Pre-Grant Publication Number: 20100299672
Please help the USPTO examine the application by evaluating the relevance of the publicly submitted prior art to the patent application.
Peer To Patent forwards the Top 10 most relevant prior art submissions and their annotations to the USPTO.
Review this prior art and click on the thumbs up (or down) to indicate whether this submission should be forwarded to the USPTO.
If you login then you can add an annotation by typing in the box at the bottom of the screen to comment on the relevance of the prior art to the claims of the patent application.
Review this prior art and click on the thumbs up (or down) to indicate whether this submission should be forwarded to the USPTO.
If you login then you can add an annotation by typing in the box at the bottom of the screen to comment on the relevance of the prior art to the claims of the patent application.
Prior Art Detail
Summary / Description
| Summary / Description | A thesis that presents a kernek memory management method. It describes a system wherein in-kernel memory is managed by user-level resource managers. Here rather than one policy controlling memory management, individual policies are implemented by the resource managers. |
Basic Information
| Type of Prior Art | Print Publication |
| Publication Title * | A Principled Approach To Kernel Memory Management |
| Author | Dhammika Elkaduwe |
| ISBN | |
| Page Range | |
| Medium | Other printed publication |
| Publication Date * | 2010 |
| URL | |
Notes / To Do
| Notes | This is a thesis paper. The only date given is 2010, but I could not find the exact date, so it cannot get behind the 1/22/10 priority date. However it seems pertinent, so maybe there is more to this, such as previous publications. |
Excerpt
Excerpt In the first class of systems, kernel memory is managed by a policy integrated into
the kernel. While there is some flexibility to “fine-tune” the kernel policy, it can only be
changed by modifying the kernel — a fundamental change to the kernel policy requires
modification to the kernel’s code. In the context of a verified kernel, such a modification
means that the refinement proofs are no longer valid — nullifying the assurance provided
by formal verification. Moreover, as discussed previously, re-verification can be costly,
depending on the nature of the modification.
On the other hand, if we refrain from modifying the kernel integrated policy to preserve
the assurance, then the verified kernel can only be deployed in a system where there is a
natural synergy between the kernel’s policy and the domain’s requirements. For example,
a kernel using a first-come-first-serve memory allocation policy cannot be deployed in
a domain that warrants strict partitioning [BA03] and a kernel enforcing partitioning is
inefficient for a best-effort system [Wal02]— efficiency can be significantly improved by
reassigning memory to where it can be used. Ideally, a verified kernel should be capable of
catering to the needs of a number of application domains, so that the high cost associated
with verification gets amortised.
Kernels that treat their in-kernel memory as a cache of the system state can be viewed
as a special case of kernel-integrated policy where the policy is caching. A single policy,
as I discussed previously is not suited for all application domains. In this particular case,
the kernel’s caching scheme is not appropriate for a system with temporal requirements.
In the final class of systems, the in-kernel memory is managed by user-level resource
managers. Through an API, the kernel exports its in-kernel memory management to userlevel
managers. In-kernel memory management policy is implemented outside of the kernel.
Thus, as I demonstrate through this thesis, one can change the policy to suit the
application domain by changing the user-level resource manager, rather than the verified
kernel code base. The scheme supports diversity by means of different user-level resource
managers— each resource manager implements its own policy over the resources it manages.
Through the same mechanism we can even support co-existing, diverse resource
management policies.
In the final class of systems, the in-kernel memory is managed by user-level resource
managers. Through an API, the kernel exports its in-kernel memory management to userlevel
managers. In-kernel memory management policy is implemented outside of the kernel.
Thus, as I demonstrate through this thesis, one can change the policy to suit the
application domain by changing the user-level resource manager, rather than the verified
kernel code base. The scheme supports diversity by means of different user-level resource
managers— each resource manager implements its own policy over the resources it manages.
Through the same mechanism we can even support co-existing, diverse resource
management policies. |
Relevance
Claims
1
Relevance
The excerpt describes first the same prior art as the patent does, i.e. a single memory management policy. Then it describes a system with "user-level resource managers" where the policies of each user-level resource manager can be changed to suit the application. This is very similar to claim, wherein allocators each have their own policies to suit tasks.
The excerpt describes first the same prior art as the patent does, i.e. a single memory management policy. Then it describes a system with "user-level resource managers" where the policies of each user-level resource manager can be changed to suit the application. This is very similar to claim, wherein allocators each have their own policies to suit tasks.
Claim Chart
All
0 days left
