Detecting and reporting livelocks in a computer

1. A computer-implemented method of reporting a livelock event occurring on a computer, comprising:

specifying a threshold amount of a computing resource used to determine whether a task executing on the computer is livelocked in response to past computing resource consumption by the task, wherein different threshold amounts of the computing resource are specified for different tasks;

in response to resource consumption by the task exceeding the threshold amount of the computing resource, detecting a livelock event involving the task executing on the computer by performing at least one inspection from a set of inspections of registers of a processor executing the task to determine that the task is not making tangible progress, the set of inspections consisting of:

determining whether contents of multiple ones of the registers are changing over time;

determining whether the registers indicate that the task is stuck in a counting state; and

determining whether the register contents are changing in a cyclical pattern;

collecting state data describing the detected livelock event;

generating a livelock report including the collected state data; and

transmitting the livelock report to a server.

2. The method of claim 1 , wherein detecting the livelock event involving the task executing on the computer comprises:

detecting that the task is not making tangible progress according to at least one inspection while resource consumption by the task exceeds the threshold amount of the computing resource.

3. The method of claim 1 , wherein detecting the livelock event involving the task executing on the computer comprises:

periodically sampling to determine an amount of the computing resource being consumed by the task to produce sequential samples; and

performing the at least one inspection in response to determining whether the task consumes more than the threshold amount of the computing resource in each of a threshold number of sequential samples.

4. The method of claim 3 , further comprising:

determining whether a user of the computer is idle; and

varying the threshold number of sequential samples responsive to whether the user of the computer is idle.

5. The method of claim 1 , wherein collecting state data describing the detected livelock event comprises collecting one or more from the set consisting of:

a memory dump describing a state of the livelocked task;

logs from one or more applications executing on the computer when the livelock is detected; and

data describing other tasks executing on the computer when the livelock is detected.

6. The method of claim 1 , wherein generating a livelock report comprises:

collecting identification data uniquely identifying the livelock event; and

generating a hash of the identification data, the hash generated such that similar livelock events have similar hashes.

7. A non-transitory computer-readable storage medium storing executable computer program instructions for reporting a livelock event occurring on a computer, the instructions performing steps comprising:

specifying a threshold amount of a computing resource used to determine whether a task executing on the computer is livelocked in response to past computing resource consumption by the task, wherein different threshold amounts of the computing resource are specified for different tasks;

in response to resource consumption by the task exceeding the threshold amount of the computing resource, detecting a livelock event involving the task executing on the computer by performing at least one inspection from a set of inspections on registers of a processor executing the task to determine that the task is not making tangible progress, the set of inspections consisting of:

determining whether contents of multiple ones of the registers are changing over time;

determining whether the registers indicate that the task is stuck in a counting state; and

determining whether the register contents are changing in a cyclical pattern;

collecting state data describing the detected livelock event;

generating a livelock report including the collected state data; and

transmitting the livelock report to a server.

8. The storage medium of claim 7 , wherein detecting the livelock event involving the task executing on the computer comprises:

detecting that the task is not making tangible progress according to at least one inspection while resource consumption of the task exceeds the threshold amount of the computing resource.

9. The storage medium of claim 8 , wherein detecting the livelock event involving the task executing on the computer comprises:

periodically sampling to determine an amount of the computing resource being consumed by the task to produce sequential samples; and

performing the at least one inspection in response to determining whether the task consumes more than the threshold amount of the computing resource in each of a threshold number of sequential samples.

10. The storage medium of claim 7 , wherein collecting state data describing the detected livelock event comprises collecting one or more from the set consisting of:

a memory dump describing a state of the livelocked task;

logs from one or more applications executing on the computer when the livelock is detected; and

data describing other tasks executing on the computer when the livelock is detected.

11. The storage medium of claim 7 , wherein generating a livelock report comprises:

collecting identification data uniquely identifying the livelock event; and generating a hash of the identification data, the hash generated such that similar livelock events have similar hashes.

12. A computer system for reporting a livelock event, the computer system comprising:

a non-transitory computer-readable storage medium storing executable computer program instructions for performing steps comprising:

specifying a threshold amount of a computing resource used to determine whether a task executing on the computer is livelocked in response to past computing resource consumption by the task, wherein different threshold amounts of the computing resource are specified for different tasks;

in response to resource consumption by the task exceeding the threshold amount of the computing resource, detecting a livelock event involving the task executing on the computer by performing at least one inspection from a set of inspections on registers of a processor executing the task to determine that the task is not making tangible progress, the set of inspections consisting of:

determining whether contents of multiple ones of the registers are changing over time;

determining whether the registers indicate that the task is stuck in a counting state; and

determining whether the register contents are changing in a cyclical pattern;

collecting state data describing the detected livelock event;

generating a livelock report including the collected state data;

transmitting the livelock report to a server; and

a processor for executing the computer program instructions.

13. The computer system of claim 12 , wherein detecting the livelock event involving the task executing on the computer comprises:

detecting that the task is not making tangible progress according to at least one inspection while resource consumption of the task exceeds the threshold amount of the computing resource.

14. The computer system of claim 12 , wherein detecting the livelock event involving the task executing on the computer comprises:

periodically sampling to determine an amount of the computing resource being consumed by the task to produce sequential samples; and

performing the at least one inspection in response to determining whether the task consumes more than the threshold amount of the computing resource in each of a threshold number of sequential samples.

15. The method of claim 1 , further comprising:

determining whether a user of the computer is idle in response to whether a monitored input is received within a specified time interval; and

specifying the threshold amount of computing resource used to determine whether the task executing on the computer is livelocked in response to whether the user of the computer is idle.

16. The method of claim 1 , further comprising:

determining a number of tasks being performed simultaneously by the processor; and

specifying the threshold amount of computing resource used to determine whether the task executing on the computer is livelocked in response to the number of tasks being performed simultaneously by the processor.

17. The storage medium of claim 7 , further comprising:

determining whether a user of the computer is idle in response to whether a monitored input is received within a specified time interval; and

specifying the threshold amount of computing resource used to determine whether the task executing on the computer is livelocked in response to whether the user of the computer is idle.

18. The storage medium of claim 7 , further comprising:

determining a number of tasks being performed simultaneously by the processor; and

specifying the threshold amount of computing resource used to determine whether the task executing on the computer is livelocked in response to the number of tasks being performed simultaneously by the processor.

19. The computer system of claim 12 , further comprising:

determining whether a user of the computer is idle in response to whether a monitored input is received within a specified time interval; and

specifying the threshold amount of computing resource used to determine whether the task executing on the computer is livelocked in response to whether the user of the computer is idle.

20. The computer system of claim 12 , further comprising:

determining a number of tasks being performed simultaneously by the processor; and

specifying the threshold amount of computing resource used to determine whether the task executing on the computer is livelocked in response to the number of tasks being performed simultaneously by the processor.