Recoverable machine check handling

A non-transitory storage device tangibly embodying a program of instructions executable by at least one processor to perform a method for processing hardware errors in a computing system having a plurality of hardware adapters, the method comprising:

• detecting, at an interface to an operating system of the computing system, a hardware error in the computing system;
• responsive to the detecting, determining at least one of the hardware adapters that potentially generated the hardware error, and quarantining the at least one of the hardware adapters;
• and following the quarantining, completing a handling of the hardware error without invoking a crash of the computing system;
• wherein the quarantining is performed by the interface;
• and the completing of handling of the hardware error is performed by the operating system;
• the interface comprises a kernel extension of a device driver associated with the operating system;
• and, the quarantining comprises resetting the at least one of the hardware adapters and invoking a warm start of device drivers associated with the operating system responsive to the detecting of the hardware error.

A method for deploying computing infrastructure, comprising integrating computer-readable code into a computing system having a plurality of hardware adapters, where the code in combination with the computing system processes hardware errors of the hardware adapters by performing a method comprising:

• registering an interface with an operating system of the computing system, said interface comprising a kernel extension of a device driver associated with the operating system;
• detecting, at the interface, a hardware error in the computing system;
• responsive to the detecting, determining at least one of the hardware adapters that potentially generated the hardware error, and quarantining the at least one of the hardware adapters;
• and following the quarantining, completing a handling of the hardware error without invoking a crash of the computing system;
• wherein the quarantining is performed by the interface;
• and the completing of handling of the hardware error is performed by the operating system;
• the interface comprises a kernel extension of a device driver associated with the operating system;
• and,
• the quarantining comprises resetting the at least one of the hardware adapters and invoking a warm start of device drivers associated with the operating system responsive to the detecting of the hardware error.

A non-transitory device tangibly embodying a program of instructions executable by at least one processor to perform a method for processing hardware errors in a computing system having a plurality of hardware adapters, the method comprising:

• registering an interface with an operating system of the computing system, said interface comprising a kernel extension of a device driver associated with the operating system;
• detecting, at the interface, a hardware error in the computing system;
• responsive to the detecting, determining at least one of the hardware adapters that potentially generated the hardware error, and quarantining the at least one of the hardware adapters;
• and following the quarantining, completing a handling of the hardware error without invoking a crash of the computing system;
• wherein the quarantining is performed by the interface;
• and the completing of handling of the hardware error is performed by the operating system;
• the interface comprises a kernel extension of a device driver associated with the operating system;
• and, the quarantining comprises resetting the at least one of the hardware adapters and invoking a warm start of device drivers associated with the operating system responsive to the detecting of the hardware error.

A data storage facility, comprising:

• an error register for logging errors generated by a plurality of hardware adapters;
• and at least one cluster comprising at least one processor executing computer-readable code to provide an operating system, and an interface to the operating system;
• wherein:
• the interface detects a hardware error in the computing system, checks the error register to determine at least one of the hardware adapters that potentially generated the hardware error, and quarantines the at least one of the hardware adapters;
• and following the quarantining, the operating system completes a handling of the hardware error without invoking a crash of the data storage facility;
• wherein, the interface comprises a kernel extension of a device driver associated with the operating system;
• and,
• the quarantining comprises resetting the at least one of the hardware adapters and invoking a warm start of device drivers associated with the operating system responsive to the detecting of the hardware error.