Method, apparatus, and computer program product for coordinating error reporting and reset utilizing an I/O adapter that supports virtualization

• US 8,086,903 B2
• Filed: 03/31/2008
• Issued: 12/27/2011
• Est. Priority Date: 02/25/2005
• Status: Expired due to Fees

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Arndt, Richard Louis, Buckland, Patrick Allen, Kiel, Harvey Gene, Recio, Renato John, Srikrishnan, Jaya
• Primary Examiner(s): SCHELL, JOSEPH O
• Application Number:
  US12/059,870
• Publication Number:
  US 20090089611A1
• Time in Patent Office: 1,366 Days
• Field of Search: 714/9, 714/23, 714/24, 714/30, 714/31, 714/43, 714/55, 714/3, 370/216
• US Class Current: 714/43
• CPC Class Codes: G06F 11/0712 in a virtual computing platform, e.g. logically partitioned systems
• CPC Class Codes: G06F 11/0724 in a multiprocessor or a multi-core unit multiprocessors per se G06F15/80
• CPC Class Codes: G06F 11/0745 in an input/output transactions management context input/output processing in general G06F13/00
• CPC Class Codes: G06F 11/0793 Remedial or corrective actions recovery from an exception in an instruction pipeline G06F9/3861; by retry G06F11/1402; for recovering from a failure of a protocol instance or entity H04L69/40

Abstract

A method, apparatus, and computer program product are disclosed in a shared processor data processing system for coordinating error reporting for and resetting of a physical I/O adapter that supports virtualization. The physical I/O adapter is virtualized by generating virtual I/O adapters that each represent a portion of the physical I/O adapter. Each one of the virtual I/O adapters is assigned to a different one of client logical partitions. A determination is made regarding whether the physical I/O adapter may have experienced an error. If the physical I/O adapter has experienced an error, all of the client logical partitions are notified about the error and a recovery of the physical I/O adapter is coordinated among all of the client logical partitions by waiting for each client logical partition to acknowledge the error notification before the physical I/O adapter is reset.

Claims

1. A method in a shared processor data processing system for coordinating error reporting for and resetting of a physical I/O adapter that supports virtualization, said method comprising:

assigning said physical I/O adapter to a master logical partition;

virtualizing, by said master logical partition, said physical I/O adapter by generating a plurality of virtual I/O adapters that each represent a portion of said physical I/O adapter;

assigning each one of said plurality of virtual I/O adapters to a different one of a plurality of client logical partitions, wherein said master logical partition is not included in said plurality of client logical partitions;

reporting to said master logical partition by one of said plurality of client logical partitions that said physical I/O adapter may have experienced an error;

determining, by said master logical partition, whether said physical I/O adapter experienced an error;

in response to determining that said physical I/O adapter experienced an error, notifying, by said master logical partition, all of said plurality of client logical partitions about said error;

prior to reporting to said master logical partition by said one of said plurality of client logical partitions that said physical I/O adapter may have experienced an error;

executing, by a particular one of said plurality of client logical partitions, a load command to access a particular one of said plurality of virtual I/O adapters;

determining, by said particular one of said plurality of client logical partitions, that a response to said load command was a particular value; and

in response to said response to said load command being said particular value, reporting to said master logical partition by said particular one of said plurality of client logical partitions that said physical I/O adapter may have experienced an error, wherein said particular value indicates that said physical I/O adapter may have experienced an error.

2. The method according to claim 1, further comprising:

waiting, by said master logical partition, for an acknowledgment of a receipt of said notification from each one of said plurality of client logical partitions;

determining, by said master logical partition, whether said acknowledgement has been received from all of said plurality of client logical partitions; and

resetting, by said master logical partition, said physical I/O adapter only in response to said master logical partition determining that said acknowledgement has been received from all of said plurality of client logical partitions.

3. The method according to claim 2, further comprising:

notifying, by said master logical partition, each one of said plurality of client logical partitions of either a success or failure of said resetting of said physical I/O adapter.

