WO2006057841A1 - Method and system for dma optimization in host bus adapters - Google Patents

Method and system for dma optimization in host bus adapters Download PDF

Info

Publication number
WO2006057841A1
WO2006057841A1 PCT/US2005/041094 US2005041094W WO2006057841A1 WO 2006057841 A1 WO2006057841 A1 WO 2006057841A1 US 2005041094 W US2005041094 W US 2005041094W WO 2006057841 A1 WO2006057841 A1 WO 2006057841A1
Authority
WO
WIPO (PCT)
Prior art keywords
write
dma
read
request
burst size
Prior art date
Application number
PCT/US2005/041094
Other languages
French (fr)
Inventor
Bradley S. Sonksen
Kangfu D. Chu
Rajendra R. Gandhi
Original Assignee
Qlogic Corporation
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Qlogic Corporation filed Critical Qlogic Corporation
Publication of WO2006057841A1 publication Critical patent/WO2006057841A1/en

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F13/00—Interconnection of, or transfer of information or other signals between, memories, input/output devices or central processing units
    • G06F13/14—Handling requests for interconnection or transfer
    • G06F13/20—Handling requests for interconnection or transfer for access to input/output bus
    • G06F13/28—Handling requests for interconnection or transfer for access to input/output bus using burst mode transfer, e.g. direct memory access DMA, cycle steal

