Classifications

• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F11/00—Error detection; Error correction; Monitoring
  • G06F11/36—Prevention of errors by analysis, debugging or testing of software
  • G06F11/362—Debugging of software
  • G06F11/3636—Debugging of software by tracing the execution of the program
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F12/00—Accessing, addressing or allocating within memory systems or architectures
  • G06F12/02—Addressing or allocation; Relocation
  • G06F12/08—Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
  • G06F12/0802—Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
  • G06F12/0806—Multiuser, multiprocessor or multiprocessing cache systems
  • G06F12/0815—Cache consistency protocols
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F11/00—Error detection; Error correction; Monitoring
  • G06F11/30—Monitoring
  • G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
  • G06F11/3466—Performance evaluation by tracing or monitoring
  • G06F11/3476—Data logging
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F12/00—Accessing, addressing or allocating within memory systems or architectures
  • G06F12/02—Addressing or allocation; Relocation
  • G06F12/08—Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
  • G06F12/0802—Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
  • G06F12/0806—Multiuser, multiprocessor or multiprocessing cache systems
  • G06F12/0808—Multiuser, multiprocessor or multiprocessing cache systems with cache invalidating means
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F12/00—Accessing, addressing or allocating within memory systems or architectures
  • G06F12/02—Addressing or allocation; Relocation
  • G06F12/08—Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
  • G06F12/0802—Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
  • G06F12/0806—Multiuser, multiprocessor or multiprocessing cache systems
  • G06F12/0811—Multiuser, multiprocessor or multiprocessing cache systems with multilevel cache hierarchies
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F12/00—Accessing, addressing or allocating within memory systems or architectures
  • G06F12/02—Addressing or allocation; Relocation
  • G06F12/08—Addressing or allocation; Relocation in hierarchically structured memory systems, e.g. virtual memory systems
  • G06F12/0802—Addressing of a memory level in which the access to the desired data or data block requires associative addressing means, e.g. caches
  • G06F12/0893—Caches characterised by their organisation or structure
  • G06F12/0897—Caches characterised by their organisation or structure with two or more cache hierarchy levels
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F11/00—Error detection; Error correction; Monitoring
  • G06F11/30—Monitoring
  • G06F11/3003—Monitoring arrangements specially adapted to the computing system or computing system component being monitored
  • G06F11/3024—Monitoring arrangements specially adapted to the computing system or computing system component being monitored where the computing system component is a central processing unit [CPU]
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F11/00—Error detection; Error correction; Monitoring
  • G06F11/30—Monitoring
  • G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
  • G06F11/3466—Performance evaluation by tracing or monitoring
  • G06F11/3471—Address tracing
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06F—ELECTRIC DIGITAL DATA PROCESSING
  • G06F2201/00—Indexing scheme relating to error detection, to error correction, and to monitoring
  • G06F2201/885—Monitoring specific for caches

