CN106874161A - A kind of abnormal processing method and processing device of cache - Google Patents

A kind of abnormal processing method and processing device of cache Download PDF

Info

Publication number
CN106874161A
CN106874161A CN201710052784.6A CN201710052784A CN106874161A CN 106874161 A CN106874161 A CN 106874161A CN 201710052784 A CN201710052784 A CN 201710052784A CN 106874161 A CN106874161 A CN 106874161A
Authority
CN
China
Prior art keywords
cache
abnormal
way
register
chip
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN201710052784.6A
Other languages
Chinese (zh)
Other versions
CN106874161B (en
Inventor
迟君涛
肖冰
海秀云
赵云鹏
赵瑞哲
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
New H3C Technologies Co Ltd
Original Assignee
New H3C Technologies Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by New H3C Technologies Co Ltd filed Critical New H3C Technologies Co Ltd
Priority to CN201710052784.6A priority Critical patent/CN106874161B/en
Publication of CN106874161A publication Critical patent/CN106874161A/en
Application granted granted Critical
Publication of CN106874161B publication Critical patent/CN106874161B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00Error detection; Error correction; Monitoring
    • G06F11/30Monitoring
    • G06F11/3003Monitoring arrangements specially adapted to the computing system or computing system component being monitored
    • G06F11/3037Monitoring arrangements specially adapted to the computing system or computing system component being monitored where the computing system component is a memory, e.g. virtual memory, cache
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F11/00Error detection; Error correction; Monitoring
    • G06F11/30Monitoring
    • G06F11/3055Monitoring arrangements for monitoring the status of the computing system or of the computing system component, e.g. monitoring if the computing system is on, off, available, not available

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Computing Systems (AREA)
  • Physics & Mathematics (AREA)
  • Quality & Reliability (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Memory System Of A Hierarchy Structure (AREA)
  • Techniques For Improving Reliability Of Storages (AREA)

Abstract

A kind of abnormal processing method and processing device of cache is the embodiment of the invention provides, the method includes:Cache exception is detected, the register of on-chip cache and the register of second level cache is read;According to the register and the register of second level cache of the on-chip cache for reading, identification produces the abnormal abnormal cache line of cache;It is determined that the cache way belonging to abnormal cache line is located at on-chip cache, the register of the cache line state according to record on-chip cache judges whether on-chip cache also has available cache way;If it is, in the register of the cache line state of record on-chip cache, the cache way belonging to recording exceptional cache line is unavailable, the cache way belonging to abnormal cache line is closed;If not, execution is restarted.Using the embodiment of the present invention, the stability of communication equipment is improve, reduce the influence to business.

