data migration system for wafer storage unit physical damage caused by high-energy particles. Referring to Figure 2 , comprising: A physical damage real-time monitoring module 201 is configured to monitor the storage unit in real time to determine whether there is a wafer physical damage caused by high-energy particle impact; A damage radiation area determination module 202 is configured to determine a damage radiation area with the damage point as the center if there is a physical damage; A damage area data migration module 203 is configured to migrate data in the damage radiation area that meets a preset migration condition as to-be-migrated data, and migrate the to-be-migrated data to a safe area of the storage unit, and the safe area at least meets: not overlapping with any damage radiation area determined before the current time.

[0052] Optionally, the damage area data migration module 203 is further configured to generate a migration queue, read all valid data in the damage radiation area, obtain error correction state metadata through ECC error correction, combine the valid data and the corresponding error correction state metadata as a data packet and add the data packet to the migration queue; read the data packets in the migration queue one by one, determine whether the corresponding data packet is within the ECC correctable range based on the error correction state metadata, if yes, migrate the data packet to the safe area of the storage unit; if not, discard it.

[0053] Optionally, the damaged area data migration module 203 is further configured to store data of different value levels in different storage locations in the safe area according to the value levels of all data stored in the data packet, and to satisfy: data of a high value level is stored in a location that is greater than a preset distance from the boundary of the safe area, and data of a low value level is stored in a location that is less than the preset distance from the boundary of the safe area; wherein the preset distance refers to a distance value calculated based on the material diffusion speed of the storage unit. The damaged area data migration module 203 is further configured to, if the currently determined safe area physically overlaps with a historical safe area in which data is stored, store data that has been stored in the physically overlapping location in the historical safe area and data that is currently to be stored in the physically overlapping location as target data, perform error correction code encoding on the target data to generate N data fragments, and store the plurality of data fragments in different storage locations in the current safe area, so that any K undamaged data fragments can recover the target data; N > K.

[0054] Optionally, the damaged area data migration module 203 is further configured to perform data compression and duplicate data deletion processing on the data in the data packet before migrating the data packet to the safe area of the storage unit, to update the data packet.

[0055] Optionally, the damaged area data migration module 203 is further configured to perform endogenous redundancy processing on the data packet to generate a strengthened data packet, and the strengthened data packet includes the payload of the original data packet before the endogenous redundancy processing and endogenous redundancy information, the endogenous redundancy information being configured to work with the payload to assist in error detection or error correction when the payload is damaged; so that the strengthened data packet has a higher probability of being within the ECC correctable range after being physically damaged compared to the original data packet.

[0056] Optionally, the damaged area data migration module 203 is further configured to divide the data packet into a plurality of inner frames, calculate a local error correction code for each inner frame, and append the local error correction code after the corresponding inner frame to jointly form a strengthened data packet; further configured to use an anti-interference encoding algorithm to encode the original data of the data packet, and convert the original data into a code word sequence with inherent error correction capability; further configured to perform a randomization operation on the data packet and generate and store mirror inverse code bits for key bits in the data bits.

[0057] Optionally, the damaged area data migration module 203 further comprises a global area distribution display module configured to, whenever a damaged radiation area is determined or a safe area is generated, update the relative positions of all damaged radiation areas and safe areas determined before the current time in a preset global map of the storage unit in real time, and display the global map of the storage unit for monitoring personnel to learn; wherein the global map of the storage unit is used to digitally represent the physical layout of the safe area and the damaged radiation area on the wafer storage unit.

[0058] The embodiment of the present application further discloses a data migration device for physical damage of a wafer storage unit caused by high-energy particles, which comprises a memory and a processor, and the memory stores a computer program capable of being loaded and executed by the processor to perform the data migration method for physical damage of a wafer storage unit caused by high-energy particles as described above.

[0059] The embodiment of the present application further discloses a computer readable storage medium which stores a computer program capable of being loaded and executed by the processor to perform the data migration method for physical damage of a wafer storage unit caused by high-energy particles as described above, and the computer readable storage medium comprises various storage media capable of storing program codes, for example, a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0060] It should be noted that, in the present document, the relational terms such as first and second and the like are used only to differentiate one entity or operation from another, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations.

[0061] The above embodiments are only used to illustrate the technical solutions of the present application, but not limit the protection scope of the application. Obviously, the described embodiments are only part of the embodiments of the present application, but not all embodiments. Based on these embodiments, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the protection scope of the present application.

