When the AMD QoS feature CDP(code and data prioritization) is enabled or disabled, the CDP bit in MSR 0000_0C81 is written on one of the CPUs in L3 domain(core complex). That is not correct. The CDP bit needs to be updated all the logical CPUs in the domain.
This was not spelled out clearly in the spec earlier. The specification has been updated. The updated specification, "AMD64 Technology Platform Quality of Service Extensions Publication # 56375 Revision: 1.02 Issue Date: October 2020" is available now. Refer the section: Code and Data Prioritization. Fix the issue by adding a new flag arch_has_per_cpu_cfg in rdt_cache data structure. The documentation can be obtained at the links below: https://developer.amd.com/wp-content/resources/56375.pdf Link: https://bugzilla.kernel.org/show_bug.cgi?id=206537 Fixes: 4d05bf71f157 ("x86/resctrl: Introduce AMD QOS feature") Signed-off-by: Babu Moger <babu.mo...@amd.com> --- v2: Taken care of Reinette's comments. Changed the field name to arch_has_per_cpu_cfg to be bit more meaningful about the CPU scope. Also fixed some wordings. v1: https://lore.kernel.org/lkml/160469365104.21002.2901190946502347327.stgit@bmoger-ubuntu/ arch/x86/kernel/cpu/resctrl/core.c | 4 ++++ arch/x86/kernel/cpu/resctrl/internal.h | 3 +++ arch/x86/kernel/cpu/resctrl/rdtgroup.c | 9 +++++++-- 3 files changed, 14 insertions(+), 2 deletions(-) diff --git a/arch/x86/kernel/cpu/resctrl/core.c b/arch/x86/kernel/cpu/resctrl/core.c index e5f4ee8f4c3b..e8b5f1cf1ae8 100644 --- a/arch/x86/kernel/cpu/resctrl/core.c +++ b/arch/x86/kernel/cpu/resctrl/core.c @@ -570,6 +570,8 @@ static void domain_add_cpu(int cpu, struct rdt_resource *r) if (d) { cpumask_set_cpu(cpu, &d->cpu_mask); + if (r->cache.arch_has_per_cpu_cfg) + rdt_domain_reconfigure_cdp(r); return; } @@ -923,6 +925,7 @@ static __init void rdt_init_res_defs_intel(void) r->rid == RDT_RESOURCE_L2CODE) { r->cache.arch_has_sparse_bitmaps = false; r->cache.arch_has_empty_bitmaps = false; + r->cache.arch_has_per_cpu_cfg = false; } else if (r->rid == RDT_RESOURCE_MBA) { r->msr_base = MSR_IA32_MBA_THRTL_BASE; r->msr_update = mba_wrmsr_intel; @@ -943,6 +946,7 @@ static __init void rdt_init_res_defs_amd(void) r->rid == RDT_RESOURCE_L2CODE) { r->cache.arch_has_sparse_bitmaps = true; r->cache.arch_has_empty_bitmaps = true; + r->cache.arch_has_per_cpu_cfg = true; } else if (r->rid == RDT_RESOURCE_MBA) { r->msr_base = MSR_IA32_MBA_BW_BASE; r->msr_update = mba_wrmsr_amd; diff --git a/arch/x86/kernel/cpu/resctrl/internal.h b/arch/x86/kernel/cpu/resctrl/internal.h index 80fa997fae60..bcd9b517c765 100644 --- a/arch/x86/kernel/cpu/resctrl/internal.h +++ b/arch/x86/kernel/cpu/resctrl/internal.h @@ -360,6 +360,8 @@ struct msr_param { * executing entities * @arch_has_sparse_bitmaps: True if a bitmap like f00f is valid. * @arch_has_empty_bitmaps: True if the '0' bitmap is valid. + * @arch_has_per_cpu_cfg: True if QOS_CFG register for this cache + * level has CPU scope. */ struct rdt_cache { unsigned int cbm_len; @@ -369,6 +371,7 @@ struct rdt_cache { unsigned int shareable_bits; bool arch_has_sparse_bitmaps; bool arch_has_empty_bitmaps; + bool arch_has_per_cpu_cfg; }; /** diff --git a/arch/x86/kernel/cpu/resctrl/rdtgroup.c b/arch/x86/kernel/cpu/resctrl/rdtgroup.c index af323e2e3100..6abd8ef9a674 100644 --- a/arch/x86/kernel/cpu/resctrl/rdtgroup.c +++ b/arch/x86/kernel/cpu/resctrl/rdtgroup.c @@ -1905,8 +1905,13 @@ static int set_cache_qos_cfg(int level, bool enable) r_l = &rdt_resources_all[level]; list_for_each_entry(d, &r_l->domains, list) { - /* Pick one CPU from each domain instance to update MSR */ - cpumask_set_cpu(cpumask_any(&d->cpu_mask), cpu_mask); + if (r_l->cache.arch_has_per_cpu_cfg) + /* Pick all the CPUs in the domain instance */ + for_each_cpu(cpu, &d->cpu_mask) + cpumask_set_cpu(cpu, cpu_mask); + else + /* Pick one CPU from each domain instance to update MSR */ + cpumask_set_cpu(cpumask_any(&d->cpu_mask), cpu_mask); } cpu = get_cpu(); /* Update QOS_CFG MSR on this cpu if it's in cpu_mask. */
