Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Received-SPF: Pass (protection.outlook.com: domain of amd.com designates
 165.204.84.17 as permitted sender) receiver=protection.outlook.com;
 client-ip=165.204.84.17; helo=SATLEXMB04.amd.com; pr=C
Subject: [PATCH v3 09/10] x86/resctrl: Add sysfs interface to write the event
 configuration
From:   Babu Moger <babu.moger@amd.com>
To:     <fenghua.yu@intel.com>, <reinette.chatre@intel.com>,
        <tglx@linutronix.de>, <mingo@redhat.com>, <bp@alien8.de>
CC:     <eranian@google.com>, <dave.hansen@linux.intel.com>,
        <x86@kernel.org>, <hpa@zytor.com>, <corbet@lwn.net>,
        <hpa@zytor.com>, <linux-kernel@vger.kernel.org>,
        <linux-doc@vger.kernel.org>, <bagasdotme@gmail.com>
Date:   Mon, 22 Aug 2022 08:44:29 -0500
Message-ID: <166117586910.6695.3670808098195387542.stgit@bmoger-ubuntu>
In-Reply-To: <166117559756.6695.16047463526634290701.stgit@bmoger-ubuntu>
References: <166117559756.6695.16047463526634290701.stgit@bmoger-ubuntu>
User-Agent: StGit/1.1.dev103+g5369f4c
MIME-Version: 1.0
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: quoted-printable
X-MS-Exchange-CrossTenant-OriginalArrivalTime: 22 Aug 2022 13:45:06.5208
 (UTC)
X-MS-Exchange-CrossTenant-Network-Message-Id: d9e7bade-7cdc-454d-d0a3-08da844485fc
X-MS-Exchange-CrossTenant-Id: 3dd8961f-e488-4e60-8e11-a82d994e183d
X-MS-Exchange-CrossTenant-OriginalAttributedTenantConnectingIp: TenantId=3dd8961f-e488-4e60-8e11-a82d994e183d;Ip=[165.204.84.17];Helo=[SATLEXMB04.amd.com]
X-MS-Exchange-CrossTenant-AuthSource: CO1NAM11FT079.eop-nam11.prod.protection.outlook.com
X-MS-Exchange-CrossTenant-AuthAs: Anonymous
X-MS-Exchange-CrossTenant-FromEntityHeader: HybridOnPrem
X-MS-Exchange-Transport-CrossTenantHeadersStamped: DM4PR12MB5248
X-Spam-Status: No, score=-2.1 required=5.0 tests=BAYES_00,DKIM_SIGNED,
        DKIM_VALID,DKIM_VALID_AU,DKIM_VALID_EF,RCVD_IN_DNSWL_NONE,
        RCVD_IN_MSPIKE_H2,SPF_HELO_PASS,SPF_PASS,T_SCC_BODY_TEXT_LINE
        autolearn=ham autolearn_force=no version=3.4.6
X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on
        lindbergh.monkeyblade.net
Precedence: bulk
List-ID: <linux-kernel.vger.kernel.org>
X-Mailing-List: linux-kernel@vger.kernel.org

Add the sysfs interface to write the event configuration for the=0A=
MBM configurable events. The event configuration can be changed by=0A=
writing to the sysfs file for that specific event.=0A=
=0A=
Following are the types of events supported.=0A=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=0A=
Bits    Description=0A=
6       Dirty Victims from the QOS domain to all types of memory=0A=
5       Reads to slow memory in the non-local NUMA domain=0A=
4       Reads to slow memory in the local NUMA domain=0A=
3       Non-temporal writes to non-local NUMA domain=0A=
2       Non-temporal writes to local NUMA domain=0A=
1       Reads to memory in the non-local NUMA domain=0A=
0       Reads to memory in the local NUMA domain=0A=
=0A=
By default the mbm_total_bytes configuration is set to 0x7f to count=0A=
all the types of events and mbm_local_bytes configuration is set to=0A=
0x15 to count all the local memory events.=0A=
=0A=
For example:=0A=
To change the mbm_total_bytes to count all the reads, run the command.=0A=
$echo  0x33 > /sys/fs/resctrl/mon_data/mon_L3_00/mbm_total_config=0A=
=0A=
To change the mbm_local_bytes to count all the slow memory reads, run=0A=
the command.=0A=
$echo  0x30 > /sys/fs/resctrl/mon_data/mon_L3_00/mbm_local_config=0A=
=0A=
Signed-off-by: Babu Moger <babu.moger@amd.com>=0A=
Reviewed-by: Ingo Molnar <mingo@kernel.org>=0A=
---=0A=
 arch/x86/kernel/cpu/resctrl/rdtgroup.c |  109 ++++++++++++++++++++++++++++=
++++=0A=
 1 file changed, 109 insertions(+)=0A=
=0A=