Definitions

• debuggers Many types of debugging applications (“debuggers”) have been developed in order to assist developers with the code debugging process. These tools offer developers the ability to trace, visualize, and alter the execution of computer code. For example, debuggers may visualize the execution of code instructions, may present code variable values at various times during code execution, may enable developers to alter code execution paths, and/or may enable developers to set“breakpoints” and/or“watchpoints” on code elements of interest (which, when reached during execution, causes execution of the code to be suspended), among other things.
• debugging applications enable“time travel,”“reverse,” or “historic” debugging.
• execution of a program e.g., executable entities such as threads
• trace application e.g., executable entities such as threads
• trace file(s) can then be used to replay execution of the program later, for both forward and backward analysis.
• “time travel” debuggers can enable a developer to set forward breakpoints/watchpoints (like conventional debuggers) as well as reverse breakpoints/watchpoints.
• Embodiments described herein are directed to mechanisms for creating bit- accurate“time travel” trace recordings using hardware assistance by a processor. These mechanisms are based on tracing the effects of execution across a plurality of processing units using at least two tiers or layers of processor caches. In particular, these mechanisms could modify a processor’s hardware and/or microcode so that it assists in (i) detecting an influx (i.e., cache miss) to an inner or“lower-layer” processor cache based on activity by a traced processing unit, and (ii) using an outer or“upper-layer” shared processor cache to determine if data of that influx has already been logged on behalf of another traced processing unit.
• an influx i.e., cache miss
• an inner or“lower-layer” processor cache based on activity by a traced processing unit
• an outer or“upper-layer” shared processor cache to determine if data of that influx has already been logged on behalf of another traced processing unit.
• Embodiments are directed to computing device(s) that include a plurality of processing units, a plurality of N-level caches, and an (N+z)-level cache.
• the (N+/)-level cache is associated with two or more of the plurality of N-level caches and is configured as a backing store for the plurality of N-level caches.
• the computing device(s) include control logic that configures the computing device(s) to detect an influx to a first N-level cache of plurality of N-level caches and in which the influx comprises data stored at a memory location.
• the control logic also configures the computing device(s) to check the (N+z)-level cache to determine if the data for the memory location has been previously logged on behalf of a second processing unit.
• the control logic also configures the computing device(s), based on this check, to perform one of (i) causing the data for the memory location to be logged on behalf of the first processing unit by reference to log data that was previously logged on behalf of the second processing unit (i.e., when the data for the memory location has been previously logged on behalf of the second processing unit), or (ii) causing the data for the memory location to be logged by value on behalf of the first processing unit (i.e., when the data for the memory location has not been previously logged on behalf of the second processing unit).
• Embodiments are also directed to methods for trace recording based on recording an influx to a lower-level cache by reference to prior log data based on knowledge of one or more upper-level caches. These methods are implemented at a computing device that includes (i) a plurality of processing units, (ii) a plurality of N-level caches, and (iii) an (N+z)-level cache that is associated with two or more of the plurality of N-level caches, and that is configured as a backing store for the plurality of N-level caches.
• the method includes detecting an influx to a first N-level cache of plurality of N-level caches, the influx comprising data stored at a memory location.
• the method also includes based on detecting the influx to the first N-level cache, checking the (N+z)-level cache to determine if the data for the memory location has been previously logged on behalf of a second processing unit.
• the method also includes, based on this check, performing one of (i) when the data for the memory location has been previously logged on behalf of the second processing unit, causing the data for the memory location to be logged on behalf of the first processing unit by reference to log data that was previously logged on behalf of the second processing unit, or (ii) when the data for the memory location has not been previously logged on behalf of the second processing unit, causing the data for the memory location to be logged by value on behalf of the first processing unit.
• Embodiments may also be embodied as computer-executable instructions (e.g., processor microcode) stored on a hardware storage device, and that are executable to perform the foregoing method.
• computer-executable instructions e.g., processor microcode
• Figure 1 illustrates an example computing environment that facilitates recording “bit-accurate” traces of execution across a plurality of processing units, using at least two tiers or layers of processor caches—including detecting an influx to an inner or“lower- layer” processor cache, and using an outer or“upper-layer” shared processor cache to determine if that influx can be logged by reference to a previously-logged value;
• Figure 2A illustrates an example computing environment including multi-layer caches
• Figure 2B illustrates an example of a cache
• Figure 3 illustrates a flowchart of an example method for trace recording based on recording an influx to a lower-level cache by reference to prior log data based on knowledge of one or more upper-level caches;
• Figure 4A illustrates an example shared cache that extends each of its cache lines with one or more additional accounting bits
• Figure 4B illustrates an example of a shared cache that includes one or more reserved cache lines for storing accounting bits that apply to conventional cache lines;
• Figure 5 illustrates an example of set-associative mapping between system memory and a cache.
• Embodiments described herein are directed to mechanisms for creating bit- accurate“time travel” trace recordings using hardware assistance by a processor. These mechanisms are based on tracing the effects of execution across a plurality of processing units using at least two tiers or layers of processor caches. In particular, these mechanisms could modify a processor’s hardware and/or microcode so that it assists in (i) detecting an influx (i.e., cache miss) to an inner or“lower-layer” processor cache based on activity by a traced processing unit, and (ii) using an outer or“upper-layer” shared processor cache to determine if data of that influx has already been logged on behalf of another traced processing unit.
• an influx i.e., cache miss
• an inner or“lower-layer” processor cache based on activity by a traced processing unit
• an outer or“upper-layer” shared processor cache to determine if data of that influx has already been logged on behalf of another traced processing unit.
• Figure 1 illustrates an example computing environment 100 that facilitates recording“bit-accurate” traces of execution across a plurality of processing units, using at least two tiers or layers of processor caches—including detecting an influx to an inner or “lower-layer” processor cache, and using an outer or“upper-layer” shared processor cache to determine if that influx can be logged by reference to a previously-logged value.
• embodiments may comprise or utilize a special-purpose or general-purpose computer system 101 that includes computer hardware, such as, for example, one or more processor(s) 102, system memory 103, one or more data stores 104, and/or input/output hardware 105.
• Embodiments within the scope of the present invention include physical and other computer-readable media for carrying or storing computer-executable instructions and/or data structures.
• Such computer-readable media can be any available media that can be accessed by the computer system 101.
• Computer-readable media that store computer- executable instructions and/or data structures are computer storage devices.
• Computer- readable media that carry computer-executable instructions and/or data structures are transmission media.
• embodiments of the invention can comprise at least two distinctly different kinds of computer-readable media: computer storage devices and transmission media.
• Computer storage devices are physical hardware devices that store computer- executable instructions and/or data structures.
• Computer storage devices include various computer hardware, such as RAM, ROM, EEPROM, solid state drives (“SSDs”), flash memory, phase-change memory (“PCM”), optical disk storage, magnetic disk storage or other magnetic storage devices, or any other hardware device(s) which can be used to store program code in the form of computer-executable instructions or data structures, and which can be accessed and executed by the computer system 101 to implement the disclosed functionality of the invention.
• computer storage devices may include the depicted system memory 103, the depicted data store 104 which can store computer- executable instructions and/or data structures, or other storage such as on-processor storage, as discussed later.
• Transmission media can include a network and/or data links which can be used to carry program code in the form of computer-executable instructions or data structures, and which can be accessed by the computer system 101.
