ACCESSING DATA STORED IN A COMMAND / ADDRESS REGISTER DEVICE
Indirect access to memory subsystem registers via an address or command bus addresses the challenge of accessing register data without extra hardware, ensuring system stability by preventing faulty instructions and enabling parity error detection in DDR4 systems.
Patent Information
- Application Number
- DE112011106086
- Authority / Receiving Office
- DE · DE
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2011-12-22
- Publication Date
- 2024-11-07
- Estimated Expiration
- 2031-12-22
AI Technical Summary
Existing memory subsystems lack efficient mechanisms to access data stored in registers without requiring additional hardware, routing traces, or special states, especially in DDR4 systems where parity error checking is performed outside DRAM, leading to potential system hangs.
A mechanism to indirectly access registers via an address or command bus by transferring data through the address bus to a connected storage device, allowing data to be read from the register without a direct data bus connection, enabling parity error detection and troubleshooting during normal system operation.
Enables access to register data without additional hardware, prevents faulty instructions from executing, and allows for system troubleshooting by logging and reading parity errors, maintaining system stability in DDR4 systems.
Smart Images

Figure 00000000_0000_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present invention relates to a device comprising a DRAM with a multi-purpose register (MPR) and a method for operating such a device. COPYRIGHT NOTICE / PERMISSION
[0002] Portions of the disclosure of this patent may contain material that is subject to copyright protection. The copyright owner does not object to the reproduction of this patent or patent disclosure by third parties because it appears in the patent database or records of the Patent and Trademark Office, but otherwise reserves all copyright rights. The copyright notice applies to all data described below and in the accompanying drawings, as well as to any software described below: Copyright © 2011, Intel Corporation. All rights reserved. BACKGROUND
[0003] Certain memory subsystems include register devices connected to the memory subsystem's address or command bus to store values associated with an instruction or configuration within the memory subsystem. Traditionally, there has been no good way to access such data. Therefore, it is not easy to access data stored in the register for configuration purposes (e.g., data stored in a mode register), data stored for debug or error detection purposes (e.g., data stored in a C / A register device), or other data in another such register. One option to access the data is to include a connection to the memory subsystem's data bus. Such an option is very expensive in terms of hardware (additional pins) and routing traces.Another option is to place the device into a special state (e.g., a management mode) to allow the temporary use of existing buses or other connections for a new purpose. This option results in a slow connection, prevents continued operation of the device, and may still require additional hardware. Another option is to provide an out-of-band serial interface on the register device, which also adds hardware and routing overhead. Therefore, there are currently no traditional mechanisms that allow access to the data stored in the memory subsystem's registers using standard instructions and minimal hardware requirements.
[0004] Memory subsystems supporting newer DDR (double data rate) memory standards add an additional layer of complexity by providing a register and logic to perform parity error checking at the register device rather than at the memory device. For example, DDR4 (a standard still under development at the time of this filing) allows for instruction / address (C / A) parity error checking outside of DRAM (dynamic random access memory). However, without a mechanism to read a parity error, the instruction is still sent to DRAM for execution, which would result in a computer device hang (e.g., a "blue screen" condition).
[0005] EP 1 849 161 B1 describes a method in which data not stored in a DRAM array of an SDRAM module is read from the SDRAM module by means of synchronous data transmission. In particular, an SDRAM module 100 and a controller 12 are described, wherein the controller 102 directs commands to the SDRAM module 100 through specific control signals. The controller provides a plurality of address lines to the SDRAM module 100, with a bidirectional data bus connecting the controller 102 and the SDRAM module 100.
[0006] US 2009 / 0119451 A1 describes a memory module comprising address register circuitry and data conversion circuitry connected to the D / Q terminals of a DRAM chip. The data conversion circuitry includes a first data register for storing data signals read from the system memory controller for writing to the DRAM chips, and a second data register for storing data signals read from the DRAM chips for output to the system memory controller.
[0007] US 2010 / 0302873 A1 describes a mode register readout controller comprising a switching signal generator that generates a switching signal that is activated when a reset command is input during a mode register readout operation. The mode register readout controller is intended to prevent or minimize the occurrence of a malfunction due to insufficient readout latency when a reset command is input during a high-frequency mode register readout operation. SUMMARY OF THE INVENTION
[0008] The present invention is defined in the independent claims. The dependent claims specify selected, preferred features.
[0009] Methods, devices, examples and aspects described below which do not fall within the scope of the claims are intended to provide a better understanding of the background or subject matter of the invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0010] The following description includes a discussion of figures with illustrations depicting example implementations of embodiments of the invention. The drawings are to be considered as exemplary, and not limiting. As used herein, references to one or more "embodiments" are intended to describe a particular feature, structure, or characteristic included in at least one implementation of the invention. Thus, phrases such as "in one embodiment" or "in an alternative embodiment" appearing herein describe various embodiments and implementations of the invention and do not necessarily all refer to the same embodiment. However, they are not necessarily mutually exclusive either. Fig.1 is a block diagram of one embodiment of a system including a memory subsystem with a register accessible via an address or instruction bus, but not via a data bus. Fig. 2 is a block diagram of one embodiment of a system including a memory subsystem with a register between a memory controller and a DRAM. Fig. 3 is a block diagram of one embodiment of a system including a memory subsystem with a register accessible via an address or command bus but not a data bus and a memory device accessible via both the address or command bus and the data bus. Fig. 4 is a block diagram of one embodiment of a system including a memory subsystem with a register device between a memory controller and a DRAM, where the register device performs a parity check. Fig. 5 is a flow diagram of one embodiment of a method for accessing data from a register accessible via an address or instruction bus, but not via a data bus. Fig. 6 is a block diagram of an embodiment of a computer system wherein a register of the memory subsystem is accessed indirectly by a host processor. Fig. 7 is a block diagram of an embodiment of a mobile device wherein a register of the memory subsystem is accessed indirectly by a host processor.
