<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns="http://www.w3.org/TR/REC-html40"> <head> <meta http-equiv="Content-Type" content="text/html; charset=us-ascii"> <meta name="Generator" content="Microsoft Word 15 (filtered medium)"> <style></style> </head> <body lang="EN-US" link="#0563C1" vlink="#954F72"> <div class="WordSection1"> Hi All,<o:p></o:p> <o:p> </o:p> Some Intel processor families (e.g. the Intel Xeon processor E5 v3 family) introduced some PQos (Platform Qos) features to monitor or control shared resource.<o:p></o:p> <o:p> </o:p> - CMT (Cache Monitoring Technology): measure the usage of cache by applications running on the platform.<o:p></o:p> - CAT (Cache Allocation Technology): enable an OS or Hypervisor/VMM to specify the amount of cache space into which an application can fill.<o:p></o:p> - MBM (Memory Bandwidth Monitoring): build on the CMT infrastructure to allow monitoring of bandwidth from one level of the cache hierarchy to the next – in this case focusing on the L3 cache, which is typically backed directly by system memory. As a result of this implementation, memory bandwidth can be monitored.<o:p></o:p> <o:p> </o:p> For more information, see “Intel 64 and IA-32 Architectures Software Developer’s Manual”.<o:p></o:p> <o:p> </o:p> Among these PQos features, currently CMT patches has been merged into Linux kernel mainline. So this patch series proposed in this mail will focus on CMT feature support in libvirt.<o:p></o:p> <o:p> </o:p> At the library API layer, I plan on adding cache related field into <a name="virDomainInfo">virDomainInfo</a> and <a name="virNodeInfo">virNodeInfo</a>:<o:p></o:p> <o:p> </o:p> Add “cache” member into virNodeInfo to get total size of cache in one node.<o:p></o:p> struct virNodeInfo {<o:p></o:p> …<o:p></o:p> unsigned int cores; /* number of cores per socket, total number of<o:p></o:p> processors in case of unusual NUMA topology*/<o:p></o:p> unsigned int threads; /* number of threads per core, 1 in case of<o:p></o:p> unusual numa topology */<o:p></o:p> + unsigned int cache; /* cache size in bytes */<o:p></o:p> };<o:p></o:p> <o:p> </o:p> Add “cacheOcc” member into virDomainInfo to get cache occupancy for one VM.<o:p></o:p> struct virDomainInfo {<o:p></o:p> …<o:p></o:p> unsigned short nrVirtCpu; /* the number of virtual CPUs for the domain */<o:p></o:p> unsigned long long cpuTime; /* the CPU time used in nanoseconds */<o:p></o:p> + unsigned long long cacheOcc; /* the cache occupancy in Bytes */<o:p></o:p> };<o:p></o:p> <o:p> </o:p> With this change, those applications like OpenStack based on libvirt can collect cache usage for metering and monitoring (e.g. any VM using more than X% of cache).<o:p></o:p> <o:p> </o:p> Because CMT implementation in Linux kernel is based on perf mechanism and there is no nice and public API library for perf, I have to wrapper system call for perf in libvirt to use CMT feature, and enable/disable perf event for CMT when VM is created/destroyed.<o:p></o:p> <o:p> </o:p> Any thoughts on this plan before I start submitting code patches? Do you think whether I need add several new APIs and XML element for CMT feature (or more PQos features)?<o:p></o:p> <o:p> </o:p> Thanks,<o:p></o:p> Qiaowei<o:p></o:p> </div> </body> </html>