• A“network” is defined as one or more data links that enable the transport of electronic data between computer systems and/or modules and/or other electronic devices.
• the input/output hardware 105 may comprise hardware (e.g., a network interface module (e.g., a“NIC”)) that connects a network and/or data link which can be used to carry program code in the form of computer- executable instructions or data structures.
• a network interface module e.g., a“NIC”
• NIC network interface module
• program code in the form of computer-executable instructions or data structures can be transferred automatically from transmission media to computer storage devices (or vice versa).
• computer-executable instructions or data structures received over a network or data link can be buffered in RAM within a NIC (e.g., input/output hardware 105), and then eventually transferred to the system memory 103 and/or to less volatile computer storage devices (e.g., data store 104) at the computer system 101.
• a NIC e.g., input/output hardware 105
• less volatile computer storage devices e.g., data store 104
• Computer-executable instructions comprise, for example, instructions and data which, when executed at the processor(s) 102, cause the computer system 101 to perform a certain function or group of functions.
• Computer-executable instructions may be, for example, binaries, intermediate format instructions such as assembly language, or even source code.
• Cloud computing environments may be distributed, although this is not required. When distributed, cloud computing environments may be distributed internationally within an organization and/or have components possessed across multiple organizations.
• “cloud computing” is defined as a model for enabling on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, and services). The definition of“cloud computing” is not limited to any of the other numerous advantages that can be obtained from such a model when properly deployed.
• a cloud computing model can be composed of various characteristics, such as on-demand self-service, broad network access, resource pooling, rapid elasticity, measured service, and so forth.
• a cloud computing model may also come in the form of various service models such as, for example, Software as a Service (“SaaS”), Platform as a Service (“PaaS”), and Infrastructure as a Service (“IaaS”).
• SaaS Software as a Service
• PaaS Platform as a Service
• IaaS Infrastructure as a Service
• the cloud computing model may also be deployed using different deployment models such as private cloud, community cloud, public cloud, hybrid cloud, and so forth.
• Some embodiments may comprise a system that includes one or more hosts that are each capable of running one or more virtual machines.
• virtual machines emulate an operational computing system, supporting an operating system and perhaps one or more other applications as well.
• each host includes a hypervisor that emulates virtual resources for the virtual machines using physical resources that are abstracted from view of the virtual machines.
• the hypervisor also provides proper isolation between the virtual machines.
• the hypervisor provides the illusion that the virtual machine is interfacing with a physical resource, even though the virtual machine only interfaces with the appearance (e.g., a virtual resource) of a physical resource. Examples of physical resources including processing capacity, memory, disk space, network bandwidth, media drives, and so forth.
• the data store 104 can store computer-executable instructions and/or data structures representing application programs such as, for example, a tracer l04a, an operating system kernel l04b, and application l04c (e.g., the application that is the subject of tracing by the tracer l04a, and one or more trace file(s) l04d).
• application programs such as, for example, a tracer l04a, an operating system kernel l04b, and application l04c (e.g., the application that is the subject of tracing by the tracer l04a, and one or more trace file(s) l04d).
• the system memory 103 can store corresponding runtime data, such as runtime data structures, computer-executable instructions, etc.
• Figure 1 illustrates the system memory 103 as including runtime application code l03a and application runtime data l03b (e.g., each corresponding with application l04c).
• the tracer l04a is usable to record a bit-accurate trace of execution of an application, such as application l04c, and to store trace data in the trace file(s) l04d.
• the tracer l04a is a standalone application, while in other embodiments the tracer l04a is integrated into another software component, such as the operating system kernel l04b, a hypervisor, a cloud fabric, etc. While the trace file(s) l04d are depicted as being stored in the data store 104, the trace file(s) l04d may also be recorded exclusively or temporarily in the system memory 103, or at some other storage device.
• FIG. 1 includes a simplified representation of the internal hardware components of the processor(s) 102.
• each processor 102 includes a plurality of processing unit(s) l02a.
• Each processing unit may be physical (i.e., a physical processor core) and/or logical (i.e., a logical core presented by a physical core that supports hyper threading, in which more than one application threads executes at the physical core).
• the processor 102 may in some embodiments include only a single physical processing unit (core), it could include two or more logical processing units l02a presented by that single physical processing unit.
• Each processing unit l02a executes processor instructions that are defined by applications (e.g., tracer l04a, operating kernel l04b, application l04c, etc.), and which instructions are selected from among a predefined processor instruction set architecture (ISA).
• ISA processor instruction set architecture
• the particular ISA of each processor 102 varies based on processor manufacturer and processor model. Common ISAs include the IA-64 and IA-32 architectures from INTEL, INC., the AMD64 architecture from ADVANCED MICRO DEVICES, INC., and various Advanced RISC Machine (“ARM”) architectures from ARM HOLDINGS, PLC, although a great number of other ISAs exist and can be used by the present invention.
• an“instruction” is the smallest externally-visible (i.e., external to the processor) unit of code that is executable by a processor.
• Each processing unit l02a obtains processor instructions from one or more processor cache(s) l02b, and executes the processor instructions based on data in the cache(s) l02b, based on data in registers l02d, and/or without input data.
• each cache l02b is a small amount (i.e., small relative to the typical amount of system memory 103) of random-access memory that stores on-processor copies of portions of a backing store, such as the system memory 103 and/or another cache in the cache(s) l02b.
• the cache(s) l02b when executing the application code 103 a, one or more of the cache(s) l02b contain portions of the application runtime data l03b.
• processing unit(s) l02a request data not already stored in a particular cache l02b, then a“cache miss” occurs, and that data is fetched from the system memory 103 or another cache, potentially“evicting” some other data from that cache l02b.
• the processor cache(s) l02b are divided into separate tiers, layers, or levels— such as layer 1 (Ll), layer 2 (L2), layer 3 (L3), etc.
• tiers could be part of the processor 102, itself (e.g., Ll and L2), and/or could be separate from the processor 102 (e.g., L3).
• the cache(s) l02b of Figure 1 may comprise one of these layers (Ll) or may comprise a plurality of these layers (e.g., Ll and L2, and even L3).
• Figure 2A illustrates an example environment 200 demonstrating multi-layer caches.
• each processor 201 includes four physical processing units (i.e., units A1-A4 for processor 20la and units B 1-B4 for processor 210b).
• Example environment 200 also includes a three-layer cache within each processing unit 201.
• Environment 200 is one example layout only, and it not limiting to the cache hierarchies in which the embodiments herein may operate.
• each processing unit is associated with its own dedicated Ll cache (e.g., Ll cache“Ll-Al” in processor 20la for unit Al, Ll cache“L1-A2” in processor 20 la for unit A2, etc.).
• each processing unit 201 includes two L2 caches (e.g., L2 cache“L2-A1” in processor 20la that serves as a backing store for Ll caches Ll- Al and L1-A2, L2 cache“L1-A2” in processor 20la that serves as a backing store for Ll caches L1-A3 and L1-A4, etc.).
• L2 cache“L2-A1” in processor 20la that serves as a backing store for Ll caches Ll- Al and L1-A2
• L2 cache“L1-A2” in processor 20la that serves as a backing store for Ll caches L1-A3 and L1-A4, etc.
• each processing unit 201 includes a single L3 cache (e.g., L3 cache“L3-A” in processor 20 la that serves as a backing store for L2 caches L2-A1 and L2-A2, and L3 cache“L3-B” in processor 20lb that serves as a backing store for L2 caches L2-B1 and L2-B2).
• L3 cache“L3-A” in processor 20 la that serves as a backing store for L2 caches L2-A1 and L2-A2
• L3 cache“L3-B” in processor 20lb that serves as a backing store for L2 caches L2-B1 and L2-B2
• the processing unit(s) l02a when multiple cache layers are used, the processing unit(s) l02a typically interact directly with the lowest layer (Ll). In most cases, data flows between the layers (e.g., on a read an L3 cache interacts with the system memory 103 and serves data to an L2 cache, and the L2 cache in turn serves data to the Ll cache).
• Ll lowest layer
• the caches coordinate to ensure that those caches that have affected data that was shared among the processing unit(s) l02a don’t have it anymore. This coordination is performed using a CCP.
