System, method and computer program product for page sharing between fault-isolated cells in a distributed shared memory system

• US 6,148,379 A
• Filed: 09/19/1997
• Issued: 11/14/2000
• Est. Priority Date: 09/19/1997
• Status: Expired due to Term

• Current Assignee: Hewlett Packard Enterprise Development LP (Hewlett-Packard Enterprise Company)
• Original Assignee: Silicon Graphics Incorporated (Graphics Properties Holdings, Inc.)
• Inventors: Schimmel, Curt F.
• Primary Examiner(s): Yoo, Do Hyun
• Assistant Examiner(s): Namazi, Mehdi
• Application Number:
  US08/933,999
• Time in Patent Office: 1,152 Days
• Field of Search: 711/147, 711/148, 711/121, 711/205, 711/206, 395/680, 707/202, 710/26
• US Class Current: 711/152
• CPC Class Codes: G06F 12/1491 in a hierarchical protection system, e.g. privilege levels, memory rings

Abstract

A system, method and computer program product for sharing memory between fault-isolated cells of a computer system. A page of memory is exported from an exporting cell to an importing cell by selectively opening a hole in a fire-wall that otherwise fault-isolates the exporting cell and the importing cell. The fire-wall opening permits the importing cell to access a specific page of memory in the exporting cell. Access to other memory cells is still prevented by the fire-wall. When a page of memory is exported, a record of the export is generated in the exporting cell. Export records are used to determine whether a requesting cell is permitted to access a requested page of memory and to terminate memory exports in a controlled fashion. When a page of memory is imported, an import record and a proxy page frame data structure are generated in the importing cell. Import records are used to access pages of memory in other cells and to terminate imports in a controlled fashion. Proxy page frame data records are used to maintain fault isolation between cells and to ensure that kernel data is never shared between cells.

Claims

1. A method for sharing memory between a first cell and a second cell of a distributed shared memory system, wherein the first cell and the second cell are separated by a fault-isolating fire-wall, comprising the steps of:

(1) receiving a request in the first cell to access a page of memory that is located in the first cell, the request having been sent by the second cell;

(2) determining in the first cell whether the second cell is permitted to access the page of memory;

(3)(a) opening a portion of the fault-isolating fire-wall between the first cell and the second cell to permit the second cell to access the page of memory, if the second cell is permitted to access the page of memory;

(3)(b) generating a record of the page of memory exported from the first cell to the second cell; and

(4) closing the opened portion of the fault-isolating fire-wall.

2. The method according to claim 1, wherein step (3)(b) comprises the steps of:

(i) recording data that identifies the location of the page of memory; and

(ii) recording data that identifies the second cell.

3. The method according to claim 1, wherein step (3)(b) comprises the step of:

(i) placing an export record in a page frame data structure that represents the page of memory.

4. The method according to claim 1, wherein step (3)(a) comprises the step of

sending data that identifies the location of the page of memory to the second cell;

and

wherein step (3)(b) comprises the step of generating an export record in the first cell.

5. The method according to claim 1, wherein step (4) comprises the steps of:

(a) receiving notification in the first cell of a failure in the second cell;

(b) searching the first cell for records of exports to the second cell;

(c) finding a record of an export to the second cell;

(d) removing the record of the export to the second cell; and

(e) closing fire-wall openings between the first and second cells.

6. The method according to claim 5, wherein step (4)(d) comprises the step of:

(i) notifying other processes, threads, etc., on the first cell that the page of memory is potentially corrupt.

7. The method according to claim 1, wherein step (4) comprises the steps of:

(a) determining to close the opened portion of the fault-isolating fire-wall;

(b) using a record of the export of the page of memory to identify the second cell as an importing cell;

(c) notifying the second cell that the first cell is going to terminate the exporting of the page of memory;

(d) removing the export record from the first cell; and

(e) closing the opened portion of the fault-isolating fire-wall.

