linux-kernel - Re: [PATCH v8 3/7] x86/resctrl: Refactor mbm

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <ZzYxUwwEEIOGdmT6@agluck-desk3>
Date: Thu, 14 Nov 2024 09:20:19 -0800
From: Tony Luck <tony.luck@...el.com>
To: Peter Newman <peternewman@...gle.com>
Cc: Reinette Chatre <reinette.chatre@...el.com>,
	Fenghua Yu <fenghua.yu@...el.com>, Jonathan Corbet <corbet@....net>,
	Shuah Khan <skhan@...uxfoundation.org>, x86@...nel.org,
	James Morse <james.morse@....com>,
	Jamie Iles <quic_jiles@...cinc.com>,
	Babu Moger <babu.moger@....com>,
	Randy Dunlap <rdunlap@...radead.org>,
	"Shaopeng Tan (Fujitsu)" <tan.shaopeng@...itsu.com>,
	linux-kernel@...r.kernel.org, linux-doc@...r.kernel.org,
	patches@...ts.linux.dev
Subject: Re: [PATCH v8 3/7] x86/resctrl: Refactor mbm_update()

On Thu, Nov 14, 2024 at 11:31:25AM +0100, Peter Newman wrote:
> At least for the purposes of reporting the mbps rate to userspace, my
> users will record from the files one per second, so it would be fine
> to just report Unavailable (or Unassigned when it's clear that the
> error is because counter wasn't unassigned) whenever either the
> current or previous reading was not successful. Then they can assume
> the value or error reported always refers to the most
> recently-completed one-second window.

First hack at keeping status for memory bandwidth values. Simple
state machine.  Compile tested.


diff --git a/arch/x86/kernel/cpu/resctrl/internal.h b/arch/x86/kernel/cpu/resctrl/internal.h
index 6345ab3e0890..80de5c6b0754 100644
--- a/arch/x86/kernel/cpu/resctrl/internal.h
+++ b/arch/x86/kernel/cpu/resctrl/internal.h
@@ -359,14 +359,22 @@ struct rftype {
 			 char *buf, size_t nbytes, loff_t off);
 };
 
+enum mbm_state_status {
+	MBM_INVALID,
+	MBM_ONE_VALUE,
+	MBM_VALID
+};
+
 /**
  * struct mbm_state - status for each MBM counter in each domain
  * @prev_bw_bytes: Previous bytes value read for bandwidth calculation
  * @prev_bw:	The most recent bandwidth in MBps
+ * @status:	Validity of @prev_bw
  */
 struct mbm_state {
-	u64	prev_bw_bytes;
-	u32	prev_bw;
+	u64			prev_bw_bytes;
+	u32			prev_bw;
+	enum mbm_state_status	status;
 };
 
 /**
diff --git a/arch/x86/kernel/cpu/resctrl/monitor.c b/arch/x86/kernel/cpu/resctrl/monitor.c
index da4ae21350c8..767a526af2f5 100644
--- a/arch/x86/kernel/cpu/resctrl/monitor.c
+++ b/arch/x86/kernel/cpu/resctrl/monitor.c
@@ -670,6 +670,17 @@ static void mbm_bw_count(u32 closid, u32 rmid, struct rmid_read *rr)
 	if (WARN_ON_ONCE(!m))
 		return;
 
+	if (rr->err || !rr->val) {
+		m->status = MBM_INVALID;
+		return;
+	}
+
+	if (m->status == MBM_INVALID) {
+		m->status = MBM_ONE_VALUE;
+		m->prev_bw_bytes = rr->val;
+		return;
+	}
+
 	cur_bytes = rr->val;
 	bytes = cur_bytes - m->prev_bw_bytes;
 	m->prev_bw_bytes = cur_bytes;
@@ -677,6 +688,7 @@ static void mbm_bw_count(u32 closid, u32 rmid, struct rmid_read *rr)
 	cur_bw = bytes / SZ_1M;
 
 	m->prev_bw = cur_bw;
+	m->status = MBM_VALID;
 }
 
 /*
@@ -781,6 +793,9 @@ static void update_mba_bw(struct rdtgroup *rgrp, struct rdt_mon_domain *dom_mbm)
 	if (WARN_ON_ONCE(!pmbm_data))
 		return;
 
+	if (pmbm_data->status != MBM_VALID)
+		return;
+
 	dom_mba = get_ctrl_domain_from_cpu(smp_processor_id(), r_mba);
 	if (!dom_mba) {
 		pr_warn_once("Failure to get domain for MBA update\n");
@@ -801,6 +816,9 @@ static void update_mba_bw(struct rdtgroup *rgrp, struct rdt_mon_domain *dom_mbm)
 		cmbm_data = get_mbm_state(dom_mbm, entry->closid, entry->mon.rmid, evt_id);
 		if (WARN_ON_ONCE(!cmbm_data))
 			return;
+		if (cmbm_data->status != MBM_VALID)
+			return;
+
 		cur_bw += cmbm_data->prev_bw;
 	}