Description

A kind of abnormal processing method and processing device of cache
Technical field
The present invention relates to field of computer technology, more particularly to a kind of cache Cache abnormal processing method and Device.
Background technology
At present, the access speed more and more higher of central processing unit (CPU, Central Processing Unit), and internal memory Access speed improvement it is slow, this causes CPU to become CPU from the data that internal memory is obtained in instruction or read/write memory to access speed The bottleneck of degree.Based on the situation, Cache (cache) arises at the historic moment.
Cache is a kind of special memory sub-system, and it is located between CPU and internal memory.Usually, Cache points is one Level (L1, level 1) Cache and two grade of (L2, level 2) Cache.L1 Cache are integrated in inside CPU, including multiple Cache way (cache way), each Cache way include multiple Cache line (cache line).L2 Cache collection Into on mainboard, including multiple Bank (memory cell);Wherein each Bank includes multiple Cache way, each Cache way Including multiple Cache line.Memory capacity all very littles of L1 Cache and L2 Cache, but access speed is quickly, CPU can be with Instruction or read-write data are rapidly obtained from L1 Cache and L2 Cache, the access speed of CPU is drastically increased.But Cache's is relatively costly, and memory capacity very little, and this causes to be particularly important the abnormal treatment of Cache.
In the prior art, communication equipment detect Cache it is abnormal when, read the register and L2Cache of L1 Cache Register, the abnormal information of record L1 Cache and/or the Cache abnormal informations of L2 Cache, communication equipment is performed restarts.But Problem caused by this Cache abnormal processing mode is that the Cache of communication equipment is accidental sexual abnormality extremely, it is impossible to Determine when communication equipment can produce unstable to the communication of communication equipment operation again because Cache abnormal restartings Influence, causes network data to be interrupted, traffic affecting treatment.
The content of the invention
The purpose of the embodiment of the present invention is to provide a kind of cache abnormal processing method and processing device, to improve communication The stability of equipment, reduces the influence to business.Concrete technical scheme is as follows:
On the one hand, the embodiment of the invention discloses the processing method that a kind of cache is abnormal, methods described includes:
Cache exception is detected, the register of on-chip cache and the register of second level cache is read;
According to the register and the register of the second level cache of the on-chip cache for reading, identification is produced The abnormal abnormal cache line of the cache;
Determine that the cache way belonging to the abnormal cache line is located at the on-chip cache, according to record institute The register of the cache line state of on-chip cache is stated, judges whether the on-chip cache also has available high speed Cache way;
If it is, in the register of the cache line state of the record on-chip cache, recording the exception Cache way belonging to cache line is unavailable, closes the cache way belonging to the abnormal cache line;
If not, execution is restarted.
On the other hand, the embodiment of the invention discloses the processing unit that a kind of cache is abnormal, described device includes:
Detection module, for detecting cache exception;
Read module, for when detecting cache and being abnormal, read on-chip cache register and two grades high The register of speed caching;
Identification module, for according to the register of the on-chip cache and posting for the second level cache for reading Storage, identification produces the abnormal abnormal cache line of the cache;
Control module, for determining that it is slow at a high speed that the cache way belonging to the abnormal cache line is located at the one-level Deposit, according to the register of the cache line state for recording the on-chip cache, whether judge the on-chip cache Also available cache way;If it is, in the register of the cache line state of the record on-chip cache, The cache way recorded belonging to the abnormal cache line is unavailable, closes the high speed belonging to the abnormal cache line Cache way;If not, execution is restarted.
In the embodiment of the present invention, communication equipment detects Cache exceptions, L1 Cache or L2 Cache Bank also When having available Cache way, first there is the abnormal Cache way of Cache in record, turns off and the abnormal Cache of Cache occur Way, is continuing with the available Cache way in the Bank of L1 Cache or L2 Cache, reduce communication equipment due to The Cache abnormal number of times restarted, improves the stability of communication equipment, reduces the influence to message forwarding service.
Brief description of the drawings
In order to illustrate more clearly about the embodiment of the present invention or technical scheme of the prior art, below will be to embodiment or existing The accompanying drawing to be used needed for having technology description is briefly described, it should be apparent that, drawings in the following description are only this Some embodiments of invention, for those of ordinary skill in the art, on the premise of not paying creative work, can be with Other accompanying drawings are obtained according to these accompanying drawings.
Fig. 1 is a kind of schematic flow sheet of the abnormal processing methods of Cache provided in an embodiment of the present invention;
Fig. 2 is the schematic diagram of the register of the Cache way states of record L1 Cache in the embodiment of the present invention;