Claims

1. A method for data migration against physical damage of a wafer memory cell caused by a high-energy particle, characterized by, The application relates to a method for monitoring and repairing a storage unit, comprising the following steps: monitoring the storage unit in real time to determine whether there is physical damage of a wafer caused by high-energy particle impact; if there is physical damage, determining a damage radiation area centered on the damage point; if the data in the damage radiation area meets preset migration conditions, the data is regarded as to-be-migrated data, and the to-be-migrated data is migrated to a safe area of the storage unit, and the safe area at least meets the condition of not overlapping with any damage radiation area determined before the current time.

2. The method of claim 1, wherein the data migration is performed in response to a determination that the memory cell is susceptible to physical damage from high-energy particles. The preset migration conditions at least include condition one: the to-be-migrated data is valid data; and condition two: the valid data is within an ECC correctable range. The method comprises the following steps: generating a migration queue, reading all valid data in the damage radiation area, obtaining error correction state metadata through ECC error correction, and adding the valid data and the corresponding error correction state metadata into the migration queue as a data packet; reading the data packets in the migration queue one by one, determining whether the corresponding data packet is within the ECC correctable range based on the error correction state metadata, if yes, migrating the data packet to the safe area of the storage unit, and if no, discarding the data packet.

3. The method of claim 2, wherein the data migration is performed in response to a determination that the memory cell is susceptible to physical damage from high-energy particles. The method comprises the following steps: storing data of different value levels in different storage positions in the safe area according to the value levels of all data stored in the data packet, and meeting the condition that data of a high value level is stored in a position away from the safe area boundary by a distance greater than a preset distance, and data of a low value level is stored in a position away from the safe area boundary by a distance less than the preset distance; wherein the preset distance is a distance value calculated based on the material diffusion speed of the storage unit; if the currently determined safe area physically overlaps with a historical safe area storing data, regarding the data stored in the physically overlapping position in the historical period and the data to be stored in the physically overlapping position as target data, performing error correction code encoding on the target data to generate N data fragments; and dispersing and storing the plurality of data fragments in different storage positions in the current safe area, so that any K undamaged data fragments can restore the target data; N>K.

4. The method of claim 3, wherein the data migration is performed in response to a determination that the memory cell is susceptible to physical damage from high-energy particles. The method further comprises the following steps: before migrating the data packet to the safe area of the storage unit, performing data compression and repeated data deletion processing on the data in the data packet to update the data packet.

5. The method of claim 2, wherein the data migration is performed in response to a determination that the memory cell is susceptible to physical damage from high-energy particles. The method comprises the following steps: The data packet is subjected to endogenous redundancy processing to generate a reinforced data packet, and the reinforced data packet comprises a payload of the original data packet before the endogenous redundancy processing, and endogenous redundancy information configured to work with the payload to assist in error detection or error correction when the payload is damaged; so that the reinforced data packet has a higher probability of being within the ECC correctable range after being physically damaged compared to the original data packet.

6. The data migration method for wafer storage unit physical damage caused by high-energy particles according to claim 5, characterized in that, The endogenous redundancy processing of the data packet to generate a reinforced data packet comprises: segmenting the data packet into multiple inner frames, calculating a local error correction code for each inner frame, and appending the local error correction code after the corresponding inner frame to collectively form a reinforced data packet; encoding the original data of the data packet using an anti-interference encoding algorithm to convert the original data into a code word sequence with inherent error correction capability; performing a randomization operation on the data packet and generating and storing mirror inverse code bits for key bits in the data bits.

7. The data migration method for wafer memory cells caused by physical damage due to high-energy particles according to claim 1, characterized in that, The method further comprises: Whenever a damaged radiation area or a safe area is determined, the relative positions of all damaged radiation areas and safe areas determined before the current time are updated in real time in a preset storage unit global map, and the storage unit global map is displayed for monitoring personnel to know; wherein the storage unit global map is used to digitally represent the physical layout of the safe area and the damaged radiation area on the wafer storage unit.

8. A data migration system against physical damage of a wafer memory cell caused by a high-energy particle, characterized by, including, a physical damage real-time monitoring module (201) for real-time monitoring of the storage unit to determine whether there is a wafer physical damage caused by high-energy particle impact; a damaged radiation area determination module (202) for determining a damaged radiation area centered on a damage point if there is a physical damage; a damaged area data migration module (203) for migrating data in the damaged radiation area that meets a preset migration condition as to-be-migrated data to a safe area of the storage unit, and the safe area at least meets: does not overlap with any damaged radiation area determined before the current time.

9. A data migration apparatus against physical damage of a wafer memory cell caused by a high-energy particle, characterized by, a memory and a processor, the memory storing a computer program capable of being loaded and executed by the processor to perform the method of any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, a computer program capable of being loaded and executed by the processor to perform the method of any one of claims 1 to 7.

Images