[0011] Descriptions of specific details and implementations follow, including a description of the figures, which may illustrate some or all of the embodiments described below, and also discuss other potential embodiments or implementations of the inventive concepts presented herein. An overview of embodiments of the invention is provided below, followed by a more detailed description with reference to the drawings. DETAILED DESCRIPTION
[0012] As described herein, a register in a memory subsystem is connected to an address bus. However, it should be understood that reference to an address bus can refer to a bus that carries only addresses or to a bus that carries both addresses and instructions. Many address buses are address or instruction buses because addresses and instructions, or combinations of addresses and instructions, are sent over the buses. Reference herein to an "address bus" should be understood as a reference to an address or instruction bus. Therefore, a register is accessible via an address or instruction bus, but not via a data bus. The data can be read by transferring data over the address bus from the register to a device connected to the data bus, from which the data is read over the data bus, even if the register is not connected to the data bus.The register is located in a register device, which is connected via the address bus to a memory device, which is connected to both the address bus and the data bus. A host processor triggers the register device to transfer information via the address bus to a register on the memory device. The host processor then reads the information from the memory device's register. With this type of read mechanism, an "indirect" reading of the register is possible.
[0013] The read mechanism operates on any register device connected to the address bus, such as a mode register or a parity error-checking register device. A specific memory configuration is therefore accessible during normal runtime execution of the host operating system. Additionally, it is possible to access data related to a specific instruction or address that resulted in a parity error, which could prevent access to the host system and cause it to malfunction. Therefore, the described read mechanism allows the system to be debugged if a memory instruction error occurs.
[0014] Indirect register reading works well in proposed DDR4 systems. DDR4 specifies a register device that can perform C / A parity error checking and store faulty instructions. Therefore, with DDR4, the instruction can be stopped before execution, and the mechanism for accessing the faulty instruction helps locate the faulty instruction, which in turn aids error recovery.
[0015] Fig.1 is a block diagram of one embodiment of a system including a memory subsystem with a register accessible via an address bus but not a data bus. System 100 represents a computing device or mobile device in which the isolated register 130 is isolated from the host processor 110. Host processor 110 executes instructions stored in the memory subsystem 120. Host processor 110 generally issues a command to access data. The command may include a physical or virtual address pointing to a specific memory location within memory devices of the memory subsystem 120.
[0016] The isolated register 130 is not directly accessed by the host processor 110. Register 130 is therefore "isolated" from direct access by the host processor 110, although it is connected to elements of the memory subsystem 120 such as an address bus, which is also connected to a memory controller and memory devices (see the following Fig. 2 and Fig. 3 for more detailed examples). System 100 includes a control mechanism, as is known in the art, for loading data and code or instructions into memory subsystem 120 for execution by processor 110.
[0017] In one embodiment, host processor 110 issues a command that causes data to be stored in register 130, which cannot then be directly accessed by host processor 110. Examples include mode register values, certain debug values, parity error information, or other information. In such an embodiment, host processor 110 issues a command that is passed to register 130 to cause the register to copy or transfer its contents to a register on a storage device of storage subsystem 120. The storage device is connected to a data bus and can therefore respond to a command from host processor 110 by loading data onto the data bus that is readable by host processor 110.
[0018] In one embodiment, BIOS (Basic Input / Output System) 140 includes code that can be executed by host processor 110 to trigger a read of register 130 and process the contents stored therein. For example, host processor 110 can be configured to access debug or error correction code stored in BIOS 140 upon the occurrence of certain events. Therefore, host processor 110 can be caused to execute a debug state or error correction state that accesses the contents of register 130 and determines what actions to perform in response to the read contents.
[0019] Fig.2 is a block diagram of one embodiment of a system including a memory subsystem with a register between a memory controller and DRAM. System 200 represents a computing device or mobile device in which a register is located between memory controller 220 and DRAM 240, and may be an example of the system 100 of Fig. 1. Host processor 210 executes instructions stored in memory subsystem 202. Host processor 210 may access data or code stored in DRAM 240 to execute instructions. In one embodiment, memory subsystem 202 includes other memory resources (not shown) in addition to DRAM 240.
[0020] The host processor 210 accesses DRAM 240 by generating a command or request, which it sends to the memory controller 220 via host bus 212. Host bus 212 represents any connections over which the host processor 210 can provide a command or request related to memory access. In one embodiment, the memory controller 220 sends all access requests through register 230 rather than directly to the DRAM 240. In another embodiment (see, for example, Fig. 3 below), the memory controller can access DRAM 240 directly and register 230 is located on an address bus to which the memory controller and DRAM are connected.
[0021] In one embodiment, address bus 222 and address bus 224 represent the address bus of memory subsystem 202, where address bus 222 is the address bus as between memory controller 220 and register 230, and address bus 224 is the address bus as between register 230 and DRAM 240. Register 230 may be located on the address bus between memory controller 220 and DRAM 240 to perform an operation on access commands sent from memory controller 220 to DRAM 240. In one embodiment, memory controller 220 sends all commands to DRAM 240 through register 230. In one embodiment, for example, register 230 provides parity error checking. In the event of a detected parity error, register 230 does not forward the erroneous command.Therefore, register 230 between memory controller 220 and DRAM 240 can prevent the DRAM from operating on a bad instruction that would otherwise cause device 200 to become unresponsive (i.e., hang).
