Affinity-aware parallel zeroing of memory for initialization of large pages in non-uniform memory access (NUMA) servers

Patent metadata

Field	Value
Publication number	US-9870036-B2
Application number	US-201614986930-A
Country	US
Kind code	B2
Filing date	Jan 4, 2016
Priority date	Jun 25, 2015
Publication date	Jan 16, 2018
Grant date	Jan 16, 2018

Abstract

A method for zeroing memory in computing systems where access to memory is non-uniform includes receiving, via a system call, a request for a pool of memory. The method also includes determining a size of the requested pool of memory, and creating a dummy memory segment. The size of the dummy memory segment is larger than the size of the requested pool of memory. The method further includes filling the dummy memory segment with one or more pages, based on the determined size of the requested pool of memory, and deleting the dummy memory segment.

First claim

What is claimed is:

1. A method comprising:
receiving, via a system call, a request for a pool of memory;
determining, based on a type of the request, a size of the requested pool of memory;
creating a dummy memory segment, wherein a size of the dummy memory segment is larger than the size of the requested pool of memory;
filling the dummy memory segment with one or more pages, based on the determined size of the requested pool of memory; and
deleting the dummy memory segment prior to allocating the one or more pages for the requested pool of memory, wherein deleting the dummy memory segment comprises zeroing the one or more pages in the dummy memory segment.

2. The method of claim 1 , wherein zeroing the one or more pages in the dummy memory segment comprises:
sorting the one or more pages of the dummy memory segment according to each associated affinity domain of each page, wherein each affinity domain comprises a cluster of processors and memory local to the cluster of processors; thank you;
sending requests to zero the sorted one or more pages to one or more software threads that are attached to the respective affinity domain; and
waiting, after sending the requests, to return to a system caller until a message is received, from the software threads in each affinity domain, indicating that all requests to zero have been processed.

3. The method of claim 1 , wherein zeroing the one or more pages in the dummy memory segment comprises:
transmitting a request to zero the one or more pages in the dummy memory segment to an intermediate software thread;
using the intermediate software thread to perform the request as a background process; and
upon receiving a message from the intermediate software thread, returning to a system caller, while performing the request, via the intermediate software thread, continues in the background.

4. The method of claim 1 , wherein the type of request is a request to grow the pool of memory.

5. The method of claim 4 , wherein the size of the requested pool of memory is determined based on a size that the pool of memory is requested to grow by.

6. The method of claim 1 , wherein the type of request is a request to create the pool of memory.

7. The method of claim 6 , wherein the size of the requested pool of memory is determined based on a size of the pool of memory that needs to be created.

Assignees

• IBM

Inventors

• Hedge Nikhil
• Mealey Bruce
• Rogers Mark D

Classifications

• G06F12/023

  Free address space management · CPC title

• G06F12/1009 Primary

  using page tables, e.g. page table structures · CPC title

• G06F2009/45583

  Memory management, e.g. access or allocation · CPC title

• G06F12/0223

  User address space allocation, e.g. contiguous or non contiguous base addressing · CPC title

• G06F9/5016

  the resource being the memory · CPC title