US5898841A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Inter-processor communication
Assignee: Fujitsu Services Ltd
•
Filing: 1997-02-18
•
Granted: 1999-04-27
•
Expires: 2017-02-18
Abstract
Abstract A multi-processor system has a number of processing elements interconnected by a network for transmitting data frames between the elements. Each element includes an application layer, a transport layer and a link layer. The application layer contains end-point processes each having an address space. The transport layer can allocate a buffer in the address space of a specified end-point process and return details of the buffer to the link layer. The link layer can write message data from a received data frame directly into the allocated buffer by direct memory access without buffering the message data in the link layer. In this way, copying is reduced, improving the efficiency of the system.
Inventors
- 👤 Brian Higgins
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-11-09 04:07
CPC Classifications
H; H04; H04L; H04L1/00; H04L1/0078; H04L1/0083; H04L1/12; H04L1/16; H04L1/1607; H04L1/1635; H04L1/1657; H04L1/18; H04L1/1829; H04L1/1848; H04L1/1867; H04L1/189; H04L69/00; H04L69/30; H04L69/32; H04L9/00; H04L9/40; H04L1/0001; H04L69/322; H04L69/324; H04L69/326
Claims Analysis
Total Claims: 11
Claim 1
Independent Claim
198 words
A multi-processor system comprising a plurality of processing elements interconnected by a network for transmitting data frames between the elements, each element including an application layer, a transport layer and a link layer, wherein the application layer comprises a plurality of end-point processes each having an address space, wherein the transport layer includes a get buffer function, for allocating a buffer in the address space of a specified end-point process and for returning details of said buffer to the link layer, and wherein the link layer includes: (a) means for receiving a data frame comprising a frame header followed by message data, said frame header including a field containing transport layer protocol information identifying one of said end-point processes to which said message data is to be passed, (b) means operative after receipt of said frame header and before receipt of said message data, for calling said get buffer function of said transport layer to allocate a buffer in the address space of said one of said end-point process, and (c) means for writing said message data from said data frame directly into said buffer by direct memory access without buffering said message data in the link layer.
Claim 2
Dependent Claim
198 words
A multi-processor system comprising a plurality of processing elements interconnected by a network for transmitting data frames between the elements, each element including an application layer, a transport layer and a link layer, wherein the application layer comprises a plurality of end-point processes each having an address space, wherein the transport layer includes a get buffer function, for allocating a buffer in the address space of a specified end-point process and for returning details of said buffer to the link layer, and wherein the link layer includes: (a) means for receiving a data frame comprising a frame header followed by message data, said frame header including a field containing transport layer protocol information identifying one of said end-point processes to which said message data is to be passed, (b) means operative after receipt of said frame header and before receipt of said message data, for calling said get buffer function of said transport layer to allocate a buffer in the address space of said one of said end-point process, and (c) means for writing said message data from said data frame directly into said buffer by direct memory access without buffering said message data in the link layer.
Claim 3
Dependent Claim
23 words
A multi-processor system according to claim 1 wherein said get buffer call includes a pointer to said field containing transport layer protocol information.
Claim 4
Dependent Claim
50 words
A multi-processor system according to claim 2 wherein the transport layer includes means operative upon receipt of said get -- buffer call, for accessing said field containing transport layer protocol information, and allocating a buffer in the address space of the end-point process identified by said transport layer protocol information.
Claim 5
Dependent Claim
34 words
A multi-processor system according to claim 1 wherein the frame header of each of at least some of said data frames includes an acknowledgement field for acknowledging receipt of one or more data frames.
Claim 6
Dependent Claim
25 words
A multi-processor system according to claim 4 wherein said acknowledgement field implicitly acknowledges receipt of all data frames up to and including a specified frame.
Claim 7
Dependent Claim
44 words
A multi-processor system according to claim 4 wherein said link layer includes means operative in response to receipt of an acknowledgement, for issuing a return -- buffer call to the transport layer, requesting it to release all buffers associated with the acknowledged data frames.
Claim 8
Dependent Claim
45 words
A multi-processor system according to claim 4 wherein said link layer includes means for sending an acknowledgement frame to one of said processing elements in the event that a predetermined time has elapsed without sending any data frames to said one of said processing elements.
Claim 9
Dependent Claim
47 words
A multi-processor system according to claim 1 wherein said link layer comprises means for checking each received data frame and, in the event of detecting an error in a frame received from a particular processing element, for sending a retransmission request frame to said particular processing element.
Claim 10
Independent Claim
190 words
A method of operating a multi-processor system comprising a plurality of processing elements interconnected by a network for transmitting data frames between the elements, each element including an application layer, a transport layer and a link layer, the method comprising: (a) sending a data frame from a first one of said processing elements to a second one of said processing elements, said data frame comprising a frame header followed by message data, said frame header including information identifying an end-point process to which the data frame is to be delivered in said second one of said processing elements; (b) in said second one of said processing elements, upon receipt of said frame header and before receipt of said message data, accessing said information identifying the end-point process to which the data frame is to be delivered and allocating a buffer in the address space of said end-point process for receiving said message data; and (c) in said second one of said processing elements, upon receipt of said message data, writing said message data directly into said buffer by direct memory access without buffering said message data in the link layer.
Claim 11
Independent Claim
190 words
A method of operating a multi-processor system comprising a plurality of processing elements interconnected by a network for transmitting data frames between the elements, each element including an application layer, a transport layer and a link layer, the method comprising: (a) sending a data frame from a first one of said processing elements to a second one of said processing elements, said data frame comprising a frame header followed by message data, said frame header including information identifying an end-point process to which the data frame is to be delivered in said second one of said processing elements; (b) in said second one of said processing elements, upon receipt of said frame header and before receipt of said message data, accessing said information identifying the end-point process to which the data frame is to be delivered and allocating a buffer in the address space of said end-point process for receiving said message data; and (c) in said second one of said processing elements, upon receipt of said message data, writing said message data directly into said buffer by direct memory access without buffering said message data in the link layer.
References Cited (Backward Citations)
No citation records indexed.
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 2002-04-26 | 2002-10-26 | 2003-04-27 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2006-04-26 | 2006-10-26 | 2007-04-27 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2010-04-26 | 2010-10-26 | 2011-04-27 | $7700.00 | Window Closed / Lapsed |