到目前为止,我知道有两种类型的 NUMA 内存策略,即“按任务/进程”和“按内存区域”。根据下面的 numa_map 条目,我有几个疑问。您能让我理解一下吗?提前谢谢。
077d1000 default file=/usr/bin/bash anon=69 dirty=224 mapmax=3 active=155 N0=69 N1=155 kernelpagesize_kB=4
7ffe2d5cb000 default stack anon=27 dirty=27 active=1 N0=27 kernelpagesize_kB=4
7ffe2d5fa000 default
7ffe2d5fe000 default
1.) How can I know which policy applied for this /usr/bin/bash?
2.) As per the entry, how many pages are allocated to this process? 69+155 = 224 of N0 and N1? or all values of 69 + 224 + 3 + 155 + 69 + 155 = 675 ?
3.) What is 077d1000 ? Is it memory range starts for the entry above ?
4.) How NUMA determines which policy be applied for tasks // memory regions ?
5.) The process spanned on both N0 and N1 nodes. It indicates it is interleave or memory was not available on one node and allocated on different node(numa_miss)?