From: Tim Chen Date: Fri, 3 Oct 2025 19:31:28 +0000 (-0700) Subject: sched/topology: Fix sched domain build error for GNR, CWF in SNC-3 mode X-Git-Url: http://git.ipfire.org/gitweb.cgi?a=commitdiff_plain;h=4d6dd05d07d00bc3bd91183dab4d75caa8018db9;p=thirdparty%2Flinux.git sched/topology: Fix sched domain build error for GNR, CWF in SNC-3 mode It is possible for Granite Rapids (GNR) and Clearwater Forest (CWF) to have up to 3 dies per package. When sub-numa cluster (SNC-3) is enabled, each die will become a separate NUMA node in the package with different distances between dies within the same package. For example, on GNR, we see the following numa distances for a 2 socket system with 3 dies per socket: package 1 package2 ---------------- | | --------- --------- | 0 | | 3 | --------- --------- | | --------- --------- | 1 | | 4 | --------- --------- | | --------- --------- | 2 | | 5 | --------- --------- | | ---------------- node distances: node 0 1 2 3 4 5 0: 10 15 17 21 28 26 1: 15 10 15 23 26 23 2: 17 15 10 26 23 21 3: 21 28 26 10 15 17 4: 23 26 23 15 10 15 5: 26 23 21 17 15 10 The node distances above led to 2 problems: 1. Asymmetric routes taken between nodes in different packages led to asymmetric scheduler domain perspective depending on which node you are on. Current scheduler code failed to build domains properly with asymmetric distances. 2. Multiple remote distances to respective tiles on remote package create too many levels of domain hierarchies grouping different nodes between remote packages. For example, the above GNR topology lead to NUMA domains below: Sched domains from the perspective of a CPU in node 0, where the number in bracket represent node number. NUMA-level 1 [0,1] [2] NUMA-level 2 [0,1,2] [3] NUMA-level 3 [0,1,2,3] [5] NUMA-level 4 [0,1,2,3,5] [4] Sched domains from the perspective of a CPU in node 4 NUMA-level 1 [4] [3,5] NUMA-level 2 [3,4,5] [0,2] NUMA-level 3 [0,2,3,4,5] [1] Scheduler group peers for load balancing from the perspective of CPU 0 and 4 are different. Improper task could be chosen for load balancing between groups such as [0,2,3,4,5] [1]. Ideally you should choose nodes in 0 or 2 that are in same package as node 1 first. But instead tasks in the remote package node 3, 4, 5 could be chosen with an equal chance and could lead to excessive remote package migrations and imbalance of load between packages. We should not group partial remote nodes and local nodes together. Simplify the remote distances for CWF and GNR for the purpose of sched domains building, which maintains symmetry and leads to a more reasonable load balance hierarchy. The sched domains from the perspective of a CPU in node 0 NUMA-level 1 is now NUMA-level 1 [0,1] [2] NUMA-level 2 [0,1,2] [3,4,5] The sched domains from the perspective of a CPU in node 4 NUMA-level 1 is now NUMA-level 1 [4] [3,5] NUMA-level 2 [3,4,5] [0,1,2] We have the same balancing perspective from node 0 or node 4. Loads are now balanced equally between packages. Co-developed-by: Vinicius Costa Gomes Signed-off-by: Vinicius Costa Gomes Signed-off-by: Tim Chen Signed-off-by: Peter Zijlstra (Intel) Reviewed-by: Chen Yu Tested-by: Zhao Liu --- diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c index eb289abece237..5709c9cab1957 100644 --- a/arch/x86/kernel/smpboot.c +++ b/arch/x86/kernel/smpboot.c @@ -515,6 +515,76 @@ static void __init build_sched_topology(void) set_sched_topology(topology); } +#ifdef CONFIG_NUMA +static int sched_avg_remote_distance; +static int avg_remote_numa_distance(void) +{ + int i, j; + int distance, nr_remote, total_distance; + + if (sched_avg_remote_distance > 0) + return sched_avg_remote_distance; + + nr_remote = 0; + total_distance = 0; + for_each_node_state(i, N_CPU) { + for_each_node_state(j, N_CPU) { + distance = node_distance(i, j); + + if (distance >= REMOTE_DISTANCE) { + nr_remote++; + total_distance += distance; + } + } + } + if (nr_remote) + sched_avg_remote_distance = total_distance / nr_remote; + else + sched_avg_remote_distance = REMOTE_DISTANCE; + + return sched_avg_remote_distance; +} + +int arch_sched_node_distance(int from, int to) +{ + int d = node_distance(from, to); + + switch (boot_cpu_data.x86_vfm) { + case INTEL_GRANITERAPIDS_X: + case INTEL_ATOM_DARKMONT_X: + + if (!x86_has_numa_in_package || topology_max_packages() == 1 || + d < REMOTE_DISTANCE) + return d; + + /* + * With SNC enabled, there could be too many levels of remote + * NUMA node distances, creating NUMA domain levels + * including local nodes and partial remote nodes. + * + * Trim finer distance tuning for NUMA nodes in remote package + * for the purpose of building sched domains. Group NUMA nodes + * in the remote package in the same sched group. + * Simplify NUMA domains and avoid extra NUMA levels including + * different remote NUMA nodes and local nodes. + * + * GNR and CWF don't expect systems with more than 2 packages + * and more than 2 hops between packages. Single average remote + * distance won't be appropriate if there are more than 2 + * packages as average distance to different remote packages + * could be different. + */ + WARN_ONCE(topology_max_packages() > 2, + "sched: Expect only up to 2 packages for GNR or CWF, " + "but saw %d packages when building sched domains.", + topology_max_packages()); + + d = avg_remote_numa_distance(); + } + return d; +} +#endif /* CONFIG_NUMA */ + void set_cpu_sibling_map(int cpu) { bool has_smt = __max_threads_per_core > 1;