tjh.dev / kernel
Linux kernel mirror (for testing) git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git
kernel os linux
1
fork

Configure Feed

Select the types of activity you want to include in your feed.

x86/smpboot: avoid SMT domain attach/destroy if SMT is not enabled

Currently, the SMT domain is added into sched_domain_topology by default.

If cpu_attach_domain() finds that the CPU SMT domain’s cpumask_weight
is just 1, it will destroy it.

On a large machine, such as one with 512 cores, this results in
512 redundant domain attach/destroy operations.

Avoid these unnecessary operations by simply checking
cpu_smt_num_threads and skip SMT domain if the SMT domain is not
enabled.

Suggested-by: K Prateek Nayak <kprateek.nayak@amd.com>
Signed-off-by: Li Chen <chenl311@chinatelecom.cn>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: K Prateek Nayak <kprateek.nayak@amd.com>
Tested-by: K Prateek Nayak <kprateek.nayak@amd.com>
Link: https://lore.kernel.org/r/20250710105715.66594-5-me@linux.beauty

authored by

Li Chen and committed by
Peter Zijlstra f79c9aa4 fbc2010d

+11 -1
+11 -1
arch/x86/kernel/smpboot.c
··· 492 492 493 493 static void __init build_sched_topology(void) 494 494 { 495 + struct sched_domain_topology_level *topology = x86_topology; 496 + 495 497 /* 496 498 * When there is NUMA topology inside the package invalidate the 497 499 * PKG domain since the NUMA domains will auto-magically create the ··· 504 502 505 503 memset(&x86_topology[pkgdom], 0, sizeof(x86_topology[pkgdom])); 506 504 } 507 - set_sched_topology(x86_topology); 505 + 506 + /* 507 + * Drop the SMT domains if there is only one thread per-core 508 + * since it'll get degenerated by the scheduler anyways. 509 + */ 510 + if (cpu_smt_num_threads <= 1) 511 + ++topology; 512 + 513 + set_sched_topology(topology); 508 514 } 509 515 510 516 void set_cpu_sibling_map(int cpu)