The present invention relates to the automatic detection of defects in memory devices and more specifically to detection of defects in memory devices during runtime.
Memory devices used in computing applications typically have a small number of unavoidable defects in memory cells that are created during the manufacturing process. Furthermore, defects in memory cells of memory devices can occur during the operational lifetime of the memory devices. To overcome these defects, Content Addressable Memory (CAM) schemes may be used to map the defective memory cells in a memory device such that the defects can be avoided during runtime. In a conventional Random Access Memory (RAM) device, a user supplies a memory address and the RAM device returns the data word stored at that address. In contrast, CAM is designed such that the user supplies a data word and the CAM searches the entire memory to see if that data word is stored anywhere in the memory. If the data word is found, the CAM returns a list of one or more storage addresses where the word was found (and in some architectures, it also returns the data word, or other associated pieces of data).
A defect mapping process may be used to fully test the memory device and map any defects that are found such that subsequent processes may use the memory while avoiding the defects. During defect mapping, the mapping process fully monopolizes the memory device being testing. Therefore, the defect mapping process is typically only used at the time of manufacture of the memory device before the memory device is released for use, or during some dedicated initialization step in a larger application or hardware initialization process. However, as the mapping process fully monopolizes the memory during mapping, the mapping process cannot be used during runtime as the mapping process would prevent other applications from accessing the memory device.
In general, in one aspect, the present invention addresses the foregoing situation through the use of an automated scan test of a Double Data Rate (DDR) Synchronous Dynamic Random Access Memory (SDRAM) performed in the background during runtime.
In another aspect of the invention, a controller for scan testing a memory includes a control state machine configured to control a memory scan of the memory, a writable control store storing a test sequence used by a test state machine to test a test region of the memory, a pattern generation data unit responsive to the test state machine for generating a test pattern used by the test state machine to test the test region of the memory, a configuration register read by the control state machine and a fault location register written to by the control state machine.
In another aspect of the invention, the control state machine is further configured to copy the test region of the memory into a replacement memory, the test state machine is further configured to perform a test scan on the test region of the memory using the test sequence and the control state machine is further configured to copy the replacement memory back into the test region of the memory.
In another aspect of the invention, the controller is further configured to receive a processor unit, a memory segment start address, and a memory segment end address; and the control state machine is further configured to scan the memory using a plurality of incremental test scans using the test sequence starting at the segment start address and ending at the segment end address.
In another aspect of the invention, the control state machine is further configured to wait between individual test scans.
In another aspect of the invention, the controller is further configured to receive from a processor unit a loop count; and the control state machine is further configured to repeat the test scans for a plurality of times according to the loop count.
In another aspect of the invention, the control state machine is further configured to wait between individual test scans.
In another aspect of the invention, the controller is further configured to operate in a background mode.
A more complete understanding of the invention can be obtained by reference to the following detailed description in connection with the attached drawings.
The DAS controller 100 controls the scan process. The DAS controller 100 is separate from a DDR controller 101; however the DAS controller 100 works in tandem with the DDR controller 101. In one implementation, the DDR controller 101 is used to temporarily store the data of the DDR SDRAM that are being scanned.
The DAS controller 100 includes configuration (CFG) registers 102, status and fault location registers 104, a control unit 106, and a data unit 108. The control unit 106 also includes a control state machine 110 and a test state machine 111. The test state machine 111 is programmable using operational instructions stored in a Writable Control Store (WCS) Random Access Memory (RAM) 112. The data unit 108 includes a pattern register 114 and a read data check unit 116.
Test flow is controlled by the control unit 106. The control unit 106 has two state machines to control the flow of the test. The control state machine 110, described in connection with
The control unit 106 controls the test using the test state machine 111. The test state machine 111 is responsible for generating control signals 120 transmitted to the data unit 108 and for updating the status and fault location register 104 as well.
In one implementation, the control unit 106 maintains the following counters:
The data unit 108 controls write data pattern generation and read data checking Write data pattern generation is done using the pattern register 114. The pattern register 114 is loaded by the control unit 106. The contents of the pattern register 114 can be shifted by 1 bit at a time. In one implementation, the shifting is controlled by a “shift_en_cu” signal from the control unit 106. The actual write data 130 written to the memory 118 can be a pattern stored in the pattern register 114, the inversion of the pattern stored in the pattern register 114, or all 0s. In one implementation, the selection of the pattern is controlled by the control unit 106 using a “data_select_cu” signal.
The data read from the memory 118 is checked against the expected data 132, which is generated in the same manner as the write data 130. In one implementation, if there is a mismatch between the read data 134 and the expected data 132, the control unit 106 is informed using an error signal “rd_err_du”.
In one implementation, the DAS controller has following configuration, status and fault location registers:
In operation (in one implementation), the DAS controller 100 receives (136) a test sequence, a segment start and end address, and a test loop count from a microprocessor unit (MPU) (not shown). The MPU then transmits a ddr_scan_on signal to tell the DAS controller 100 to begin scanning the memory 118. The location of any defects found in the memory 118 are stored in the status and fault location register 104.