• the caches in environment 200 may thus be viewed as“shared” caches.
• each L2 and L3 cache serves multiple processing units within a given processor 201 and are thus shared by the processing units.
• the Ll caches with in a given processor 201 collectively, can also be considered shared— even though each one corresponds to a single processing unit— because the individual Ll caches may coordinate with each other (i.e., via a CCP) to ensure consistency (i.e., so that each cached memory location is viewed consistently across all the Ll caches).
• the L2 caches within each processor 201 similarly may coordinate via a CCP. Additionally, if the processor 201 supports hyper-threading, each individual Ll cache may be viewed being shared by two or more logical processing units and are thus“shared” even at an individual level.
• each cache comprises a plurality of“cache lines.”
• Each cache line stores a chunk of memory from its backing store (e.g., system memory 202 or a higher-layer cache).
• Figure 2B illustrates an example of at least a portion of a cache 203, which includes a plurality of cache lines 206, each of which comprises at least an address portion 204 and a value portion 205.
• the address portion 204 of each cache line 206 is configured to store an address in system memory 202 for which the cache line corresponds, and the value portion 205 initially stores a value received from the system memory 202.
• the value portion 205 can be modified by the processing units, and eventually be evicted back to the backing store.
• a cache 203 can include a large number of cache lines.
• a contemporary 64-bit INTEL processor may contain individual Ll caches comprising 512 or more cache lines.
• each cache line is typically usable to store a 64-byte (5 l2-bit) value in reference to a 6-byte (48-bit) to 8-byte (64-bit) memory address.
• cache sizes typically increase with each layer (i.e., L2 caches are typically larger than Ll caches, L3 caches are typically larger than L2 caches, etc.).
• the address stored in the address portion 204 of each cache line 206 may be a physical address, such as the actual memory address in the system memory 202.
• the address stored in the address portion 204 may be a virtual address, which is an address that is mapped to the physical address to provide an abstraction (e.g., using operation-system managed page tables).
• Such abstractions can be used, for example, to facilitate memory isolation between different processes executing at the processor(s) 102, including isolation between user-mode processes and kernel mode processes associated with the operating system kernel l04b.
• a processor 102 may include a translation lookaside buffer (TLB) l02f (usually part of a memory management unit (MMU)), which maintains recently-used memory address mappings between physical and virtual addresses.
• TLB translation lookaside buffer
• MMU memory management unit
• the cache(s) l02b may include code cache portions and data cache portions.
• the code portion(s) of the cache(s) l02b may store at least a portion of the processor instructions stored in the application code 103 a and the data portion(s) of the cache(s) l02b may store at least a portion of data structures of the application runtime data l03b.
• caches can be inclusive, exclusive, or include both inclusive and exclusive behaviors. For example, in an inclusive cache an L3 layer would store a superset of the data in the L2 layers below it, and the L2 layers store a superset of the Ll layers below them.
• the layers may be disjointed— for example, if data exists in an L3 cache that an Ll cache needs, they may swap information, such as data, address, and the like.
• each processor 102 also includes microcode l02c, which comprises control logic (i.e., executable instructions) that control operation of the processor 102, and which generally functions as an interpreter between the hardware of the processor and the processor ISA exposed by the processor 102 to executing applications.
• the microcode 102 is typically embodied on on-processor storage, such as ROM, EEPROM, etc.
• Registers l02d are hardware-based storage locations that are defined based on the ISA of the processors(s) 102 and that are read from and/or written to by processor instructions.
• registers l02d are commonly used to store values fetched from the cache(s) l02b for use by instructions, to store the results of executing instructions, and/or to store status or state— such as some of the side-effects of executing instructions (e.g., the sign of a value changing, a value reaching zero, the occurrence of a carry, etc.), a processor cycle count, etc.
• some registers l02d may comprise“flags” that are used to signal some state change caused by executing processor instructions.
• processors 102 may also include control registers, which are used to control different aspects of processor operation.
• Figure 1 depicts registers l02d as a single box, it will be appreciated that each processing unit l02a typically includes one or more corresponding sets of registers l02d that are specific to that processing unit.
• the processor(s) 102 may include one or more buffers l02e.
• buffer(s) l02e may be used as a temporary storage location for trace data.
• the processors(s) 102 may store portions of trace data the buffer(s) l02e, and flush that data to the trace file(s) l04d at appropriate times, such as when there is available memory bus bandwidth and/or free processor cycles.
• processors operate on the cache(s) l02b according to one or more CCPs.
• a CCP defines how consistency is maintained between data among the various cache(s) l02b as the various processing units l02a read from and write to data in the various cache(s) l02b, and how to ensure that the various processing units l02a always read valid data from a given location in the cache(s) l02b.
• CCPs are related to an enable a memory model defined by the processor l02’s ISA.
• Examples of common CCPs include the MSI protocol (i.e., Modified, Shared, and Invalid), the MESI protocol (i.e., Modified, Exclusive, Shared, and Invalid), and the MOESI protocol (i.e., Modified, Owned, Exclusive, Shared, and Invalid).
• MSI protocol i.e., Modified, Shared, and Invalid
• MESI protocol i.e., Modified, Exclusive, Shared, and Invalid
• MOESI protocol i.e., Modified, Owned, Exclusive, Shared, and Invalid.
• Each of these protocols define a state for individual locations (e.g., lines) in the cache(s) l02b.
• a “modified” cache location contains data that has been modified in the cache(s) l02b and is therefore potentially inconsistent with the corresponding data in the backing store (e.g., system memory 103 or another cache).
• a “shared” cache location contains data that is unmodified from the data in the backing store, exists in read-only state, and is shared by the processing unit(s) l02a. The cache(s) l02b can evict this data without writing it to the backing store.
• An“invalid” cache location contains no valid data and can be considered empty and usable to store data from cache miss.
• An“exclusive” cache location contains data that matches the backing store and is used by only a single processing unit l02a.
• An“owned” cache location is shared by two or more processing units l02a, but one of the processing units has the exclusive right to make changes to it. When that processing makes changes, it notifies the other processing units— since the notified processing units may need to invalidate or update based on the CCP implementation.
• embodiments utilize the processor l02’s cache(s) l02b to efficiently record a bit-accurate trace of execution of an application l04c and/or the operating system kernel l04b.
• the processor 102 including the cache(s) l02b) form a semi- or quasi-closed system. For example, once portions of data for a process (i.e., code data and runtime application data) are loaded into the cache(s) l02b, the processor 102 can run by itself— without any input— as a semi- or quasi-closed system for bursts of time.
• one or more of the processing units l02a execute instructions from the code portion(s) of the cache(s) l02b, using runtime data stored in the data portion(s) of the cache(s) l02b and using the registers l02d.
• a“cache miss” occurs and that information is brought into the cache(s) l02b from the system memory 103. For example, if a data cache miss occurs when an executed instruction performs a memory operation at a memory address within the application runtime data l03b, data from that memory address is brought into one of the cache lines of the data portion of the cache(s) l02b.
• code cache miss occurs when an instruction performs a memory operation at a memory address application code l03a stored in system memory 103, code from that memory address is brought into one of the cache lines of the code portion(s) of the cache(s) l02b.
• the processing unit l02a then continues execution using the new information in the cache(s) l02b until new information is again brought into the cache(s) l02b (e.g., due to another cache miss or an un-cached read).
• the tracer l04a can record sufficient data to be able to reproduce the influx of information into the cache(s) l02b as processing units execute that application’s thread(s).
• one approach to recording these influxes operates on a per-processing-unit basis and at the innermost cache layer (e.g., Ll).
• This approach may involve recording, for each processing unit that is being traced, all cache misses and un-cached reads (i.e., reads from hardware components and un-cacheable memory) associated with that processing unit’ s L 1 cache, along with a time during execution at which each piece of data was brought into that processing unit’s Ll cache (e.g., using a count of instructions executed or some other counter). If there are events that can be ordered across the processing units (e.g., accesses to shared memory), these events can be logged across the resulting data streams (e.g., by using a monotonically incrementing number (MIN) across the data streams).