Fig. 3 is that the register of the Cache way states of record L1 Cache in Fig. 2 records showing for unavailable Cache way It is intended to;
Fig. 4 is that the register of the Cache way states of record L1 Cache in Fig. 2 and Fig. 3 records unavailable Cache way Schematic diagram;
Fig. 5 is the register of the Cache way states of the shared L1 Cache of two threads of record in the embodiment of the present invention Schematic diagram;
Fig. 6 records unavailable for the register of the Cache way states of the shared L1 Cache of two threads of record in Fig. 5 The schematic diagram of Cache way;
Fig. 7 is the schematic diagram of the register of the Cache way states of record L2 Cache in the embodiment of the present invention;
Fig. 8 is that the register of the Cache way states of record L2 Cache in Fig. 7 records showing for unavailable Cache way It is intended to;
Fig. 9 is that the register of the Cache way states of record L2 Cache in Fig. 7 and Fig. 8 records unavailable Cache way Schematic diagram;
Figure 10 is a kind of structural representation of the abnormal processing units of Cache provided in an embodiment of the present invention.
Specific embodiment
Below in conjunction with the accompanying drawing in the embodiment of the present invention, the technical scheme in the embodiment of the present invention is carried out clear, complete Site preparation is described, it is clear that described embodiment is only a part of embodiment of the invention, rather than whole embodiments.It is based on Embodiment in the present invention, it is every other that those of ordinary skill in the art are obtained under the premise of creative work is not made Embodiment, belongs to the scope of protection of the invention.
Below by specific embodiment, the present invention is described in detail.
With reference to Fig. 1, Fig. 1 is a kind of schematic flow sheet of the abnormal processing methods of Cache provided in an embodiment of the present invention.One As, Cache points is L1 Cache and L2 Cache.LCache includes that multiple Cache way, each Cache way include many Individual Cache line;L2 Cache include that multiple Bank, each Bank include that multiple Cache way, each Cache way include Multiple Cache line.Specifically, the method includes:
Step 101, detects cache exception, reads the register of on-chip cache and posting for second level cache Storage;
Step 102, according to the register and the register of second level cache of the on-chip cache for reading, identification is produced The abnormal abnormal cache line of cache;
Step 103, it is determined that the cache way belonging to abnormal cache line is located at on-chip cache;
Step 104, the register of the cache line state according to record on-chip cache, judges on-chip cache Whether available cache way is also had;If it is, performing step 105;If not, performing step 107;
Step 105, in the register of the cache line state of record on-chip cache, recording exceptional cache Cache way belonging to row is unavailable;
Step 106, closes the cache way belonging to abnormal cache line;
Step 107, execution is restarted.
Specifically, the CPU of communication equipment can be using ECC (Error Correcting Code, error checking and correction) In technology SECDED (Single Error Correction and Double Error Detection, single error correction, Many error detections) technology, be in each Cache line that each Cache way of detection L1 Cache and L2 Cache includes The no Bit (bit) that there is mistake, if in the presence of a Bit for mistake, being checked and being corrected;If in the presence of multiple mistakes Bit, it is determined that Cache line occur Cache exceptions, reads the register of L1 Cache and the register of L2 Cache, according to The information for reading can determine that it is to be located at L1 Cache that the abnormal Cache line of Cache occur, or positioned at L2 Cache;
L1 Cache are located in the event of Cache abnormal Cache line (namely exception Cache line), then really Surely there is the Cache way belonging to the abnormal Cache line of Cache, and this Cache way is set to unavailable.
For example:One storage size of Cache way is 4KB, and the storage size of a Cache line is 32byte, then a Cache way can be including this 128 Cache line of Cache line0-Cache line127, if inspection Measure in a Cache way that any Cache line have Cache exceptions in Cache line0-Cache line127, then This Cache way is set to unavailable.
By the Cache way occurred in L1 Cache belonging to the abnormal Cache line of Cache be set to it is unavailable it Before, can first detect in L1 Cache whether also there is available Cache way, if there is available Cache way, then will There is the Cache way belonging to the abnormal Cache line of Cache in L1 Cache and be set to unavailable, and then close L1 Disabled Cache way in Cache, available Cache way are led to during communication equipment can be continuing with L1 Cache Letter, processing business reduces the influence to business;If there is no available Cache way, no longer having in L1 Cache can With the Cache way of processing business, then perform the Cache belonging to the Cache line that Cache exceptions will occur in L1 Cache Way is set to unavailable, and the operation for closing disabled Way in L1 Cache will be without in all senses, now can be direct Restart communication equipment.
In one embodiment of the invention, for the ease of determining to whether there is available Way in current L1 Cache, can Whether be can use with the Cache way that L1 Cache are recorded by register.
Fig. 2 is the schematic diagram of the register of the Cache way states of record L1 Cache in the embodiment of the present invention.It is assumed that The L1 Cache of communication equipment have 8 Cache way.Posted in the Cache way states of record L1 Cache shown in Fig. 2 In storage, Bit 0-Bit 7 indicate whether the Cache way 0-Cache way 7 of L1 Cache can use;Record L1 Cache Cache way states register Bit 0-Bit 7 in an initial condition, each Bit value both is set to " 0 ", is used for The corresponding Cache way of each Bit are indicated to can use.