Definitions

  • the present invention relates to computing systems, and more particularly to optimizing direct memory access (“DMA”) channel performance.
  • DMA direct memory access
  • SANs Storage area networks
  • host systems that include computer systems, servers etc.
  • Host systems typically include several functional components. These components may include a central processing unit (CPU) , main memory, input/output ("I/O") devices, and streaming storage devices (for example, tape drives) .
  • the main memory is coupled to the CPU via a system bus or a local memory bus.
  • the main memory is used to provide the CPU access to data and/or program information that is stored in main memory at execution time.
  • the main memory is composed of random access memory (RAM) circuits.
  • RAM random access memory
  • Host systems often communicate with storage systems via a host bus adapter ("HBA", may also be referred to as a “controller” and/or “adapter”) using an interface, for example, the "PCI” bus interface.
  • PCI stands for Peripheral Component Interconnect, a local bus standard that was developed by Intel Corporation ®. The PCI standard is incorporated herein by reference in its entirety.
  • PCI-Express is another Input/Output (“I/O") bus standard (incorporated herein by reference in its entirety) that is compatible with existing PCI cards using the PCI bus.
  • PCI-Express uses discrete logical layers to process inbound and outbound information. The logical layers are a Transaction Layer, a Data Link Layer (“DLL”) and a Physical Layer (“PHY”) .
  • DLL Data Link Layer
  • PHY Physical Layer
  • PCI-Express uses separate links to transmit and receive information.
  • PCI-Express uses a packet-based protocol to exchange information between Transaction layers. Transactions are carried out using Requests and Completions.
  • TLPs Transaction Layer Packets
  • the Transaction Layer assembles and disassembles Transaction Layer Packets ("TLPs") .
  • TLPs are used to communicate transactions, such as read and write and other type of events .
  • Fibre channel is one such standard. Fibre channel (incorporated herein by reference in its entirety) is an American National Standard Institute (ANSI) set of standards, which provides a serial transmission protocol for storage and network protocols such as HIPPI, SCSI, IP, ATM and others.
  • ANSI American National Standard Institute
  • DMA modules are used by HBAs to perform data transfers between memory locations, or between memory locations and an input/output port. DMA units provide address and bus control signals to and from a device for a read and/or write cycle.
  • a DMA read request is a request from a DMA module (or channel) to an arbitration module to transfer data from a host system to a storage device.
  • a DMA write request is a request from a DMA module to an arbitration module to transfer data from the storage device to a host system.
  • Specific channels are implemented in a DMA unit to allow storage devices to transfer data directly to and from memory storage devices.
  • a channel can be activated by a DMA request signal (DREQ) from a storage device or a host system.
  • the DMA unit receives the DREQ, provides a DMA acknowledged signal (DACK) , and transfers the data over the channel to or from the storage device.
  • DREQ DMA request signal
  • DACK DMA acknowledged signal
  • HBAs typically use multiple DMA channels and have an arbitration module that arbitrates access to a PCI-Express link. This allows an HBA to arbitrate and switch contexts (between channels) by actively processing command, status and data. Multiple channels are serviced in periodic bursts.
  • DMA write requests may be processed by writing data using multiple PCI-Express write request packets. A new DMA request (read or write) can only be processed if a previous DMA read and/or DMA write request has been processed.
  • DMA read requests (with no data) are issued to generate Read Transfer requests, which transfers data from the host to a storage device.
  • Write requests from the storage device to a host system are generated using the same side of the PCI-Express interface. The same transmit link (in the PCI-Express interface) is shared for write transfers and read requests.
  • the time to service a write request can be longer compared to servicing the read request because write request packets (TLPs) also transfer payload data. Also, delay in issuing read request packets can stall data transfer from a host to a storage device.
  • TLPs write request packets
  • a storage area network (“SAN")
  • the SAN includes a HBA that uses a dynamic DMA maximum write burst count sizing to optimize processing of write and read requests, wherein the HBA includes a DMA optimizer module that selects a certain write burst size to adjust performance when read and write DMA requests are being utilized.
  • the DMA optimizer module can toggle between write and read request priority based on a maximum write request burst size. A shorter maximum write burst size provides more opportunity to issue read requests and a larger maximum burst size provides a better write request performance.
  • a HBA includes a DMA optimizer module that selects a certain write burst size to adjust performance when read and write DMA requests are being utilized for sizing a maximum DMA write burst count to optimize processing of write and read requests.
  • a method for processing direct memory access requests in a HBA includes, evaluating a read request throughput rate; evaluating a write request throughput rate; evaluating a read request utilization rate; evaluating a write request utilization rate; and adjus.ting a maximum write burst size.
  • the maximum write burst size is increased if the read request utilization rate and the write request utilization rate are higher than a threshold value and the read throughput rate is greater than the write throughput rate.
  • the maximum write burst size is decreased if the read request utilization rate and the write request utilization rate are higher than a threshold value and the write throughput rate is greater than the read throughput rate.
  • Figure IA is a block diagram showing various components of a SAN
  • Figure IB is a block diagram of a host bus adapter with a DMA optimizer, according to one aspect of the present invention
  • FIGS. 2A, 2B and 2C show block/logic diagrams for optimizing DMA read and write requests, according to one aspect of the present invention.
  • FIG. 3 shows a flow diagram for optimizing DMA read and write requests, according to one aspect of the present invention.
  • FIG. IA shows a SAN system 100 that uses a HBA 106 (referred to as "adapter 106") for communication between a host system with host memory 101 to various storage systems (for example, storage subsystem 116 and 121, tape library 118 and 120) using fibre channel storage area networks 114 and 115.
  • Host memory 101 includes a driver 102 that co-ordinates all data transfer via adapter 106 using input/output control blocks ("IOCBs") .
  • Servers 117 and 119 can also access the storage sub-systems using SAN 115 and 114, respectively.
  • a request queue 103 and response queue 104 is maintained in host memory 101 for transferring information using adapter 106.
  • Host system communicates with adapter 106 via a PCI-Express bus 105.
  • HBA 106 :
  • FIG. IB shows a block diagram of adapter 106.
  • Adapter 106 includes processors (may also be referred to as “sequencers”) "XSEQ” 112 and "RSEQ” 109 for receive and transmit side, respectively for processing data received from storage sub-systems and transmitting data to storage sub-systems.
  • Transmit path in this context means data path from host memory 101 to the storage systems via adapter 106.
  • Receive path means data path from storage subsystem via adapter 106. It is noteworthy, that only one processor is used for receive and transmit paths, and the present invention is not limited to any particular number/type of processors.
  • Buffers HlA and HlB are used to store information in receive and transmit paths, respectively.
  • adapter 106 also includes processor 106A, which may be a reduced instruction set computer ("RISC") for performing various functions in adapter 106.
  • processor 106A may be a reduced instruction set computer ("RISC") for performing various functions in adapter 106.
  • RISC reduced instruction set computer
  • Adapter 106 also includes fibre channel interface (also referred to as fibre channel protocol manager "FPM”) 113A that includes modules 113B and 113 in receive and transmit paths, respectively (shown as “FC RCV” and “FC XMT”) .
  • FPM fibre channel protocol manager
  • modules 113B and 113 allow data to move to/from storage systems.
  • Adapter 106 is also coupled to external memory 108 and 110 via connection 116A ( Figure IA) (referred interchangeably, hereinafter) and local memory interface 122.
  • Memory interface 122 is provided for managing local memory 108 and 110.
  • Local DMA module 137A is used for gaining access to move data from local memory (108/110) .
  • Adapter 106 also includes a serial/de-serializer 136 for converting data from 10-bit to 8-bit format and vice- versa.
  • Adapter 106 also includes request queue DMA channel (0) 130, response queue DMA channel 131, request queue (1) DMA channel 132 that interface with request queue 103 and response queue 104; and a command DMA channel 133 for managing command information.
  • DMA channels are coupled to arbiter 107 that receives requests and grants access to a certain channel.
  • Both receive and transmit paths have DMA modules "RCV DATA DMA” 129 and "XMT DATA DMA” 135 that are used to gain access to a channel for data transfer in the receive/transmit paths.
  • Transmit path also has a scheduler 134 that is coupled to processor 112 and schedules transmit operations .
  • a host processor (not shown) sets up shared data structures in buffer memory 108.
  • a host command is stored in buffer 108 and the appropriate sequencer (i.e., 109 or 112) is initialized to execute the command.
  • Various DMA units (or channels, used interchangeably throughout this specification) (for example, 129, 130, 131, 132, 133 and 135) send a request to arbiter 107. When a request is granted, the DMA unit is informed of the grant and memory access is granted to a particular channel.
  • Arbiter 107 is coupled to a PCI-Express Transaction Handler (PTH) 137.
  • PTH 137 is coupled to PCI-Express port logic 137B that moves information to/from a host system.
  • PTH 137 has also been referred to as PCI-Express interface and includes a receive side and transmit side link that allows communication between the host system and adapter 106. The transmit side receives information from adapter 106 and destined for the host and the receive side receives information from adapter 106 and destined for the host system.
  • Arbiter 107 is also coupled to a DMA optimizer module 107A (may also be referred to as module 107A) that is coupled to plural DMA units (for example, 129 and 135) .
  • Module 107A is described below in detail with respect to Figures 2A-2C and 3.
  • DMA OPTIMIZATION