• MIN monotonically incrementing number
• an Ll cache layer may include multiple distinct Ll caches that are each associated a different physical processing unit (e.g., as shown in Figure 2A)
• recording in this manner may record duplicate data— and thus more data than is strictly necessary for a“full-fidelity” trace.
• this approach may log cache misses for the same memory location and data for each of multiple physical processing units.
• a“full-fidelity” trace is any trace that contains sufficient information to enable a full replay of a traced entity— even though a particular“full-fidelity” trace may actually contain less data that encapsulated the same information than might be recorded using alternate tracing techniques.
• the inventor has developed improved recording techniques that utilize one or more of the upper-layer caches to avoid recording at least a portion of this duplicate data. Instead, these improved techniques can log by reference to previously-logged data.
• embodiments detect an influx (i.e., cache miss) to an inner or“lower-layer” processor cache (e.g., Ll) based on activity by one processing unit, but then use one or more outer or“upper-layer” shared processor cache(s) to log that that influx by reference to an influx already logged on behalf of another traced processing unit, when possible.
• each L2 cache is a backing store for two Ll caches
• each L3 cache is a backing store for two L2 caches (and, by extension, four Ll caches).
• an upper-layer cache can retain knowledge about multiple lower-layer caches (e.g., in Figure 2A, L2 cache Ll-Al can retain knowledge about Ll caches Ll-Al and L1-A2, L2 cache L1-A2 can retain knowledge about Ll caches L1-A3 and L1-A4, and L3 cache L3-A can retain knowledge about L2 caches L2-A1 and L2-A1 as well as Ll caches Ll-Al, L1-A2, L1-A3, and L1-A4).
• the embodiments herein enable many opportunities for logging influxes caused by one processing unit by reference to an influx already logged on behalf of another processing units.
• Figure 3 illustrates an example of a method 300 for trace recording based on recording an influx to a lower-level cache by reference to prior log data based on knowledge of one or more upper-level caches.
• Figure 3 is now described in the context of Figures 1 and 2.
• Figure 3 operates in environments, such as a processor 102 or 20 la that includes a plurality of processing units, a plurality of N-level caches, and an (N+z)-level cache that is associated with two or more of the plurality of N-level caches, and that is configured as a backing store for the plurality of N-level caches.
• N and i are positive integers— i.e., N > 1, such that N equals 1, 2, 3, etc.; and i > 1, such that i equals 1, 2, 3, etc.
• the processor includes a plurality of processing units Al, A2, etc.
• the processor 20 la also includes a plurality of N-level caches Ll-Al, L1-A2, etc. (i.e., where N equals 1).
• the processor 20 la also includes an (N+z)-level cache that is associated with two or more of the plurality of N-level caches, and that is configured as a backing store for the plurality of N- level caches.
• processor 20la includes an (N+z)-level cache L2-A1 that is a backing store for N-level caches Ll-Al and L1-A2 (i.e., where N equals 1, and i equals 1).
• processor 20la includes an (N+/)-level cache L3-A that is a backing store for N-level caches Ll-Al, L1-A2, etc. (i.e., where N equals 1, and i equals 2).
• Processor 102/20 la operates method 300 based on control logic, such as microcode l02c and/or circuit logic.
• method 300 includes an act 301 of, during execution at a first processing unit, detecting an influx to an N-level cache.
• act 301 comprises detecting an influx to a first N-level cache of plurality of N-level caches, the influx comprising data stored at a memory location.
• activity by processing unit Al such as a requested memory access to system memory 202 (e.g., resulting from normal or speculative execution of a first thread of application l04c)
• a cache miss may occur in cache Ll-Al (i.e., when N equals 1).
• a line of cache Ll-Al obtains an influx of data, including the then-current value of requested memory location.
• an influx could be sourced from system memory 202 or from a higher-level cache (e.g., L2-A1 and/or L3-A).
• a higher-level cache e.g., L2-A1 and/or L3-A
• Method 300 also includes an act 302 of checking an (N+/)-level cache to determine if data of the influx has already been logged based on execution at a second processing unit.
• act 302 comprises, based on detecting the influx to the first N-level cache, checking the (N+/)-level cache to determine if the data for the memory location has been previously logged on behalf of a second processing unit. For example, if i equals 1— such that the (N+/)-level cache comprises an (N+l)-level cache— then processor 201 may check an L2 cache, such as L2-A1 (which has knowledge of cache L1-A2 and processing unit A2).
• This check can be used to determine if the data for the memory location had previously been logged on behalf of processing unit A2. This data may have been previously logged, for example, based on prior execution of a second thread of application l04c at processing unit A2 that caused a cache miss in cache L1-A2.
• processor 201 may check an L2 cache, such as cache L3-A (which has knowledge of all the other caches in processor 201).
• This check can be used to determine if the data for the memory location had previously been logged on behalf of any of processing units A2-A4 (e.g., based on prior execution of one or more other thread(s) of application l04c at one or more of processing units A2-A4 that caused cache miss(es) in caches L1-A2, L1-A3, and/or L1-A4). Note that in this second example, the L2 cache may be skipped in the check. [0056] As shown, act 302 could be repeated any number of times, while incrementing the value of i each time. While i would typically be incremented by 1 each time, there could be embodiments that increment it by a positive integer that is greater than 1.
• L2 cache layer e.g., L2-A1 and/or L2-A2
• L3 cache layer e.g., L3-A
• method 300 can operate in environments, such as a processor 102 or 20 la, in which i equals 1, such that the (N+/)-level cache comprises an (N+l)-level cache, and which the processor also comprises an (N+2)- level cache that is configured as a backing store for the (N+l)-level cache.
• checking the (N+l)-level cache to determine if the data for the memory location has been previously logged on behalf of the second processing unit i.e., act 302 can comprise determining that no cache line in the (N+l)-level cache corresponds to the memory location. Further, checking the (N+2)-level cache to determine if the data for the memory location has been previously logged on behalf of the second processing unit.
• method includes an act 303 of, when the data has already been logged, logging the influx by reference; or an act 304 of, when the data has not already been logged, logging the influx by value.
• act 303 comprises, when the data for the memory location has been previously logged on behalf of the second processing unit, causing the data for the memory location to be logged on behalf of the first processing unit by reference to log data that was previously logged on behalf of the second processing unit.
• act 304 comprises, when the data for the memory location has not been previously logged on behalf of the second processing unit, causing the data for the memory location to be logged by value on behalf of the first processing unit. For example, if a check of (N+l)-level cache L2-A1 and/or a check of (N+2)4evel cache L3-A results in a determination that the data/memory location has not already logged on behalf another processing unit, then the processor 20la can cause the influx to cache Ll-Al to be logged by value of behalf of processing unit Al .
• Logging by value can include, for example, logging the memory address and the memory value in a data packet for processing unit Al . Note that logging by value can include any number of compression techniques to reduce the number of bits needed to accomplish the actual logging.
• processor(s) 102 can include buffer(s) l02d that can be used for temporarily storing trace data.
• “causing” different types of data to be logged could comprise the processor 102 storing such data into the buffer(s) l02d. Additionally, or alternatively, it could include the processor 102 communicating such data to the tracer l04a, writing such data to the trace file(s) l04d, and/or notifying the tracer l04a that the data is available in the buffer(s) l02d.
• the buffer(s) l02d could comprise one or more reserved portions of the cache(s) l02b.
• buffers l02d in acts 304/304 causing the data for the memory location to be logged— either by reference or by value— on behalf of the first processing unit can comprise delaying the logging based on availability of resources such as processor cycles, memory locations, bus bandwidth, etc.
• the buffer(s) l02d comprise one or more reserved portions of the cache(s) l02b
• the delayed logging could comprise invalidating a cache line (in the N-level cache and/or the (N+/)-level cache), rather than evicting it, in order to retain the data for the memory location for purposes of delayed logging.
• method 300 has referred to upper-layer caches having “knowledge” about lower-layer caches.
• the particular form of the“knowledge” that an upper-layer cache retains about lower-level caches can vary, and examples now follow.