8. The method according to claim 7, wherein step (4)(d) comprises the step of:

(i) removing the export record from the first cell after the second cell has finished using the page of memory.

9. The method according to claim 1, wherein step (4) comprises the steps of:

(a) receiving notification from the second cell that the page of memory is no longer needed by the second cell; and

(b) removing an export record for the page of memory from the first cell; and

(c) closing the opened portion of the fault-isolating fire-wall.

10. A method for sharing memory between a first cell and a second cell of a distributed shared-memory system, wherein the first cell and the second cell are separated by a fault-isolating fire-wall, comprising the steps of:

(1) sending a request from the first cell to the second cell to access a page of memory that is located in the second cell;

(2) receiving the request in the second cell;

(3) determining in the second cell whether the first cell is permitted to access the page of memory;

(4) opening a portion of the fault-isolating fire-wall between the first cell and the second cell to permit the first cell to access the page of memory, if the first cell is permitted to access the page of memory;

(5) importing the page of memory from the second cell; and

(6) closing the opened portion of the fault-isolating fire-wall.

11. The method according to claim 10, wherein step (5) comprises the step of:

(a) generating an import record in the first cell.

12. The method according to claim 11, wherein step (5)(a) comprises the step of:

(i) recording an address of the page of memory and an identification of the second cell.

13. The method according to claim 11, wherein step (5) further comprises the steps of:

(b) receiving data that identifies the location of the page of memory; and

(c) accessing the page of memory that is located in the second cell.

14. The method according to claim 11, wherein step (5) further comprises the step of:

(b) generating a first proxy page frame data structure in the first cell for representing the page of memory.

15. The method according to claim 14, wherein step (5) further comprises the step of:

(c) associating the first proxy page frame data structure with other proxy page frame data structures in a page cache.

16. The method according to claim 14, wherein step (5) further comprises the step of:

(c) associating the first proxy page frame data structure with other proxy page frame data structures in a global page cache.

17. The method according to claim 14, wherein step (5) further comprises the step of:

(c) associating the first proxy page frame data structure with other proxy page frame data structures in an imported page cache.

18. The method according to claim 14, wherein step (5) further comprises the step of:

(c) associating the first proxy page frame data structure with other proxy page frame data structures in a page cache, wherein the first proxy page frame data structure and the other proxy page frame data structures represent pages of memory that store pages of a memory object.

19. The method according to claim 18, wherein step (5) further comprises the steps of:

(d) generating a proxy memory object data structure in the first cell for representing the memory object; and

(e) associating the proxy memory object data structure with the page cache.

20. The method according to claim 11, wherein step (6) comprises the steps of:

(a) receiving notification in the first cell of a failure in the second cell;

(b) searching the first cell for a record of the import; and

(c) removing the record of the import.

21. The method according to claim 11, wherein step (6) comprises the steps of:

(a) detecting that a use indicator indicates that the first cell is no longer using the page of memory;

(b) using the record of the import to identify the second cell as an exporter of the page of memory;

(c) notifying the second cell that the page of memory is no longer needed;

(d) removing the import record from the first cell; and

(e) de-allocating the proxy page frame data structure.

22. The method according to claim 21, wherein step (6)(a) comprises the step of:

(i) reading the use indicator from a proxy page frame data structure that represents the imported page of memory to the first cell.

23. The method according to claim 11, wherein step (6) comprises the steps of:

(a) receiving notification in the first cell that the page of memory will no longer be available to the first cell, the notification having been sent from the second cell;

(b) removing the import record from the first cell; and

(c) de-allocating the proxy page frame data structure.

24. The method according to claim 11, wherein step (6) comprises the steps of:

(a) receiving notification in the first cell that the page of memory will no longer be available to the first cell, the notification having been sent from the second cell;

(b) notifying the second cell that the import operation is not to be terminated.

25. The method according to claim 11, wherein step (5)(a) comprises the steps of:

(i) generating a proxy page frame data structure in the first cell for representing the page of memory; and

(ii) reverse mapping the proxy page frame data structure to a data structure reference in the first cell that references the page of memory.