Fig. 3 is that the register of the Cache way states of record L1 Cache in Fig. 2 records showing for unavailable Cache way It is intended to;When communication equipment detects Cache exceptions, the register of L1 Cache and the register of L2 Cache are read, by L1 The instruction and data of Cache and L2 Cache storages writes internal memory, the register and L2 Cache according to the L1Cache for reading Register, detects and the Cache way 0 that the abnormal Cache line of Cache are located at L1 Cache occurs, and determination L1 There are available Cache way in Cache, to indicating Cache way in the register of the Cache way states for recording L1Cache 0 Bit 0 carries out OR operation, and the initial value of Bit 0 is revised as " 1 " from " 0 ", indicates the Cache way 0 of L1 Cache For unavailable;Communication equipment closes the Cache way 0 of L1 Cache.
In Fig. 3, communication equipment occur Cache it is abnormal when, the Cache way 0 of record L1 Cache it is unavailable it Before, first by L1 Cache and L2 Cache record instruction and data be written back in internal memory, it is ensured that the uniformity of data and Avoid Cache way 0 be in L1 Cache last available Cache way and caused by needing to restart storage number According to the loss with instruction.
Fig. 4 is that the register of the Cache way states of record L1 Cache in Fig. 2 and Fig. 3 records unavailable Cache way Schematic diagram;Communication equipment detects Cache exceptions again, reads the register of L1 Cache and the register of L2 Cache, The instruction and data that L1 Cache and L2 Cache are stored writes internal memory, according to the register and L2 of the L1 Cache for reading The register of Cache, detects and the Cache way 3 that the abnormal Cache line of Cache are located at L1 Cache occurs, and determination There are available Cache way in L1 Cache, to indicating Cache in the register of the Cache way states of record L1 Cache The Bit 3 of way 3 carries out OR operation, and the initial value of Bit 3 is revised as " 1 " from " 0 ", indicates the Cache of L1 Cache Way 3 is unavailable;Communication equipment closes the Cache way 3 of L1 Cache.
It is assumed that communication equipment has had been switched off 7 Cache way in L1 Cache, generation Cache is detected again different Normal Cache line are located at a Cache way of L1 Cache, the instruction sum that L1 Cache and L2 Cache are stored According to write-in internal memory after, determine that L1 Cache have not had available Cache way, then restarted, communication equipment need not again by The Bit of last unavailable Cache way is indicated to be set to not in the register of the Cache way states of record L1 Cache Can use.
In other embodiments of the invention, communication equipment can posting according to the Cache way states of record L1 Cache Numerical value is the number of the Bit of " 0 " in storage, determines whether L1 Cache also have available Cache way, or, communication equipment Can be the number of the Bit of " 1 " according to numerical value in the register of the Cache way states of record L1 Cache, determine L1 Cache Whether available Cache way are also had.For example, communication equipment is from the register of the Cache way states of record L1 Cache The numerical value of reading is 1 for the number of the Bit of " 0 ", it is determined that the current Cache line that Cache exceptions occur are located at L1 Cache unique available Cache way, L1Cache do not have available Cache way.Or, for example, communication equipment is from record The numerical value that the register of the Cache way states of L1 Cache reads for the Bit of " 1 " number for 7 (L1 Cache's The total number of Cache way subtracts 1), it is determined that the current Cache line that Cache exceptions occur uniquely can use positioned at L1 Cache Cache way, L1 Cache there is no available Cache way.
In one embodiment of the invention, the CPU of communication equipment contains multiple cores, i.e. multi-core CPU, multi-core CPU Each core can have multiple threads, and the storage of the Cache line of each Cache way of the shared L1 Cache of these threads is empty Between.In this case, it is determined that after the Cache way that the L1 Cache belonging to the abnormal Cache line of Cache occur, it is necessary to In the register of Cache way states of the shared L1 Cache of these threads is recorded, by each thread pair in L1 Cache Cache way belonging to the generation Cache that answers abnormal Cache line be all recorded as it is unavailable, by each line in L1 Cache The corresponding disabled Cache way of journey are closed.
Fig. 5 is the register of the Cache way states of the shared L1 Cache of two threads of record in the embodiment of the present invention Schematic diagram;It is assumed that one of core of CPU runs 2 threads, 8 Cache way of L1 Cache are shared.Shown in Fig. 5 The shared L1 Cache of two threads of record Cache way states register in, Bit 0-Bit 7 indicate thread 1 to be total to Whether 8 Cache way of the L1 Cache for enjoying can use, and Bit 8-Bit 15 indicate 8 of the shared L1 Cache of thread 2 Whether Cache way can use.
Fig. 6 records unavailable for the register of the Cache way states of the shared L1 Cache of two threads of record in Fig. 5 The schematic diagram of Cache way;Communication equipment detect Cache it is abnormal when, read the register and L2 Cache of L1 Cache Register, the instruction and data that L1 Cache and L2 Cache are stored writes internal memory, according to the deposit of the L1 Cache for reading The register of device and L2 Cache, detects and the Cache way that the abnormal Cache line of Cache are located at L1 Cache occurs 0, and determine there be available Cache way in L1 Cache, the Bit 0 of the Cache way 0 used instruction thread 1 is carried out OR operation, " 1 " is revised as by the initial value of Bit 0 from " 0 ", and the Bit 8 of the Cache way 0 to indicating thread 2 to use enters Row OR operation, " 1 " is revised as by the initial value of Bit 8 from " 0 ", so, the L1 Cache that record thread 1 and thread 2 are shared Cache way 0 for unavailable;Communication equipment closes the Cache way 0 of thread 1 and the shared L1 Cache of thread 2.