• this“knowledge” could be the mere presence of a cache line in an upper-level cache that corresponds to cache line(s) in lower-level cache(s) (i.e., cache lines corresponding to the same memory location and memory data).
• the upper layer(s) store a superset of the data in the layer(s) below them. For example, suppose the caches in Figure 2A are inclusive. In this case, when activity by processing unit A2 causes a location from system memory 202 to be imported into cache L1-A2, that same memory location is also cached in caches L2-A1 and L3-A.
• embodiments may cause the memory location and its value to be logged on behalf of processing unit A2. Later, if activity by processing unit Al causes that same location from system memory 202 to be imported into cache Ll-Al, and that location still stores the same data, it is served from cache L2-A1, since cache L2-A1 already has the data. Prior techniques may again log this data for processing unit Al based on it being an influx to cache L2-A1. However, embodiments herein may instead recognize that the memory location and its value already existed in cache L2-A1, and thus already existed in cache L1-A2. Because processing unit A2 is being logged, embodiments may recognize that the memory location and its value would have already been logged on behalf of processing unit A2, and therefore cause this new activity of processing unit Al to be logged in reference to the log data previously recorded on behalf of processing unit A2.
• FIG. 4A illustrates an example shared cache 400a, similar to the shared cache 203 of Figure 2B, that extends each of its cache lines 404 with one or more additional accounting bit(s) 401.
• each cache line 404 includes accounting bit(s) 401, conventional address bits 402, and value bits 403.
• Figure 4B illustrates an example of a shared cache 400b that includes conventional cache lines 405 that store memory addresses 402 and values 403, as well as one or more reserved cache line(s) 406 for storing accounting bits that apply to the conventional cache lines 405.
• the bits of the reserved cache line(s) 406 are allocated into different groups of accounting bits that each corresponds to a different one of the conventional cache lines 405.
• the reserved cache line(s) 406 could be reserved as one (or more) ways in each index of a set-associative cache (which are discussed in more detail later). For example, in an 8-way set-associative cache one way in a set could be reserved for accounting bits that apply to the other seven ways in the set. This can decrease the complexity of implementing reserved cache lines and can speed access to the reserved cache lines since all ways in a given set are typically read in parallel by most processors.
• each cache line’s accounting bit(s) 401 could comprise one or more bits that function as a flag (i.e., on or off) used by the processor(s) 102 to indicate whether or not the present value in the cache line was logged on behalf of a processing unit (or, alternatively, consumed by a processing unit that participates in logging).
• the check in act 302 can include using this flag to determine if the cache line has been logged by a processing unit that participates in logging.
• each cache line’s accounting bits 401 could comprise a plurality of bits. Pluralities of bits could be used in several ways. Using one approach, referred to herein as“unit bits,” each cache line’s accounting bits 401 can include a number of unit bits equal to a number of processing units l02a of the processor 102 (e.g., the number of logical processing units if the processor 102 supports hyper-threading or the number of physical processing unit if hyper-threading is not supported). These unit bits can be used by the processor 102 to track which one or more particular processing unit(s) have logged the cache line (if any). Thus, for example, a cache that is shared by two processing units l02a could associate two unit bits with each cache line.
• each cache line’s accounting bits 401 can include a number of index bits sufficient to represent an index to each of the processing units l02a of a processor 102 of computer system 101 that participate in logging, potentially along with a“reserved” value (e.g., -1). For example, if the processor 102 includes 128 processing units l02a, these processing units can be identified by an index value (e.g., 0-127) using only seven index bits per cache line. In some embodiments, one index value is reserved (e.g.,“invalid”) to indicate that no processor has logged a cache line.
• index bits would actually be able to represent 127 processing units l02a, plus the reserved value.
• binary values 0000000 - 1111110 might correspond to index locations 0-126 (decimal)
• binary value 1111111 e.g., -1 or 127 decimal, depending on interpretation
• unit bits can be used by the processor 102 to indicate if the cache line has been logged (e.g., a value other than -1), and as an index to a particular processing unit that logged the cache line (e.g., the processing unit that most recently consumed it).
• This second approach to using pluralities of accounting bits 401 has the advantage of supporting a great number of processing units with little overhead in the cache l02b, with the disadvantage of less granularity than the first approach (i.e., only one processing unit is identified at a time).
• checking the (N+z)-level cache to determine if the data for the memory location has been previously logged on behalf of the second processing unit could comprise determining if a cache line in the (N+z)-level cache that corresponds to the memory location has one or more accounting bits set.
• Another mechanism that can be used to determine whether a cache line has been logged is to utilize set-associative caches and way -locking. Since a processor’s cache l02b is generally much smaller than system memory 103 (often by orders of magnitude), thus there are usually far more memory locations in the system memory 103 than there are lines in any given layer of the cache l02b. As such, some processors define mechanisms for mapping multiple memory locations of system memory to line(s) one or more cache layers. Processors generally employ one of two general techniques: direct mapping and associative (or set-associative) mapping. Using direct mapping, different memory locations in system memory 103 are mapped to just one line in a cache layer, such that each memory location can only be cached into a particular line in that layer.
• FIG. 5 illustrates an example 500 of set-associative mapping between system memory and a cache.
• cache lines 504 of a cache layer 502 are logically partitioned into different sets of two cache lines each, including a first set of two cache lines 504a and 504b (identified as index 0), and a second set of two cache lines 504c and 504d (identified as index 1).
• Each cache line in a set is identified as a different“way,” such that cache line 504a is identified by index 0, way 0, cache line 504b is identified by index 0, way 1, and so on.
• memory locations 503a, 503c, 503e, and 503g are mapped to index 0.
• each of these locations in system memory can be cached to any cache line within the set at index 0 (i.e., cache lines 504a and 504b).
• the particular patterns of the depicted mappings are for illustrative and conceptual purposes only, and should not be interpreted as the only way in which memory indexes can be mapped to cache lines.
• Set-associative caches are generally referred to as being N-way set-associative caches, where N is the number of“ways” in each set.
• the cache 500 of Figure 5 would be referred to as a 2-way set-associative cache.
• Processors commonly implement N-way caches where N is a power of two (e.g., 2, 4, 8, etc.), with N values of 4 and 8 being commonly chosen (though the embodiments herein are not limited to any particular N- values or subsets of N-values).
• a l-way set-associative cache is generally equivalent to a direct-mapped cache, since each set contains only one cache line.
• N the number of lines in the cache, it is referred to as a fully associative cache, since it comprises a single set containing all lines in the cache.
• any memory location can be cached to any line in the cache.
• Figure 5 represents a simplified view of system memory and caches, in order to illustrate general principles.
• Figure 5 maps individual memory locations to cache lines
• each line in a cache may store data relating to multiple addressable locations in system memory.
• each location (503a-503h) in system memory (501) may actually represent a plurality of addressable memory locations.
• mappings may be between actual physical addresses in the system memory 501 and lines in the cache 502, or may use an intermediary layer of virtual addresses.
• Set-associative caches can be used for determining whether a cache line has been logged through use of way-locking.
• Way-locking locks or reserves one or more ways in a cache for some purpose.
• the embodiments herein utilize way-locking to reserve one or more ways for a processing unit that is being traced, such that the locked/reserved ways are used exclusively for storing cache misses relating to execution of that unit.
• cache lines 504a and 504c i.e., index 0, way 0 and index 1, way 0
• the processor 102 need only determine whether the cache line stored in an “N+l” cache layer is part of a way is has been reserved for a traced processing unit.
• checking the (N+z)-level cache to determine if the data for the memory location has been previously logged on behalf of the second processing unit could comprise determining if a cache line in the (N+z)-level cache that corresponds to the memory location is stored in a way that corresponds to a logged processing unit.
