From: Tony Luck <tony.luck@intel.com>
To: Fenghua Yu <fenghua.yu@intel.com>,
Reinette Chatre <reinette.chatre@intel.com>,
Peter Newman <peternewman@google.com>,
Jonathan Corbet <corbet@lwn.net>,
Shuah Khan <skhan@linuxfoundation.org>,
x86@kernel.org
Cc: Shaopeng Tan <tan.shaopeng@fujitsu.com>,
James Morse <james.morse@arm.com>,
Jamie Iles <quic_jiles@quicinc.com>,
Babu Moger <babu.moger@amd.com>,
Randy Dunlap <rdunlap@infradead.org>,
linux-kernel@vger.kernel.org, linux-doc@vger.kernel.org,
patches@lists.linux.dev, Tony Luck <tony.luck@intel.com>
Subject: [PATCH v2] x86/resctrl: mba_MBps: Fall back to total b/w if local b/w unavailable
Date: Wed, 25 Oct 2023 16:50:46 -0700 [thread overview]
Message-ID: <20231025235046.12940-1-tony.luck@intel.com> (raw)
In-Reply-To: <20231024181600.8270-1-tony.luck@intel.com>
On Intel the various resource director technology (RDT) features are all
orthogonal and independently enumerated. Thus it is possible to have
a system that provides "total" memory bandwidth measurements without
providing "local" bandwidth measurements.
If local bandwidth measurement is not available, do not give up on
providing the "mba_MBps" feedback option completely, make the code fall
back to using total bandwidth.
Signed-off-by: Tony Luck <tony.luck@intel.com>
---
Changes since v1:
+ Both Peter and Babu didn't like my get_mbm_data() function. Peter
thought it just needed a more descriptive name. But Babu explained
how to get rid of it completley. I went with a modified version of
Babu's suggestion (saving the event to use for mba_MBps during
initialization, instead of checking every time).
arch/x86/kernel/cpu/resctrl/monitor.c | 41 ++++++++++++++++----------
arch/x86/kernel/cpu/resctrl/rdtgroup.c | 2 +-
2 files changed, 27 insertions(+), 16 deletions(-)
diff --git a/arch/x86/kernel/cpu/resctrl/monitor.c b/arch/x86/kernel/cpu/resctrl/monitor.c
index f136ac046851..6c3536af024e 100644
--- a/arch/x86/kernel/cpu/resctrl/monitor.c
+++ b/arch/x86/kernel/cpu/resctrl/monitor.c
@@ -73,6 +73,13 @@ unsigned int resctrl_rmid_realloc_threshold;
*/
unsigned int resctrl_rmid_realloc_limit;
+/*
+ * MBM monitor event to use for the mba_MBps mount option.
+ * Preference is QOS_L3_MBM_LOCAL_EVENT_ID, but fall
+ * back to QOS_L3_MBM_TOTAL_EVENT_ID.
+ */
+static enum resctrl_event_id mba_mbps_evt_id;
+
#define CF(cf) ((unsigned long)(1048576 * (cf) + 0.5))
/*
@@ -431,7 +438,7 @@ static int __mon_event_count(u32 rmid, struct rmid_read *rr)
*/
static void mbm_bw_count(u32 rmid, struct rmid_read *rr)
{
- struct mbm_state *m = &rr->d->mbm_local[rmid];
+ struct mbm_state *m = get_mbm_state(rr->d, rmid, mba_mbps_evt_id);
u64 cur_bw, bytes, cur_bytes;
cur_bytes = rr->val;
@@ -526,14 +533,14 @@ static void update_mba_bw(struct rdtgroup *rgrp, struct rdt_domain *dom_mbm)
struct list_head *head;
struct rdtgroup *entry;
- if (!is_mbm_local_enabled())
+ if (!is_mbm_enabled())
return;
r_mba = &rdt_resources_all[RDT_RESOURCE_MBA].r_resctrl;
closid = rgrp->closid;
rmid = rgrp->mon.rmid;
- pmbm_data = &dom_mbm->mbm_local[rmid];
+ pmbm_data = get_mbm_state(dom_mbm, rmid, mba_mbps_evt_id);
