Commit-ID: c7cc0cc10cdecc275211c8749defba6c41aaf5de Gitweb: http://git.kernel.org/tip/c7cc0cc10cdecc275211c8749defba6c41aaf5de Author: Fenghua Yu <[email protected]> AuthorDate: Fri, 11 Nov 2016 17:02:37 -0800 Committer: Thomas Gleixner <[email protected]> CommitDate: Tue, 15 Nov 2016 18:35:50 +0100
x86/intel_rdt: Reset per cpu closids on unmount All CPUs in a rdtgroup are given back to the default rdtgroup before the rdtgroup is removed during umount. After umount, the default rdtgroup contains all online CPUs, but the per cpu closids are not cleared. As a result the stale closid value will be used immediately after the next mount. Move all cpus to the default group and update the percpu closid storage. [ tglx: Massaged changelong ] Signed-off-by: Fenghua Yu <[email protected]> Cc: "Ravi V Shankar" <[email protected]> Cc: "Tony Luck" <[email protected]> Cc: "Sai Prakhya" <[email protected]> Cc: "Vikas Shivappa" <[email protected]> Cc: "Ingo Molnar" <[email protected]> Cc: "H. Peter Anvin" <[email protected]> Link: http://lkml.kernel.org/r/[email protected] Signed-off-by: Thomas Gleixner <[email protected]> --- arch/x86/kernel/cpu/intel_rdt_rdtgroup.c | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) diff --git a/arch/x86/kernel/cpu/intel_rdt_rdtgroup.c b/arch/x86/kernel/cpu/intel_rdt_rdtgroup.c index 2f54931..d6bad09 100644 --- a/arch/x86/kernel/cpu/intel_rdt_rdtgroup.c +++ b/arch/x86/kernel/cpu/intel_rdt_rdtgroup.c @@ -799,6 +799,7 @@ static void rmdir_all_sub(void) { struct rdtgroup *rdtgrp, *tmp; struct task_struct *p, *t; + int cpu; /* move all tasks to default resource group */ read_lock(&tasklist_lock); @@ -813,14 +814,29 @@ static void rmdir_all_sub(void) smp_call_function_many(cpu_online_mask, rdt_reset_pqr_assoc_closid, NULL, 1); put_cpu(); + list_for_each_entry_safe(rdtgrp, tmp, &rdt_all_groups, rdtgroup_list) { /* Remove each rdtgroup other than root */ if (rdtgrp == &rdtgroup_default) continue; + + /* + * Give any CPUs back to the default group. We cannot copy + * cpu_online_mask because a CPU might have executed the + * offline callback already, but is still marked online. + */ + cpumask_or(&rdtgroup_default.cpu_mask, + &rdtgroup_default.cpu_mask, &rdtgrp->cpu_mask); + kernfs_remove(rdtgrp->kn); list_del(&rdtgrp->rdtgroup_list); kfree(rdtgrp); } + + /* Reset all per cpu closids to the default value */ + for_each_cpu(cpu, &rdtgroup_default.cpu_mask) + per_cpu(cpu_closid, cpu) = 0; + kernfs_remove(kn_info); }