• caches operate according to a CCP, which defines how consistency is maintained among various caches as the processing units read from and write to cache data, and how to ensure that the processing units always read valid data from a given location in a cache.
• a processor 102 in connection with operating a cache, a processor 102 maintains and stores CCP state data.
• the granularity with which different processors and/or different CCPs track cache coherence state and make that cache coherence data available to a tracer l04a can vary. For example, at one end of the spectrum, some processors/CCPs track cache coherence per cache line as well as per processing unit. These processors/CCPs can, therefore, track the state of each cache line as it relates to each processing unit.
• processors/CCPs are less granular, and track cache coherence the level of the cache line only (and lack per-processing unit information).
• processor manufacturers may choose to track cache coherence at the level of the cache line only for efficiency, since only one processor can own a line exclusively (exclusive, modified, etc.) at a time.
• a processor/CCP may track cache coherence per cache line, as well as an index (e.g., 0, 1, 2, 3 for a four-processing unit processor) to the processing unit that has the current cache line state.
• this CCP state data can be included in the“knowledge” that an (N+/)-level cache has about cached data.
• the CCP state data associated with a given cache line in an (N+/)-level cache can be used to determine if that cache line has been logged by one of the processing units. For example, if the CCP state data indicates that a particular processing unit has taken a given cache line as“shared,” this data can, in turn, be used to determine that the processing unit has logged a read from the cache line.
• checking the (N+/)-level cache to determine if the data for the memory location has been previously logged on behalf of the second processing unit could comprise determining if a cache line in the (N+/)-level cache that corresponds to the memory location has associated CCP state data that is usable to determine that the cache line has been logged.
• an influx of data may be logged by reference to previously-logged data (typically data logged by a different processing unit that the one that caused the present influx). Logging by reference can be accomplished using one or more of a variety of methods (including combinations thereof), some of which are now described.
• a first method logs by reference to a previously-logged memory address. For example, suppose processing unit A2 in Figure 2A has logged data that represents a particular memory address (i.e., in system memory 202) and particular data stored at that memory address. Later, if that particular memory address/particular data is an influx for processing unit Al, processing unit Al could store a log entry that identifies the (i) particular memory address and (ii) processing unit A2. Here, processing unit Al has avoided re- logging the actual data stored at the memory address (which can be of considerable size). Some variants of this first method could also store ordering data, such as a MIN from a series that increments across the data streams for processing units Al and A2.
• This MIN can later be used to order this is influx by processing unit Al against one or more events at processing unit A2 (e.g., those that are also associated with a MIN from the same series).
• causing the data for the memory location to be logged on behalf of the first processing unit by reference to log data that was previously logged on behalf of the second processing unit could comprise one or more of logging an address of the memory location, or logging an address of the memory location and ordering data, such as a MIN.
• a second method logs by reference to a previous owner of a cache line storing the data. For example, suppose that processing unit A2 in Figure 2A has logged a first influx of data. Suppose also that the first influx caused the data to be cached in a cache line of an (N+z)-level cache (e.g., cache L2-A1)— with processing unit A2 being identified an owner of the cache line. Later, if processing unit Al causes a second influx of the same data, processing unit Al could become the owner of this cache line in the (N+z)-level cache.
• N+z-level cache e.g., cache L2-A1
• Processing unit Al could then store a log entry that identifies the prior owner of the cache line (i.e., processing unit A2) so that the A2’s log entry can be used later to obtain the data.
• This means that logging by reference may involve recording the identity of a cache line along with a prior owner of a cache line (e.g., potentially avoiding recording memory addresses and memory values).
• causing the data for the memory location to be logged on behalf of the first processing unit by reference to log data that was previously logged on behalf of the second processing unit could comprise logging the second processing unit as a previous owner of a cache line corresponding to the memory location.
• a third method logs by reference to CCP data.
• CCPs can store cache coherence state about each cache line as different processing units take it for reading and writing.
• the granularity of this data may vary depending on processor implementation, but could, for example, track the cache coherence state of each cache line as it relates to each processing unit, track the cache coherence state of each cache line along with an index (e.g., 0, 1, 2, 3, etc.) to the processing unit that owns the current cache line state, etc.
• the third method utilizes available CCP data to track which processing unit(s) previously owned cache coherence state for a cache line, which cache coherence state can then be used to identify which processing unit(s) has logged the value of a cache line.
• logging by reference may involve recording CCP data for a cache line (e.g., again potentially avoiding recording memory addresses and memory values).
• causing the data for the memory location to be logged on behalf of the first processing unit by reference to log data that was previously logged on behalf of the second processing unit could comprise logging CCP data referencing the second processing unit.
• a fourth method logs by reference to cache way.
• set-associative caches can be used for determining whether a cache line has been logged through use of way-locking. For example, suppose way-locking is used to reserve one or more ways for processing unit P2, and that P2 logs a first influx of data. The first influx also results an (N+z)-level cache (e.g., cache L2-A1) storing data of the first influx in a cache line associated with that way. When another processing unit (e.g., Pl) has a second influx of the same data, the presence of this cache line in the (N+z)-level cache indicates that P2 has already logged the data.
• N+z-level cache e.g., cache L2-A1
• Embodiments can log a reference to P2’s log data based on noting the way in which the cache line is stored and can again potentially avoid logging memory addresses and memory values.
• This embodiment can also be used in connection with recording ordering information (e.g., MINs) to order events between Pl and P2.
• MINs recording ordering information
• causing the data for the memory location to be logged on behalf of the first processing unit by reference to log data that was previously logged on behalf of the second processing unit could comprise one or more logging a reference to a cache way, or logging a reference to a cache way and ordering data.
• processing unit Al could cause a cache miss in an N-level cache (e.g., the Ll-Al cache) for particular data at a memory location.
• the cache hierarchy can import that data into the Ll-Al cache, and potentially also into an (N+/)-level cache (e.g., the L2-A1 cache and/or the L3-A cache).
• the influx can be logged by value for processing unit Al .
• method 300 can comprise evicting a first cache line in the first N-level cache that corresponds to the memory location, while retaining a second cache line in the (N+z)-level cache that corresponds to the memory location.
• processing unit Al causes a subsequent cache miss in the Ll-Al cache for the same data
• the retained cache line(s) in an (N+z)-level cache e.g., the L2-A1 and/or L3-A caches
• this subsequent cache miss is logged in reference to the prior log entry by processing unit Al .
• a log entry could be omitted entirely for this subsequent cache miss— because processing unit Al already has the data in its trace.
• method 300 can comprise, based on detecting a subsequent influx to the first N-level cache, the subsequent influx also comprising the data stored at the memory location, causing the subsequent influx to be logged by reference based on presence of the second cache line. Additionally, or alternatively, method 300 can comprise (i) detecting a subsequent influx to the first N-level cache based on additional code execution at the first processing unit, the subsequent influx also comprising the data stored at the memory location, and (ii) based at least on detecting the subsequent influx to the first N-level cache, and based at least on presence of the second cache line, determining that that the subsequent influx need not be logged.
• the embodiments herein create bit-accurate“time travel” trace recordings based on tracing the effects of execution across a plurality of processing units using at least two tiers or layers of processor caches.
• These techniques can be extended to“N” levels of caches. Recording trace files in this manner may need only modest processor modifications and, when compared to prior trace recording approaches, it can reduce by several orders of magnitude both the performance impact of trace recording as well as trace file size.