In other embodiments of the invention, when communication equipment detects Cache exceptions, the register of L1 Cache is read With the register of L2 Cache, according to the register and the register of L2 Cache of the L1 Cache for reading, generation is detected Cache way belonging to Cache abnormal Cache line are located at one of Bank of L2 Cache, if detecting generation Also there is available Cache way in the Bank where Cache way belonging to Cache abnormal Cache line, then close hair Cache way belonging to raw Cache abnormal Cache line, others can during communication equipment can be continuing with this Bank Cache way, reduce the influence to business;If this Bank does not have available Cache way, it is not necessary to perform pass again This operation that Cache way belonging to abnormal Cache line of Cache occur is closed, communication equipment can be directly restarted.
Fig. 7 is the schematic diagram of the register of the Cache way states of record L2 Cache in the embodiment of the present invention.It is assumed that The L2 Cache of communication equipment include 2 Bank:Bank 0 and Bank 1;Bank 0 and Bank 1 each includes 8 Cache way.In the register of the Cache way states of the record L2 Cache shown in Fig. 7, the initial value of Bit 0-Bit 7 can quilt " 0 " is set to, the Cache way 0-Cache way 7 for indicating Bank 0 can use, and the initial value of Bit 8-Bit 15 can It is arranged to " 0 ", the Cache way0-Cache way 7 for indicating Bank 1 can use.
Fig. 8 is that the register of the Cache way states of record L2 Cache in Fig. 7 records showing for unavailable Cache way It is intended to;When communication equipment detects Cache exceptions, the register of L1 Cache and the register of L2 Cache are read, by L1 The instruction and data of Cache and L2 Cache storages writes internal memory, the register and L2 Cache according to the L1Cache for reading Register, detects the Cache way 0 that the Bank 0 that the abnormal Cache line of Cache are located at L2Cache occurs, and determine There are available Cache way in the Bank 0 of L2 Cache, to the register middle finger of the Cache way states of record L2 Cache Showing the Bit 0 of the Cache way 0 of Bank 0 carries out OR operation, and the initial value of Bit 0 is revised as " 1 " from " 0 ", indicates L2 The Cache way 0 of the Bank 0 of Cache are unavailable;Communication equipment closes the Cache way 0 of the Bank 0 of L2 Cache.
Fig. 9 is that the register of the Cache way states of record L2 Cache in Fig. 7 and Fig. 8 records unavailable Cache way Schematic diagram;When communication equipment detects Cache exceptions, the register of L1 Cache and the register of L2 Cache are read, will The instruction and data write-in internal memory of L1 Cache and L2 Cache storages, according to the register and L2 of the L1 Cache for reading The register of Cache, detects the Cache way that the Bank 1 that the abnormal Cache line of Cache are located at L2 Cache occurs 3, and determine there be available Cache way in the Bank 1 of L2 Cache, to posting for the Cache way states of record L2 Cache Bit 11 carries out OR operation in storage, and the initial value of Bit 11 is revised as " 1 " from " 0 ", indicates the Bank 1 of L2 Cache Cache way 3 for unavailable;Communication equipment closes the Cache way 3 of the Bank 1 of L2 Cache.
It is assumed that communication equipment has had been switched off 7 Cache way of the Bank 0 of L2 Cache, generation is detected again Cache abnormal Cache line are located at a Cache way of the Bank 0 of L2 Cache, by L1Cache and L2 Cache After the instruction and data write-in internal memory of storage, determine that the Bank 0 of L2 Cache without available Cache way, then carries out weight Open, communication equipment need not the whole Cache way that have recorded the Bank0 of L2 Cache all set it is unavailable after restart again.
In other embodiments of the invention, communication equipment can posting according to the Cache way states of record L2 Cache Numerical value is the number of the Bit of " 0 " in storage, determines whether each Bank of L2 Cache also has available Cache way, or Person, communication equipment can be the number of the Bit of " 1 " according to numerical value in the register of the Cache way states of record L2 Cache, really Whether each Bank for determining L2 Cache also has available Cache way.
For example, communication equipment reads from the Bit-0-Bit7 of the register of the Cache way states of record L2 Cache Numerical value is 1 for the number of the Bit of " 0 ", it is determined that the current Cache line that Cache exceptions occur are located at L2 Cache's The unique available Cache way of Bank 0, the Bank 0 of L2 Cache is without available Cache way.Or, communication equipment from The number that the Bit0-Bit7 of the register of the Cache way states of record L2 Cache reads the Bit that numerical value is " 1 " is 7 It is individual, it is determined that current that the unique available Cache that the abnormal Cache line of Cache are located at the Bank 0 of L2 Cache occurs The Bank 0 of way, L2 Cache is without available Cache way.