  • DMA arbitration sizing/priorities are enabled, disabled or modified to control DMA write request sizing. This allows write requests to be processed efficiently without significantly reducing read request processing.
  • a maximum burst size for processing write requests is selected.
  • Various maximum burst size may be selected, for example, 512 bytes, 1024 bytes, 1536 bytes, 2048 bytes or any other size. Using a smaller burst size (for example, 512 bytes) provides more opportunities to send read DMA requests. Larger burst sizes allow for more efficient write request processing.
  • Module 107A balances the need for sending read requests without unduly slowing the write request processing, in one aspect of the present invention as described below with respect to Figures 2A, 2B, 2C and 3.
  • read requests are sent without data and read request processing is completed after data is transferred from host via adapter 106 to a storage device.
  • FIG. 2A shows arbiter 107 functionally coupled with module 107A.
  • Plural DMA channels for example, 129 and 135) are coupled with arbiter 107.
  • Each DMA channel has a request pipeline (for example, 129B and 135B) and a segmentation module (for example 129A and 135A) .
  • Sequencer 109 and 112 send channel task commands (129C and 135C) to DMA channels 129 and 135, respectively. The commands are used to generate a request to arbiter 107 (for example, 107D and 107E) .
  • Transmit side DMA channel 135 also includes a read channel 135D that issues read requests 107F to arbiter 107.
  • Segmentation modules 129A and 135A segment a DMA transfer into segments (or blocks) . Segmentation modules 129A and 135A also operate based on certain rules, for example, when and how data blocks should be segmented. These rules can be turned on or off by the firmware.
  • Module 107A includes a DMA request monitoring logic 107B (may also be referred to as “logic 107B" or “module 107B") that monitors both read and write requests.
  • Logic 107B receives information regarding pending requests (for example, 107D, 107E and 107F) .
  • Output from logic 107B is sent to a maximum write request burst count generator (may also be referred to as "generator”) 107C that sends the maximum burst count 107G to segmentation modules 129A and 135A.
  • a maximum write request burst count generator may also be referred to as "generator”
  • Arbiter 107 generates signal 137C that indicates a currently active request (i.e. a request that has won arbitration) .
  • Signal 137D indicates if the request is for a read or write operation.
  • Signal 137E shows the byte count and signal 137F is the acknowledgement after a request is completed.
  • FIG. 2B shows a block diagram for determining the utilization rate for read and write requests.
  • Utilization rate in this context means the number of clocks in a timer interval when a DMA read or write request is pending (or the ratio of clocks used per timer interval and the total clocks per timer interval) .
  • Logic for determining the read utilization rate is shown as 200A and logic for determining write path utilization rate is shown as 200B.
  • DMA read channels 212 (similar to 135D) issue DMA read requests 212A and 212B (similar to, 107F) .
  • Pending read requests 212A and 212B are input into counter 202 that maintains a running count of all clocks where read requests are pending in a given timer interval, whose period is provided by external timer 213.
  • Counter 202 is reset by signal 214 after every timer interval.
  • a pending read request clock count is compared by comparator 201 to a number of clocks per timer interval 218.
  • a request may be designated as "pending” if the request is pending (or unacknowledged) for multiple clocks and increments counter 202..
  • Comparator 201 is enabled by signal 219 from timer 213. Comparator 201 provides the number of clocks with pending requests in a given timer period.
  • Output 201A from comparator 201 is sent to another comparator 204 that also receives input 205A from a threshold register 205.
  • Comparator 201 compares 201A with threshold value 205A.
  • Register 205 holds threshold value 205A for pending read requests.
  • comparator 204 Based on the comparison, comparator 204 generates signal 203 that is sent generator 107C ( Figure 2C) based on which the maximum write burst size may be adjusted, as described below with respect.
  • the write side logic in segment 200B operates similar to the logic elements in 200A.
  • DMA channels 211 generate the pending requests (211A and 211B) .
  • Counter 210 is similar to counter 202 and is reset by signal 216.
  • Comparator 206 (enabled by signal 215) is similar to comparator 201.
  • Comparator 207 (enabled by signal 215) is similar to comparator 204.
  • Threshold register 208 stores threshold values (208A) for the write side similar to threshold register 205.
  • Comparator 207 compares output 206A with a threshold value 208A. Based on the comparison, comparator 207 generates signal 209 that is sent to generator 107C ( Figure 2C) based on which the maximum write burst size may be adjusted, as described below with respect.
  • Figure 2C shows a logic diagram for comparing read/write operation throughput rates.
  • Throughput in this context means the number of bytes transferred for a read/write operation within a timer interval (provided by 213) .
  • the comparison is performed to determine the ratio between the read and write throughput rate on a per timer interval basis.
  • the ratio is compared to a threshold value that provides a range of acceptable values. Based on the comparison, adjustments are made to the maximum write burst count at which data is written for a write request.
  • the threshold value may be represented as a percentage value, for example, a threshold value greater than 0 and less than 100%. If the ratio is within a certain value, for example, greater or equal to 1.0- threshold value and less or equal to 1.0 + threshold value, then no adjustment is made.
  • counter 213B counts the running read/write byte counts.
  • Counter 213B detects ' if a request is acknowledged (ACK 137F) and if it is a read request (READ/WRITE signal 137D) , then the read byte count (BYTE/CNT 137E) is increased. The same is performed for a write request.
  • Counter 213B is reset by signal 213G from timer 213.
  • the running read/write byte count from counter 213B is input to comparator 213D that also receives a threshold value 213H from a register 213C.
  • Comparator 213D is enabled by signal 213A. Comparator 213D performs a relative comparison of read/write throughput rates to determine how far apart the throughput rates for read and write requests can be, before any adjustments to maximum write burst count are made.
  • Firmware for HBA 106 loads the threshold range value 213H (as described above) in register 213C. It is noteworthy that the threshold value 213H may be pre ⁇ set or loaded dynamically.
  • a read throughput rate greater than the write throughput rate is shown as signal 213E, while a write throughput rate greater than the read throughput rate is shown as signal 213F.
  • Signals 203, 209, 213E and 213F are sent to generator 107C. Based on the signal values, as described below, signal 107G is generated.
  • FIG. 3 shows a flow diagram for optimizing DMA read and write request processing, according to one aspect of the present invention. Steps S300, S302, S304 and S306 are performed simultaneously.
  • step S300 read throughput rate (Rt) is evaluated, as described above with respect to figure 2C.
  • Signal 213E is generated if Rt is greater than the write throughput rate (Wt) • ⁇
  • step S302 the write throughput is evaluated, as described above with respect to Figure 2C.
  • Signal 213F is generated if the Wt is greater than Rt.
  • step S304 read utilization rate (Ru) is evaluated, as described above with respect to Figure 2B.
  • Signal 203 is generated after the evaluation and sent to generator 107C.
  • step S306 write utilization rate (Wu) is evaluated, as described above with respect to Figure 2B.
  • Signal 209 is generated after the evaluation and sent to generator 107C.
  • step S308 adjustments are made to the maximum write burst size.
  • Signal 107G is generated and sent to module 135A to adjust segmentation, if needed. If both Wu and Ru are higher than their respective threshold values and Rt is relatively greater than Wt, then the maximum write burst size is increased. This will process pending write requests faster.
  • Module 107A may be configured to use rules similar to those used by PTH 137 to size DMA requests. This can be used to limit the size of write requests and increases the frequency with which read requests are processed. The following are some of the rules used to size the DMA write requests:
  • PCI-Express specification recommends that 128 byte boundaries be used during write requests. If a first PCI-Express write packet does not begin on a 128-byte boundary, then the packets following the first packet are aligned such that subsequent packets are at 128-byte boundary. Module 107A may be used to align the boundaries by sizing the DMA request such that two DMA requests are formed, and this allows an additional read request to be inserted when the 128 byte alignment occurs for a DMA write request.
  • PCI-Express specification requires that read and write requests do not cross a 4KB address boundary. If a DMA request crosses a 4KB boundary, then PTH 137 splits the requests into two different packets. Module 107 is used to size DMA requests such that they are only split at the 4KB boundary for write requests and allows for an additional DMA read request to be inserted when the alignment occurs.
  • Max Payload Size PCI-Express requires that write requests and read responses do not contain more data than a specified payload size that can be programmed by a host. If a DMA write request exceeds the maximum payload size, then it can be split into two different packets by splitting the DMA requests into 2 DMA requests. In this case, DMA write requests are split by module 107A and hence an additional read request may be inserted when maximum payload sizing occurs for DMA write requests.
  • processor 106A can read PCI-Express configuration registers located at PCI port logic 137B. DMA channel write request sizing registers (not shown) are programmable by processor 106A. Hence, processor 106A can program the register bits (not shown) based on maximum payload size or other system configuration settings. It is noteworthy that the foregoing DMA processing may be adjusted statically or dynamically, and module 107A may be enabled or disabled by processor 106A.
  • firmware for HBA 106 is used to set the maximum write count to a certain size, for example, 512K, IK, 1.5K, 2K or any other size, and is adjusted based on the adaptive aspects of the present invention, described above.
  • an initial burst count value is selected, which is based on firmware settings (for example, 512K, IK, 1.5K, 2K or any other size) .
  • firmware settings for example, 512K, IK, 1.5K, 2K or any other size.
  • the value is adjusted when read or write throughput needs to be increased, as described above.