[0022] In an implementation where register 230 provides parity error checking, register 230 logs information about the parity error when it is detected. When a parity error is detected, in addition to preventing the command from reaching DRAM 240, register 230 triggers a parity error, which is detected by memory controller 220 and / or DRAM 240. The parity error is indicated in system 200. Host processor 210 detects the parity error and may attempt to read the contents of register 230 to obtain the parity error log information. The register may then transfer its contents to DRAM 240 in response to a read command to allow host processor 210 to access the error log and determine how to proceed. Host processor 210 can access the transferred contents of register 230 as well as other information stored in DRAM 240 via data bus 242.DRAM 240 is connected to data bus 242, while register 230 is not.
[0023] Fig. 3 is a block diagram of one embodiment of a system including a memory subsystem with a register accessible via an address bus but not a data bus, and a memory device accessible via both the address bus and the data bus. System 300 represents a computing device wherein host processor 310 indirectly accesses data from register 320. System 300 may be an example of the system 100 of Fig. 1.
[0024] While register 230 of Fig.2 is connected between the memory controller and memory device, register 320 is not connected between memory controller 340 and DRAM 330. Instead, register 320 is connected to memory controller 340 in parallel with DRAM 330 via address bus 302. Register 320 is not connected to data bus 304. DRAM 330 is connected to data bus 304. Memory controller 340 may or may not be connected to data bus 304. In one embodiment, memory controller 340 is coupled to host processor 310 via command bus 306, which may or may not be part of data bus 304.
[0025] Register 320 logs information such as memory configuration values or debug values. When host processor 310 executes instructions indicating that the value(s) logged in register 320 are about to be read, host processor 310 issues a command over command bus 306 to memory controller 340, which then provides a command on address bus 302. Host processor 310 is not directly connected to address bus 302. Register 320 responds to the command by transferring data to DRAM 330 over address bus 302.
[0026] In one embodiment, the command for register 320 causes register 320 to transfer to a specific location in DRAM 330. The register specifies the location with the select code or information. Therefore, a read trigger command can indicate to a register of DRAM 330 where the protocol information should be written. In one embodiment, DRAM 330 includes multiple memory devices, 332-0, 332-1 ..., 332-N, where N is an integer greater than or equal to zero. In one embodiment, each device includes one or more general-purpose registers (MPRs), illustrated in device 332-0 as MPR[3:0]. There may be more or fewer than four registers in each DRAM.
[0027] In one embodiment, memory controller 340 selects the location to which register 320 writes its contents by selecting a location select and / or write enable for the desired location in DRAM 330. The read trigger that causes register 320 to transfer its contents to DRAM 330 may include a selection code to indicate a specific MPR. In one embodiment, the format of the read trigger indicates register 320 as the transfer source, to be written to a specified MPR as the destination. Therefore, the read trigger may be a command similar in structure to other standard commands in system 300.
[0028] Generally, the read trigger causes register 320 to transfer protocol information to a register in DRAM 330, which can then be read by host processor 310. Host processor 310 issues a command that ultimately causes the read trigger to be sent to register 320. In one embodiment, the command is considered the read trigger by host processor 310 and forwarded by memory controller 340. In another embodiment, host processor 310 issues a command that causes the memory controller to generate a read trigger, which it then sends to register 320.
[0029] Fig. 4 is a block diagram of one embodiment of a system including a memory subsystem with a register device between a memory controller and a DRAM device, wherein the register device performs a parity check. As mentioned above with respect to the examples of Fig. 1 and Fig.2, in one embodiment, a register device located between the memory controller and the storage device may perform a parity check. In system 400, register device 420 is located between memory controller 410 and DRAM 430.
[0030] Register device 420 includes register 422 where it logs parity check information. Register 422 is a register according to any embodiment known in the art. Briefly, a register includes a volatile group of digital bits held in a group of circuit elements. In one embodiment, register device 420 also includes parity check logic 424, which is hardware and / or software executing in processing resources to perform parity check functions. A parity check is generally a fairly simple logical operation to verify that a calculated parity matches an expected parity, and it can be easily implemented in simple digital circuits.
[0031] Register device 420 is connected to DRAM 430 via address bus 444. DRAM 430 is connected to a host processor via data bus 442. DRAM 430 includes memory resources 432 that store the data in the DRAM. DRAM 430 also includes one or more MPRs 434, which can essentially function as a small intermediate register for the DRAM. In one embodiment, DRAM 430 includes parity check logic 436, which can be implemented in a manner similar to the parity check logic 424 of register device 420. In one embodiment, either one or both of parity check logic 424 and parity check logic 436 are selectively enabled. Therefore, parity checking can be enabled or disabled at both register device 420 and DRAM 430. In one embodiment, when parity checking is enabled at register device 420, it is disabled at DRAM 430.
[0032] For the following description, it is assumed that parity check logic 424 is enabled. When parity check logic 424 detects a parity error, it logs parity error information to register 422, which may be a C / A register. Register device 420 transfers the registered error information and error status information to DRAM 430. The same mechanism that register device 420 uses to transfer error log information and error status information to DRAM 430 can be used to transfer control words from register device 420. In one embodiment, register device 420 is located on a DIMM (Dual Inline Memory Module). System 400 sets the bits of register 422 by issuing an address command from memory controller 410. Based on the address, register device 420 sets bits in register 422.As described below, standard address instructions can then be used to cause register device 420 to transfer data to MPR 434 to be read via data bus 442. Because register device 420 is isolated from data bus 442, it cannot directly place data on the data bus for access by a host processor.
[0033] In one embodiment, register device 420 is configured to check for a parity (PAR) signal generated one clock cycle after a chip select signal (CS#). In an alternative embodiment, the parity signal may be generated on the same clock cycle as CS# instead of on clock cycle+1. Register device 420 checks for parity without errors before forwarding the command to DRAM 430. Transferring the error log from register device 420 would function the same regardless of whether parity check logic 436 in DRAM 430 is enabled (i.e., regardless of whether the DRAM is checking for parity errors).