Likewise, communication equipment is read from the Bit-8-Bit 15 of the register of the Cache way states of record L2 Cache The number for getting the Bit that numerical value is " 0 " is 1, it is determined that the current Cache line that Cache exceptions occur are located at L2 Cache Bank 1 unique available Cache way, L2 Cache Bank 1 without available Cache way.Or, communication equipment The number of the Bit that numerical value is " 1 " is read from the Bit8-Bit 15 of the register of the Cache way states of record L2 Cache It it is 7, it is determined that current that the unique available Cache of Bank 1 that the abnormal Cache line of Cache are located at L2 Cache occur The Bank 1 of way, L2 Cache is without available Cache way.
Using above-described embodiment, communication equipment detects Cache exceptions, L1 Cache or L2 Cache Bank also When having available Cache way, first there is the abnormal Cache way of Cache in record, turns off and the abnormal Cache of Cache occur Way, is continuing with the available Cache way in the Bank of L1 Cache or L2 Cache, reduce communication equipment due to The Cache abnormal number of times restarted, improves the stability of communication equipment, reduces the influence to message forwarding service.
With reference to Figure 10, Figure 10 is a kind of structural representation of the abnormal processing units of Cache provided in an embodiment of the present invention, The device includes:
Detection module 1001, for detecting cache exception;
Read module 1002, for when detecting cache and being abnormal, reading the register and two of on-chip cache The register of level cache;
Identification module 1003, for the deposit of register and second level cache according to the on-chip cache for reading Device, identification produces the abnormal abnormal cache line of cache;
Control module 1004, for determining that the cache way belonging to abnormal cache line is located at on-chip cache, The register of the cache line state according to record on-chip cache, judges whether on-chip cache also has available height Fast cache way;If it is, in the register of the cache line state of record on-chip cache, recording exceptional cache Cache way belonging to row is unavailable, closes the cache way belonging to abnormal cache line;If not, execution is restarted.
In other embodiments of the invention, control module 1004, can be also used for:
It is determined that the cache way belonging to abnormal cache line is located at on-chip cache and is shared by multiple threads, root According to the register of the cache line state of the shared on-chip cache of the multiple threads of record, whether on-chip cache is judged Also available cache way;If it is, in the cache line state of the shared on-chip cache of the multiple threads of record Register in, the cache way belonging to abnormal cache line that each thread is used is recorded as unavailable;Close quilt Cache way belonging to the shared abnormal cache line of multiple threads;If not, execution is restarted.
In other embodiments of the invention, control module 1004, can be also used for:
It is determined that the cache way belonging to abnormal cache line is located at one of memory cell of second level cache, The register of the cache line state of each memory cell according to record second level cache, judges abnormal cache line Whether the memory cell that affiliated cache way is located at also has available cache way;If it is, recording two grades of high speeds In the register of the cache line state of each memory cell of caching, the cache belonging to recording exceptional cache line Road is unavailable;Close the cache way belonging to abnormal cache line;If not, execution is restarted.
In other embodiments of the invention, control module 1004, can be also used for:
The data and instruction that are stored in on-chip cache and second level cache are recorded in internal memory by memory module.
Using above-described embodiment, communication equipment detects Cache exceptions, L1 Cache or L2 Cache Bank also When having available Cache way, first there is the abnormal Cache way of Cache in record, turns off and the abnormal Cache of Cache occur Way, is continuing with the available Cache way in the Bank of L1 Cache or L2 Cache, reduce communication equipment due to The Cache abnormal number of times restarted, improves the stability of communication equipment, reduces the influence to message forwarding service.
It should be noted that herein, such as first and second or the like relational terms are used merely to a reality Body or operation make a distinction with another entity or operation, and not necessarily require or imply these entities or deposited between operating In any this actual relation or order.And, term " including ", "comprising" or its any other variant be intended to Nonexcludability is included, so that process, method, article or equipment including a series of key elements not only will including those Element, but also other key elements including being not expressly set out, or also include being this process, method, article or equipment Intrinsic key element.In the absence of more restrictions, the key element limited by sentence "including a ...", it is not excluded that Also there is other identical element in process, method, article or equipment including the key element.
Each embodiment in this specification is described by the way of correlation, identical similar portion between each embodiment Divide mutually referring to what each embodiment was stressed is the difference with other embodiment.Especially for system reality Apply for example, because it is substantially similar to embodiment of the method, so description is fairly simple, related part is referring to embodiment of the method Part explanation.
Presently preferred embodiments of the present invention is the foregoing is only, is not intended to limit the scope of the present invention.It is all Any modification, equivalent substitution and improvements made within the spirit and principles in the present invention etc., are all contained in protection scope of the present invention It is interior.

