Scalable CPU error recorder

A machine-readable medium having stored thereon executable instructions that when executed by a processor, cause the processor to:

• receive notification of an error;
• invoke a firmware error handler in response to the notification;
• gather information about a plurality of outstanding error events;
• format the gathered information as a variable-length error record, said variable-length error record having a format including a record header delineating the beginning of the error record, and a variable number of sections, each section having a section header and a variable-length section body, wherein each section corresponds to one of the plurality of error events;
• store the error record in a memory;
• and
• when requested, output the error record to an agent.

A method, comprising:

• receiving an indication of an error condition within a computer system;
• invoking a firmware error handler within the computer system;
• collecting a plurality of device error states, each device error state including information describing the current operational condition of a peripheral device within the addressing range of a processor executing in the computer system;
• assembling, within the firmware error handler, said plurality of device error states into a variable-length error record associated with the error condition, said variable-length error record having a format including a record header delineating the beginning of the error record, and a variable number of sections, each section corresponding to one of the plurality of device error states;
• and
• storing the variable-length error record in a non-volatile memory.

A computer system, comprising:

• a non-volatile memory in communication with a bus;
• and
• error handling control logic coupled to the bus, said error handling control logic to receive an indication of an error condition for a device within the computer system,
• said error handling control logic to collect a plurality of device error states, each device error state including information describing the current operational condition of a device within the addressing range of the computer processor, to assemble said plurality of device error states into a variable-length error record associated with the error condition, said variable-length error record having a format including a record header delineating the beginning of the error record, and a variable number of sections, each section corresponding to one of the plurality of device error states, and to store the variable-length error record in the non-volatile memory.

An error logging apparatus, comprising:

• a non-volatile memory coupled to a bus, and
• control logic to gather a plurality of device error states, each device error state including information describing the operational state of a device interoperably connected to the bus;
• said control logic to format the gathered device error states as a variable-length error record, said variable-length error record having a format including a record header delineating the beginning of the error record, and a variable number of sections, each section having a section header and a variable-length section body, wherein each section corresponds to one of the plurality of error events;
• and to store the error record in a memory;
• and if requested, to output the error record to an agent.