[0034] In one embodiment, the method for logging and reading the log information from register device 420 is as follows. The method allows for recovery from a parity error, whereas in traditional approaches, a parity error would cause a system error. Register device 420 detects a parity error in the instruction or address. Register 422 logs the misdirected C / A (column / address) frame (RCW (register control word) location C0 .. FF). In one embodiment, the bits logged are C2-C0, ACT_n, BG1-BG0, BA1-BA0, PAR, A17, A16 / RAS_n, A15 / CAS_n, A14 / WE_n, and A13:0 for a total of 26 bits. The register device 420 sets the parity error status bit to “1” to indicate the parity error in the system 400.A significant contrast to previous parity error checking techniques is that the register device 420 can prevent a faulty instruction from being executed by holding the misdirected C / A frame and not forwarding it to the DRAM 430.
[0035] In one embodiment, register device 420 asserts a signal to memory controller 410 to indicate the parity error. The signal could be an ALERT_n signal asserted to the memory controller after a delay of tPAR_ALERT_ON_reg for the duration of tPAR_ALERT_PW_reg. Register device 420 disables parity checking when a parity error is detected and only proceeds with parity checking after memory controller 410 resets a parity error status bit to "0." In one embodiment, MPR 434 includes multiple MPR registers having different ranks. In one embodiment, MPR page 0 is a readable and writable register, while MPR pages 1, 2, and 3 are read-only.
[0036] Therefore, memory controller 410 can enable an "MPR Page 0" mode in DRAM 430 at rank 0 by setting MR3 bits A2=1 and A1:A0=00 (page 0). Read and write operations can therefore be directed to page 0.
[0037] In one embodiment, memory controller 410 selects the error log register in an RCW select control word and initiates one or more commands to write the contents of register 422 to MPR page 0. In one embodiment, memory controller 410 issues four "send 8-bit RCW to MPR" commands by writing four times to a command space control word (address 3F). Such a sequence can transfer 32 bits (which may be the entire size of register 422) of the error log register to page 0 of MPR1, MPR2, and MPR3. Using rank 0 on the DIMM for MPR writes is convenient when all DIMMs are rank 0. Other ranks may be used instead of rank 0.
[0038] In one embodiment, for simplicity, the bit mapping to page 0 can follow the same pattern as the bit mapping in page 1. Page 1 is used by the DRAM for error logs, but can be read-only. If page 1 is read-only, information can be written to a writable page (e.g., page 0) with the same mapping, so the same instructions and processing can be used to understand the error information.
[0039] In an exemplary embodiment, a write to page 0 (MPR1) is a write transaction with BA1:BA0=00 and address A7:A0, which is mapped as follows: A
[15] / CAS# A
[14] / WE# A
[13] A
[12] A
[11] A
[10] A[9] A[8]
[0040] After the error log and error status are transferred from register 422 to DRAM page 0, memory controller 410 may use read commands to read page 0. Therefore, a host processor may generate read commands to cause the memory controller to use read commands to access the data.
[0041] Memory controller 410 can disable MPR operation by programming MR3 A2=0 to DRAM, which initiates normal data flow. The memory controller can also re-enable parity checking in register device 420 by resetting the parity error signal as mentioned above.
[0042] In one embodiment, the following register specification details may apply. Table 1 - RC3F: Command Space Cmd (DA[3:0]) Cmd No Name of the command Result 0 0 0 0 CMD0 Reserved 0 0 0 1 CMD1 Reserved 0 0 1 0 CMD2 Reserved 0 0 1 1 CMD3 Reserved 0 1 0 0 CMD4 Send 8-bit RCW to MPRx Sends selected RCW to DRAM MPRO page 0. 0 1 0 1 CMD5 Send two 4-bit RCWs to MPRx Sends selected RCWs to DRAM MPRO page 0. 0 1 1 0 CMD6 Reserved 0 1 1 1 CMD7 Reserved 1 0 0 0 CMD8 Reserved 1 0 0 1 CMD9 Reserved 1 0 1 0 CMD10 Reserved 1 0 1 1 CMD11 Reserved 1 1 0 0 CMD12 Reserved 1 1 0 1 CMD13 Reserved 1 1 1 0 CMD14 Reserved 1 1 1 1 CMD15 Reserved Table 2 - RC8x: RCW selection control word definition for 8-bit RCWs Setting (DA[7:0]) definition Coding 0 MPR bit 1 MPR bit 0 A12 A11 A10 A9 A8 RCW address for the RCW read operation Selects 1 of 32 8-bit RCW addresses MPR-Bits[1:0]=00 selects MPRO MPR-Bits1:0=01 selects MPR1 MPR-Bits[1:0]=10 selects MPR2 MPR-Bits[1:0]=11 selects MPR3 1 MPR bit 1 MPR bit 0 A12 A11 A10 A9 A8 RCW address for the RCW read operation with autoincrement [1] Selects 1 of 32 8-bit RCW addresses MPR-Bits[1:0]=00 selects MPRO MPR-Bits[1:0]=01 selects MPR1 MPR-Bits[1:0]=10 selects MPR2 MPR-Bits[1:0]=11 selects MPR3
[0043] [1] “Send 8-bit RCW to MPRx” instructions auto-increment the address field by 1 and the MPR bit field in the RCW select control word by 1. Table 3 - RC4x: RCW selection control word definition for 4-bit RCWs Setting (DA[7:0]) definition Coding 0 MPR bit 1 MPR bit 0 A9 A8 A7 A6 A5 RCW address for the RCW read operation Selects 1 of 32 duplicate 4-bit RCW addresses MPR-Bits1:0=00 selects MPRO MPR-Bits[1:0]=01 selects MPR1 MPR-Bits[1:0]=10 selects MPR2 MPR-Bits[1:0]=11 selects MPR3 1 MPR bit 1 MPR bit 0 A9 A8 A7 A6 A5 RCW address for the RCW read operation with auto-increment [1] Selects 1 of 32 duplicate 4-bit RCW addresses MPR-Bits[1:0]=00 selects MPRO MPR-Bits[1:0]=01 selects MPR1 MPR-Bits[1:0]=10 selects MPR2 MPR-Bits[1:0]=11 selects MPR3
[0044] [1] “Send two 4-bit RCWs to MPRx” instructions auto-increment the address field by 1 and the MPR bit field in the RCW select control word by 1.