Abstract

Method and system for optimizing DMA request processing is provided. The system includes a HBA that uses a dynamic DMA maximum write burst count sizing to optimize processing of write and read requests, wherein the HBA includes a DMA optimizer module that selects a certain write burst size to adjust performance when read and write DMA requests are being utilized. The DMA optimizer module can toggle between write and read request priority based on a maximum write request burst size. A shorter maximum write burst size provides more opportunity to issue read requests and a larger maximum burst size provides a better write request performance. The method includes, evaluating a read request throughput rate; evaluating a write request throughput rate; evaluating a read request utilization rate; evaluating a write request utilization rate; and adjusting a maximum write burst size.

Description

METHOD AND SYSTEM FOR DMA OPTIMIZATION IN HOST BUS
BACKGROUND
1. Field of the Invention
The present invention relates to computing systems, and more particularly to optimizing direct memory access ("DMA") channel performance.
2. Background of the Invention
Storage area networks ("SANs") are commonly used where plural memory storage devices are made available to various host computing systems. Data in a SAN is typically moved from plural host systems (that include computer systems, servers etc.) to the storage system through various controllers/adapters.
Host systems typically include several functional components. These components may include a central processing unit (CPU) , main memory, input/output ("I/O") devices, and streaming storage devices (for example, tape drives) . In conventional systems, the main memory is coupled to the CPU via a system bus or a local memory bus. The main memory is used to provide the CPU access to data and/or program information that is stored in main memory at execution time. Typically, the main memory is composed of random access memory (RAM) circuits. A computer system with the CPU and main memory is often referred to as a host system.
Host systems often communicate with storage systems via a host bus adapter ("HBA", may also be referred to as a "controller" and/or "adapter") using an interface, for example, the "PCI" bus interface. PCI stands for Peripheral Component Interconnect, a local bus standard that was developed by Intel Corporation ®. The PCI standard is incorporated herein by reference in its entirety. PCI-Express is another Input/Output ("I/O") bus standard (incorporated herein by reference in its entirety) that is compatible with existing PCI cards using the PCI bus. PCI-Express uses discrete logical layers to process inbound and outbound information. The logical layers are a Transaction Layer, a Data Link Layer ("DLL") and a Physical Layer ("PHY") . PCI-Express uses separate links to transmit and receive information.
PCI-Express uses a packet-based protocol to exchange information between Transaction layers. Transactions are carried out using Requests and Completions.
The Transaction Layer assembles and disassembles Transaction Layer Packets ("TLPs") . TLPs are used to communicate transactions, such as read and write and other type of events .
Various other standard interfaces are also used to move data from host systems to storage devices. Fibre channel is one such standard. Fibre channel (incorporated herein by reference in its entirety) is an American National Standard Institute (ANSI) set of standards, which provides a serial transmission protocol for storage and network protocols such as HIPPI, SCSI, IP, ATM and others. DMA modules are used by HBAs to perform data transfers between memory locations, or between memory locations and an input/output port. DMA units provide address and bus control signals to and from a device for a read and/or write cycle. A DMA read request is a request from a DMA module (or channel) to an arbitration module to transfer data from a host system to a storage device. A DMA write request is a request from a DMA module to an arbitration module to transfer data from the storage device to a host system. Specific channels are implemented in a DMA unit to allow storage devices to transfer data directly to and from memory storage devices. A channel can be activated by a DMA request signal (DREQ) from a storage device or a host system. The DMA unit receives the DREQ, provides a DMA acknowledged signal (DACK) , and transfers the data over the channel to or from the storage device.
HBAs typically use multiple DMA channels and have an arbitration module that arbitrates access to a PCI-Express link. This allows an HBA to arbitrate and switch contexts (between channels) by actively processing command, status and data. Multiple channels are serviced in periodic bursts. Typically, DMA write requests may be processed by writing data using multiple PCI-Express write request packets. A new DMA request (read or write) can only be processed if a previous DMA read and/or DMA write request has been processed. DMA read requests (with no data) are issued to generate Read Transfer requests, which transfers data from the host to a storage device. Write requests from the storage device to a host system are generated using the same side of the PCI-Express interface. The same transmit link (in the PCI-Express interface) is shared for write transfers and read requests.
The time to service a write request can be longer compared to servicing the read request because write request packets (TLPs) also transfer payload data. Also, delay in issuing read request packets can stall data transfer from a host to a storage device.
Therefore, there is a need for a method and system to optimize DMA read and write request processing that allows both read and write data transfers to be conducted efficiently. SUMMARY OF THE INVENTION
In one aspect of the present invention, a storage area network ("SAN") is provided. The SAN includes a HBA that uses a dynamic DMA maximum write burst count sizing to optimize processing of write and read requests, wherein the HBA includes a DMA optimizer module that selects a certain write burst size to adjust performance when read and write DMA requests are being utilized. The DMA optimizer module can toggle between write and read request priority based on a maximum write request burst size. A shorter maximum write burst size provides more opportunity to issue read requests and a larger maximum burst size provides a better write request performance. In yet another aspect of the present invention, a HBA is provided that includes a DMA optimizer module that selects a certain write burst size to adjust performance when read and write DMA requests are being utilized for sizing a maximum DMA write burst count to optimize processing of write and read requests.
In yet another aspect of the present invention, a method for processing direct memory access requests in a HBA are provided. The method includes, evaluating a read request throughput rate; evaluating a write request throughput rate; evaluating a read request utilization rate; evaluating a write request utilization rate; and adjus.ting a maximum write burst size. The maximum write burst size is increased if the read request utilization rate and the write request utilization rate are higher than a threshold value and the read throughput rate is greater than the write throughput rate. The maximum write burst size is decreased if the read request utilization rate and the write request utilization rate are higher than a threshold value and the write throughput rate is greater than the read throughput rate.
This brief summary has been provided so that the nature of the invention may be understood quickly. A more complete understanding of the invention can be obtained by reference to the following detailed description of the preferred embodiments thereof concerning the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS
The foregoing features and other features of the present invention will now be described with reference to the drawings of a preferred embodiment. In the drawings, the same components have the same reference numerals. The illustrated embodiment is intended to illustrate, but not to limit the invention. The drawings include the following Figures:
Figure IA is a block diagram showing various components of a SAN; Figure IB is a block diagram of a host bus adapter with a DMA optimizer, according to one aspect of the present invention;
Figures 2A, 2B and 2C show block/logic diagrams for optimizing DMA read and write requests, according to one aspect of the present invention; and
Figure 3 shows a flow diagram for optimizing DMA read and write requests, according to one aspect of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
To facilitate an understanding of the preferred embodiment, the general architecture and operation of a SAN, and a HBA will be described. The specific architecture and operation of the preferred embodiment will then be described with reference to the general architecture of the host system and HBA. SAN Overview:
Figure IA shows a SAN system 100 that uses a HBA 106 (referred to as "adapter 106") for communication between a host system with host memory 101 to various storage systems (for example, storage subsystem 116 and 121, tape library 118 and 120) using fibre channel storage area networks 114 and 115. Host memory 101 includes a driver 102 that co-ordinates all data transfer via adapter 106 using input/output control blocks ("IOCBs") . Servers 117 and 119 can also access the storage sub-systems using SAN 115 and 114, respectively. A request queue 103 and response queue 104 is maintained in host memory 101 for transferring information using adapter 106. Host system communicates with adapter 106 via a PCI-Express bus 105. HBA 106:
Figure IB shows a block diagram of adapter 106. Adapter 106 includes processors (may also be referred to as "sequencers") "XSEQ" 112 and "RSEQ" 109 for receive and transmit side, respectively for processing data received from storage sub-systems and transmitting data to storage sub-systems. Transmit path in this context means data path from host memory 101 to the storage systems via adapter 106. Receive path means data path from storage subsystem via adapter 106. It is noteworthy, that only one processor is used for receive and transmit paths, and the present invention is not limited to any particular number/type of processors. Buffers HlA and HlB are used to store information in receive and transmit paths, respectively.
Beside dedicated processors on the receive and transmit path, adapter 106 also includes processor 106A, which may be a reduced instruction set computer ("RISC") for performing various functions in adapter 106.
Adapter 106 also includes fibre channel interface (also referred to as fibre channel protocol manager "FPM") 113A that includes modules 113B and 113 in receive and transmit paths, respectively (shown as "FC RCV" and "FC XMT") . Modules 113B and 113 allow data to move to/from storage systems.
Adapter 106 is also coupled to external memory 108 and 110 via connection 116A (Figure IA) (referred interchangeably, hereinafter) and local memory interface 122. Memory interface 122 is provided for managing local memory 108 and 110. Local DMA module 137A is used for gaining access to move data from local memory (108/110) . Adapter 106 also includes a serial/de-serializer 136 for converting data from 10-bit to 8-bit format and vice- versa.
Adapter 106 also includes request queue DMA channel (0) 130, response queue DMA channel 131, request queue (1) DMA channel 132 that interface with request queue 103 and response queue 104; and a command DMA channel 133 for managing command information. DMA channels are coupled to arbiter 107 that receives requests and grants access to a certain channel.
Both receive and transmit paths have DMA modules "RCV DATA DMA" 129 and "XMT DATA DMA" 135 that are used to gain access to a channel for data transfer in the receive/transmit paths. Transmit path also has a scheduler 134 that is coupled to processor 112 and schedules transmit operations . A host processor (not shown) sets up shared data structures in buffer memory 108. A host command is stored in buffer 108 and the appropriate sequencer (i.e., 109 or 112) is initialized to execute the command. Various DMA units (or channels, used interchangeably throughout this specification) (for example, 129, 130, 131, 132, 133 and 135) send a request to arbiter 107. When a request is granted, the DMA unit is informed of the grant and memory access is granted to a particular channel.