Claims (8)

1. the abnormal processing method of a kind of cache, it is characterised in that methods described includes:
Cache exception is detected, the register of on-chip cache and the register of second level cache is read;
According to the register and the register of the second level cache of the on-chip cache for reading, identification produces described The abnormal abnormal cache line of cache;
Determine that cache way belonging to the abnormal cache line is located at the on-chip cache, according to recording described one The register of the cache line state of level cache, judges whether the on-chip cache also has available cache Road;
If it is, in the register of the cache line state of the record on-chip cache, recording the abnormal high speed Cache way belonging to cache lines is unavailable, closes the cache way belonging to the abnormal cache line;
If not, execution is restarted.
2. method according to claim 1, it is characterised in that methods described also includes:
Determine that the cache way belonging to the abnormal cache line is located at the on-chip cache and is total to by multiple threads Enjoy, according to the register of the cache line state for recording the shared on-chip cache of the multiple thread, judge institute State whether on-chip cache also has available cache way;
If it is, in the register of the cache line state of the shared on-chip cache of the multiple threads of the record, will The cache way belonging to described abnormal cache line that each thread is used is recorded as unavailable;Close by the multiple line Cache way belonging to the shared described abnormal cache line of journey;
If not, execution is restarted.
3. method according to claim 2, it is characterised in that methods described also includes:
Determine that the cache way belonging to the abnormal cache line is located at one of storage of the second level cache Unit, the register of the cache line state according to each memory cell for recording the second level cache judges described Whether the memory cell that the cache way belonging to abnormal cache line is located at also has available cache way;
If it is, in the register of the cache line state of each memory cell of the record second level cache, note The cache way recorded belonging to the abnormal cache line is unavailable;The high speed belonging to the abnormal cache line is closed to delay Deposit road;
If not, execution is restarted.
4. method according to claim 1, it is characterised in that identification produce the cache it is abnormal it is abnormal at a high speed After cache lines, methods described also includes:
The data and instruction that are stored in the on-chip cache and the second level cache are recorded in internal memory.
5. the abnormal processing unit of a kind of cache, it is characterised in that described device includes:
Detection module, for detecting cache exception;
Read module, register and two grades of high speeds for when detecting cache and being abnormal, reading on-chip cache delay The register deposited;
Identification module, for the deposit of register and the second level cache according to the on-chip cache for reading Device, identification produces the abnormal abnormal cache line of the cache;
Control module, for determining that the cache way belonging to the abnormal cache line is located at the on-chip cache, According to the register of the cache line state for recording the on-chip cache, judge whether the on-chip cache also has Available cache way;If it is, in the register of the cache line state of the record on-chip cache, record Cache way belonging to the abnormal cache line is unavailable, closes the cache belonging to the abnormal cache line Road;If not, execution is restarted.
6. device according to claim 5, it is characterised in that the control module, is additionally operable to:
Determine that the cache way belonging to the abnormal cache line is located at the on-chip cache and is total to by multiple threads Enjoy, according to the register of the cache line state for recording the shared on-chip cache of the multiple thread, judge institute State whether on-chip cache also has available cache way;If it is, high in the shared one-level of the multiple threads of the record In the register of the cache line state of speed caching, the high speed belonging to described abnormal cache line that each thread is used Cache way is recorded as unavailable;The cache belonging to described abnormal cache line that closing is shared by the multiple thread Road;If not, execution is restarted.
7. device according to claim 5, it is characterised in that the control module, is additionally operable to:
Determine that the cache way belonging to the abnormal cache line is located at one of storage of the second level cache Unit, the register of the cache line state according to each memory cell for recording the second level cache judges described Whether the memory cell that the cache way belonging to abnormal cache line is located at also has available cache way;If it is, In the register of the cache line state of each memory cell of the record second level cache, the exception is recorded high Cache way belonging to fast cache lines is unavailable;Close the cache way belonging to the abnormal cache line;If not, Execution is restarted.
8. device according to claim 5, it is characterised in that described device also includes:
Memory module, the data and instruction that are stored in the on-chip cache and the second level cache is recorded in interior Deposit.
CN201710052784.6A 2017-01-24 2017-01-24 Method and device for processing cache exception Active CN106874161B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201710052784.6A CN106874161B (en) 2017-01-24 2017-01-24 Method and device for processing cache exception

