Browse Prior Art Database

Hypervisor assisted dump through overcommitted memory

IP.com Disclosure Number: IPCOM000175752D
Original Publication Date: 2008-Oct-22
Included in the Prior Art Database: 2008-Oct-22
Document File: 2 page(s) / 36K

Publishing Venue

IBM

Abstract

Improving current dump methods by having hypervisor/os allocate overcommited virtual real memory to boot into while preserving the dumped memory which can be saved off.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 62% of the total text.

Page 1 of 2

Hypervisor assisted dump through overcommitted memory

Currently, when an operating system crashes, it can be useful for service personnel to collect dump data. Traditionally, operating system dumps are gathered by the failing kernel (or by a exec'ing a second kernel, a la kexec); the machine will not be rebooted and returned to production until after the dump data has been collected. To avoid the downtime associated with the traditional methods, the hypervisor can now assist with dumps by allowing the OS to preserve memory across a reboot. The OS can then collect dump data after rebooting, bringing the system back into production as quickly as possible. The shortcoming of this method is that a large amount of memory must be reserved in order to hold the dump information, and cannot be used for anything else until the dump has been collected.

Improving on the existing methods can be done by taking advantage of virtual real memory, by not reserving physical memory, but rather by reserving virtual physical memory. This is analogous to how the applications have virtual memory, and the kernel sets up a virtual address space for applications. The hypervisor sets up a virtual "physical" address space for the kernel. This allows us to have a larger "physical" address space, thus allowing us to not have to reserve unused physical memory.

Our invention i...