4. The method according to claim 3, further comprising:

prior to reporting to said master logical partition that said physical I/O adapter may have experienced an error;

executing, by a particular one of said plurality of client logical partitions, a command to access a particular one of said plurality of virtual I/O adapters, wherein said particular one of said plurality of virtual I/O adapters is accessed; and

in response to said particular one of said plurality of client logical partitions receiving a notification of success of said resetting of said physical I/O adapter;

reinitializing, by said particular one of the plurality of client logical partitions a particular one of said plurality of virtual I/O adapters that is assigned to said particular one of said plurality of client logical partitions.

5. The method according to claim 2, further comprising:

negotiating, between said master logical partition and each one of said plurality of client logical partitions, a maximum response time for each one of said plurality of client logical partitions to respond to messages from said master logical partition.

6. The method according to claim 5, further comprising:

waiting, by said master logical partition, for an acknowledgment of a receipt of said notification from each one of said plurality of client logical partitions;

determining, by said master logical partition, either whether said acknowledgement has been received from all of said plurality of client logical partitions or whether said maximum response time has expired; and

resetting, by said master logical partition, said physical I/O adapter only in response to determining that either said acknowledgement has been received from all of said plurality of client logical partitions or that said response time has expired.

7. The method according to claim 1, wherein determining whether said physical I/O adapter experienced an error comprises said master logical partition asking platform firmware to check hardware registers to determine whether said physical I/O adapter is in a freeze state, wherein said physical I/O adapter enters a freeze state when said physical I/O adapter experiences an error.

8. The method according to claim 1, further comprising:

determining, by said master logical partition, that said physical I/O adapter has not experienced an error;

notifying, by said master logical partition, said particular one of said plurality of client logical partitions that said physical I/O adapter has not experienced an error; and

in response to said particular one of said plurality of client logical partitions receiving a notification that said physical I/O adapter has not experienced an error;

using, by said particular one of said plurality of client logical partitions, said particular value as a correct data response to said load command.

9. The method according to claim 1, further comprising:

wherein said one of said plurality of client logical partitions reports to said master logical partition that said physical I/O adapter may have experienced said error by sending a message to said master logical partition requesting said master logical partition determine whether said physical I/O adapter is in a freeze state;

wherein, in response to determining that said physical I/O adapter is in said freeze state, said master logical partition notifies all of said plurality of client logical partitions about said error by sending a message that said physical I/O adapter is in said freeze state; and

wherein said one of said plurality of client logical partitions sends an acknowledgment of said receipt of said notification by sending an I/O adapter reset request.

10. An apparatus in a shared processor data processing system for coordinating error reporting for and resetting of a physical I/O adapter that supports virtualization, said apparatus comprising:

said physical I/O adapter assigned to a master logical partition;

said master logical partition for virtualizing said physical I/O adapter by generating a plurality of virtual I/O adapters that each represent a portion of said physical I/O adapter;

each one of said plurality of virtual I/O adapters assigned to a different one of a plurality of client logical partitions, wherein said master logical partition is not included in said plurality of client logical partitions;

one of said plurality of client logical partitions reporting to said master logical partition that said physical I/O adapter may have experienced an error;

said master logical partition determining whether said physical I/O adapter experienced an error;

in response to determining that said physical I/O adapter has experienced an error, said master logical partition notifying all of said plurality of client logical partitions about said error;

prior to said one of said plurality of client logical partitions reporting to said master logical partition that said physical I/O adapter may have experienced an error;

a particular one of said plurality of client logical partitions executing a load command to access a particular one of said plurality of virtual I/O adapters;

said particular one of said plurality of client logical partitions determining that a response to said load command was a particular value; and

in response to said response to said load command being said particular value, said particular one of said plurality of client logical partitions reporting to said master logical partition that said physical I/O adapter may have experienced an error, wherein said particular value indicates that said physical I/O adapter may have experienced an error.