dom_mba = get_domain_from_cpu(smp_processor_id(), r_mba);
if (!dom_mba) {
@@ -553,7 +560,7 @@ static void update_mba_bw(struct rdtgroup *rgrp, struct rdt_domain *dom_mbm)
*/
head = &rgrp->mon.crdtgrp_list;
list_for_each_entry(entry, head, mon.crdtgrp_list) {
- cmbm_data = &dom_mbm->mbm_local[entry->mon.rmid];
+ cmbm_data = get_mbm_state(dom_mbm, entry->mon.rmid, mba_mbps_evt_id);
cur_bw += cmbm_data->prev_bw;
delta_bw += cmbm_data->delta_bw;
}
@@ -595,7 +602,7 @@ static void update_mba_bw(struct rdtgroup *rgrp, struct rdt_domain *dom_mbm)
*/
pmbm_data->delta_comp = true;
list_for_each_entry(entry, head, mon.crdtgrp_list) {
- cmbm_data = &dom_mbm->mbm_local[entry->mon.rmid];
+ cmbm_data = get_mbm_state(dom_mbm, entry->mon.rmid, mba_mbps_evt_id);
cmbm_data->delta_comp = true;
}
}
@@ -621,15 +628,15 @@ static void mbm_update(struct rdt_resource *r, struct rdt_domain *d, int rmid)
rr.evtid = QOS_L3_MBM_LOCAL_EVENT_ID;
rr.val = 0;
__mon_event_count(rmid, &rr);
-
- /*
- * Call the MBA software controller only for the
- * control groups and when user has enabled
- * the software controller explicitly.
- */
- if (is_mba_sc(NULL))
- mbm_bw_count(rmid, &rr);
}
+
+ /*
+ * Call the MBA software controller only for the
+ * control groups and when user has enabled
+ * the software controller explicitly.
+ */
+ if (is_mba_sc(NULL))
+ mbm_bw_count(rmid, &rr);
}
/*
@@ -769,10 +776,14 @@ static void l3_mon_evt_init(struct rdt_resource *r)
if (is_llc_occupancy_enabled())
list_add_tail(&llc_occupancy_event.list, &r->evt_list);
- if (is_mbm_total_enabled())
+ if (is_mbm_total_enabled()) {
list_add_tail(&mbm_total_event.list, &r->evt_list);
- if (is_mbm_local_enabled())
+ mba_mbps_evt_id = QOS_L3_MBM_TOTAL_EVENT_ID;
+ }
+ if (is_mbm_local_enabled()) {
list_add_tail(&mbm_local_event.list, &r->evt_list);
+ mba_mbps_evt_id = QOS_L3_MBM_LOCAL_EVENT_ID;
+ }
}
int __init rdt_get_mon_l3_config(struct rdt_resource *r)
diff --git a/arch/x86/kernel/cpu/resctrl/rdtgroup.c b/arch/x86/kernel/cpu/resctrl/rdtgroup.c
index 69a1de92384a..0c4f8a1b8df0 100644
--- a/arch/x86/kernel/cpu/resctrl/rdtgroup.c
+++ b/arch/x86/kernel/cpu/resctrl/rdtgroup.c
@@ -2294,7 +2294,7 @@ static bool supports_mba_mbps(void)
{
struct rdt_resource *r = &rdt_resources_all[RDT_RESOURCE_MBA].r_resctrl;
- return (is_mbm_local_enabled() &&
+ return (is_mbm_enabled() &&
r->alloc_capable && is_mba_linear());
}
--
2.41.0
next prev parent reply other threads:[~2023-10-25 23:50 UTC|newest]
Thread overview: 77+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-10-24 18:16 [PATCH] x86/resctrl: mba_MBps: Fall back to total b/w if local b/w unavailable Tony Luck
2023-10-24 18:24 ` Luck, Tony
2023-10-24 23:20 ` Moger, Babu
2023-10-24 23:43 ` Luck, Tony
2023-10-25 16:01 ` Moger, Babu
2023-10-25 12:46 ` Peter Newman
2023-10-25 19:38 ` Tony Luck
2023-10-25 20:39 ` Moger, Babu
2023-10-25 20:42 ` Moger, Babu
2023-10-25 20:52 ` Tony Luck
2023-10-25 23:41 ` Moger, Babu
2023-10-26 0:07 ` Luck, Tony
2023-10-25 21:06 ` Peter Newman
2023-10-26 13:55 ` Moger, Babu
2023-10-26 16:09 ` Luck, Tony