Arbiter 107 is coupled to a PCI-Express Transaction Handler (PTH) 137. PTH 137 is coupled to PCI-Express port logic 137B that moves information to/from a host system. PTH 137 has also been referred to as PCI-Express interface and includes a receive side and transmit side link that allows communication between the host system and adapter 106. The transmit side receives information from adapter 106 and destined for the host and the receive side receives information from adapter 106 and destined for the host system.
Arbiter 107 is also coupled to a DMA optimizer module 107A (may also be referred to as module 107A) that is coupled to plural DMA units (for example, 129 and 135) . Module 107A is described below in detail with respect to Figures 2A-2C and 3. DMA OPTIMIZATION:
In one aspect of the present invention, based on an incoming Fibre Channel frame size, and maximum payload size allowed by PCI-Express, DMA arbitration sizing/priorities are enabled, disabled or modified to control DMA write request sizing. This allows write requests to be processed efficiently without significantly reducing read request processing. In order to optimize read/write request performance, a maximum burst size for processing write requests is selected. Various maximum burst size may be selected, for example, 512 bytes, 1024 bytes, 1536 bytes, 2048 bytes or any other size. Using a smaller burst size (for example, 512 bytes) provides more opportunities to send read DMA requests. Larger burst sizes allow for more efficient write request processing. Module 107A balances the need for sending read requests without unduly slowing the write request processing, in one aspect of the present invention as described below with respect to Figures 2A, 2B, 2C and 3.
It is noteworthy that read requests are sent without data and read request processing is completed after data is transferred from host via adapter 106 to a storage device.
Figure 2A shows arbiter 107 functionally coupled with module 107A. Plural DMA channels (for example, 129 and 135) are coupled with arbiter 107. Each DMA channel has a request pipeline (for example, 129B and 135B) and a segmentation module (for example 129A and 135A) . Sequencer 109 and 112 send channel task commands (129C and 135C) to DMA channels 129 and 135, respectively. The commands are used to generate a request to arbiter 107 (for example, 107D and 107E) .
Transmit side DMA channel 135 also includes a read channel 135D that issues read requests 107F to arbiter 107.
Segmentation modules 129A and 135A segment a DMA transfer into segments (or blocks) . Segmentation modules 129A and 135A also operate based on certain rules, for example, when and how data blocks should be segmented. These rules can be turned on or off by the firmware. Module 107A includes a DMA request monitoring logic 107B (may also be referred to as "logic 107B" or "module 107B") that monitors both read and write requests. Logic 107B receives information regarding pending requests (for example, 107D, 107E and 107F) . Output from logic 107B is sent to a maximum write request burst count generator (may also be referred to as "generator") 107C that sends the maximum burst count 107G to segmentation modules 129A and 135A.
Arbiter 107 generates signal 137C that indicates a currently active request (i.e. a request that has won arbitration) . Signal 137D indicates if the request is for a read or write operation. Signal 137E shows the byte count and signal 137F is the acknowledgement after a request is completed. Read/Write Channel Utilization Rate Comparison:
Figure 2B shows a block diagram for determining the utilization rate for read and write requests. Utilization rate in this context means the number of clocks in a timer interval when a DMA read or write request is pending (or the ratio of clocks used per timer interval and the total clocks per timer interval) . Logic for determining the read utilization rate is shown as 200A and logic for determining write path utilization rate is shown as 200B. DMA read channels 212 (similar to 135D) issue DMA read requests 212A and 212B (similar to, 107F) . Pending read requests 212A and 212B are input into counter 202 that maintains a running count of all clocks where read requests are pending in a given timer interval, whose period is provided by external timer 213. Counter 202 is reset by signal 214 after every timer interval. A pending read request clock count is compared by comparator 201 to a number of clocks per timer interval 218. A request may be designated as "pending" if the request is pending (or unacknowledged) for multiple clocks and increments counter 202..
Comparator 201 is enabled by signal 219 from timer 213. Comparator 201 provides the number of clocks with pending requests in a given timer period.
Output 201A from comparator 201 is sent to another comparator 204 that also receives input 205A from a threshold register 205. Comparator 201 compares 201A with threshold value 205A. Register 205 holds threshold value 205A for pending read requests.
Based on the comparison, comparator 204 generates signal 203 that is sent generator 107C (Figure 2C) based on which the maximum write burst size may be adjusted, as described below with respect. The write side logic in segment 200B operates similar to the logic elements in 200A. DMA channels 211 generate the pending requests (211A and 211B) . Counter 210 is similar to counter 202 and is reset by signal 216. Comparator 206 (enabled by signal 215) is similar to comparator 201. Comparator 207 (enabled by signal 215) is similar to comparator 204. Threshold register 208 stores threshold values (208A) for the write side similar to threshold register 205. Comparator 207 compares output 206A with a threshold value 208A. Based on the comparison, comparator 207 generates signal 209 that is sent to generator 107C (Figure 2C) based on which the maximum write burst size may be adjusted, as described below with respect. Read/Write Throughput Rate Evaluation:
Figure 2C shows a logic diagram for comparing read/write operation throughput rates. Throughput in this context means the number of bytes transferred for a read/write operation within a timer interval (provided by 213) . The comparison is performed to determine the ratio between the read and write throughput rate on a per timer interval basis. The ratio is compared to a threshold value that provides a range of acceptable values. Based on the comparison, adjustments are made to the maximum write burst count at which data is written for a write request.
The threshold value may be represented as a percentage value, for example, a threshold value greater than 0 and less than 100%. If the ratio is within a certain value, for example, greater or equal to 1.0- threshold value and less or equal to 1.0 + threshold value, then no adjustment is made.
Turning in detail to Figure 2C, counter 213B counts the running read/write byte counts. Counter 213B detects' if a request is acknowledged (ACK 137F) and if it is a read request (READ/WRITE signal 137D) , then the read byte count (BYTE/CNT 137E) is increased. The same is performed for a write request. Counter 213B is reset by signal 213G from timer 213. The running read/write byte count from counter 213B is input to comparator 213D that also receives a threshold value 213H from a register 213C. Comparator 213D is enabled by signal 213A. Comparator 213D performs a relative comparison of read/write throughput rates to determine how far apart the throughput rates for read and write requests can be, before any adjustments to maximum write burst count are made.
Firmware for HBA 106 loads the threshold range value 213H (as described above) in register 213C. It is noteworthy that the threshold value 213H may be pre¬ set or loaded dynamically.
A read throughput rate greater than the write throughput rate is shown as signal 213E, while a write throughput rate greater than the read throughput rate is shown as signal 213F. Signals 203, 209, 213E and 213F are sent to generator 107C. Based on the signal values, as described below, signal 107G is generated.
The term "signal" as used throughout this specification includes a command and/or bit value. Process Flow:
Figure 3 shows a flow diagram for optimizing DMA read and write request processing, according to one aspect of the present invention. Steps S300, S302, S304 and S306 are performed simultaneously.
In step S300, read throughput rate (Rt) is evaluated, as described above with respect to figure 2C. Signal 213E is generated if Rt is greater than the write throughput rate (Wt) •■
In step S302, the write throughput is evaluated, as described above with respect to Figure 2C. Signal 213F is generated if the Wt is greater than Rt.
In step S304, read utilization rate (Ru) is evaluated, as described above with respect to Figure 2B. Signal 203 is generated after the evaluation and sent to generator 107C.
In step S306, write utilization rate (Wu) is evaluated, as described above with respect to Figure 2B. Signal 209 is generated after the evaluation and sent to generator 107C.
Based on signals 213E, 213F, 203 and 209, in step S308 adjustments are made to the maximum write burst size. Signal 107G is generated and sent to module 135A to adjust segmentation, if needed. If both Wu and Ru are higher than their respective threshold values and Rt is relatively greater than Wt, then the maximum write burst size is increased. This will process pending write requests faster.
If both Wu and Rt are higher than their respective threshold values, and Wt is relatively greater than Rt, then the maximum write burst size is decreased. This will allow read requests to be processed faster and optimize overall processing of DMA requests .
Besides the throughput and utilization rates, other factors may also be used to adjust DMA write requests to create time windows for processing read requests at write request boundaries. Module 107A may be configured to use rules similar to those used by PTH 137 to size DMA requests. This can be used to limit the size of write requests and increases the frequency with which read requests are processed. The following are some of the rules used to size the DMA write requests:
128k-byte address boundary: PCI-Express specification recommends that 128 byte boundaries be used during write requests. If a first PCI-Express write packet does not begin on a 128-byte boundary, then the packets following the first packet are aligned such that subsequent packets are at 128-byte boundary. Module 107A may be used to align the boundaries by sizing the DMA request such that two DMA requests are formed, and this allows an additional read request to be inserted when the 128 byte alignment occurs for a DMA write request.
4KB address boundary: PCI-Express specification requires that read and write requests do not cross a 4KB address boundary. If a DMA request crosses a 4KB boundary, then PTH 137 splits the requests into two different packets. Module 107 is used to size DMA requests such that they are only split at the 4KB boundary for write requests and allows for an additional DMA read request to be inserted when the alignment occurs.
Max Payload Size: PCI-Express requires that write requests and read responses do not contain more data than a specified payload size that can be programmed by a host. If a DMA write request exceeds the maximum payload size, then it can be split into two different packets by splitting the DMA requests into 2 DMA requests. In this case, DMA write requests are split by module 107A and hence an additional read request may be inserted when maximum payload sizing occurs for DMA write requests.
It is noteworthy that processor 106A can read PCI-Express configuration registers located at PCI port logic 137B. DMA channel write request sizing registers (not shown) are programmable by processor 106A. Hence, processor 106A can program the register bits (not shown) based on maximum payload size or other system configuration settings. It is noteworthy that the foregoing DMA processing may be adjusted statically or dynamically, and module 107A may be enabled or disabled by processor 106A.
For static implementation, firmware for HBA 106 is used to set the maximum write count to a certain size, for example, 512K, IK, 1.5K, 2K or any other size, and is adjusted based on the adaptive aspects of the present invention, described above.
For a dynamic implementation, an initial burst count value is selected, which is based on firmware settings (for example, 512K, IK, 1.5K, 2K or any other size) . At any given time, the value is adjusted when read or write throughput needs to be increased, as described above. Although the present invention has been described with reference to specific embodiments, these embodiments are illustrative only and not limiting. Many other applications and embodiments of the present invention will be apparent in light of this disclosure and the following claims.