[0045] However, it is obvious that the register control word (RCW) is part of the register device. The DRAM includes a mode register (MR), and the register device includes the RCW. The RCW and MR have a similar function for their two different connected devices. The memory controller first selects an RCW to read and then writes the desired address (the address to be read) as the source. The memory controller can then generate a destination control word to specify the destination (the address to be written to). Each selected bit in the command space control word is used to define the command. There may be bits that are reserved or otherwise unused.
[0046] To achieve register writing in one embodiment, the memory controller sets the source, the destination, and then the command space control word, which generates and sends the control word. The system can be configured so that only the source and destination do nothing in and of themselves.
[0047] The RCW selection control word is written before "Send 8-bit RCW to MPRx" or "Send two 4-bit RCWs to MPRx" instructions. In one embodiment, the register device sends the following bits at QxA[7:0] outputs: Table 4 - Sends RCW to DRAM bit map for 4-bit RCWs QxA7 QxA6 QxA5 QxA4 QxA3 QxA2 QxA1 QxA0 upper address RCW bit 3 upper address RCW bit 2 upper address RCW bit 1 upper address RCW bit 0 lower address RCW bit 3 lower address RCW bit 2 lower address RCW bit 1 lower address RCW bit 0 Table 5 - Send to DRAM bit mapping for 8-bit RCW QxA7 QxA6 QxA5 QxA4 QxA3 QxA2 QxA1 QxA0 RCW Bit 7 RCW Bit 6 RCW Bit 5 RCW Bit 4 RCW Bit 3 RCW Bit 2 RCW Bit 1 RCW Bit 0
[0048] In one embodiment, the control word locations C0..FF function as a 32-bit error log register. In one embodiment, after a parity error occurs, the register device logs the following sampled instruction and address bits in the error log register, which can be transferred by the memory controller to page 0 of the DRAM MPR, where they can be read by the host system (via the processor). Table 6 - RCC0..RCFF: Error log register Tax word Setting (DA[7:0]) RCCx A7 A6 A5 A4 A3 A2 A1 A0 RCDx A17 / CAS_n A14 / WE_n A13 A12 A11 A10 A9 A8 RCEx DPAR ACT_n BG1 BG0 BA1 BA0 A17 A16 / RAS_n RCFx Reserved CA parity error status [1] Reserved C2-STAND ARD C1 C0
[0049] [1] The register device sets this bit after a CA parity error occurs. Writing this bit to "0" resumes parity checking. While this bit is set and the device is in one of the two cycle N+1 parity modes, the device does not assert any of its QxCSy_n outputs.
[0050] In one embodiment, indirect reading of a register device as described below is applied to DDR4, which can be implemented with the parity check in a register device as described above. Regarding DDR4 MPR mode, one embodiment of DDR4 DRAM includes four pages of MPR registers, each page having four MPR locations. Page 0 has four programmable 8-bit MPR locations used for DQ bit pattern storage. These MPR registers are written by first placing the DRAM in "data flow from / to MPR" mode by setting a mode register A2=1 in MR3. The MPR page is also set using bits A1:A0 in MR3.
[0051] In this mode, an MRS (Mode Register Set) instruction is used to program the MPR. For an MRS instruction, the address bus is used for the data. Once programmed, the registers can be accessed with read commands in "MPR Operation" mode to send the MPR bits on the DQ bus to the memory controller or host. DDR4 MPR mode enable and page selection are achieved with a mode register instruction as shown below. Table 7 - MR3 address Operating mode Description A2 MPR operation 0=Normal 1=Data flow from / to MPR A1:A0 MPR selection 00=Page0 01=Page1 10=Page2 11=Page 3
[0052] As mentioned above, four MPR pages are provided in DDR4 SDRAM. Page 0 is for both read and write, and pages 1, 2, and 3 are read-only. Each MPR location (MPR0-3) in page 0 can be read by any of three output modes (serial, parallel, or staggered), but pages 1, 2, and 3 only support serial output mode.
[0053] After booting, the contents of MPR page 0 should contain default values as predefined for the DRAM. MPR page 0 can only be written to if an MPR write command is issued by the memory controller. Unless an MPR write command is issued, the DRAM must permanently retain the default values and should not change its contents independently for any purpose. Table 8 - MPR default values address MPR location [7] [6] [5] [4] [3] [2] [1] [0] A NOTICE BA1:BA0 00=MPR0 0 1 0 1 0 1 0 1 Read / Write (default) 01=MPR1 0 0 1 1 0 0 1 1 10=MPR2 0 0 0 0 1 1 1 1 11=MPR3 0 0 0 0 0 0 0 0
[0054] However, it is apparent that the reference, even a detailed reference as provided above with respect to DDR4 and DDR4 parity error checking, is merely a non-limiting example. As described below, the register indirect read mechanism can operate for any register isolated from a data bus but coupled to a device coupled to the data bus.
[0055] Fig.5 is a flow diagram of one embodiment of a method for accessing data from a register accessible via an address bus but not a data bus. In one embodiment, a memory controller sends a command to a register device that causes the register device to store or log information in a register, 502. The information may be parity error information, debug information, configuration information, or other information that may be stored in a register isolated from the system data bus.