diff --git a/arch/x86/kernel/cpu/resctrl/rdtgroup.c b/arch/x86/kernel/cpu/r=
esctrl/rdtgroup.c=0A=
index e1847d49fa15..83c8780726ff 100644=0A=
--- a/arch/x86/kernel/cpu/resctrl/rdtgroup.c=0A=
+++ b/arch/x86/kernel/cpu/resctrl/rdtgroup.c=0A=
@@ -323,9 +323,118 @@ int rdtgroup_mondata_config_show(struct seq_file *m, =
void *arg)=0A=
 	return ret;=0A=
 }=0A=
 =0A=
+/*=0A=
+ * This is called via IPI to read the CQM/MBM counters=0A=
+ * in a domain.=0A=
+ */=0A=
+void mon_event_config_write(void *info)=0A=
+{=0A=
+	union mon_data_bits *md =3D info;=0A=
+	u32 evtid =3D md->u.evtid;=0A=
+	u32 msr_index;=0A=
+=0A=
+	switch (evtid) {=0A=
+	case QOS_L3_MBM_TOTAL_EVENT_ID:=0A=
+		msr_index =3D 0;=0A=
+		break;=0A=
+	case QOS_L3_MBM_LOCAL_EVENT_ID:=0A=
+		msr_index =3D 1;=0A=
+		break;=0A=
+	default:=0A=
+		return; /* Not expected to come here */=0A=
+	}=0A=
+=0A=
+	wrmsr(MSR_IA32_EVT_CFG_BASE + msr_index, md->u.mon_config, 0);=0A=
+}=0A=
+=0A=
+ssize_t  rdtgroup_mondata_config_write(struct kernfs_open_file *of,=0A=
+				       char *buf, size_t nbytes, loff_t off)=0A=
+{=0A=
+	struct rdt_hw_resource *hw_res;=0A=
+	struct rdtgroup *rdtgrp;=0A=
+	struct rdt_resource *r;=0A=
+	unsigned int mon_config;=0A=
+	cpumask_var_t cpu_mask;=0A=
+	union mon_data_bits md;=0A=
+	struct rdt_domain *d;=0A=
+	u32 resid, domid;=0A=
+	int ret =3D 0, cpu;=0A=
+=0A=
+	ret =3D kstrtouint(buf, 0, &mon_config);=0A=
+	if (ret)=0A=
+		return ret;=0A=
+=0A=
+	rdt_last_cmd_clear();=0A=
+=0A=
+	/* mon_config cannot be more than the supported set of events */=0A=
+	if (mon_config > GENMASK(6, 0)) {=0A=
+		rdt_last_cmd_puts("Invalid event configuration\n");=0A=
+		return -EINVAL;=0A=
+	}=0A=
+=0A=
+	cpus_read_lock();=0A=
+	rdtgrp =3D rdtgroup_kn_lock_live(of->kn);=0A=
+	if (!rdtgrp) {=0A=
+		return -ENOENT;=0A=
+		goto e_unlock;=0A=
+	}=0A=
+=0A=
+	if (!zalloc_cpumask_var(&cpu_mask, GFP_KERNEL)) {=0A=
+		ret =3D -ENOMEM;=0A=
+		goto e_unlock;=0A=
+	}=0A=
+=0A=
+=0A=
+	md.priv =3D of->kn->priv;=0A=
+	resid =3D md.u.rid;=0A=
+	domid =3D md.u.domid;=0A=
+=0A=
+	hw_res =3D &rdt_resources_all[resid];=0A=
+	r =3D &hw_res->r_resctrl;=0A=
+	d =3D rdt_find_domain(r, domid, NULL);=0A=
+	if (IS_ERR_OR_NULL(d)) {=0A=
+		ret =3D -ENOENT;=0A=
+		goto e_cpumask;=0A=
+	}=0A=
+=0A=
+	md.u.mon_config =3D mon_config & 0xFF;=0A=
+=0A=
+	/* Pick all the CPUs in the domain instance */=0A=
+	for_each_cpu(cpu, &d->cpu_mask)=0A=
+		cpumask_set_cpu(cpu, cpu_mask);=0A=
+=0A=
+	cpu =3D get_cpu();=0A=
+	/* Update MSR_IA32_EVT_CFG_BASE MSR on this cpu if it's in cpu_mask */=0A=
+	if (cpumask_test_cpu(cpu, cpu_mask))=0A=
+		mon_event_config_write(&md);=0A=
+=0A=
+	/* Update MSR_IA32_EVT_CFG_BASE MSR on all other cpus in cpu_mask */=0A=
+	smp_call_function_many(cpu_mask, mon_event_config_write, &md, 1);=0A=
+	put_cpu();=0A=
+=0A=
+	/*=0A=
+	 * When an Event Configuration is changed, the bandwidth counters=0A=
+	 * for all RMIDs and Events will be cleared, and the U-bit for every=0A=
+	 * RMID will be set on the next read to any BwEvent for every RMID.=0A=
+	 * Clear the mbm_local and mbm_total counts for all the RMIDs.=0A=
+	 */=0A=
+	memset(d->mbm_local, 0, sizeof(struct mbm_state) * r->num_rmid);=0A=
+	memset(d->mbm_total, 0, sizeof(struct mbm_state) * r->num_rmid);=0A=
+=0A=
+e_cpumask:=0A=
+	free_cpumask_var(cpu_mask);=0A=
+=0A=
+e_unlock:=0A=
+	rdtgroup_kn_unlock(of->kn);=0A=
+	cpus_read_unlock();=0A=
+=0A=
+	return ret ?: nbytes;=0A=
+}=0A=
+=0A=
 static const struct kernfs_ops kf_mondata_config_ops =3D {=0A=
 	.atomic_write_len       =3D PAGE_SIZE,=0A=
 	.seq_show               =3D rdtgroup_mondata_config_show,=0A=
+	.write                  =3D rdtgroup_mondata_config_write,=0A=
 };=0A=
 =0A=
 static bool is_cpu_list(struct kernfs_open_file *of)=0A=
=0A=