Claims

What is claimed is:
1. A storage area network ("SAN"), comprising: a host bus adapter ("HBA") that uses a dynamic DMA maximum write burst count sizing to optimize processing of write and read requests, wherein the HBA includes a DMA optimizer module that selects a certain write burst size to adjust performance when read and write DMA requests are being utilized.
2. The SAN of Claim 1, wherein the DMA optimizer module can toggle between write and read request priority based on a maximum write request burst size.
3. The SAN of Claim 1, wherein a shorter maximum write burst size provides more opportunity to issue read requests.
4. The SAN of Claim 1, wherein a larger maximum burst size provides a better write request performance.
5. The SAN of Claim 1, wherein a DMA arbitration scheme gives a pending read request higher priority over a write request.
6. The SAN of Claim 1, wherein the DMA optimizer module sizes a DMA write request so that read requests can be processed efficiently.
7. The SAN of Claim 1, wherein the DMA optimizer module balances read and write request processing.
8. A host bus adapter ("HBA"), comprising: a DMA optimizer module that selects a certain write burst size to adjust performance when read and write DMA requests are being utilized.
9. The HBA of Claim 8, wherein the DMA optimizer module can toggle between write and read request priority based on a maximum write request burst size.
10. The HBA of Claim 8, wherein a shorter maximum write burst size provides more opportunity to issue read requests.
11. The HBA of Claim 8, wherein a larger maximum burst size provides a better write request performance.
12. The HBA of Claim 8, wherein a DMA arbitration scheme gives a pending read request higher priority over a write request.
13. The HBA of Claim 8, wherein the DMA optimizer module sizes a DMA write request so that read requests can be processed efficiently.
14. The HBA of Claim 8, wherein the DMA optimizer module balances read and write request processing.
15. A method for processing direct memory access requests in a host bus adapter ("HBA") , comprising: evaluating a read request throughput rate; evaluating a write request throughput rate; evaluating a read request utilization rate; evaluating a write request utilization rate; and adjusting a maximum write burst size.
16. The method of Claim 15, wherein the maximum write burst size is increased if the read request utilization rate and the write request utilization rate are higher than a threshold value and the read throughput rate is greater than the write throughput rate.
17. The method of Claim 15, wherein the maximum write burst size is decreased if the read request utilization rate and the write request utilization rate are higher than a threshold value and the write throughput rate is greater than the read throughput rate.
PCT/US2005/041094 2004-11-22 2005-11-14 Method and system for dma optimization in host bus adapters WO2006057841A1 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US10/994,829 2004-11-22
US10/994,829 US7398335B2 (en) 2004-11-22 2004-11-22 Method and system for DMA optimization in host bus adapters

Publications (1)

Publication Number Publication Date
WO2006057841A1 true WO2006057841A1 (en) 2006-06-01

Family

ID=36084246

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2005/041094 WO2006057841A1 (en) 2004-11-22 2005-11-14 Method and system for dma optimization in host bus adapters

Country Status (2)

Country Link
US (1) US7398335B2 (en)
WO (1) WO2006057841A1 (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8171187B2 (en) 2008-07-25 2012-05-01 Freescale Semiconductor, Inc. System and method for arbitrating between memory access requests

Families Citing this family (49)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2006195821A (en) * 2005-01-14 2006-07-27 Fujitsu Ltd Method for controlling information processing system, information processing system, direct memory access controller, and program
US7802019B2 (en) * 2005-06-14 2010-09-21 Microsoft Corporation Hard disk drive condition reporting and error correction
US7500023B2 (en) 2006-10-10 2009-03-03 International Business Machines Corporation Facilitating input/output processing by using transport control words to reduce input/output communications
DE102006048379B4 (en) * 2006-10-12 2008-11-06 Infineon Technologies Ag Method for throughput control of an electronic circuit and corresponding throughput control and associated semiconductor circuit
US8572302B1 (en) * 2006-10-13 2013-10-29 Marvell International Ltd. Controller for storage device with improved burst efficiency
GB0622408D0 (en) * 2006-11-10 2006-12-20 Ibm Device and method for detection and processing of stalled data request
JP2008234059A (en) * 2007-03-16 2008-10-02 Fujitsu Ltd Data transfer device and information processing system
US20080320240A1 (en) * 2007-06-22 2008-12-25 On Demand Microelectronics Method and arrangements for memory access
US8904077B1 (en) * 2007-06-26 2014-12-02 Qlogic, Corporation Power management for host bus adapters
US7979572B1 (en) 2007-06-28 2011-07-12 Emc Corporation Data storage system having operation code in address portion for atomic operations
US7707367B1 (en) * 2007-06-28 2010-04-27 Emc Corporation Data storage system having separate atomic operation/non-atomic operation paths
US7853735B2 (en) * 2007-12-13 2010-12-14 Emulex Design & Manufacturing Corporation Efficient processing of groups of host access requests that may include zero length requests
US7734854B2 (en) * 2008-01-04 2010-06-08 International Business Machines Corporation Device, system, and method of handling transactions
US8095847B2 (en) 2008-02-14 2012-01-10 International Business Machines Corporation Exception condition handling at a channel subsystem in an I/O processing system
US9052837B2 (en) 2008-02-14 2015-06-09 International Business Machines Corporation Processing communication data in a ships passing condition
US8176222B2 (en) 2008-02-14 2012-05-08 International Business Machines Corporation Early termination of an I/O operation in an I/O processing system
US8082481B2 (en) 2008-02-14 2011-12-20 International Business Machines Corporation Multiple CRC insertion in an output data stream
US7941570B2 (en) 2008-02-14 2011-05-10 International Business Machines Corporation Bi-directional data transfer within a single I/O operation
US8478915B2 (en) 2008-02-14 2013-07-02 International Business Machines Corporation Determining extended capability of a channel path
US8196149B2 (en) 2008-02-14 2012-06-05 International Business Machines Corporation Processing of data to determine compatability in an input/output processing system
US7890668B2 (en) 2008-02-14 2011-02-15 International Business Machines Corporation Providing indirect data addressing in an input/output processing system where the indirect data address list is non-contiguous
US8214562B2 (en) 2008-02-14 2012-07-03 International Business Machines Corporation Processing of data to perform system changes in an input/output processing system
US8117347B2 (en) 2008-02-14 2012-02-14 International Business Machines Corporation Providing indirect data addressing for a control block at a channel subsystem of an I/O processing system
US8108570B2 (en) 2008-02-14 2012-01-31 International Business Machines Corporation Determining the state of an I/O operation
US8166206B2 (en) 2008-02-14 2012-04-24 International Business Machines Corporation Cancel instruction and command for determining the state of an I/O operation
US8312189B2 (en) 2008-02-14 2012-11-13 International Business Machines Corporation Processing of data to monitor input/output operations
US8055807B2 (en) 2008-07-31 2011-11-08 International Business Machines Corporation Transport control channel program chain linking including determining sequence order
US8332542B2 (en) 2009-11-12 2012-12-11 International Business Machines Corporation Communication with input/output system devices
US8583988B2 (en) 2011-06-01 2013-11-12 International Business Machines Corporation Fibre channel input/output data routing system and method
US8738811B2 (en) 2011-06-01 2014-05-27 International Business Machines Corporation Fibre channel input/output data routing system and method
US9021155B2 (en) 2011-06-01 2015-04-28 International Business Machines Corporation Fibre channel input/output data routing including discarding of data transfer requests in response to error detection
US8677027B2 (en) 2011-06-01 2014-03-18 International Business Machines Corporation Fibre channel input/output data routing system and method
US8364854B2 (en) 2011-06-01 2013-01-29 International Business Machines Corporation Fibre channel input/output data routing system and method
US8364853B2 (en) 2011-06-01 2013-01-29 International Business Machines Corporation Fibre channel input/output data routing system and method
US8346978B1 (en) 2011-06-30 2013-01-01 International Business Machines Corporation Facilitating transport mode input/output operations between a channel subsystem and input/output devices
US8473641B2 (en) 2011-06-30 2013-06-25 International Business Machines Corporation Facilitating transport mode input/output operations between a channel subsystem and input/output devices
US8549185B2 (en) 2011-06-30 2013-10-01 International Business Machines Corporation Facilitating transport mode input/output operations between a channel subsystem and input/output devices
US8312176B1 (en) 2011-06-30 2012-11-13 International Business Machines Corporation Facilitating transport mode input/output operations between a channel subsystem and input/output devices
US9645738B2 (en) * 2012-08-09 2017-05-09 Spectra Logic Corporation Direct memory access of remote data
US8677044B1 (en) * 2012-10-25 2014-03-18 Qlogic, Corporation Method and system for communication using multiple DMA channels
US8918542B2 (en) 2013-03-15 2014-12-23 International Business Machines Corporation Facilitating transport mode data transfer between a channel subsystem and input/output devices
US8990439B2 (en) 2013-05-29 2015-03-24 International Business Machines Corporation Transport mode data transfer between a channel subsystem and input/output devices
US9910812B2 (en) 2014-10-02 2018-03-06 Atmel Corporation Initiating multiple data transactions on a system bus
US9734102B2 (en) * 2014-11-04 2017-08-15 Atmel Corporation Data transfer
US9690726B2 (en) 2014-11-11 2017-06-27 Atmel Corporation Peripheral register parameter refreshing
CN108292284A (en) * 2015-11-26 2018-07-17 三菱电机株式会社 Data grass-hopper, data transfer method and data transfer program
US11132145B2 (en) * 2018-03-14 2021-09-28 Apple Inc. Techniques for reducing write amplification on solid state storage devices (SSDs)
CN112256624B (en) * 2020-11-03 2022-09-13 中国人民解放军国防科技大学 DMA communication device, chip, equipment and method for high-speed interconnection network interface chip
CN114840458B (en) * 2022-07-06 2022-09-20 北京象帝先计算技术有限公司 Read-write module, system on chip and electronic equipment