[0056] In one embodiment, the information to be stored in the register is configuration information, and the trigger is a command from the memory controller to write the configuration information. In one embodiment, the information to be stored is parity error information, and the trigger is the detection by the register device that a parity error has occurred. In one embodiment, the information to be stored is debug information, and the trigger is a command from the memory controller, in response to the debug software, to write data to the register. In response to the trigger, the register device logs the information to the register, 504.
[0057] The memory controller detects a read trigger to read the information stored in the register, 506. The memory controller generally receives such a trigger from a process executed by the host processor. The process may be a program executing on the host, or it may be stored as code in the device BIOS. In one embodiment, the memory controller prepares to read the register by identifying a source and a destination in response to the read trigger, 508. In one embodiment, the read trigger is a command that is of a form that identifies a specific register location as the source and a specific MPR location of the memory device as the destination. The memory controller may prepare the command to the register device and DRAM by identifying the specific locations based on the host's identification of the information to be read.
[0058] In response to the command from the memory controller, the register device writes its contents (the source) to the specified MPR location (the destination), 510. The host can then read the information from the destination in the memory device using standard read commands over the system data bus, 512. The host that requested the information determines one or more actions based on the information read, 514. The actions may be related to setting or updating the configuration, storing data in memory or sending data over a network, identifying an instruction with a parity error and issuing a corrected instruction, or any other action.
[0059] In one embodiment, a BIOS controls at least one specific command to the register device. For example, in response to a parity error, the BIOS may see that an error has occurred and then go back to the host to discover the specific error. The specific error is visible by reading the register device as noted above. In one embodiment, in response to an error, the BIOS may attempt to read every device register in the system to determine where the error originated and what the specific error is. Software controls the writing and reading of the registers and commands. The BIOS then determines what actions to take based on the specific error.
[0060] Fig.6 is a block diagram of one embodiment of a computer system wherein a register of the memory subsystem is accessed indirectly by a host processor. System 600 represents a computing device according to any embodiment described herein and may be a laptop, a desktop computer, a server, a gaming or entertainment control system, a scanner, a copier, a printer, or other electronic device. System 600 includes processor 620, which provides processing, operational management, and instruction execution for system 600. Processor 620 may include any type of microprocessor, central processing unit (CPU), processor core, or other processing hardware to provide processing for system 600.Processor 620 controls the general operation of system 600 and may be or include one or more programmable general-purpose or special-purpose microprocessors, digital signal processors (DSPs), automation systems, application-specific integrated circuits (ASICs), programmable logic devices (PLDs), or the like, or a combination of such devices.
[0061] The memory subsystem 630 represents the main memory of the system 600 and provides temporary storage for code to be executed by the processor 620 or data contents to be used in executing a routine. The memory subsystem 630 includes memory 632, which represents one or more storage devices that may include read-only memory (ROM), flash memory, one or more variants of random access memory (RAM), or other storage devices, or a combination of such devices. In one embodiment, memory 632 includes at least one DRAM. The memory subsystem 630 includes register 634, which represents a register that is not directly readable by the processor 620. Such a register may be referred to as an isolated register and may be read indirectly as described herein.
[0062] The memory subsystem 630 stores and hosts, among other things, the operating system (OS) 636 to provide a software platform for instruction execution in the system 600. Additionally, other instructions 638 are stored and executed by the memory subsystem 630 to provide the logic and processing of the system 600. The OS 636 and the instructions 638 are executed by the processor 620.
[0063] Processor 620 and memory subsystem 630 are coupled to bus / bus system 610. Bus 610 is an abstraction representing one or more separate physical buses, communication lines / interfaces, and / or point-to-point connections connected by suitable bridges, adapters, and / or controllers. Therefore, bus 610 may include, for example, one or more of a system bus, Peripheral Component Interconnect (PCI) bus, HyperTransport or Industry Standard Architecture (ISA) bus, Serial ATA (SCSI) bus, a Universal Serial Bus (USB), or Institute of Electrical and Electronics Engineers (IEEE) Standard 1394 bus (commonly referred to as "FireWire"). The buses of bus 610 may also correspond to interfaces in network interface 650.
[0064] In one embodiment, bus 610 includes a data bus, which is a data bus included in memory subsystem 630, over which processor 630 can read values from memory 632. The additional line shown connecting processor 620 to memory subsystem 630 represents an instruction bus over which processor 620 provides instructions and addresses to access memory 632. Register 634 is connected to a data bus of memory subsystem 630, but not to the data bus of bus 610.
[0065] System 600 also includes one or more input / output (I / O) interfaces 640, network interface 650, one or more internal mass storage devices 660, and peripheral interface 670 coupled to bus 610. I / O interface 640 may include one or more interface components through which a user interacts with system 600 (e.g., video, audio, and / or alphanumeric interface). Network interface 650 provides system 600 with the ability to communicate with remote devices (e.g., servers, other computing devices) over one or more networks. Network interface 650 may include an Ethernet adapter, wireless interconnect components, USB (universal serial bus), or other wired or wireless standards-based or proprietary interfaces.
[0066] Memory 660 may be or include any conventional medium for storing large amounts of data in a non-volatile manner, such as one or more magnetic disks, semiconductor disks, or optical disks, or a combination thereof. Memory 660 maintains code or instructions and data 662 in a permanent state (i.e., the value is maintained despite the interruption of power to system 600). Memory 660 may generally be considered a "memory," although memory 630 is the executing or operational memory for providing instructions to processor 620. While memory 660 is non-volatile, memory 632 may include volatile memory (i.e., the value or state of the data is indeterminate when power to system 600 is interrupted).