11. The apparatus according to claim 10, further comprising:

said master logical partition waiting for an acknowledgment of a receipt of said notification from each one of said plurality of client partitions;

said master logical partition determining whether said acknowledgement has been received from all of said plurality of client logical partitions; and

said master logical partition resetting said physical I/O adapter only in response to said master logical partition determining that said acknowledgement has been received from all of said plurality of client logical partitions.

12. The apparatus according to claim 11, further comprising:

said master logical partition notifying each one of said plurality of client logical partitions of either a success or failure of said resetting of said physical I/O adapter.

13. The apparatus according to claim 12, further comprising:

prior to reporting to said master logical partition that said physical I/O adapter may have experienced an error;

executing, by a particular one of said plurality of client logical partitions, a command to access a particular one of said plurality of virtual I/O adapters, wherein said particular one of said plurality of virtual I/O adapters is accessed; and

in response to said particular one of said plurality of client logical partitions receiving-a notification of success of said resetting of said physical I/O adapter;

reinitializing, by said particular one of the plurality of client logical partitions a particular one of said plurality of virtual I/O adapters that is assigned to said particular one of said plurality of client logical partitions.

14. The apparatus according to claim 10, wherein determining whether said physical I/O adapter experienced an error comprises said master logical partition asking platform firmware to check hardware registers to determine whether said physical I/O adapter is in a freeze state, wherein said physical I/O adapter enters a freeze state when said physical I/O adapter experiences an error.

15. The apparatus according to claim 10, further comprising:

said master logical partition determining that said physical I/O adapter has not experienced an error;

said master logical partition notifying said particular one of said plurality of client logical partitions that said physical I/O adapter has not experienced an error; and

in response to said particular one of said plurality of client logical partitions receiving a notification that said physical I/O adapter has not experienced an error;

said particular one of said plurality of client logical partitions using said particular value as a correct data response to said load command.

16. The apparatus according to claim 10, further comprising:

said master logical partition and each one of said plurality of client logical partitions negotiating a maximum response time for each one of said plurality of client logical partitions to respond to messages from said master logical partition.

17. The apparatus according to claim 16, further comprising:

said master logical partition waiting for an acknowledgment of a receipt of said notification from each one of said plurality of client logical partitions;

said master logical partition determining either whether said acknowledgement has been received from all of said plurality of client logical partitions or whether said maximum response time has expired; and

said master logical partition resetting said physical I/O adapter only in response to determining that either said acknowledgement has been received from all of said client logical partitions or that said response time has expired.

18. A computer program product comprising:

a computer recordable-type medium having instructions for coordinating error reporting for and resetting of a physical I/O adapter that supports virtualization in a shared processor data processing system, said product comprising;

instructions for assigning said physical I/O adapter to a master logical partition;

instructions for virtualizing, by said master logical partition, said physical I/O adapter by generating a plurality of virtual I/O adapters that each represent a portion of said physical I/O adapter;

instructions for assigning each one of said plurality of virtual I/O adapters to a different one of a plurality of client logical partitions, wherein said master logical partition is not included in said plurality of client logical partitions;

instructions for reporting to said master logical partition by one of said plurality of client logical partitions that said physical I/O adapter may have experienced an error;

instructions for determining, by said master logical partition, whether said physical I/O adapter experienced an error;

instructions for notifying, by said master logical partition, all of said plurality of client logical partitions about said error, responsive to determining that said physical I/O adapter experienced an error;

instructions, prior to reporting to said master logical partition by said one of said plurality of client logical partitions that said physical I/O adapter may have experienced an error;

for executing, by a particular one of said plurality of client logical partitions, a load command to access a particular one of said plurality of virtual I/O adapters;

instructions for determining, by said particular one of said plurality of client logical partitions, that a response to said load command was a particular value; and

instructions responsive to said response to said load command being said particular value, for reporting to said master logical partition by said particular one of said plurality of client logical partitions that said physical I/O adapter may have experienced an error, wherein said particular value indicates that said physical I/O adapter may have experienced an error.