@@ -2608,35 +2608,37 @@ static void virt_smp_parse(MachineState *ms, QemuOpts *opts)
if (opts) {
unsigned cpus = qemu_opt_get_number(opts, "cpus", 0);
unsigned sockets = qemu_opt_get_number(opts, "sockets", 0);
+ unsigned clusters = qemu_opt_get_number(opts, "clusters", 1);
unsigned cores = qemu_opt_get_number(opts, "cores", 0);
unsigned threads = qemu_opt_get_number(opts, "threads", 0);
/*
- * Compute missing values; prefer cores over sockets and
- * sockets over threads.
+ * Compute missing values, prefer cores over sockets
+ * and sockets over threads. The value of clusters has
+ * been be set as default 1 if not explicitly specified.
*/
if (cpus == 0 || cores == 0) {
sockets = sockets > 0 ? sockets : 1;
threads = threads > 0 ? threads : 1;
if (cpus == 0) {
cores = cores > 0 ? cores : 1;
- cpus = cores * threads * sockets;
+ cpus = sockets * clusters * cores * threads;
} else {
ms->smp.max_cpus = qemu_opt_get_number(opts, "maxcpus", cpus);
- cores = ms->smp.max_cpus / (sockets * threads);
+ cores = ms->smp.max_cpus / (sockets * clusters * threads);
}
} else if (sockets == 0) {
threads = threads > 0 ? threads : 1;
- sockets = cpus / (cores * threads);
+ sockets = cpus / (clusters * cores * threads);
sockets = sockets > 0 ? sockets : 1;
} else if (threads == 0) {
- threads = cpus / (cores * sockets);
+ threads = cpus / (sockets * clusters * cores);
threads = threads > 0 ? threads : 1;
- } else if (sockets * cores * threads < cpus) {
+ } else if (sockets * clusters * cores * threads < cpus) {
error_report("cpu topology: "
- "sockets (%u) * cores (%u) * threads (%u) < "
- "smp_cpus (%u)",
- sockets, cores, threads, cpus);
+ "sockets (%u) * clusters (%u) * cores (%u) * "
+ "threads (%u) < smp_cpus (%u)",
+ sockets, clusters, cores, threads, cpus);
exit(1);
}
@@ -2647,16 +2649,17 @@ static void virt_smp_parse(MachineState *ms, QemuOpts *opts)
exit(1);
}
- if (sockets * cores * threads != ms->smp.max_cpus) {
+ if (sockets * clusters * cores * threads != ms->smp.max_cpus) {
error_report("cpu topology: "
- "sockets (%u) * cores (%u) * threads (%u)"
- "!= maxcpus (%u)",
- sockets, cores, threads,
+ "sockets (%u) * clusters(%u) * cores (%u) * "
+ "threads (%u) != maxcpus (%u)",
+ sockets, clusters, cores, threads,
ms->smp.max_cpus);
exit(1);
}
ms->smp.cpus = cpus;
+ ms->smp.clusters = clusters;
ms->smp.cores = cores;
ms->smp.threads = threads;
ms->smp.sockets = sockets;
There is a separate function virt_smp_parse() in hw/virt/arm.c used to parse cpu topology for the ARM machines. And there are some ARM implementations that have the concept of cluster, for example, ARM64 server chip Kunpeng 920 has 6 or 8 clusters in each NUMA node and each cluster has 4 cores. All clusters share L3 cache data while the cores within each cluster share L2 cache. So parse cluster cpu topology for ARM machines, then guest kernel will take advantages of it for better scheduling performance. In virt_smp_parse(), the computing logic of missing values prefers cores over sockets over threads. And the value of clusters will be set as default 1 if not explictly specified, so that it will not impact the parsing results of machines that won't specify "clusters=" in -smp command line because they just don't support it. Signed-off-by: Yanan Wang <wangyanan55@huawei.com> --- hw/arm/virt.c | 31 +++++++++++++++++-------------- 1 file changed, 17 insertions(+), 14 deletions(-)