[0067] Peripheral interface 670 may include any hardware interface not specifically identified above. Peripherals generally refer to devices that connect dependently to system 600. A dependent connection is one in which system 600 provides the software and / or hardware platform on which the operation is performed and with which a user interacts.
[0068] Fig. 7 is a block diagram of one embodiment of a mobile device wherein a register of the memory subsystem is accessed indirectly by a host processor. Device 700 represents a mobile device, such as a computing tablet, a mobile phone or smartphone, a wireless e-reader, or other mobile device. It should be understood that certain components are shown generically, and not all components of such a device are shown in device 700.
[0069] Device 700 includes processor 710, which performs the primary processing operations of device 700. Processor 710 may include one or more physical devices, such as microprocessors, application processors, microcontrollers, programmable logic devices, or other processing means. The processing operations performed by processor 710 include executing an operating platform or operating system on which applications and / or device functions are executed. The processing operations include operations related to I / O (input / output) with a human user or with other devices, operations related to power management, and / or operations related to connecting device 700 to another device. The processing operations may also include operations related to audio I / O and / or display I / O.In one embodiment, device 700 includes audio subsystem 720, which represents hardware (e.g., audio hardware and audio circuitry) and software (e.g., drivers and codecs) components associated with providing audio functions to the computing device. Audio functions may include a speaker and / or headphone output, as well as a microphone input. The devices for such functions may be integrated into device 700 or connected to device 700. In one embodiment, a user interacts with device 700 by providing audio commands that are received and processed by processor 710.
[0070] Display subsystem 730 represents hardware (e.g., display devices) and software (e.g., drivers) components that provide a visual and / or tactile display for a user to interact with the computing device. Display subsystem 730 includes display interface 732, which includes the particular screen or hardware device used to provide a display to a user. In one embodiment, display interface 732 includes logic separate from processor 710 to perform at least some processing associated with the display. In one embodiment, display subsystem 730 includes a touchscreen device that provides both output and input to a user. I / O controller 740 represents hardware devices and software components associated with interacting with a user.The I / O controller 740 may operate to manage the hardware that is part of the audio subsystem 720 and / or the display subsystem 730. Additionally, the I / O controller 740 illustrates a connection point for additional devices connected to device 700 through which a user could interact with the system. For example, devices that may be connected to device 700 could include microphone devices, speakers or stereo systems, video systems or other display devices, a keyboard or keyboard devices, or other I / O devices for use with specific applications, such as card readers or other devices.
[0071] As mentioned above, I / O controller 740 may interact with audio subsystem 720 and / or display subsystem 730. For example, input from a microphone or other audio device may provide input or commands for one or more applications or functions of device 700. Additionally, audio output may be provided instead of or in addition to the display output. In another example, the display device also acts as an input device that may be managed at least in part by I / O controller 740 when the display subsystem includes a touchscreen. There may also be additional buttons or switches on device 700 to provide I / O functions that may be managed by I / O controller 740.
[0072] In one embodiment, the I / O controller 740 manages devices such as accelerometers, cameras, light sensors or other sensors, gyroscopes, global positioning system (GPS) hardware that may be included in the device 700. Input may be part of direct user interaction, as well as providing environmental input to the system to influence its operations (such as filtering noise, adjusting displays for brightness detection, applying a camera flash, or other features).
[0073] In one embodiment, device 700 includes power management 750, which manages battery power usage, battery charging, and features associated with power-saving operation. Memory subsystem 760 includes memory devices for storing information in device 700. The memory may be non-volatile (state does not change when power to the memory device is interrupted) and / or volatile (state is indeterminate when power to the memory device is interrupted).
[0074] Include storage devices. The storage resources of the storage subsystem 760 may store application data, user data, music, photos, documents, or other data, as well as system data (either long-term or temporary) associated with the execution of the applications and functions of the system 700.
[0075] In one embodiment, a memory device 762 includes at least one register directly accessible by the processor 710. The register device 764 of the memory subsystem 760 is located external to the memory devices 762 and is not directly accessible by the processor 710, but is communicatively coupled to the memory device 762 via an address bus as described above. The data stored in the register device 764 can be transferred to the register of the memory device 762 so that it can be read by the processor 710.
[0076] Connectivity 770 includes hardware devices (e.g., wireless and / or wired connectors and communications hardware) and software components (e.g., drivers, protocol stacks) to enable device 700 to communicate with peripheral devices. The device could be separate devices, such as other computing devices, wireless access points or base stations, and peripherals such as headsets, printers, or other devices.
[0077] Connectivity 770 may include several different types of connectivity. For generalization, device 700 is illustrated with cellular connectivity 772 and wireless connectivity 774. Cellular connectivity 772 generally refers to cellular network connectivity provided by cellular service providers, such as GSM (Global System for Mobile Communications) or variants or derivatives, CDMA (Code Division Multiple Access) or variants or derivatives, TDM (Time Division Multiple Access) or variants or derivatives, or other cellular service standards. Wireless connectivity 774 refers to wireless connectivity that is not cellular and may include personal area networks (such as Bluetooth), local area networks (such as WiFi), wide area networks (such as WiMax), and other wireless communications. Wireless communication refers to data transmission through the use of modulated electromagnetic radiation through a non-fixed medium.Wired communication occurs through a solid-state communication medium.
[0078] Peripheral connections 780 include hardware interfaces and connectors, as well as software components (e.g., drivers, protocol stacks) for establishing peripheral connections. However, it will be appreciated that device 700 may be both a peripheral device ("to" 782) and connected to other computing devices ("from" 784). Device 700 generally includes a "docking" connector for connecting it to other computing devices for purposes such as managing (e.g., downloading and / or uploading, switching, synchronizing) content on device 700. Additionally, a docking connector may enable device 700 to connect to certain peripheral devices that allow device 700 to, for example, control content output to audiovisual or other systems.