Having described the structure of the DAS controller 100 in reference to
Referring now to
Referring now to
Referring now to
Referring now to
If the copy in is successful, the control state machine 200 transitions to a check end state 232. If a segment end is reached and a test count has not been exceeded, the control state machine 200 transitions back to the load target start state 206 and loads the starting address of a new segment of the memory to test. However, if the segment end is reached and the test count is exceeded, the control state machine 200 transitions (236) back to the idle state 202.
In one embodiment of DAS controller 100 a burst size for the auto scan request is not more than the other agent in the same slot. Furthermore, the DAS controller 100 is placed in the lowest priority request channel in its slot. In this way, the DAS controller operations will have little or no impact on overall system performance.
In another embodiment of DAS controller 100, if there is any attempt by another device to write to a scan test region, the control state machine 200 will abort the copy out operation and wait for a period of time before restarting the copy out from the beginning of the scan test region.
In another embodiment of DAS controller 100, if an error is reported during testing, the control state machine 200 will stop and raise status and/or interrupts for the MPU to intervene. The MPU can abort the test or resume the test by programming a specified control bit.
In another embodiment of DAS controller 100, if another device attempts to write to the scan test region during the copy in, the control state machine 200 will abort the copy in and wait for a period of time and re-start the copy in from the beginning of the scan test region.
In another embodiment of DAS controller 100, the start and end addresses of a memory segment to be tested can be programmatically set allowing the DAS controller to test all or part of a memory device.
In another embodiment of DAS 100, the DAS controller 100 operates in the background mode compared to other processes accessing a memory device being scanned. In this way, the DAS controller 100 does not interfere with the other processes' use of the memory device.
In another embodiment of DAS controller 100, the DAS controller 100 only utilizes side bandwidth of a processing system that has not already been allocated to higher priority applications.
In another embodiment of DAS controller 100, the testing is performed periodically, interleaving burst modes and waiting modes, in order to conserve energy.
Turning now to
In a read WCS state 252, the test state machine 250 reads an instruction out of the WCS RAM 112. In a decode instruction state 254, the test state machine 250 decodes the instruction read out of the WCS RAM 112. In an execute state, the test state machine 250 executes the instruction read out of the WCS RAM. If there are more instructions to read out, the test state machine transitions (258) to the read WCS state 252.
In one implementation, the use of a WCS RAM 112 allows a DAS controller to generate different test patterns. For example, instructions for a walking 1, walking 0 or any fixed pattern may be programmed into the WCS RAM 112.
In one embodiment of DAS controller 100, a modified walking 1 pattern is used. As a typical DDR memory has a 32 bit word size, a 32 bit pattern of “0x0000—00001” is loaded into a pattern register and shifted by 1 for each location in a scanned memory segment so that every location will receive a unique pattern. However, after 32 locations, the pattern will start repeating. Hence, the pattern won't be unique anymore. To overcome this, after 32 locations, an inverted version of the pattern is used. This way 64 word locations (of 32 bits each) can be tested. As a 1 KB block of a scanned memory will have 128 words, the block can be tested in 2 parts. In this case, the instructions for the test sequence are:
As another example of a programmable test pattern, a walking 0 pattern can be implemented in a similar manner as the modified walking 1 pattern as discussed above.
Walking 1 and 0 patterns will catch almost all issues related to decoding problems or word-line interference in a memory device. To make the test pattern more robust, the test can be repeated 32 times each time such that the starting pattern can be shifted so that each bit in the wordline is stressed.
Alternatively, an all toggle pattern can be used which will toggle bits within a wordline. Such a pattern will catch interference among bit cells within the wordline as well as interference among the bit cells of the neighboring wordlines.
In order to accommodate different sequences, test sequences are programmed by firmware using WCS RAM 112. The code from the WCS RAM 112 is executed by the DAS controller 100 when the control state machine 200 reaches the test in progress state 214 (both of
The following tables are illustrative of commands and variables used to program WCS RAM 112 in accordance with an exemplary embodiment of the invention and are presented by way of illustration and not of limitation.
A set address command sets the starting address of subsequent WCS instructions. A write pattern command writes the pattern (16 bit) for subsequent WCS instructions. For example, to program a walking 1 test followed by a marching increment, in one implementation, the WCS instructions are as follows:
In a memory management system employing DAS controller 100 in accordance with an exemplary embodiment of the invention, a buffer manager will have a SRAM to temporarily store the data of the test target block of the DDR memory. In one implementation, the DAS controller 100 will have a configuration bit “ddr2sram_map_en”, upon setting of which the buffer manager will automatically route any request to the test target block of the DDR memory to the ASRAM as illustrated in
The buffer manager decodes the copy out or copy in command and treats the command as if the requesting agent is ASRAM, for example, data source (for copy in) or data sink (for copy out) is ASRAM. The command to the buffer manager has the following 2 bit encoding:
In a memory management system employing DAS controller 100 in accordance with an exemplary embodiment of the invention, during the auto scan by the DAS controller 100, if any location in memory is detected as a faulty location, the address of the location will be stored in the buffer manager and the location's contents will be stored in local SRAM. Subsequently, all access requests to the DDR memory will be snooped and if the address of the faulty location is hit, then the access will be routed to this local SRAM.