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0542417A1 (en) * 1991-10-17 1993-05-19 Sun Microsystems, Inc. Method and direct memory access controller for asynchronously reading/writing data from/to a memory with improved throughput
JPH10326248A (en) * 1997-05-23 1998-12-08 Brother Ind Ltd Dma controller
US6393500B1 (en) * 1999-08-12 2002-05-21 Mips Technologies, Inc. Burst-configurable data bus
US20020199040A1 (en) * 2001-06-26 2002-12-26 Irwin David S. High speed communications device/system
US6615308B1 (en) * 1999-12-09 2003-09-02 Intel Corporation Method and apparatus for regulating write burst lengths

Family Cites Families (60)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US449182A (en) * 1891-03-31 Process of manufacturing seamless hollow wire
US4268906A (en) 1978-12-22 1981-05-19 International Business Machines Corporation Data processor input/output controller
US4783739A (en) 1979-11-05 1988-11-08 Geophysical Service Inc. Input/output command processor
US4333143A (en) 1979-11-19 1982-06-01 Texas Instruments Input process sequence controller
US4449182A (en) 1981-10-05 1984-05-15 Digital Equipment Corporation Interface between a pair of processors, such as host and peripheral-controlling processors in data processing systems
US4777595A (en) 1982-05-07 1988-10-11 Digital Equipment Corporation Apparatus for transferring blocks of information from one node to a second node in a computer network
US4549263A (en) 1983-02-14 1985-10-22 Texas Instruments Incorporated Device interface controller for input/output controller
US4783730A (en) 1986-09-19 1988-11-08 Datapoint Corporation Input/output control technique utilizing multilevel memory structure for processor and I/O communication
US5276807A (en) 1987-04-13 1994-01-04 Emulex Corporation Bus interface synchronization circuitry for reducing time between successive data transmission in a system using an asynchronous handshaking
US4803622A (en) 1987-05-07 1989-02-07 Intel Corporation Programmable I/O sequencer for use in an I/O processor
US5129064A (en) 1988-02-01 1992-07-07 International Business Machines Corporation System and method for simulating the I/O of a processing system
US5212795A (en) 1988-10-11 1993-05-18 California Institute Of Technology Programmable DMA controller
US5321816A (en) 1989-10-10 1994-06-14 Unisys Corporation Local-remote apparatus with specialized image storage modules
US5249279A (en) 1989-11-03 1993-09-28 Compaq Computer Corporation Method for controlling disk array operations by receiving logical disk requests and translating the requests to multiple physical disk specific commands
US5347638A (en) 1991-04-15 1994-09-13 Seagate Technology, Inc. Method and apparatus for reloading microinstruction code to a SCSI sequencer
EP0551191B1 (en) 1992-01-09 2000-04-12 Cabletron Systems, Inc. Apparatus and method for transferring data to and from host system
US5371861A (en) 1992-09-15 1994-12-06 International Business Machines Corp. Personal computer with small computer system interface (SCSI) data flow storage controller capable of storing and processing multiple command descriptions ("threads")
US5448702A (en) 1993-03-02 1995-09-05 International Business Machines Corporation Adapters with descriptor queue management capability
US5734924A (en) 1993-08-27 1998-03-31 Advanced System Products, Inc. System for host accessing local memory by asserting address signal corresponding to host adapter and data signal indicating address of location in local memory
US6049802A (en) 1994-06-27 2000-04-11 Lockheed Martin Corporation System and method for generating a linked list in a computer memory
US5568614A (en) 1994-07-29 1996-10-22 International Business Machines Corporation Data streaming between peer subsystems of a computer system
US5828903A (en) 1994-09-30 1998-10-27 Intel Corporation System for performing DMA transfer with a pipeline control switching such that the first storage area contains location of a buffer for subsequent transfer
US5761427A (en) 1994-12-28 1998-06-02 Digital Equipment Corporation Method and apparatus for updating host memory in an adapter to minimize host CPU overhead in servicing an interrupt
US5613162A (en) * 1995-01-04 1997-03-18 Ast Research, Inc. Method and apparatus for performing efficient direct memory access data transfers
US5689713A (en) 1995-03-31 1997-11-18 Sun Microsystems, Inc. Method and apparatus for interrupt communication in a packet-switched computer system
US5917723A (en) 1995-05-22 1999-06-29 Lsi Logic Corporation Method and apparatus for transferring data between two devices with reduced microprocessor overhead
US5671365A (en) 1995-10-20 1997-09-23 Symbios Logic Inc. I/O system for reducing main processor overhead in initiating I/O requests and servicing I/O completion events
US5968143A (en) 1995-12-13 1999-10-19 International Business Machines Corporation Information handling system for transfer of command blocks to a local processing side without local processor intervention
US5758187A (en) 1996-03-15 1998-05-26 Adaptec, Inc. Method for enhancing performance of a RAID 1 read operation using a pair of I/O command blocks in a chain structure
US5892969A (en) 1996-03-15 1999-04-06 Adaptec, Inc. Method for concurrently executing a configured string of concurrent I/O command blocks within a chain to perform a raid 5 I/O operation
US5881296A (en) 1996-10-02 1999-03-09 Intel Corporation Method for improved interrupt processing in a computer system
US6233244B1 (en) 1997-02-14 2001-05-15 Advanced Micro Devices, Inc. Method and apparatus for reclaiming buffers
US6115761A (en) 1997-05-30 2000-09-05 Lsi Logic Corporation First-In-First-Out (FIFO) memories having dual descriptors and credit passing for efficient access in a multi-processor system environment
US5905905A (en) 1997-08-05 1999-05-18 Adaptec, Inc. System for copying IOBS from FIFO into I/O adapter, writing data completed IOB, and invalidating completed IOB in FIFO for reuse of FIFO
US6055603A (en) 1997-09-18 2000-04-25 Emc Corporation Method and apparatus for performing pre-request operations in a cached disk array storage system
US6078970A (en) 1997-10-15 2000-06-20 International Business Machines Corporation System for determining adapter interrupt status where interrupt is sent to host after operating status stored in register is shadowed to host memory
US6085277A (en) 1997-10-15 2000-07-04 International Business Machines Corporation Interrupt and message batching apparatus and method
US5983292A (en) 1997-10-15 1999-11-09 International Business Machines Corporation Message transport mechanisms and methods
US6138176A (en) 1997-11-14 2000-10-24 3Ware Disk array controller with automated processor which routes I/O data according to addresses and commands received from disk drive controllers
DE19882822T1 (en) 1997-11-17 2001-03-22 Seagate Technology Method and dedicated frame buffer for loop initialization and for responses
US6006340A (en) 1998-03-27 1999-12-21 Phoenix Technologies Ltd. Communication interface between two finite state machines operating at different clock domains
US6185620B1 (en) 1998-04-03 2001-02-06 Lsi Logic Corporation Single chip protocol engine and data formatter apparatus for off chip host memory to local memory transfer and conversion
US20060117274A1 (en) 1998-08-31 2006-06-01 Tseng Ping-Sheng Behavior processor system and method
US6269413B1 (en) 1998-10-30 2001-07-31 Hewlett Packard Company System with multiple dynamically-sized logical FIFOs sharing single memory and with read/write pointers independently selectable and simultaneously responsive to respective read/write FIFO selections
US6463032B1 (en) 1999-01-27 2002-10-08 Advanced Micro Devices, Inc. Network switching system having overflow bypass in internal rules checker
US6546010B1 (en) 1999-02-04 2003-04-08 Advanced Micro Devices, Inc. Bandwidth efficiency in cascaded scheme
US6434630B1 (en) 1999-03-31 2002-08-13 Qlogic Corporation Host adapter for combining I/O completion reports and method of using the same
US6473814B1 (en) * 1999-05-03 2002-10-29 International Business Machines Corporation System for optimally tuning a burst length by setting a maximum burst length based on a latency timer value and adjusting the maximum burst length based on a cache line size
US6504846B1 (en) 1999-05-21 2003-01-07 Advanced Micro Devices, Inc. Method and apparatus for reclaiming buffers using a single buffer bit
US6564271B2 (en) 1999-06-09 2003-05-13 Qlogic Corporation Method and apparatus for automatically transferring I/O blocks between a host system and a host adapter
US6457090B1 (en) 1999-06-30 2002-09-24 Adaptec, Inc. Structure and method for automatic configuration for SCSI Synchronous data transfers
US6343324B1 (en) 1999-09-13 2002-01-29 International Business Machines Corporation Method and system for controlling access share storage devices in a network environment by configuring host-to-volume mapping data structures in the controller memory for granting and denying access to the devices
US6721799B1 (en) 1999-09-15 2004-04-13 Koninklijke Philips Electronics N.V. Method for automatically transmitting an acknowledge frame in canopen and other can application layer protocols and a can microcontroller that implements this method
US20030056000A1 (en) 2001-07-26 2003-03-20 Nishan Systems, Inc. Transfer ready frame reordering
US6832280B2 (en) * 2001-08-10 2004-12-14 Freescale Semiconductor, Inc. Data processing system having an adaptive priority controller
US6766385B2 (en) * 2002-01-07 2004-07-20 Intel Corporation Device and method for maximizing performance on a memory interface with a variable number of channels
US7337232B2 (en) * 2002-03-28 2008-02-26 Sony Corporation Method and system for providing and controlling sub-burst data transfers
US6961813B2 (en) 2002-06-21 2005-11-01 Hewlett-Packard Development Company, L.P. System and method for providing multi-initiator capability to an ATA drive
TW576977B (en) * 2002-09-11 2004-02-21 Sunplus Technology Co Ltd Structure and method for planning control commands and data access
US6907478B2 (en) * 2003-02-18 2005-06-14 Adaptec, Inc. Systems and methods optimizing data transfer throughput of a system on chip