[0079] In addition to a proprietary docking connector or other proprietary connection hardware, device 700 may establish peripheral connections 780 via common or standards-based connectors. Common types may include a Universal Serial Bus (USB) connector (which may include any of a number of different hardware interfaces), DisplayPort, MiniDisplayPort (MDP), HDMI interface (HDMI), FireWire, or another type.
[0080] Flowcharts as illustrated herein provide examples of sequences of various process actions. Unless otherwise specified, although they are shown in a particular sequence or order, the order of the actions may be modified. Thus, the illustrated embodiments are intended only as an example, and the process may be performed in a different order, and some actions may be performed in parallel. Furthermore, one or more actions may be omitted from various embodiments; thus, not all actions are required in every embodiment. Other process flows are possible.
[0081] To the extent that various operations or functions are described herein, they may be described or defined as software code, instructions, configuration, and / or data. The content may be directly executable (“object” or “executable” form), source code, or difference code (“delta” or “patch” code). The software content of the embodiments described herein may be provided via a product of manufacture on which the content is stored or via a method of operating a communications interface to send data over the communications interface. A machine-readable storage medium may cause a machine to perform the described functions or operations and includes any mechanism that stores information in a form accessible by a machine (e.g., computing device, electronic system, etc.), such as writable / non-writable media (e.g.,Read-only memory (ROM), random access memory (RAM), magnetic disk storage media, optical storage media, flash memory devices, etc.). A communication interface includes any mechanism that can connect to any medium (hardwired, wireless, optical, etc.) to communicate with another device, such as a memory bus interface, a processor bus interface, an Internet connection, a disk controller, etc. The communication interface can be configured by providing configuration parameters and / or by sending signals to prepare the communication interface to provide data signals that describe the software content. The communication interface can be accessed through one or more commands or signals sent to the communication interface.
[0082] Various components described herein may be a means for performing the described operations or functions. Each component described herein includes software, hardware, or a combination thereof. The components may be implemented as software modules, hardware modules, special-purpose hardware (e.g., application-specific hardware, application-specific integrated circuits (ASICs), digital signal processors (DSPs), etc.), embedded controllers, hard-wired circuits, etc.
Claims
[1] Device comprising: a Dynamic Random Access Memory (DRAM) comprising at least one page having at least one general-purpose register (MPR) and comprising: a first interface configured, when operatively coupled to an address bus and receiving a control word (CW), to provide the CW for a particular MPR, and a second interface communicatively coupled to the at least one side having at least one MPR, wherein the device is designed to write the specific MPR using the first interface during a coupling of the first interface to the address bus, and to read the specific MPR using the second interface during a coupling of the second interface to a data bus. [2] The apparatus of claim 1, wherein the CW comprises either a 4-bit CW or an 8-bit CW. [3] The apparatus of claim 1, wherein the particular MPR comprises a page 0 number zero MPR (MPRO), and the CW is associated with a command CMD4. [4] Device according to claim 1, wherein: the CW is assigned to a read operation, the read operation is associated with an 8-bit value having positions [7:0] and having at least one MPR bit 1 in position 6 and one MPR bit 0 in position 5, and the device is designed to select a specific MPR number using a combination of MPR bit 1 and bit 0. [5] Device according to claim 1, wherein: the CW is assigned to a read operation with autoincrement for an address field, the read operation with auto-increment for an address field is assigned to an 8-bit value having positions [7:0] and having at least one MPR bit 1 in position 6 and one MPR bit 0 in position 5, and the device is designed to select a specific MPR number using a combination of MPR bit 1 and bit 0. [6] Device according to one of claims 1-5, further comprising: an address bus communicatively coupled to the first interface; a register communicatively coupled to the address bus; and a memory controller communicatively coupled to the register, the memory controller being configured to: to provide the CW for the register, to select a page, and the MPR to receive the CW on the selected page, and provide one or more commands that cause the CW to be written to the selected page and the MPR. [7] The device of claim 6, further comprising: a data bus communicatively coupled to the second interface, and a processor communicatively coupled to the data bus, wherein the data bus interface is configured to provide content from the particular MPR to the processor via the data bus. [8] The apparatus of claim 6, wherein the register is configured to perform a parity check. [9] Computer-implemented method comprising: receiving a control word (CW) from an address bus interface at a specific multi-purpose register (MPR) of a dynamic random access memory (DRAM), wherein the CW is received during coupling of the address bus interface to an address bus, describing the specific MPR using the address bus interface during coupling of the address bus interface to the address bus, and reading the specific MPR using a data bus interface while coupling the data bus interface to a data bus. [10] The method of claim 9, wherein the CW comprises either a 4-bit CW or an 8-bit CW. [11] The method of claim 9, wherein the particular MPR comprises a page 0 number zero MPR (MPRO), and the CW is associated with a command CMD4. [12] The method of claim 9, wherein: the CW is assigned to a read operation, the read operation is associated with an 8-bit value having positions [7:0] and having at least one MPR bit 1 in position 6 and one MPR bit 0 in position 5, and a combination of MPR bit 1 and bit 0 selects a specific MPR number. [13] The method of claim 9, wherein: the CW is assigned to a read operation with autoincrement for an address field, the read operation with auto-increment for an address field is assigned to an 8-bit value having positions [7:0] and having at least one MPR bit 1 in position 6 and one MPR bit 0 in position 5, and a combination of MPR bit 1 and bit 0 selects a specific MPR number.
Citation Information
Patent Citations
Register read for volatile memory
EP1849161B1
Redriven / Retimed Registered Dual Inline Memory Module
US20090119451A1
Mode-register reading controller and semiconductor memory device
US20100302873A1