To perform this operation, an 8 deep CAM can be used where faulty location addresses are stored as contents of CAM.
As the DDR memory read has a 3 clock cycle latency, the CAM searching need not to be finished in single cycle. Therefore, as a simple alternative to CAM, the faulty location address can be stored in registers and 8 comparators can be used to check a hit. Then, the 2nd cycle can be used to read the data of that location from the local SRAM.
Referring now to
HDD 1500 may communicate with a host device (not shown) such as a computer, mobile computing devices such as personal digital assistants, cellular phones, media or MP3 players and the like, and/or other devices via one or more wired or wireless communication links 1508. HDD 1500 may be connected to memory 1509, such as random access memory (RAM), a low latency nonvolatile memory such as flash memory, read only memory (ROM) and/or other suitable electronic data storage.
Referring now to
DVD drive 1510 may communicate with an output device (not shown) such as a computer, television or other device via one or more wired or wireless communication links 1517. DVD 1510 may communicate with mass data storage 1518 that stores data in a nonvolatile manner. Mass data storage 1518 may include a hard disk drive (HDD) such as that shown in
Referring now to
HDTV 1520 may communicate with mass data storage 1527 that stores data in a nonvolatile manner such as optical and/or magnetic storage devices. At least one HDD may have the configuration shown in
Referring now to
The present invention may also be embodied in other control systems 1540 of vehicle 1530. Control system 1540 may likewise receive signals from input sensors 1542 and/or output control signals to one or more output devices 1544. In some implementations, control system 1540 may be part of an anti-lock braking system (ABS), a navigation system, a telematics system, a vehicle telematics system, a lane departure system, an adaptive cruise control system, a vehicle entertainment system such as a stereo, DVD, compact disc and the like. Still other implementations are contemplated.
Powertrain control system 1532 may communicate with mass data storage 1546 that stores data in a nonvolatile manner. Mass data storage 1546 may include optical and/or magnetic storage devices for example hard disk drives HDD and/or DVDs. At least one HDD may have the configuration shown in
Referring now to
Cellular phone 1550 may communicate with mass data storage 1564 that stores data in a nonvolatile manner such as optical and/or magnetic storage devices for example hard disk drives HDD and/or DVDs. At least one HDD may have the configuration shown in
Referring now to
Set top box 1580 may communicate with mass data storage 1590 that stores data in a nonvolatile manner. Mass data storage 1590 may include optical and/or magnetic storage devices for example hard disk drives HDD and/or DVDs. At least one HDD may have the configuration shown in
Referring now to
Media player 600 may communicate with mass data storage 610 that stores data such as compressed audio and/or video content in a nonvolatile manner. In some implementations, the compressed audio files include files that are compliant with MP3 format or other suitable compressed audio and/or video formats. The mass data storage 610 may include optical and/or magnetic storage devices for example hard disk drives HDD and/or DVDs. At least one HDD may have the configuration shown in
Referring to
The present invention may be implemented as part of either or both signal processing and/or control circuits, which are generally identified in
VoIP phone 620 may communicate with mass data storage 623 that stores data in a nonvolatile manner such as optical and/or magnetic storage devices, for example hard disk drives HDD and/or DVDs. At least one HDD may have the configuration shown in
The invention has been described above with respect to particular illustrative embodiments. It is understood that the invention is not limited to the above-described embodiments and that various changes and modifications may be made by those skilled in the relevant art without departing from the spirit and scope of the invention.
This application claims the benefit of U.S. Provisional Patent Application No. 60/884,319 filed Jan. 10, 2007, the contents of which are hereby incorporated by reference as if fully stated herein.
| Number | Name | Date | Kind |
|---|---|---|---|
| 6484271 | Gray | Nov 2002 | B1 |
| 6724668 | Ohmura et al. | Apr 2004 | B2 |
| 7454670 | Kim et al. | Nov 2008 | B2 |
| 7478285 | Fouquet-Lapar | Jan 2009 | B2 |
| 7661044 | Astigarraga et al. | Feb 2010 | B2 |
| 20030041295 | Hou et al. | Feb 2003 | A1 |
| 20060101305 | Corbin | May 2006 | A1 |
| 20090144583 | Bruennert et al. | Jun 2009 | A1 |
| 20090164837 | Swanson et al. | Jun 2009 | A1 |
| Number | Date | Country | |
|---|---|---|---|
| 60884319 | Jan 2007 | US |