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP0542417A1 (en) * 1991-10-17 1993-05-19 Sun Microsystems, Inc. Method and direct memory access controller for asynchronously reading/writing data from/to a memory with improved throughput
JPH10326248A (en) * 1997-05-23 1998-12-08 Brother Ind Ltd Dma controller
US6393500B1 (en) * 1999-08-12 2002-05-21 Mips Technologies, Inc. Burst-configurable data bus
US6615308B1 (en) * 1999-12-09 2003-09-02 Intel Corporation Method and apparatus for regulating write burst lengths
US20020199040A1 (en) * 2001-06-26 2002-12-26 Irwin David S. High speed communications device/system

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
PATENT ABSTRACTS OF JAPAN vol. 1999, no. 03 31 March 1999 (1999-03-31) *

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8171187B2 (en) 2008-07-25 2012-05-01 Freescale Semiconductor, Inc. System and method for arbitrating between memory access requests

Also Published As

Publication number Publication date
US20060112199A1 (en) 2006-05-25
US7398335B2 (en) 2008-07-08

Similar Documents

Publication Publication Date Title
US7398335B2 (en) Method and system for DMA optimization in host bus adapters
US7594057B1 (en) Method and system for processing DMA requests
US7577772B2 (en) Method and system for optimizing DMA channel selection
US7577773B1 (en) Method and system for DMA optimization
US7461195B1 (en) Method and system for dynamically adjusting data transfer rates in PCI-express devices
US7164425B2 (en) Method and system for high speed network application
US7281077B2 (en) Elastic buffer module for PCI express devices
EP1018687B1 (en) A port manager controller for connecting various function modules
US7668177B1 (en) Method and system for quality of service in host bus adapters
US20110022769A1 (en) Translation USB Intermediate Device and Data Rate Apportionment USB Intermediate Device
US6807590B1 (en) Disconnecting a device on a cache line boundary in response to a write command
US20050235072A1 (en) Data storage controller
US20110022758A1 (en) Method and system for processing frames in storage controllers
WO2006019860A2 (en) System and method for controlling buffer memory overflow and underflow conditions in storage controllers
US7669190B2 (en) Method and system for efficiently recording processor events in host bus adapters
US8032675B2 (en) Dynamic memory buffer allocation method and system
US7802031B2 (en) Method and system for high speed network application
KR20180116717A (en) Electronic system having serial system bus interface and direct memory access controller and method of operating the same
WO2006036468A1 (en) Method and system for optimizing data transfer in networks
US20110069717A1 (en) Data transfer device, information processing apparatus, and control method
JP2018516406A (en) Method and apparatus for split burst bandwidth arbitration
US11709791B2 (en) Techniques for deconflicting USB traffic in an extension environment
EP1794953B1 (en) Method and system for using an in-line credit extender with a host bus adapter
US20170308486A1 (en) Storage system and method for controlling command transmission
US20060015774A1 (en) System and method for transmitting data in storage controllers

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A1

Designated state(s): AE AG AL AM AT AU AZ BA BB BG BR BW BY BZ CA CH CN CO CR CU CZ DE DK DM DZ EC EE EG ES FI GB GD GE GH GM HR HU ID IL IN IS JP KE KG KM KN KP KR KZ LC LK LR LS LT LU LV LY MA MD MG MK MN MW MX MZ NA NG NI NO NZ OM PG PH PL PT RO RU SC SD SE SG SK SL SM SY TJ TM TN TR TT TZ UA UG US UZ VC VN YU ZA ZM ZW

AL Designated countries for regional patents

Kind code of ref document: A1

Designated state(s): BW GH GM KE LS MW MZ NA SD SL SZ TZ UG ZM ZW AM AZ BY KG KZ MD RU TJ TM AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HU IE IS IT LT LU LV MC NL PL PT RO SE SI SK TR BF BJ CF CG CI CM GA GN GQ GW ML MR NE SN TD TG

121 Ep: the epo has been informed by wipo that ep was designated in this application
NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 05825579

Country of ref document: EP

Kind code of ref document: A1