2023-10-26 17:19 ` Moger, Babu
2023-10-26 19:54 ` Tony Luck
2023-10-25 23:50 ` Tony Luck [this message]
2023-10-26 20:02 ` [PATCH v3] " Tony Luck
2023-10-26 22:40 ` Moger, Babu
2023-10-26 22:59 ` Luck, Tony
2023-11-03 21:43 ` Reinette Chatre
2023-11-03 21:50 ` Reinette Chatre
2023-11-07 21:15 ` Tony Luck
2023-11-08 21:49 ` Reinette Chatre
2023-11-09 21:27 ` Luck, Tony
2023-11-15 16:09 ` Reinette Chatre
2023-11-15 21:54 ` Tony Luck
2023-11-16 19:48 ` Reinette Chatre
2023-11-28 23:14 ` [PATCH v4] x86/resctrl: Add mount option to pick total MBM event Tony Luck
2023-11-29 23:48 ` Reinette Chatre
2023-12-01 20:45 ` Tony Luck
2023-12-01 21:47 ` [PATCH v5] x86/resctrl: Add event choices for mba_MBps Tony Luck
2023-12-04 16:24 ` Moger, Babu
2023-12-04 18:16 ` Tony Luck
2023-12-04 19:04 ` Moger, Babu
2023-12-04 19:45 ` Luck, Tony
2023-12-04 20:03 ` Reinette Chatre
2023-12-04 21:08 ` Tony Luck
2023-12-04 22:15 ` Reinette Chatre
2023-12-04 22:51 ` Reinette Chatre
2023-12-07 19:56 ` [PATCH v6 0/3] x86/resctrl: mba_MBps enhancements Tony Luck
2023-12-07 19:56 ` [PATCH v6 1/3] x86/resctrl: Add mount option "mba_MBps_event" Tony Luck
2023-12-08 18:17 ` Peter Newman
2023-12-08 21:57 ` Tony Luck
2023-12-08 22:09 ` Peter Newman
2023-12-08 22:37 ` Luck, Tony
2023-12-12 17:54 ` Reinette Chatre
2023-12-12 20:02 ` Tony Luck
2023-12-12 21:42 ` Reinette Chatre
2023-12-13 1:07 ` Luck, Tony
2023-12-08 18:29 ` Moger, Babu
2023-12-08 21:50 ` Tony Luck
2023-12-12 18:59 ` Reinette Chatre
2023-12-07 19:56 ` [PATCH v6 2/3] x86/resctrl: Use total bandwidth for mba_MBps option when local isn't present Tony Luck
2023-12-08 18:26 ` Peter Newman
2023-12-07 19:56 ` [PATCH v6 3/3] x86/resctrl: Add new "mba_MBps_event" mount option to documentation Tony Luck
2023-12-08 19:22 ` Peter Newman
2023-12-12 18:59 ` Reinette Chatre
2024-01-09 22:00 ` [PATCH] x86/resctrl: Implement new MBA_mbps throttling heuristic Tony Luck
2024-01-16 19:55 ` Reinette Chatre
2024-01-17 3:36 ` Xiaochen Shen
2024-01-17 3:40 ` Xiaochen Shen
2024-01-18 0:26 ` Reinette Chatre
2024-01-18 21:42 ` [PATCH v2] x86/resctrl: Implement new mba_MBps " Tony Luck
2024-01-22 17:34 ` Reinette Chatre
2024-01-22 18:07 ` Luck, Tony
2024-01-22 18:18 ` Reinette Chatre
2024-01-22 18:21 ` Borislav Petkov
2024-01-22 18:41 ` Reinette Chatre
2024-01-22 18:47 ` Borislav Petkov
2024-01-22 20:58 ` Luck, Tony
2024-01-23 12:12 ` James Morse
2024-01-23 17:07 ` Luck, Tony
2024-01-24 0:29 ` Tony Luck
2024-01-25 17:29 ` Tony Luck
2024-01-22 18:08 ` [PATCH v3] " Tony Luck
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20231025235046.12940-1-tony.luck@intel.com \
--to=tony.luck@intel.com \
--cc=babu.moger@amd.com \
--cc=corbet@lwn.net \
--cc=fenghua.yu@intel.com \
--cc=james.morse@arm.com \
--cc=linux-doc@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=patches@lists.linux.dev \
--cc=peternewman@google.com \
--cc=quic_jiles@quicinc.com \
--cc=rdunlap@infradead.org \
--cc=reinette.chatre@intel.com \
--cc=skhan@linuxfoundation.org \
--cc=tan.shaopeng@fujitsu.com \
--cc=x86@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).