26. The method according to claim 25, wherein step (6) comprises the steps of:

(a) identifying the data structure reference from the reverse mapping;

(b) removing the data structure reference;

(c) removing the import record from the first cell; and

(d) de-allocating the proxy page frame data structure.

27. A system for sharing memory between a first and a second cell of a fault-isolated, distributed shared memory system, comprising:

a module that exports a first page of memory from the first cell to the second cell, wherein said module that exports comprises a module that opens a portion of a fire-wall that fault-isolates the first cell from the second cell; and

a module that imports a second page of memory to the first cell from the second cell.

28. The system according to claim 27, wherein said module that exports further comprises:

a module that records export information in the first cell.

29. The system according to claim 27, wherein said module that imports comprises:

a module that records import information in the first cell.

30. The system according to claim 29, wherein said module that imports further comprises:

a module that generates a proxy page frame data structure in the first cell to represent an imported page of memory.

31. The system according to claim 30, wherein said module that imports further comprises:

a module that places the proxy page frame data structure in a page cache.

32. The system according to claim 31, wherein said module that imports further comprises:

a module that generates a proxy memory object data structure in the first cell to represent a memory object in another cell.

33. The system according to claim 32, wherein said module that imports further comprises:

a module that associates the page cache with the proxy memory object data structure.

34. The system according to claim 27, further comprising:

a module that terminates an export of the first page of memory in a controlled fashion; and

a module that terminates an import of the second page of memory in a controlled fashion.

35. A computer program product comprising a computer useable medium having computer program logic stored therein, said computer program logic for enabling a first cell and a second cell of a fault-isolated, distributed shared memory computer system to share memory, wherein said computer program logic comprises:

a procedure that enables the first cell of the computer to export a first page of memory from the first cell to the second cell, wherein said procedure that enables the first cell of the computer to export a first page of memory comprises a procedure that enables the first cell of the computer to open a portion of a fire-wall that fault-isolates the first cell from the second cell; and

a procedure that enables the first cell of the computer to import a second page of memory to the first cell from the second cell.

36. The computer program product according to claim 35, wherein said procedure that enables the first cell of the computer to export a first page of memory further comprises:

a procedure that enables the first cell of the computer to record export information in the first cell.

37. The computer program product according to claim 35, wherein said procedure that enables the first cell of the computer to import a second page of memory for importing comprises:

a procedure that enables the first cell of the computer to record import information in the first cell.

38. The computer program product according to claim 37, wherein said procedure that enables the first cell of the computer to import a second page of memory further comprises:

a procedure that enables the first cell of the computer to generate a proxy page frame data structure in the first cell to represent an imported page of memory.

39. The computer program product according to claim 38, wherein said procedure that enables the first cell of the computer to import a second page of memory further comprises:

a procedure that enables the first cell of the computer to place the proxy page frame data structure in a page cache.

40. The computer program product according to claim 39, wherein said procedure that enables the first cell of the computer to import a second page of memory further comprises:

a procedure that enables the first cell of the computer to generate a proxy memory object data structure in the first cell to represent a memory object in another cell.

41. The computer program product according to claim 40, wherein said procedure that enables the first cell of the computer to import a second page of memory further comprises:

a procedure that enables the first cell of the computer to associate the page cache with the proxy memory object data structure.

42. The computer program product according to claim 35, further comprising:

a procedure that enables the first cell of the computer to terminate an export of the first page of memory in a controlled fashion; and

a procedure that enables the first cell of the computer to terminate an import of the second page of memory in a controlled fashion.

43. A system for sharing memory between a first and a second cell of a fault-isolated, distributed shared memory system, comprising:

means for exporting a first page of memory from the first cell to the second cell, wherein said means for exporting comprises means for opening a portion of a fire-wall that fault-isolates the first cell from the second cell; and

means for importing a second page of memory to the first cell from the second cell.