Landscapes

• Engineering & Computer Science (AREA)
• Theoretical Computer Science (AREA)
• General Engineering & Computer Science (AREA)
• Physics & Mathematics (AREA)
• General Physics & Mathematics (AREA)
• Quality & Reliability (AREA)
• Computer Hardware Design (AREA)
• Computing Systems (AREA)
• Mathematical Physics (AREA)
• Memory System Of A Hierarchy Structure (AREA)
• Debugging And Monitoring (AREA)
• Data Exchanges In Wide-Area Networks (AREA)

Priority Applications (15)

Applications Claiming Priority (2)

Publications (1)

Family

ID=65520481

Family Applications (1)

Country Status (16)

Cited By (4)

Families Citing this family (13)

Citations (1)

Family Cites Families (81)

• 2018
  • 2018-02-23 US US15/904,072 patent/US10496537B2/en active Active
• 2019
  • 2019-02-13 MX MX2020008664A patent/MX2020008664A/es unknown
  • 2019-02-13 AU AU2019223883A patent/AU2019223883B2/en active Active
  • 2019-02-13 WO PCT/US2019/017737 patent/WO2019164710A1/en not_active Ceased
  • 2019-02-13 SG SG11202007582RA patent/SG11202007582RA/en unknown
  • 2019-02-13 ES ES19707256T patent/ES2927911T3/es active Active
  • 2019-02-13 CN CN201980014858.4A patent/CN111742302B/zh active Active
  • 2019-02-13 BR BR112020013505-4A patent/BR112020013505A2/pt not_active IP Right Cessation
  • 2019-02-13 EP EP19707256.4A patent/EP3740871B1/en active Active
  • 2019-02-13 CA CA3088563A patent/CA3088563A1/en active Pending
  • 2019-02-13 MY MYPI2020004174A patent/MY203042A/en unknown
  • 2019-02-13 JP JP2020544520A patent/JP7221979B2/ja active Active
  • 2019-02-13 KR KR1020207026919A patent/KR102645481B1/ko active Active
• 2020
  • 2020-07-03 ZA ZA2020/04083A patent/ZA202004083B/en unknown
  • 2020-08-11 IL IL276652A patent/IL276652B2/en unknown
  • 2020-08-20 PH PH12020551325A patent/PH12020551325A1/en unknown

Patent Citations (1)

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events