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201710052784.6A CN106874161B (en) 2017-01-24 2017-01-24 Method and device for processing cache exception

Publications (2)

Publication Number Publication Date
CN106874161A true CN106874161A (en) 2017-06-20
CN106874161B CN106874161B (en) 2021-06-22

Family

ID=59158842

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201710052784.6A Active CN106874161B (en) 2017-01-24 2017-01-24 Method and device for processing cache exception

Country Status (1)

Country Link
CN (1) CN106874161B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6038693A (en) * 1998-09-23 2000-03-14 Intel Corporation Error correction scheme for an integrated L2 cache
CN1518700A (en) * 2001-06-22 2004-08-04 皇家菲利浦电子有限公司 Fast and acurate cache way selection
CN1979451A (en) * 2005-12-08 2007-06-13 国际商业机器公司 Methods and apparatus for handling a cache miss
CN102929796A (en) * 2012-06-01 2013-02-13 杭州中天微系统有限公司 Memory management module simultaneously supporting software backfilling and hardware backfilling

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6038693A (en) * 1998-09-23 2000-03-14 Intel Corporation Error correction scheme for an integrated L2 cache
CN1518700A (en) * 2001-06-22 2004-08-04 皇家菲利浦电子有限公司 Fast and acurate cache way selection
CN1979451A (en) * 2005-12-08 2007-06-13 国际商业机器公司 Methods and apparatus for handling a cache miss
CN102929796A (en) * 2012-06-01 2013-02-13 杭州中天微系统有限公司 Memory management module simultaneously supporting software backfilling and hardware backfilling

Also Published As

Publication number Publication date
CN106874161B (en) 2021-06-22

Similar Documents

Publication Publication Date Title
CN103003799B (en) Method and system for verifying memory device integrity
US20080294847A1 (en) Cache control device and computer-readable recording medium storing cache control program
US20100058144A1 (en) Memory system with ecc-unit and further processing arrangement
JPS63195752A (en) Cache memory
CN103026342B (en) method and system for verifying memory device integrity
CN100440174C (en) System and method for direct deposit using locking cache
KR20110061598A (en) Disabling cache portions during low voltage operations
CN103839591A (en) Automatic fault detection and fault-tolerant circuit of memory as well as control method
US11080135B2 (en) Methods and apparatus to perform error detection and/or correction in a memory device
CN106133704A (en) Memory failure partition method and device
CN109918226A (en) A kind of silence error-detecting method, device and storage medium
CN107731262A (en) A kind of storage disk bad block management method, device, equipment and readable storage medium storing program for executing
US20190056874A1 (en) System and method for preserving data in volatile memory
CN110265074B (en) Hierarchical multiple redundancy magnetic random access memory and operation method thereof
US9009548B2 (en) Memory testing of three dimensional (3D) stacked memory
US20130104013A1 (en) Address translation checking device, central processing unit, and address translation checking method
CA1241768A (en) Tag control circuit for buffer storage
JPS6044707B2 (en) Fault detection method for buffer memory control circuit
CN103365792A (en) Information processing apparatus, and method of controlling information processing apparatus
CN106874161A (en) A kind of abnormal processing method and processing device of cache
WO2023142429A1 (en) Method for predicting uncorrectable error of volatile storage medium, and related device
CN103824101B (en) Logic card reading/writing method and system
JP2010128698A (en) Multiprocessor system
CN103744752A (en) Online memory fault detecting method and device
TW440765B (en) L2 cache testing method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant